diff options
Diffstat (limited to 'java/src/com/android/inputmethod/latin')
73 files changed, 2661 insertions, 4335 deletions
diff --git a/java/src/com/android/inputmethod/latin/AbstractDictionaryWriter.java b/java/src/com/android/inputmethod/latin/AbstractDictionaryWriter.java index 463d09344..1aee22baf 100644 --- a/java/src/com/android/inputmethod/latin/AbstractDictionaryWriter.java +++ b/java/src/com/android/inputmethod/latin/AbstractDictionaryWriter.java @@ -21,21 +21,19 @@ import android.util.Log; import com.android.inputmethod.latin.makedict.DictEncoder; import com.android.inputmethod.latin.makedict.UnsupportedFormatException; -import com.android.inputmethod.latin.makedict.Ver3DictEncoder; +import com.android.inputmethod.latin.makedict.Ver2DictEncoder; import java.io.File; import java.io.IOException; import java.util.Map; -// TODO: Quit extending Dictionary after implementing dynamic binary dictionary. -abstract public class AbstractDictionaryWriter extends Dictionary { +abstract public class AbstractDictionaryWriter { /** Used for Log actions from this class */ private static final String TAG = AbstractDictionaryWriter.class.getSimpleName(); private final Context mContext; - public AbstractDictionaryWriter(final Context context, final String dictType) { - super(dictType); + public AbstractDictionaryWriter(final Context context) { mContext = context; } @@ -55,20 +53,18 @@ abstract public class AbstractDictionaryWriter extends Dictionary { // TODO: Remove lastModifiedTime after making binary dictionary support forgetting curve. abstract public void addBigramWords(final String word0, final String word1, - final int frequency, final boolean isValid, - final long lastModifiedTime); + final int frequency, final boolean isValid, final long lastModifiedTime); abstract public void removeBigramWords(final String word0, final String word1); abstract protected void writeDictionary(final DictEncoder dictEncoder, final Map<String, String> attributeMap) throws IOException, UnsupportedFormatException; - public void write(final String fileName, final Map<String, String> attributeMap) { - final String tempFileName = fileName + ".temp"; - final File file = new File(mContext.getFilesDir(), fileName); - final File tempFile = new File(mContext.getFilesDir(), tempFileName); + public void write(final File file, final Map<String, String> attributeMap) { + final String tempFilePath = file.getAbsolutePath() + ".temp"; + final File tempFile = new File(tempFilePath); try { - final DictEncoder dictEncoder = new Ver3DictEncoder(tempFile); + final DictEncoder dictEncoder = new Ver2DictEncoder(tempFile); writeDictionary(dictEncoder, attributeMap); tempFile.renameTo(file); } catch (IOException e) { diff --git a/java/src/com/android/inputmethod/latin/AssetFileAddress.java b/java/src/com/android/inputmethod/latin/AssetFileAddress.java index 875192554..fd6c24dfe 100644 --- a/java/src/com/android/inputmethod/latin/AssetFileAddress.java +++ b/java/src/com/android/inputmethod/latin/AssetFileAddress.java @@ -16,6 +16,8 @@ package com.android.inputmethod.latin; +import com.android.inputmethod.latin.utils.FileUtils; + import java.io.File; /** @@ -52,4 +54,12 @@ public final class AssetFileAddress { if (!f.isFile()) return null; return new AssetFileAddress(filename, offset, length); } + + public boolean pointsToPhysicalFile() { + return 0 == mOffset; + } + + public void deleteUnderlyingFile() { + FileUtils.deleteRecursively(new File(mFilename)); + } } diff --git a/java/src/com/android/inputmethod/latin/BinaryDictionary.java b/java/src/com/android/inputmethod/latin/BinaryDictionary.java index fd296988e..95ac3e203 100644 --- a/java/src/com/android/inputmethod/latin/BinaryDictionary.java +++ b/java/src/com/android/inputmethod/latin/BinaryDictionary.java @@ -26,6 +26,7 @@ import com.android.inputmethod.latin.settings.NativeSuggestOptions; import com.android.inputmethod.latin.utils.CollectionUtils; import com.android.inputmethod.latin.utils.JniUtils; import com.android.inputmethod.latin.utils.StringUtils; +import com.android.inputmethod.latin.utils.UnigramProperty; import java.io.File; import java.util.ArrayList; @@ -57,6 +58,21 @@ public final class BinaryDictionary extends Dictionary { @UsedForTesting public static final String MAX_BIGRAM_COUNT_QUERY = "MAX_BIGRAM_COUNT"; + public static final int NOT_A_VALID_TIMESTAMP = -1; + + // Format to get unigram flags from native side via getUnigramPropertyNative(). + private static final int FORMAT_UNIGRAM_PROPERTY_OUTPUT_FLAG_COUNT = 4; + private static final int FORMAT_UNIGRAM_PROPERTY_IS_NOT_A_WORD_INDEX = 0; + private static final int FORMAT_UNIGRAM_PROPERTY_IS_BLACKLISTED_INDEX = 1; + private static final int FORMAT_UNIGRAM_PROPERTY_HAS_BIGRAMS_INDEX = 2; + private static final int FORMAT_UNIGRAM_PROPERTY_HAS_SHORTCUTS_INDEX = 3; + + // Format to get unigram historical info from native side via getUnigramPropertyNative(). + private static final int FORMAT_UNIGRAM_PROPERTY_OUTPUT_HISTORICAL_INFO_COUNT = 3; + private static final int FORMAT_UNIGRAM_PROPERTY_TIMESTAMP_INDEX = 0; + private static final int FORMAT_UNIGRAM_PROPERTY_LEVEL_INDEX = 1; + private static final int FORMAT_UNIGRAM_PROPERTY_COUNT_INDEX = 2; + private long mNativeDict; private final Locale mLocale; private final long mDictSize; @@ -123,8 +139,13 @@ public final class BinaryDictionary extends Dictionary { private static native boolean needsToRunGCNative(long dict, boolean mindsBlockByGC); private static native void flushWithGCNative(long dict, String filePath); private static native void closeNative(long dict); + private static native int getFormatVersionNative(long dict); private static native int getProbabilityNative(long dict, int[] word); private static native int getBigramProbabilityNative(long dict, int[] word0, int[] word1); + private static native void getUnigramPropertyNative(long dict, int[] word, + int[] outCodePoints, boolean[] outFlags, int[] outProbability, + int[] outHistoricalInfo, ArrayList<int[]> outShortcutTargets, + ArrayList<Integer> outShortcutProbabilities); private static native int getSuggestionsNative(long dict, long proximityInfo, long traverseSession, int[] xCoordinates, int[] yCoordinates, int[] times, int[] pointerIds, int[] inputCodePoints, int inputSize, int commitPoint, @@ -133,10 +154,14 @@ public final class BinaryDictionary extends Dictionary { int[] outputAutoCommitFirstWordConfidence); private static native float calcNormalizedScoreNative(int[] before, int[] after, int score); private static native int editDistanceNative(int[] before, int[] after); - private static native void addUnigramWordNative(long dict, int[] word, int probability); + private static native void addUnigramWordNative(long dict, int[] word, int probability, + int[] shortcutTarget, int shortcutProbability, boolean isNotAWord, + boolean isBlacklisted, int timestamp); private static native void addBigramWordsNative(long dict, int[] word0, int[] word1, - int probability); + int probability, int timestamp); private static native void removeBigramWordsNative(long dict, int[] word0, int[] word1); + private static native int addMultipleDictionaryEntriesNative(long dict, + LanguageModelParam[] languageModelParams, int startIndex); private static native int calculateProbabilityNative(long dict, int unigramProbability, int bigramProbability); private static native String getPropertyNative(long dict, String query); @@ -235,6 +260,10 @@ public final class BinaryDictionary extends Dictionary { return mNativeDict != 0; } + public int getFormatVersion() { + return getFormatVersionNative(mNativeDict); + } + public static float calcNormalizedScore(final String before, final String after, final int score) { return calcNormalizedScoreNative(StringUtils.toCodePointArray(before), @@ -274,23 +303,55 @@ public final class BinaryDictionary extends Dictionary { return getBigramProbabilityNative(mNativeDict, codePoints0, codePoints1); } - // Add a unigram entry to binary dictionary in native code. - public void addUnigramWord(final String word, final int probability) { + @UsedForTesting + public UnigramProperty getUnigramProperty(final String word) { + if (TextUtils.isEmpty(word)) { + return null; + } + final int[] codePoints = StringUtils.toCodePointArray(word); + final int[] outCodePoints = new int[MAX_WORD_LENGTH]; + final boolean[] outFlags = new boolean[FORMAT_UNIGRAM_PROPERTY_OUTPUT_FLAG_COUNT]; + final int[] outProbability = new int[1]; + final int[] outHistoricalInfo = + new int[FORMAT_UNIGRAM_PROPERTY_OUTPUT_HISTORICAL_INFO_COUNT]; + final ArrayList<int[]> outShortcutTargets = CollectionUtils.newArrayList(); + final ArrayList<Integer> outShortcutProbabilities = CollectionUtils.newArrayList(); + getUnigramPropertyNative(mNativeDict, codePoints, outCodePoints, outFlags, outProbability, + outHistoricalInfo, outShortcutTargets, outShortcutProbabilities); + return new UnigramProperty(codePoints, + outFlags[FORMAT_UNIGRAM_PROPERTY_IS_NOT_A_WORD_INDEX], + outFlags[FORMAT_UNIGRAM_PROPERTY_IS_BLACKLISTED_INDEX], + outFlags[FORMAT_UNIGRAM_PROPERTY_HAS_BIGRAMS_INDEX], + outFlags[FORMAT_UNIGRAM_PROPERTY_HAS_SHORTCUTS_INDEX], outProbability[0], + outHistoricalInfo[FORMAT_UNIGRAM_PROPERTY_TIMESTAMP_INDEX], + outHistoricalInfo[FORMAT_UNIGRAM_PROPERTY_LEVEL_INDEX], + outHistoricalInfo[FORMAT_UNIGRAM_PROPERTY_COUNT_INDEX], + outShortcutTargets, outShortcutProbabilities); + } + + // Add a unigram entry to binary dictionary with unigram attributes in native code. + public void addUnigramWord(final String word, final int probability, + final String shortcutTarget, final int shortcutProbability, final boolean isNotAWord, + final boolean isBlacklisted, final int timestamp) { if (TextUtils.isEmpty(word)) { return; } final int[] codePoints = StringUtils.toCodePointArray(word); - addUnigramWordNative(mNativeDict, codePoints, probability); + final int[] shortcutTargetCodePoints = (shortcutTarget != null) ? + StringUtils.toCodePointArray(shortcutTarget) : null; + addUnigramWordNative(mNativeDict, codePoints, probability, shortcutTargetCodePoints, + shortcutProbability, isNotAWord, isBlacklisted, timestamp); } - // Add a bigram entry to binary dictionary in native code. - public void addBigramWords(final String word0, final String word1, final int probability) { + // Add a bigram entry to binary dictionary with timestamp in native code. + public void addBigramWords(final String word0, final String word1, final int probability, + final int timestamp) { if (TextUtils.isEmpty(word0) || TextUtils.isEmpty(word1)) { return; } final int[] codePoints0 = StringUtils.toCodePointArray(word0); final int[] codePoints1 = StringUtils.toCodePointArray(word1); - addBigramWordsNative(mNativeDict, codePoints0, codePoints1, probability); + addBigramWordsNative(mNativeDict, codePoints0, codePoints1, probability, timestamp); } // Remove a bigram entry form binary dictionary in native code. @@ -303,10 +364,71 @@ public final class BinaryDictionary extends Dictionary { removeBigramWordsNative(mNativeDict, codePoints0, codePoints1); } + public static class LanguageModelParam { + public final int[] mWord0; + public final int[] mWord1; + // TODO: this needs to be a list of shortcuts + public final int[] mShortcutTarget; + public final int mUnigramProbability; + public final int mBigramProbability; + public final int mShortcutProbability; + public final boolean mIsNotAWord; + public final boolean mIsBlacklisted; + public final int mTimestamp; + + // Constructor for unigram. TODO: support shortcuts + public LanguageModelParam(final String word, final int unigramProbability, + final int timestamp) { + mWord0 = null; + mWord1 = StringUtils.toCodePointArray(word); + mShortcutTarget = null; + mUnigramProbability = unigramProbability; + mBigramProbability = NOT_A_PROBABILITY; + mShortcutProbability = NOT_A_PROBABILITY; + mIsNotAWord = false; + mIsBlacklisted = false; + mTimestamp = timestamp; + } + + // Constructor for unigram and bigram. + public LanguageModelParam(final String word0, final String word1, + final int unigramProbability, final int bigramProbability, + final int timestamp) { + mWord0 = StringUtils.toCodePointArray(word0); + mWord1 = StringUtils.toCodePointArray(word1); + mShortcutTarget = null; + mUnigramProbability = unigramProbability; + mBigramProbability = bigramProbability; + mShortcutProbability = NOT_A_PROBABILITY; + mIsNotAWord = false; + mIsBlacklisted = false; + mTimestamp = timestamp; + } + } + + public void addMultipleDictionaryEntries(final LanguageModelParam[] languageModelParams) { + if (!isValidDictionary()) return; + int processedParamCount = 0; + while (processedParamCount < languageModelParams.length) { + if (needsToRunGC(true /* mindsBlockByGC */)) { + flushWithGC(); + } + processedParamCount = addMultipleDictionaryEntriesNative(mNativeDict, + languageModelParams, processedParamCount); + if (processedParamCount <= 0) { + return; + } + } + + } + private void reopen() { close(); final File dictFile = new File(mDictFilePath); - mNativeDict = openNative(dictFile.getAbsolutePath(), 0 /* startOffset */, + // WARNING: Because we pass 0 as the offset and file.length() as the length, this can + // only be called for actual files. Right now it's only called by the flush() family of + // functions, which require an updatable dictionary, so it's okay. But beware. + loadDictionary(dictFile.getAbsolutePath(), 0 /* startOffset */, dictFile.length(), true /* isUpdatable */); } diff --git a/java/src/com/android/inputmethod/latin/BinaryDictionaryFileDumper.java b/java/src/com/android/inputmethod/latin/BinaryDictionaryFileDumper.java index 722a82961..b4382bc2c 100644 --- a/java/src/com/android/inputmethod/latin/BinaryDictionaryFileDumper.java +++ b/java/src/com/android/inputmethod/latin/BinaryDictionaryFileDumper.java @@ -98,7 +98,7 @@ public final class BinaryDictionaryFileDumper { * This creates a URI builder able to build a URI pointing to the dictionary * pack content provider for a specific dictionary id. */ - private static Uri.Builder getProviderUriBuilder(final String path) { + public static Uri.Builder getProviderUriBuilder(final String path) { return new Uri.Builder().scheme(ContentResolver.SCHEME_CONTENT) .authority(DictionaryPackConstants.AUTHORITY).appendPath(path); } @@ -339,15 +339,25 @@ public final class BinaryDictionaryFileDumper { Log.e(TAG, "Could not copy a word list. Will not be able to use it."); // If we can't copy it we should warn the dictionary provider so that it can mark it // as invalid. - wordListUriBuilder.appendQueryParameter(QUERY_PARAMETER_DELETE_RESULT, - QUERY_PARAMETER_FAILURE); + reportBrokenFileToDictionaryProvider(providerClient, clientId, wordlistId); + } + + public static boolean reportBrokenFileToDictionaryProvider( + final ContentProviderClient providerClient, final String clientId, + final String wordlistId) { try { + final Uri.Builder wordListUriBuilder = getContentUriBuilderForType(clientId, + providerClient, QUERY_PATH_DATAFILE, wordlistId /* extraPath */); + wordListUriBuilder.appendQueryParameter(QUERY_PARAMETER_DELETE_RESULT, + QUERY_PARAMETER_FAILURE); if (0 >= providerClient.delete(wordListUriBuilder.build(), null, null)) { - Log.e(TAG, "In addition, we were unable to delete it."); + Log.e(TAG, "Unable to delete a word list."); } } catch (RemoteException e) { - Log.e(TAG, "In addition, communication with the dictionary provider was cut", e); + Log.e(TAG, "Communication with the dictionary provider was cut", e); + return false; } + return true; } // Ideally the two following methods should be merged, but AssetFileDescriptor does not @@ -432,8 +442,9 @@ public final class BinaryDictionaryFileDumper { // Actually copy the file final byte[] buffer = new byte[FILE_READ_BUFFER_SIZE]; - for (int readBytes = input.read(buffer); readBytes >= 0; readBytes = input.read(buffer)) + for (int readBytes = input.read(buffer); readBytes >= 0; readBytes = input.read(buffer)) { output.write(buffer, 0, readBytes); + } input.close(); } @@ -478,8 +489,7 @@ public final class BinaryDictionaryFileDumper { * @param context the context for resources and providers. * @param clientId the client ID to use. */ - public static void initializeClientRecordHelper(final Context context, - final String clientId) { + public static void initializeClientRecordHelper(final Context context, final String clientId) { try { final ContentProviderClient client = context.getContentResolver(). acquireContentProviderClient(getProviderUriBuilder("").build()); diff --git a/java/src/com/android/inputmethod/latin/Constants.java b/java/src/com/android/inputmethod/latin/Constants.java index 9a9653094..c260434d5 100644 --- a/java/src/com/android/inputmethod/latin/Constants.java +++ b/java/src/com/android/inputmethod/latin/Constants.java @@ -70,38 +70,47 @@ public final class Constants { public static final class ExtraValue { /** - * The subtype extra value used to indicate that the subtype keyboard layout is capable - * for typing ASCII characters. + * The subtype extra value used to indicate that this subtype is capable of + * entering ASCII characters. */ public static final String ASCII_CAPABLE = "AsciiCapable"; /** - * The subtype extra value used to indicate that the subtype keyboard layout is capable - * for typing EMOJI characters. + * The subtype extra value used to indicate that this subtype is enabled + * when the default subtype is not marked as ascii capable. + */ + public static final String ENABLED_WHEN_DEFAULT_IS_NOT_ASCII_CAPABLE = + "EnabledWhenDefaultIsNotAsciiCapable"; + + /** + * The subtype extra value used to indicate that this subtype is capable of + * entering emoji characters. */ public static final String EMOJI_CAPABLE = "EmojiCapable"; + /** - * The subtype extra value used to indicate that the subtype require network connection - * to work. + * The subtype extra value used to indicate that this subtype requires a network + * connection to work. */ public static final String REQ_NETWORK_CONNECTIVITY = "requireNetworkConnectivity"; /** - * The subtype extra value used to indicate that the subtype display name contains "%s" - * for replacement mark and it should be replaced by this extra value. + * The subtype extra value used to indicate that the display name of this subtype + * contains a "%s" for printf-like replacement and it should be replaced by + * this extra value. * This extra value is supported on JellyBean and later. */ public static final String UNTRANSLATABLE_STRING_IN_SUBTYPE_NAME = "UntranslatableReplacementStringInSubtypeName"; /** - * The subtype extra value used to indicate that the subtype keyboard layout set name. + * The subtype extra value used to indicate this subtype keyboard layout set name. * This extra value is private to LatinIME. */ public static final String KEYBOARD_LAYOUT_SET = "KeyboardLayoutSet"; /** - * The subtype extra value used to indicate that the subtype is additional subtype + * The subtype extra value used to indicate that this subtype is an additional subtype * that the user defined. This extra value is private to LatinIME. */ public static final String IS_ADDITIONAL_SUBTYPE = "isAdditionalSubtype"; diff --git a/java/src/com/android/inputmethod/latin/ContactsBinaryDictionary.java b/java/src/com/android/inputmethod/latin/ContactsBinaryDictionary.java index 47891c6b7..ae41e74e3 100644 --- a/java/src/com/android/inputmethod/latin/ContactsBinaryDictionary.java +++ b/java/src/com/android/inputmethod/latin/ContactsBinaryDictionary.java @@ -33,6 +33,7 @@ import android.util.Log; import com.android.inputmethod.latin.utils.StringUtils; +import java.io.File; import java.util.List; import java.util.Locale; @@ -44,7 +45,8 @@ public class ContactsBinaryDictionary extends ExpandableBinaryDictionary { private static final String TAG = ContactsBinaryDictionary.class.getSimpleName(); private static final String NAME = "contacts"; - private static boolean DEBUG = false; + private static final boolean DEBUG = false; + private static final boolean DEBUG_DUMP = false; /** * Frequency for contacts information into the dictionary @@ -71,8 +73,8 @@ public class ContactsBinaryDictionary extends ExpandableBinaryDictionary { private final boolean mUseFirstLastBigrams; public ContactsBinaryDictionary(final Context context, final Locale locale) { - super(context, getFilenameWithLocale(NAME, locale.toString()), Dictionary.TYPE_CONTACTS, - false /* isUpdatable */); + super(context, getDictNameWithLocale(NAME, locale), locale, + Dictionary.TYPE_CONTACTS, false /* isUpdatable */); mLocale = locale; mUseFirstLastBigrams = useFirstLastBigramsForLocale(locale); registerObserver(context); @@ -82,6 +84,11 @@ public class ContactsBinaryDictionary extends ExpandableBinaryDictionary { loadDictionary(); } + // Dummy constructor for tests. + public ContactsBinaryDictionary(final Context context, final Locale locale, final File file) { + this(context, locale); + } + private synchronized void registerObserver(final Context context) { // Perform a managed query. The Activity will handle closing and requerying the cursor // when needed. @@ -168,6 +175,10 @@ public class ContactsBinaryDictionary extends ExpandableBinaryDictionary { if (isValidName(name)) { addName(name); ++count; + } else { + if (DEBUG_DUMP) { + Log.d(TAG, "Invalid name: " + name); + } } cursor.moveToNext(); } @@ -204,6 +215,9 @@ public class ContactsBinaryDictionary extends ExpandableBinaryDictionary { if (Character.isLetter(name.codePointAt(i))) { int end = getWordEndPosition(name, len, i); String word = name.substring(i, end); + if (DEBUG_DUMP) { + Log.d(TAG, "addName word = " + word); + } i = end - 1; // Don't add single letter words, possibly confuses // capitalization of i. diff --git a/java/src/com/android/inputmethod/latin/Dictionary.java b/java/src/com/android/inputmethod/latin/Dictionary.java index fa79f5af7..e04524843 100644 --- a/java/src/com/android/inputmethod/latin/Dictionary.java +++ b/java/src/com/android/inputmethod/latin/Dictionary.java @@ -52,13 +52,10 @@ public abstract class Dictionary { public static final String TYPE_CONTACTS = "contacts"; // User dictionary, the system-managed one. public static final String TYPE_USER = "user"; - // User history dictionary internal to LatinIME. This assumes bigram prediction for now. + // User history dictionary internal to LatinIME. public static final String TYPE_USER_HISTORY = "history"; - // Personalization binary dictionary internal to LatinIME. + // Personalization dictionary. public static final String TYPE_PERSONALIZATION = "personalization"; - // Personalization prediction dictionary internal to LatinIME's Java code. - public static final String TYPE_PERSONALIZATION_PREDICTION_IN_JAVA = - "personalization_prediction_in_java"; public final String mDictType; public Dictionary(final String dictType) { diff --git a/java/src/com/android/inputmethod/latin/DictionaryFactory.java b/java/src/com/android/inputmethod/latin/DictionaryFactory.java index 828e54f14..e09c309ea 100644 --- a/java/src/com/android/inputmethod/latin/DictionaryFactory.java +++ b/java/src/com/android/inputmethod/latin/DictionaryFactory.java @@ -16,6 +16,7 @@ package com.android.inputmethod.latin; +import android.content.ContentProviderClient; import android.content.Context; import android.content.res.AssetFileDescriptor; import android.content.res.Resources; @@ -64,6 +65,10 @@ public final class DictionaryFactory { useFullEditDistance, locale, Dictionary.TYPE_MAIN); if (readOnlyBinaryDictionary.isValidDictionary()) { dictList.add(readOnlyBinaryDictionary); + } else { + readOnlyBinaryDictionary.close(); + // Prevent this dictionary to do any further harm. + killDictionary(context, f); } } } @@ -75,6 +80,51 @@ public final class DictionaryFactory { } /** + * Kills a dictionary so that it is never used again, if possible. + * @param context The context to contact the dictionary provider, if possible. + * @param f A file address to the dictionary to kill. + */ + private static void killDictionary(final Context context, final AssetFileAddress f) { + if (f.pointsToPhysicalFile()) { + f.deleteUnderlyingFile(); + // Warn the dictionary provider if the dictionary came from there. + final ContentProviderClient providerClient; + try { + providerClient = context.getContentResolver().acquireContentProviderClient( + BinaryDictionaryFileDumper.getProviderUriBuilder("").build()); + } catch (final SecurityException e) { + Log.e(TAG, "No permission to communicate with the dictionary provider", e); + return; + } + if (null == providerClient) { + Log.e(TAG, "Can't establish communication with the dictionary provider"); + return; + } + final String wordlistId = + DictionaryInfoUtils.getWordListIdFromFileName(new File(f.mFilename).getName()); + if (null != wordlistId) { + // TODO: this is a reasonable last resort, but it is suboptimal. + // The following will remove the entry for this dictionary with the dictionary + // provider. When the metadata is downloaded again, we will try downloading it + // again. + // However, in the practice that will mean the user will find themselves without + // the new dictionary. That's fine for languages where it's included in the APK, + // but for other languages it will leave the user without a dictionary at all until + // the next update, which may be a few days away. + // Ideally, we would trigger a new download right away, and use increasing retry + // delays for this particular id/version combination. + // Then again, this is expected to only ever happen in case of human mistake. If + // the wrong file is on the server, the following is still doing the right thing. + // If it's a file left over from the last version however, it's not great. + BinaryDictionaryFileDumper.reportBrokenFileToDictionaryProvider( + providerClient, + context.getString(R.string.dictionary_pack_client_id), + wordlistId); + } + } + } + + /** * Initializes a main dictionary collection from a dictionary pack, with default flags. * * This searches for a content provider providing a dictionary pack for the specified diff --git a/java/src/com/android/inputmethod/latin/DictionaryWriter.java b/java/src/com/android/inputmethod/latin/DictionaryWriter.java index 3df2a2b63..89ef96d7f 100644 --- a/java/src/com/android/inputmethod/latin/DictionaryWriter.java +++ b/java/src/com/android/inputmethod/latin/DictionaryWriter.java @@ -18,8 +18,6 @@ package com.android.inputmethod.latin; import android.content.Context; -import com.android.inputmethod.keyboard.ProximityInfo; -import com.android.inputmethod.latin.SuggestedWords.SuggestedWordInfo; import com.android.inputmethod.latin.makedict.DictEncoder; import com.android.inputmethod.latin.makedict.FormatSpec; import com.android.inputmethod.latin.makedict.FusionDictionary; @@ -37,14 +35,14 @@ import java.util.Map; * An in memory dictionary for memorizing entries and writing a binary dictionary. */ public class DictionaryWriter extends AbstractDictionaryWriter { - private static final int BINARY_DICT_VERSION = 3; + private static final int BINARY_DICT_VERSION = 2; private static final FormatSpec.FormatOptions FORMAT_OPTIONS = - new FormatSpec.FormatOptions(BINARY_DICT_VERSION, true /* supportsDynamicUpdate */); + new FormatSpec.FormatOptions(BINARY_DICT_VERSION, false /* supportsDynamicUpdate */); private FusionDictionary mFusionDictionary; - public DictionaryWriter(final Context context, final String dictType) { - super(context, dictType); + public DictionaryWriter(final Context context) { + super(context); clear(); } @@ -52,7 +50,7 @@ public class DictionaryWriter extends AbstractDictionaryWriter { public void clear() { final HashMap<String, String> attributes = CollectionUtils.newHashMap(); mFusionDictionary = new FusionDictionary(new PtNodeArray(), - new FusionDictionary.DictionaryOptions(attributes, false, false)); + new FusionDictionary.DictionaryOptions(attributes)); } /** @@ -92,18 +90,4 @@ public class DictionaryWriter extends AbstractDictionaryWriter { } dictEncoder.writeDictionary(mFusionDictionary, FORMAT_OPTIONS); } - - @Override - public ArrayList<SuggestedWordInfo> getSuggestions(final WordComposer composer, - final String prevWord, final ProximityInfo proximityInfo, - boolean blockOffensiveWords, final int[] additionalFeaturesOptions) { - // This class doesn't support suggestion. - return null; - } - - @Override - public boolean isValidWord(String word) { - // This class doesn't support dictionary retrieval. - return false; - } } diff --git a/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java b/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java index eb8650e6f..f785835b8 100644 --- a/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java +++ b/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java @@ -17,23 +17,26 @@ package com.android.inputmethod.latin; import android.content.Context; -import android.os.SystemClock; import android.util.Log; import com.android.inputmethod.annotations.UsedForTesting; import com.android.inputmethod.keyboard.ProximityInfo; +import com.android.inputmethod.latin.BinaryDictionary.LanguageModelParam; import com.android.inputmethod.latin.makedict.FormatSpec; -import com.android.inputmethod.latin.personalization.DynamicPersonalizationDictionaryWriter; import com.android.inputmethod.latin.SuggestedWords.SuggestedWordInfo; import com.android.inputmethod.latin.utils.AsyncResultHolder; import com.android.inputmethod.latin.utils.CollectionUtils; +import com.android.inputmethod.latin.utils.FileUtils; import com.android.inputmethod.latin.utils.PrioritizedSerialExecutor; import java.io.File; import java.util.ArrayList; import java.util.HashMap; +import java.util.Locale; import java.util.Map; import java.util.concurrent.ConcurrentHashMap; +import java.util.concurrent.CountDownLatch; +import java.util.concurrent.TimeUnit; import java.util.concurrent.atomic.AtomicBoolean; import java.util.concurrent.atomic.AtomicReference; @@ -52,10 +55,7 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { /** Whether to print debug output to log */ private static boolean DEBUG = false; - - // TODO: Remove. - /** Whether to call binary dictionary dynamically updating methods. */ - public static boolean ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE = true; + private static final boolean DBG_STRESS_TEST = false; private static final int TIMEOUT_FOR_READ_OPS_IN_MILLISECONDS = 100; @@ -64,22 +64,19 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { */ protected static final int MAX_WORD_LENGTH = Constants.DICTIONARY_MAX_WORD_LENGTH; - private static final int DICTIONARY_FORMAT_VERSION = 3; - - private static final String SUPPORTS_DYNAMIC_UPDATE = - FormatSpec.FileHeader.ATTRIBUTE_VALUE_TRUE; + private static final int DICTIONARY_FORMAT_VERSION = FormatSpec.VERSION4; /** * A static map of update controllers, each of which records the time of accesses to a single * binary dictionary file and tracks whether the file is regenerating. The key for this map is - * the filename and the value is the shared dictionary time recorder associated with that - * filename. + * the dictionary name and the value is the shared dictionary time recorder associated with + * that dictionary name. */ private static final ConcurrentHashMap<String, DictionaryUpdateController> - sFilenameDictionaryUpdateControllerMap = CollectionUtils.newConcurrentHashMap(); + sDictNameDictionaryUpdateControllerMap = CollectionUtils.newConcurrentHashMap(); private static final ConcurrentHashMap<String, PrioritizedSerialExecutor> - sFilenameExecutorMap = CollectionUtils.newConcurrentHashMap(); + sDictNameExecutorMap = CollectionUtils.newConcurrentHashMap(); /** The application context. */ protected final Context mContext; @@ -95,18 +92,24 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { protected AbstractDictionaryWriter mDictionaryWriter; /** - * The name of this dictionary, used as the filename for storing the binary dictionary. Multiple - * dictionary instances with the same filename is supported, with access controlled by - * DictionaryTimeRecorder. + * The name of this dictionary, used as a part of the filename for storing the binary + * dictionary. Multiple dictionary instances with the same name is supported, with access + * controlled by DictionaryUpdateController. */ - private final String mFilename; + private final String mDictName; + + /** Dictionary locale */ + private final Locale mLocale; /** Whether to support dynamically updating the dictionary */ private final boolean mIsUpdatable; + /** Dictionary file */ + private final File mDictFile; + // TODO: remove, once dynamic operations is serialized /** Controls updating the shared binary dictionary file across multiple instances. */ - private final DictionaryUpdateController mFilenameDictionaryUpdateController; + private final DictionaryUpdateController mDictNameDictionaryUpdateController; // TODO: remove, once dynamic operations is serialized /** Controls updating the local binary dictionary for this instance. */ @@ -132,45 +135,70 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { */ protected abstract boolean hasContentChanged(); + protected boolean matchesExpectedBinaryDictFormatVersionForThisType(final int formatVersion) { + // This class is using format 2 because it's used by the User and Contacts dictionary + // only, which right now use format 2 (dicts using format 4 use Decaying*, which overrides + // this method). + // TODO: Migrate these dicts to ver4 format, and remove this function. + return formatVersion == FormatSpec.VERSION2; + } + + public boolean isValidDictionary() { + return mBinaryDictionary.isValidDictionary(); + } + + private File getDictFile() { + return mDictFile; + } + /** - * Gets the dictionary update controller for the given filename. + * Gets the dictionary update controller for the given dictionary name. */ private static DictionaryUpdateController getDictionaryUpdateController( - String filename) { - DictionaryUpdateController recorder = sFilenameDictionaryUpdateControllerMap.get(filename); + final String dictName) { + DictionaryUpdateController recorder = sDictNameDictionaryUpdateControllerMap.get(dictName); if (recorder == null) { - synchronized(sFilenameDictionaryUpdateControllerMap) { + synchronized(sDictNameDictionaryUpdateControllerMap) { recorder = new DictionaryUpdateController(); - sFilenameDictionaryUpdateControllerMap.put(filename, recorder); + sDictNameDictionaryUpdateControllerMap.put(dictName, recorder); } } return recorder; } /** - * Gets the executor for the given filename. + * Gets the executor for the given dictionary name. */ - private static PrioritizedSerialExecutor getExecutor(final String filename) { - PrioritizedSerialExecutor executor = sFilenameExecutorMap.get(filename); + private static PrioritizedSerialExecutor getExecutor(final String dictName) { + PrioritizedSerialExecutor executor = sDictNameExecutorMap.get(dictName); if (executor == null) { - synchronized(sFilenameExecutorMap) { + synchronized(sDictNameExecutorMap) { executor = new PrioritizedSerialExecutor(); - sFilenameExecutorMap.put(filename, executor); + sDictNameExecutorMap.put(dictName, executor); } } return executor; } + /** + * Shutdowns all executors and removes all executors from the executor map for testing. + */ + @UsedForTesting + public static void shutdownAllExecutors() { + synchronized(sDictNameExecutorMap) { + for (final PrioritizedSerialExecutor executor : sDictNameExecutorMap.values()) { + executor.shutdown(); + sDictNameExecutorMap.remove(executor); + } + } + } + private static AbstractDictionaryWriter getDictionaryWriter(final Context context, - final String dictType, final boolean isDynamicPersonalizationDictionary) { + final boolean isDynamicPersonalizationDictionary) { if (isDynamicPersonalizationDictionary) { - if (ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE) { - return null; - } else { - return new DynamicPersonalizationDictionaryWriter(context, dictType); - } + return null; } else { - return new DictionaryWriter(context, dictType); + return new DictionaryWriter(context); } } @@ -178,26 +206,37 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { * Creates a new expandable binary dictionary. * * @param context The application context of the parent. - * @param filename The filename for this binary dictionary. Multiple dictionaries with the same - * filename is supported. + * @param dictName The name of the dictionary. Multiple instances with the same + * name is supported. + * @param locale the dictionary locale. * @param dictType the dictionary type, as a human-readable string * @param isUpdatable whether to support dynamically updating the dictionary. Please note that * dynamic dictionary has negative effects on memory space and computation time. */ - public ExpandableBinaryDictionary(final Context context, final String filename, - final String dictType, final boolean isUpdatable) { + public ExpandableBinaryDictionary(final Context context, final String dictName, + final Locale locale, final String dictType, final boolean isUpdatable) { + this(context, dictName, locale, dictType, isUpdatable, + new File(context.getFilesDir(), dictName + DICT_FILE_EXTENSION)); + } + + // Creates an instance that uses a given dictionary file. + public ExpandableBinaryDictionary(final Context context, final String dictName, + final Locale locale, final String dictType, final boolean isUpdatable, + final File dictFile) { super(dictType); - mFilename = filename; + mDictName = dictName; mContext = context; + mLocale = locale; mIsUpdatable = isUpdatable; + mDictFile = dictFile; mBinaryDictionary = null; - mFilenameDictionaryUpdateController = getDictionaryUpdateController(filename); + mDictNameDictionaryUpdateController = getDictionaryUpdateController(dictName); // Currently, only dynamic personalization dictionary is updatable. - mDictionaryWriter = getDictionaryWriter(context, dictType, isUpdatable); + mDictionaryWriter = getDictionaryWriter(context, isUpdatable); } - protected static String getFilenameWithLocale(final String name, final String localeStr) { - return name + "." + localeStr + DICT_FILE_EXTENSION; + protected static String getDictNameWithLocale(final String name, final Locale locale) { + return name + "." + locale.toString(); } /** @@ -205,23 +244,20 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { */ @Override public void close() { - getExecutor(mFilename).execute(new Runnable() { + getExecutor(mDictName).execute(new Runnable() { @Override public void run() { if (mBinaryDictionary!= null) { mBinaryDictionary.close(); mBinaryDictionary = null; } - if (mDictionaryWriter != null) { - mDictionaryWriter.close(); - } } }); } protected void closeBinaryDictionary() { // Ensure that no other threads are accessing the local binary dictionary. - getExecutor(mFilename).execute(new Runnable() { + getExecutor(mDictName).execute(new Runnable() { @Override public void run() { if (mBinaryDictionary != null) { @@ -234,19 +270,23 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { protected Map<String, String> getHeaderAttributeMap() { HashMap<String, String> attributeMap = new HashMap<String, String>(); - attributeMap.put(FormatSpec.FileHeader.SUPPORTS_DYNAMIC_UPDATE_ATTRIBUTE, - SUPPORTS_DYNAMIC_UPDATE); - attributeMap.put(FormatSpec.FileHeader.DICTIONARY_ID_ATTRIBUTE, mFilename); + attributeMap.put(FormatSpec.FileHeader.DICTIONARY_ID_ATTRIBUTE, mDictName); + attributeMap.put(FormatSpec.FileHeader.DICTIONARY_LOCALE_ATTRIBUTE, mLocale.toString()); + attributeMap.put(FormatSpec.FileHeader.DICTIONARY_VERSION_ATTRIBUTE, + String.valueOf(TimeUnit.MILLISECONDS.toSeconds(System.currentTimeMillis()))); return attributeMap; } protected void clear() { - getExecutor(mFilename).execute(new Runnable() { + getExecutor(mDictName).execute(new Runnable() { @Override public void run() { - if (ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE && mDictionaryWriter == null) { + if (mDictionaryWriter == null) { mBinaryDictionary.close(); - final File file = new File(mContext.getFilesDir(), mFilename); + final File file = getDictFile(); + if (file.exists() && !FileUtils.deleteRecursively(file)) { + Log.e(TAG, "Can't remove a file: " + file.getName()); + } BinaryDictionary.createEmptyDictFile(file.getAbsolutePath(), DICTIONARY_FORMAT_VERSION, getHeaderAttributeMap()); mBinaryDictionary = new BinaryDictionary( @@ -286,8 +326,7 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { * Check whether GC is needed and run GC if required. */ protected void runGCIfRequired(final boolean mindsBlockByGC) { - if (!ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE) return; - getExecutor(mFilename).execute(new Runnable() { + getExecutor(mDictName).execute(new Runnable() { @Override public void run() { runGCIfRequiredInternalLocked(mindsBlockByGC); @@ -296,18 +335,17 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { } private void runGCIfRequiredInternalLocked(final boolean mindsBlockByGC) { - if (!ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE) return; // Calls to needsToRunGC() need to be serialized. if (mBinaryDictionary.needsToRunGC(mindsBlockByGC)) { - if (setIsRegeneratingIfNotRegenerating()) { + if (setProcessingLargeTaskIfNot()) { // Run GC after currently existing time sensitive operations. - getExecutor(mFilename).executePrioritized(new Runnable() { + getExecutor(mDictName).executePrioritized(new Runnable() { @Override public void run() { try { mBinaryDictionary.flushWithGC(); } finally { - mFilenameDictionaryUpdateController.mIsRegenerating.set(false); + mDictNameDictionaryUpdateController.mProcessingLargeTask.set(false); } } }); @@ -318,23 +356,19 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { /** * Dynamically adds a word unigram to the dictionary. May overwrite an existing entry. */ - protected void addWordDynamically(final String word, final String shortcutTarget, - final int frequency, final int shortcutFreq, final boolean isNotAWord) { + protected void addWordDynamically(final String word, final int frequency, + final String shortcutTarget, final int shortcutFreq, final boolean isNotAWord, + final boolean isBlacklisted, final int timestamp) { if (!mIsUpdatable) { - Log.w(TAG, "addWordDynamically is called for non-updatable dictionary: " + mFilename); + Log.w(TAG, "addWordDynamically is called for non-updatable dictionary: " + mDictName); return; } - getExecutor(mFilename).execute(new Runnable() { + getExecutor(mDictName).execute(new Runnable() { @Override public void run() { - if (ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE) { - runGCIfRequiredInternalLocked(true /* mindsBlockByGC */); - mBinaryDictionary.addUnigramWord(word, frequency); - } else { - // TODO: Remove. - mDictionaryWriter.addUnigramWord(word, shortcutTarget, frequency, shortcutFreq, - isNotAWord); - } + runGCIfRequiredInternalLocked(true /* mindsBlockByGC */); + mBinaryDictionary.addUnigramWord(word, frequency, shortcutTarget, shortcutFreq, + isNotAWord, isBlacklisted, timestamp); } }); } @@ -343,23 +377,17 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { * Dynamically adds a word bigram in the dictionary. May overwrite an existing entry. */ protected void addBigramDynamically(final String word0, final String word1, - final int frequency, final boolean isValid) { + final int frequency, final int timestamp) { if (!mIsUpdatable) { Log.w(TAG, "addBigramDynamically is called for non-updatable dictionary: " - + mFilename); + + mDictName); return; } - getExecutor(mFilename).execute(new Runnable() { + getExecutor(mDictName).execute(new Runnable() { @Override public void run() { - if (ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE) { - runGCIfRequiredInternalLocked(true /* mindsBlockByGC */); - mBinaryDictionary.addBigramWords(word0, word1, frequency); - } else { - // TODO: Remove. - mDictionaryWriter.addBigramWords(word0, word1, frequency, isValid, - 0 /* lastTouchedTime */); - } + runGCIfRequiredInternalLocked(true /* mindsBlockByGC */); + mBinaryDictionary.addBigramWords(word0, word1, frequency, timestamp); } }); } @@ -370,18 +398,48 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { protected void removeBigramDynamically(final String word0, final String word1) { if (!mIsUpdatable) { Log.w(TAG, "removeBigramDynamically is called for non-updatable dictionary: " - + mFilename); + + mDictName); return; } - getExecutor(mFilename).execute(new Runnable() { + getExecutor(mDictName).execute(new Runnable() { @Override public void run() { - if (ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE) { - runGCIfRequiredInternalLocked(true /* mindsBlockByGC */); - mBinaryDictionary.removeBigramWords(word0, word1); - } else { - // TODO: Remove. - mDictionaryWriter.removeBigramWords(word0, word1); + runGCIfRequiredInternalLocked(true /* mindsBlockByGC */); + mBinaryDictionary.removeBigramWords(word0, word1); + } + }); + } + + public interface AddMultipleDictionaryEntriesCallback { + public void onFinished(); + } + + /** + * Dynamically add multiple entries to the dictionary. + */ + protected void addMultipleDictionaryEntriesDynamically( + final ArrayList<LanguageModelParam> languageModelParams, + final AddMultipleDictionaryEntriesCallback callback) { + if (!mIsUpdatable) { + Log.w(TAG, "addMultipleDictionaryEntriesDynamically is called for non-updatable " + + "dictionary: " + mDictName); + return; + } + getExecutor(mDictName).execute(new Runnable() { + @Override + public void run() { + final boolean locked = setProcessingLargeTaskIfNot(); + try { + mBinaryDictionary.addMultipleDictionaryEntries( + languageModelParams.toArray( + new LanguageModelParam[languageModelParams.size()])); + } finally { + if (callback != null) { + callback.onFinished(); + } + if (locked) { + mDictNameDictionaryUpdateController.mProcessingLargeTask.set(false); + } } } }); @@ -393,49 +451,23 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { final boolean blockOffensiveWords, final int[] additionalFeaturesOptions, final int sessionId) { reloadDictionaryIfRequired(); - if (isRegenerating()) { + if (processingLargeTask()) { return null; } - final ArrayList<SuggestedWordInfo> suggestions = CollectionUtils.newArrayList(); final AsyncResultHolder<ArrayList<SuggestedWordInfo>> holder = new AsyncResultHolder<ArrayList<SuggestedWordInfo>>(); - getExecutor(mFilename).executePrioritized(new Runnable() { + getExecutor(mDictName).executePrioritized(new Runnable() { @Override public void run() { - if (ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE) { - if (mBinaryDictionary == null) { - holder.set(null); - return; - } - final ArrayList<SuggestedWordInfo> binarySuggestion = - mBinaryDictionary.getSuggestionsWithSessionId(composer, prevWord, - proximityInfo, blockOffensiveWords, additionalFeaturesOptions, - sessionId); - holder.set(binarySuggestion); - } else { - final ArrayList<SuggestedWordInfo> inMemDictSuggestion = - composer.isBatchMode() ? null : - mDictionaryWriter.getSuggestionsWithSessionId(composer, - prevWord, proximityInfo, blockOffensiveWords, - additionalFeaturesOptions, sessionId); - // TODO: Remove checking mIsUpdatable and use native suggestion. - if (mBinaryDictionary != null && !mIsUpdatable) { - final ArrayList<SuggestedWordInfo> binarySuggestion = - mBinaryDictionary.getSuggestionsWithSessionId(composer, prevWord, - proximityInfo, blockOffensiveWords, - additionalFeaturesOptions, sessionId); - if (inMemDictSuggestion == null) { - holder.set(binarySuggestion); - } else if (binarySuggestion == null) { - holder.set(inMemDictSuggestion); - } else { - binarySuggestion.addAll(inMemDictSuggestion); - holder.set(binarySuggestion); - } - } else { - holder.set(inMemDictSuggestion); - } + if (mBinaryDictionary == null) { + holder.set(null); + return; } + final ArrayList<SuggestedWordInfo> binarySuggestion = + mBinaryDictionary.getSuggestionsWithSessionId(composer, prevWord, + proximityInfo, blockOffensiveWords, additionalFeaturesOptions, + sessionId); + holder.set(binarySuggestion); } }); return holder.get(null, TIMEOUT_FOR_READ_OPS_IN_MILLISECONDS); @@ -456,11 +488,11 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { } protected boolean isValidWordInner(final String word) { - if (isRegenerating()) { + if (processingLargeTask()) { return false; } final AsyncResultHolder<Boolean> holder = new AsyncResultHolder<Boolean>(); - getExecutor(mFilename).executePrioritized(new Runnable() { + getExecutor(mDictName).executePrioritized(new Runnable() { @Override public void run() { holder.set(isValidWordLocked(word)); @@ -484,7 +516,7 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { * dictionary exists, this method will generate one. */ protected void loadDictionary() { - mPerInstanceDictionaryUpdateController.mLastUpdateRequestTime = SystemClock.uptimeMillis(); + mPerInstanceDictionaryUpdateController.mLastUpdateRequestTime = System.currentTimeMillis(); reloadDictionaryIfRequired(); } @@ -494,12 +526,22 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { */ private void loadBinaryDictionary() { if (DEBUG) { - Log.d(TAG, "Loading binary dictionary: " + mFilename + " request=" - + mFilenameDictionaryUpdateController.mLastUpdateRequestTime + " update=" - + mFilenameDictionaryUpdateController.mLastUpdateTime); + Log.d(TAG, "Loading binary dictionary: " + mDictName + " request=" + + mDictNameDictionaryUpdateController.mLastUpdateRequestTime + " update=" + + mDictNameDictionaryUpdateController.mLastUpdateTime); + } + if (DBG_STRESS_TEST) { + // Test if this class does not cause problems when it takes long time to load binary + // dictionary. + try { + Log.w(TAG, "Start stress in loading: " + mDictName); + Thread.sleep(15000); + Log.w(TAG, "End stress in loading"); + } catch (InterruptedException e) { + } } - final File file = new File(mContext.getFilesDir(), mFilename); + final File file = getDictFile(); final String filename = file.getAbsolutePath(); final long length = file.length(); @@ -511,7 +553,7 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { // swapping in the new one. // TODO: Ensure multi-thread assignment of mBinaryDictionary. final BinaryDictionary oldBinaryDictionary = mBinaryDictionary; - getExecutor(mFilename).executePrioritized(new Runnable() { + getExecutor(mDictName).executePrioritized(new Runnable() { @Override public void run() { mBinaryDictionary = newBinaryDictionary; @@ -533,29 +575,31 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { */ private void writeBinaryDictionary() { if (DEBUG) { - Log.d(TAG, "Generating binary dictionary: " + mFilename + " request=" - + mFilenameDictionaryUpdateController.mLastUpdateRequestTime + " update=" - + mFilenameDictionaryUpdateController.mLastUpdateTime); + Log.d(TAG, "Generating binary dictionary: " + mDictName + " request=" + + mDictNameDictionaryUpdateController.mLastUpdateRequestTime + " update=" + + mDictNameDictionaryUpdateController.mLastUpdateTime); } if (needsToReloadBeforeWriting()) { mDictionaryWriter.clear(); loadDictionaryAsync(); - mDictionaryWriter.write(mFilename, getHeaderAttributeMap()); + mDictionaryWriter.write(getDictFile(), getHeaderAttributeMap()); } else { - if (ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE) { - if (mBinaryDictionary == null || !mBinaryDictionary.isValidDictionary()) { - final File file = new File(mContext.getFilesDir(), mFilename); - BinaryDictionary.createEmptyDictFile(file.getAbsolutePath(), - DICTIONARY_FORMAT_VERSION, getHeaderAttributeMap()); - } else { - if (mBinaryDictionary.needsToRunGC(false /* mindsBlockByGC */)) { - mBinaryDictionary.flushWithGC(); - } else { - mBinaryDictionary.flush(); - } + if (mBinaryDictionary == null || !isValidDictionary() + // TODO: remove the check below + || !matchesExpectedBinaryDictFormatVersionForThisType( + mBinaryDictionary.getFormatVersion())) { + final File file = getDictFile(); + if (file.exists() && !FileUtils.deleteRecursively(file)) { + Log.e(TAG, "Can't remove a file: " + file.getName()); } + BinaryDictionary.createEmptyDictFile(file.getAbsolutePath(), + DICTIONARY_FORMAT_VERSION, getHeaderAttributeMap()); } else { - mDictionaryWriter.write(mFilename, getHeaderAttributeMap()); + if (mBinaryDictionary.needsToRunGC(false /* mindsBlockByGC */)) { + mBinaryDictionary.flushWithGC(); + } else { + mBinaryDictionary.flush(); + } } } } @@ -568,12 +612,12 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { * the current binary dictionary from file. */ protected void setRequiresReload(final boolean requiresRebuild) { - final long time = SystemClock.uptimeMillis(); + final long time = System.currentTimeMillis(); mPerInstanceDictionaryUpdateController.mLastUpdateRequestTime = time; - mFilenameDictionaryUpdateController.mLastUpdateRequestTime = time; + mDictNameDictionaryUpdateController.mLastUpdateRequestTime = time; if (DEBUG) { - Log.d(TAG, "Reload request: " + mFilename + ": request=" + time + " update=" - + mFilenameDictionaryUpdateController.mLastUpdateTime); + Log.d(TAG, "Reload request: " + mDictName + ": request=" + time + " update=" + + mDictNameDictionaryUpdateController.mLastUpdateTime); } } @@ -582,7 +626,7 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { */ public final void reloadDictionaryIfRequired() { if (!isReloadRequired()) return; - if (setIsRegeneratingIfNotRegenerating()) { + if (setProcessingLargeTaskIfNot()) { reloadDictionary(); } } @@ -594,13 +638,14 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { return mBinaryDictionary == null || mPerInstanceDictionaryUpdateController.isOutOfDate(); } - private boolean isRegenerating() { - return mFilenameDictionaryUpdateController.mIsRegenerating.get(); + private boolean processingLargeTask() { + return mDictNameDictionaryUpdateController.mProcessingLargeTask.get(); } - // Returns whether the dictionary can be regenerated. - private boolean setIsRegeneratingIfNotRegenerating() { - return mFilenameDictionaryUpdateController.mIsRegenerating.compareAndSet( + // Returns whether the dictionary is being used for a large task. If true, we should not use + // this dictionary for latency sensitive operations. + private boolean setProcessingLargeTaskIfNot() { + return mDictNameDictionaryUpdateController.mProcessingLargeTask.compareAndSet( false /* expect */ , true /* update */); } @@ -611,13 +656,13 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { private final void reloadDictionary() { // Ensure that only one thread attempts to read or write to the shared binary dictionary // file at the same time. - getExecutor(mFilename).execute(new Runnable() { + getExecutor(mDictName).execute(new Runnable() { @Override public void run() { try { - final long time = SystemClock.uptimeMillis(); + final long time = System.currentTimeMillis(); final boolean dictionaryFileExists = dictionaryFileExists(); - if (mFilenameDictionaryUpdateController.isOutOfDate() + if (mDictNameDictionaryUpdateController.isOutOfDate() || !dictionaryFileExists) { // If the shared dictionary file does not exist or is out of date, the // first instance that acquires the lock will generate a new one. @@ -626,31 +671,44 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { // rebuild the binary dictionary. Empty dictionaries are supported (in // the case where loadDictionaryAsync() adds nothing) in order to // provide a uniform framework. - mFilenameDictionaryUpdateController.mLastUpdateTime = time; + mDictNameDictionaryUpdateController.mLastUpdateTime = time; writeBinaryDictionary(); loadBinaryDictionary(); } else { // If not, the reload request was unnecessary so revert // LastUpdateRequestTime to LastUpdateTime. - mFilenameDictionaryUpdateController.mLastUpdateRequestTime = - mFilenameDictionaryUpdateController.mLastUpdateTime; + mDictNameDictionaryUpdateController.mLastUpdateRequestTime = + mDictNameDictionaryUpdateController.mLastUpdateTime; } } else if (mBinaryDictionary == null || mPerInstanceDictionaryUpdateController.mLastUpdateTime - < mFilenameDictionaryUpdateController.mLastUpdateTime) { + < mDictNameDictionaryUpdateController.mLastUpdateTime) { // Otherwise, if the local dictionary is older than the shared dictionary, // load the shared dictionary. loadBinaryDictionary(); } - if (mBinaryDictionary != null && !mBinaryDictionary.isValidDictionary()) { - // Binary dictionary is not valid. Regenerate the dictionary file. - mFilenameDictionaryUpdateController.mLastUpdateTime = time; - writeBinaryDictionary(); - loadBinaryDictionary(); - } - mPerInstanceDictionaryUpdateController.mLastUpdateTime = time; + // If we just loaded the binary dictionary, then mBinaryDictionary is not + // up-to-date yet so it's useless to test it right away. Schedule the check + // for right after it's loaded instead. + getExecutor(mDictName).executePrioritized(new Runnable() { + @Override + public void run() { + if (mBinaryDictionary != null && !(isValidDictionary() + // TODO: remove the check below + && matchesExpectedBinaryDictFormatVersionForThisType( + mBinaryDictionary.getFormatVersion()))) { + // Binary dictionary or its format version is not valid. Regenerate + // the dictionary file. writeBinaryDictionary will remove the + // existing files if appropriate. + mDictNameDictionaryUpdateController.mLastUpdateTime = time; + writeBinaryDictionary(); + loadBinaryDictionary(); + } + mPerInstanceDictionaryUpdateController.mLastUpdateTime = time; + } + }); } finally { - mFilenameDictionaryUpdateController.mIsRegenerating.set(false); + mDictNameDictionaryUpdateController.mProcessingLargeTask.set(false); } } }); @@ -658,28 +716,13 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { // TODO: cache the file's existence so that we avoid doing a disk access each time. private boolean dictionaryFileExists() { - final File file = new File(mContext.getFilesDir(), mFilename); - return file.exists(); - } - - /** - * Load the dictionary to memory. - */ - protected void asyncLoadDictionaryToMemory() { - getExecutor(mFilename).executePrioritized(new Runnable() { - @Override - public void run() { - if (!ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE) { - loadDictionaryAsync(); - } - } - }); + return getDictFile().exists(); } /** * Generate binary dictionary using DictionaryWriter. */ - protected void asyncFlashAllBinaryDictionary() { + protected void asyncFlushBinaryDictionary() { final Runnable newTask = new Runnable() { @Override public void run() { @@ -687,37 +730,32 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { } }; final Runnable oldTask = mUnfinishedFlushingTask.getAndSet(newTask); - getExecutor(mFilename).replaceAndExecute(oldTask, newTask); + getExecutor(mDictName).replaceAndExecute(oldTask, newTask); } /** - * For tracking whether the dictionary is out of date and the dictionary is regenerating. - * Can be shared across multiple dictionary instances that access the same filename. + * For tracking whether the dictionary is out of date and the dictionary is used in a large + * task. Can be shared across multiple dictionary instances that access the same filename. */ private static class DictionaryUpdateController { public volatile long mLastUpdateTime = 0; public volatile long mLastUpdateRequestTime = 0; - public volatile AtomicBoolean mIsRegenerating = new AtomicBoolean(); + public volatile AtomicBoolean mProcessingLargeTask = new AtomicBoolean(); public boolean isOutOfDate() { return (mLastUpdateRequestTime > mLastUpdateTime); } } - // TODO: Implement native binary methods once the dynamic dictionary implementation is done. + // TODO: Implement BinaryDictionary.isInDictionary(). @UsedForTesting public boolean isInDictionaryForTests(final String word) { final AsyncResultHolder<Boolean> holder = new AsyncResultHolder<Boolean>(); - getExecutor(mFilename).executePrioritized(new Runnable() { + getExecutor(mDictName).executePrioritized(new Runnable() { @Override public void run() { if (mDictType == Dictionary.TYPE_USER_HISTORY) { - if (ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE) { - holder.set(mBinaryDictionary.isValidWord(word)); - } else { - holder.set(((DynamicPersonalizationDictionaryWriter) mDictionaryWriter) - .isInBigramListForTests(word)); - } + holder.set(mBinaryDictionary.isValidWord(word)); } } }); @@ -725,12 +763,33 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { } @UsedForTesting - public void shutdownExecutorForTests() { - getExecutor(mFilename).shutdown(); + public void waitAllTasksForTests() { + final CountDownLatch countDownLatch = new CountDownLatch(1); + getExecutor(mDictName).execute(new Runnable() { + @Override + public void run() { + countDownLatch.countDown(); + } + }); + try { + countDownLatch.await(); + } catch (InterruptedException e) { + Log.e(TAG, "Interrupted while waiting for finishing dictionary operations.", e); + } } @UsedForTesting - public boolean isTerminatedForTests() { - return getExecutor(mFilename).isTerminated(); + protected void runAfterGcForDebug(final Runnable r) { + getExecutor(mDictName).executePrioritized(new Runnable() { + @Override + public void run() { + try { + mBinaryDictionary.flushWithGC(); + r.run(); + } finally { + mDictNameDictionaryUpdateController.mProcessingLargeTask.set(false); + } + } + }); } } diff --git a/java/src/com/android/inputmethod/latin/ExpandableDictionary.java b/java/src/com/android/inputmethod/latin/ExpandableDictionary.java deleted file mode 100644 index 95c9bcab9..000000000 --- a/java/src/com/android/inputmethod/latin/ExpandableDictionary.java +++ /dev/null @@ -1,894 +0,0 @@ -/* - * Copyright (C) 2009 The Android Open Source Project - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -package com.android.inputmethod.latin; - -import android.text.TextUtils; -import android.util.Log; - -import com.android.inputmethod.annotations.UsedForTesting; -import com.android.inputmethod.keyboard.ProximityInfo; -import com.android.inputmethod.latin.SuggestedWords.SuggestedWordInfo; -import com.android.inputmethod.latin.utils.CollectionUtils; -import com.android.inputmethod.latin.utils.UserHistoryForgettingCurveUtils.ForgettingCurveParams; - -import java.util.ArrayList; -import java.util.LinkedList; - -/** - * Class for an in-memory dictionary that can grow dynamically and can - * be searched for suggestions and valid words. - */ -// TODO: Remove after binary dictionary supports dynamic update. -public class ExpandableDictionary extends Dictionary { - private static final String TAG = ExpandableDictionary.class.getSimpleName(); - /** - * The weight to give to a word if it's length is the same as the number of typed characters. - */ - private static final int FULL_WORD_SCORE_MULTIPLIER = 2; - - private char[] mWordBuilder = new char[Constants.DICTIONARY_MAX_WORD_LENGTH]; - private int mMaxDepth; - private int mInputLength; - - private static final class Node { - char mCode; - int mFrequency; - boolean mTerminal; - Node mParent; - NodeArray mChildren; - ArrayList<char[]> mShortcutTargets; - boolean mShortcutOnly; - LinkedList<NextWord> mNGrams; // Supports ngram - } - - private static final class NodeArray { - Node[] mData; - int mLength = 0; - private static final int INCREMENT = 2; - - NodeArray() { - mData = new Node[INCREMENT]; - } - - void add(final Node n) { - if (mLength + 1 > mData.length) { - Node[] tempData = new Node[mLength + INCREMENT]; - if (mLength > 0) { - System.arraycopy(mData, 0, tempData, 0, mLength); - } - mData = tempData; - } - mData[mLength++] = n; - } - } - - public interface NextWord { - public Node getWordNode(); - public int getFrequency(); - public ForgettingCurveParams getFcParams(); - public int notifyTypedAgainAndGetFrequency(); - } - - private static final class NextStaticWord implements NextWord { - public final Node mWord; - private final int mFrequency; - public NextStaticWord(Node word, int frequency) { - mWord = word; - mFrequency = frequency; - } - - @Override - public Node getWordNode() { - return mWord; - } - - @Override - public int getFrequency() { - return mFrequency; - } - - @Override - public ForgettingCurveParams getFcParams() { - return null; - } - - @Override - public int notifyTypedAgainAndGetFrequency() { - return mFrequency; - } - } - - private static final class NextHistoryWord implements NextWord { - public final Node mWord; - public final ForgettingCurveParams mFcp; - - public NextHistoryWord(Node word, ForgettingCurveParams fcp) { - mWord = word; - mFcp = fcp; - } - - @Override - public Node getWordNode() { - return mWord; - } - - @Override - public int getFrequency() { - return mFcp.getFrequency(); - } - - @Override - public ForgettingCurveParams getFcParams() { - return mFcp; - } - - @Override - public int notifyTypedAgainAndGetFrequency() { - return mFcp.notifyTypedAgainAndGetFrequency(); - } - } - - private NodeArray mRoots; - - private int[][] mCodes; - - public ExpandableDictionary(final String dictType) { - super(dictType); - clearDictionary(); - mCodes = new int[Constants.DICTIONARY_MAX_WORD_LENGTH][]; - } - - public int getMaxWordLength() { - return Constants.DICTIONARY_MAX_WORD_LENGTH; - } - - /** - * Add a word with an optional shortcut to the dictionary. - * @param word The word to add. - * @param shortcutTarget A shortcut target for this word, or null if none. - * @param frequency The frequency for this unigram. - * @param shortcutFreq The frequency of the shortcut (0~15, with 15 = whitelist). Ignored - * if shortcutTarget is null. - */ - public void addWord(final String word, final String shortcutTarget, final int frequency, - final int shortcutFreq) { - if (word.length() >= Constants.DICTIONARY_MAX_WORD_LENGTH) { - return; - } - addWordRec(mRoots, word, 0, shortcutTarget, frequency, shortcutFreq, null); - } - - /** - * Add a word, recursively searching for its correct place in the trie tree. - * @param children The node to recursively search for addition. Initially, the root of the tree. - * @param word The word to add. - * @param depth The current depth in the tree. - * @param shortcutTarget A shortcut target for this word, or null if none. - * @param frequency The frequency for this unigram. - * @param shortcutFreq The frequency of the shortcut (0~15, with 15 = whitelist). Ignored - * if shortcutTarget is null. - * @param parentNode The parent node, for up linking. Initially null, as the root has no parent. - */ - private void addWordRec(final NodeArray children, final String word, final int depth, - final String shortcutTarget, final int frequency, final int shortcutFreq, - final Node parentNode) { - final int wordLength = word.length(); - if (wordLength <= depth) return; - final char c = word.charAt(depth); - // Does children have the current character? - final int childrenLength = children.mLength; - Node childNode = null; - for (int i = 0; i < childrenLength; i++) { - final Node node = children.mData[i]; - if (node.mCode == c) { - childNode = node; - break; - } - } - final boolean isShortcutOnly = (null != shortcutTarget); - if (childNode == null) { - childNode = new Node(); - childNode.mCode = c; - childNode.mParent = parentNode; - childNode.mShortcutOnly = isShortcutOnly; - children.add(childNode); - } - if (wordLength == depth + 1) { - // Terminate this word - childNode.mTerminal = true; - if (isShortcutOnly) { - if (null == childNode.mShortcutTargets) { - childNode.mShortcutTargets = CollectionUtils.newArrayList(); - } - childNode.mShortcutTargets.add(shortcutTarget.toCharArray()); - } else { - childNode.mShortcutOnly = false; - } - childNode.mFrequency = Math.max(frequency, childNode.mFrequency); - if (childNode.mFrequency > 255) childNode.mFrequency = 255; - return; - } - if (childNode.mChildren == null) { - childNode.mChildren = new NodeArray(); - } - addWordRec(childNode.mChildren, word, depth + 1, shortcutTarget, frequency, shortcutFreq, - childNode); - } - - @Override - public ArrayList<SuggestedWordInfo> getSuggestions(final WordComposer composer, - final String prevWord, final ProximityInfo proximityInfo, - final boolean blockOffensiveWords, final int[] additionalFeaturesOptions) { - if (composer.size() > 1) { - if (composer.size() >= Constants.DICTIONARY_MAX_WORD_LENGTH) { - return null; - } - final ArrayList<SuggestedWordInfo> suggestions = - getWordsInner(composer, prevWord, proximityInfo); - return suggestions; - } else { - if (TextUtils.isEmpty(prevWord)) return null; - final ArrayList<SuggestedWordInfo> suggestions = CollectionUtils.newArrayList(); - runBigramReverseLookUp(prevWord, suggestions); - return suggestions; - } - } - - private ArrayList<SuggestedWordInfo> getWordsInner(final WordComposer codes, - final String prevWordForBigrams, final ProximityInfo proximityInfo) { - final ArrayList<SuggestedWordInfo> suggestions = CollectionUtils.newArrayList(); - mInputLength = codes.size(); - if (mCodes.length < mInputLength) mCodes = new int[mInputLength][]; - final InputPointers ips = codes.getInputPointers(); - final int[] xCoordinates = ips.getXCoordinates(); - final int[] yCoordinates = ips.getYCoordinates(); - // Cache the codes so that we don't have to lookup an array list - for (int i = 0; i < mInputLength; i++) { - // TODO: Calculate proximity info here. - if (mCodes[i] == null || mCodes[i].length < 1) { - mCodes[i] = new int[ProximityInfo.MAX_PROXIMITY_CHARS_SIZE]; - } - final int x = xCoordinates != null && i < xCoordinates.length ? - xCoordinates[i] : Constants.NOT_A_COORDINATE; - final int y = xCoordinates != null && i < yCoordinates.length ? - yCoordinates[i] : Constants.NOT_A_COORDINATE; - proximityInfo.fillArrayWithNearestKeyCodes(x, y, codes.getCodeAt(i), mCodes[i]); - } - mMaxDepth = mInputLength * 3; - getWordsRec(mRoots, codes, mWordBuilder, 0, false, 1, 0, -1, suggestions); - for (int i = 0; i < mInputLength; i++) { - getWordsRec(mRoots, codes, mWordBuilder, 0, false, 1, 0, i, suggestions); - } - return suggestions; - } - - @Override - public synchronized boolean isValidWord(final String word) { - final Node node = searchNode(mRoots, word, 0, word.length()); - // If node is null, we didn't find the word, so it's not valid. - // If node.mShortcutOnly is true, then it exists as a shortcut but not as a word, - // so that means it's not a valid word. - // If node.mShortcutOnly is false, then it exists as a word (it may also exist as - // a shortcut, but this does not matter), so it's a valid word. - return (node == null) ? false : !node.mShortcutOnly; - } - - public boolean removeBigram(final String word0, final String word1) { - // Refer to addOrSetBigram() about word1.toLowerCase() - final Node firstWord = searchWord(mRoots, word0.toLowerCase(), 0, null); - final Node secondWord = searchWord(mRoots, word1, 0, null); - LinkedList<NextWord> bigrams = firstWord.mNGrams; - NextWord bigramNode = null; - if (bigrams == null || bigrams.size() == 0) { - return false; - } else { - for (NextWord nw : bigrams) { - if (nw.getWordNode() == secondWord) { - bigramNode = nw; - break; - } - } - } - if (bigramNode == null) { - return false; - } - return bigrams.remove(bigramNode); - } - - /** - * Returns the word's frequency or -1 if not found - */ - @UsedForTesting - public int getWordFrequency(final String word) { - // Case-sensitive search - final Node node = searchNode(mRoots, word, 0, word.length()); - return (node == null) ? -1 : node.mFrequency; - } - - public NextWord getBigramWord(final String word0, final String word1) { - // Refer to addOrSetBigram() about word0.toLowerCase() - final Node firstWord = searchWord(mRoots, word0.toLowerCase(), 0, null); - final Node secondWord = searchWord(mRoots, word1, 0, null); - LinkedList<NextWord> bigrams = firstWord.mNGrams; - if (bigrams == null || bigrams.size() == 0) { - return null; - } else { - for (NextWord nw : bigrams) { - if (nw.getWordNode() == secondWord) { - return nw; - } - } - } - return null; - } - - private static int computeSkippedWordFinalFreq(final int freq, final int snr, - final int inputLength) { - // The computation itself makes sense for >= 2, but the == 2 case returns 0 - // anyway so we may as well test against 3 instead and return the constant - if (inputLength >= 3) { - return (freq * snr * (inputLength - 2)) / (inputLength - 1); - } else { - return 0; - } - } - - /** - * Helper method to add a word and its shortcuts. - * - * @param node the terminal node - * @param word the word to insert, as an array of code points - * @param depth the depth of the node in the tree - * @param finalFreq the frequency for this word - * @param suggestions the suggestion collection to add the suggestions to - * @return whether there is still space for more words. - */ - private boolean addWordAndShortcutsFromNode(final Node node, final char[] word, final int depth, - final int finalFreq, final ArrayList<SuggestedWordInfo> suggestions) { - if (finalFreq > 0 && !node.mShortcutOnly) { - // Use KIND_CORRECTION always. This dictionary does not really have a notion of - // COMPLETION against CORRECTION; we could artificially add one by looking at - // the respective size of the typed word and the suggestion if it matters sometime - // in the future. - suggestions.add(new SuggestedWordInfo(new String(word, 0, depth + 1), finalFreq, - SuggestedWordInfo.KIND_CORRECTION, this /* sourceDict */, - SuggestedWordInfo.NOT_AN_INDEX /* indexOfTouchPointOfSecondWord */, - SuggestedWordInfo.NOT_A_CONFIDENCE /* autoCommitFirstWordConfidence */)); - if (suggestions.size() >= Suggest.MAX_SUGGESTIONS) return false; - } - if (null != node.mShortcutTargets) { - final int length = node.mShortcutTargets.size(); - for (int shortcutIndex = 0; shortcutIndex < length; ++shortcutIndex) { - final char[] shortcut = node.mShortcutTargets.get(shortcutIndex); - suggestions.add(new SuggestedWordInfo(new String(shortcut, 0, shortcut.length), - finalFreq, SuggestedWordInfo.KIND_SHORTCUT, this /* sourceDict */, - SuggestedWordInfo.NOT_AN_INDEX /* indexOfTouchPointOfSecondWord */, - SuggestedWordInfo.NOT_A_CONFIDENCE /* autoCommitFirstWordConfidence */)); - if (suggestions.size() > Suggest.MAX_SUGGESTIONS) return false; - } - } - return true; - } - - /** - * Recursively traverse the tree for words that match the input. Input consists of - * a list of arrays. Each item in the list is one input character position. An input - * character is actually an array of multiple possible candidates. This function is not - * optimized for speed, assuming that the user dictionary will only be a few hundred words in - * size. - * @param roots node whose children have to be search for matches - * @param codes the input character codes - * @param word the word being composed as a possible match - * @param depth the depth of traversal - the length of the word being composed thus far - * @param completion whether the traversal is now in completion mode - meaning that we've - * exhausted the input and we're looking for all possible suffixes. - * @param snr current weight of the word being formed - * @param inputIndex position in the input characters. This can be off from the depth in - * case we skip over some punctuations such as apostrophe in the traversal. That is, if you type - * "wouldve", it could be matching "would've", so the depth will be one more than the - * inputIndex - * @param suggestions the list in which to add suggestions - */ - // TODO: Share this routine with the native code for BinaryDictionary - private void getWordsRec(final NodeArray roots, final WordComposer codes, final char[] word, - final int depth, final boolean completion, final int snr, final int inputIndex, - final int skipPos, final ArrayList<SuggestedWordInfo> suggestions) { - final int count = roots.mLength; - final int codeSize = mInputLength; - // Optimization: Prune out words that are too long compared to how much was typed. - if (depth > mMaxDepth) { - return; - } - final int[] currentChars; - if (codeSize <= inputIndex) { - currentChars = null; - } else { - currentChars = mCodes[inputIndex]; - } - - for (int i = 0; i < count; i++) { - final Node node = roots.mData[i]; - final char c = node.mCode; - final char lowerC = toLowerCase(c); - final boolean terminal = node.mTerminal; - final NodeArray children = node.mChildren; - final int freq = node.mFrequency; - if (completion || currentChars == null) { - word[depth] = c; - if (terminal) { - final int finalFreq; - if (skipPos < 0) { - finalFreq = freq * snr; - } else { - finalFreq = computeSkippedWordFinalFreq(freq, snr, mInputLength); - } - if (!addWordAndShortcutsFromNode(node, word, depth, finalFreq, suggestions)) { - // No space left in the queue, bail out - return; - } - } - if (children != null) { - getWordsRec(children, codes, word, depth + 1, true, snr, inputIndex, - skipPos, suggestions); - } - } else if ((c == Constants.CODE_SINGLE_QUOTE - && currentChars[0] != Constants.CODE_SINGLE_QUOTE) || depth == skipPos) { - // Skip the ' and continue deeper - word[depth] = c; - if (children != null) { - getWordsRec(children, codes, word, depth + 1, completion, snr, inputIndex, - skipPos, suggestions); - } - } else { - // Don't use alternatives if we're looking for missing characters - final int alternativesSize = skipPos >= 0 ? 1 : currentChars.length; - for (int j = 0; j < alternativesSize; j++) { - final int addedAttenuation = (j > 0 ? 1 : 2); - final int currentChar = currentChars[j]; - if (currentChar == Constants.NOT_A_CODE) { - break; - } - if (currentChar == lowerC || currentChar == c) { - word[depth] = c; - - if (codeSize == inputIndex + 1) { - if (terminal) { - final int finalFreq; - if (skipPos < 0) { - finalFreq = freq * snr * addedAttenuation - * FULL_WORD_SCORE_MULTIPLIER; - } else { - finalFreq = computeSkippedWordFinalFreq(freq, - snr * addedAttenuation, mInputLength); - } - if (!addWordAndShortcutsFromNode(node, word, depth, finalFreq, - suggestions)) { - // No space left in the queue, bail out - return; - } - } - if (children != null) { - getWordsRec(children, codes, word, depth + 1, - true, snr * addedAttenuation, inputIndex + 1, - skipPos, suggestions); - } - } else if (children != null) { - getWordsRec(children, codes, word, depth + 1, - false, snr * addedAttenuation, inputIndex + 1, - skipPos, suggestions); - } - } - } - } - } - } - - public int setBigramAndGetFrequency(final String word0, final String word1, - final int frequency) { - return setBigramAndGetFrequency(word0, word1, frequency, null /* unused */); - } - - public int setBigramAndGetFrequency(final String word0, final String word1, - final ForgettingCurveParams fcp) { - return setBigramAndGetFrequency(word0, word1, 0 /* unused */, fcp); - } - - /** - * Adds bigrams to the in-memory trie structure that is being used to retrieve any word - * @param word0 the first word of this bigram - * @param word1 the second word of this bigram - * @param frequency frequency for this bigram - * @param fcp an instance of ForgettingCurveParams to use for decay policy - * @return returns the final bigram frequency - */ - private int setBigramAndGetFrequency(final String word0, final String word1, - final int frequency, final ForgettingCurveParams fcp) { - if (TextUtils.isEmpty(word0)) { - Log.e(TAG, "Invalid bigram previous word: " + word0); - return frequency; - } - // We don't want results to be different according to case of the looked up left hand side - // word. We do want however to return the correct case for the right hand side. - // So we want to squash the case of the left hand side, and preserve that of the right - // hand side word. - final String word0Lower = word0.toLowerCase(); - if (TextUtils.isEmpty(word0Lower) || TextUtils.isEmpty(word1)) { - Log.e(TAG, "Invalid bigram pair: " + word0 + ", " + word0Lower + ", " + word1); - return frequency; - } - final Node firstWord = searchWord(mRoots, word0Lower, 0, null); - final Node secondWord = searchWord(mRoots, word1, 0, null); - LinkedList<NextWord> bigrams = firstWord.mNGrams; - if (bigrams == null || bigrams.size() == 0) { - firstWord.mNGrams = CollectionUtils.newLinkedList(); - bigrams = firstWord.mNGrams; - } else { - for (NextWord nw : bigrams) { - if (nw.getWordNode() == secondWord) { - return nw.notifyTypedAgainAndGetFrequency(); - } - } - } - if (fcp != null) { - // history - firstWord.mNGrams.add(new NextHistoryWord(secondWord, fcp)); - } else { - firstWord.mNGrams.add(new NextStaticWord(secondWord, frequency)); - } - return frequency; - } - - /** - * Searches for the word and add the word if it does not exist. - * @return Returns the terminal node of the word we are searching for. - */ - private Node searchWord(final NodeArray children, final String word, final int depth, - final Node parentNode) { - final int wordLength = word.length(); - final char c = word.charAt(depth); - // Does children have the current character? - final int childrenLength = children.mLength; - Node childNode = null; - for (int i = 0; i < childrenLength; i++) { - final Node node = children.mData[i]; - if (node.mCode == c) { - childNode = node; - break; - } - } - if (childNode == null) { - childNode = new Node(); - childNode.mCode = c; - childNode.mParent = parentNode; - children.add(childNode); - } - if (wordLength == depth + 1) { - // Terminate this word - childNode.mTerminal = true; - return childNode; - } - if (childNode.mChildren == null) { - childNode.mChildren = new NodeArray(); - } - return searchWord(childNode.mChildren, word, depth + 1, childNode); - } - - private void runBigramReverseLookUp(final String previousWord, - final ArrayList<SuggestedWordInfo> suggestions) { - // Search for the lowercase version of the word only, because that's where bigrams - // store their sons. - final Node prevWord = searchNode(mRoots, previousWord.toLowerCase(), 0, - previousWord.length()); - if (prevWord != null && prevWord.mNGrams != null) { - reverseLookUp(prevWord.mNGrams, suggestions); - } - } - - // Local to reverseLookUp, but do not allocate each time. - private final char[] mLookedUpString = new char[Constants.DICTIONARY_MAX_WORD_LENGTH]; - - /** - * reverseLookUp retrieves the full word given a list of terminal nodes and adds those words - * to the suggestions list passed as an argument. - * @param terminalNodes list of terminal nodes we want to add - * @param suggestions the suggestion collection to add the word to - */ - private void reverseLookUp(final LinkedList<NextWord> terminalNodes, - final ArrayList<SuggestedWordInfo> suggestions) { - Node node; - int freq; - for (NextWord nextWord : terminalNodes) { - node = nextWord.getWordNode(); - freq = nextWord.getFrequency(); - int index = Constants.DICTIONARY_MAX_WORD_LENGTH; - do { - --index; - mLookedUpString[index] = node.mCode; - node = node.mParent; - } while (node != null && index > 0); - - // If node is null, we have a word longer than MAX_WORD_LENGTH in the dictionary. - // It's a little unclear how this can happen, but just in case it does it's safer - // to ignore the word in this case. - if (freq >= 0 && node == null) { - suggestions.add(new SuggestedWordInfo(new String(mLookedUpString, index, - Constants.DICTIONARY_MAX_WORD_LENGTH - index), - freq, SuggestedWordInfo.KIND_CORRECTION, this /* sourceDict */, - SuggestedWordInfo.NOT_AN_INDEX /* indexOfTouchPointOfSecondWord */, - SuggestedWordInfo.NOT_A_CONFIDENCE /* autoCommitFirstWordConfidence */)); - } - } - } - - /** - * Recursively search for the terminal node of the word. - * - * One iteration takes the full word to search for and the current index of the recursion. - * - * @param children the node of the trie to search under. - * @param word the word to search for. Only read [offset..length] so there may be trailing chars - * @param offset the index in {@code word} this recursion should operate on. - * @param length the length of the input word. - * @return Returns the terminal node of the word if the word exists - */ - private Node searchNode(final NodeArray children, final CharSequence word, final int offset, - final int length) { - final int count = children.mLength; - final char currentChar = word.charAt(offset); - for (int j = 0; j < count; j++) { - final Node node = children.mData[j]; - if (node.mCode == currentChar) { - if (offset == length - 1) { - if (node.mTerminal) { - return node; - } - } else { - if (node.mChildren != null) { - Node returnNode = searchNode(node.mChildren, word, offset + 1, length); - if (returnNode != null) return returnNode; - } - } - } - } - return null; - } - - public void clearDictionary() { - mRoots = new NodeArray(); - } - - private static char toLowerCase(final char c) { - char baseChar = c; - if (c < BASE_CHARS.length) { - baseChar = BASE_CHARS[c]; - } - if (baseChar >= 'A' && baseChar <= 'Z') { - return (char)(baseChar | 32); - } else if (baseChar > 127) { - return Character.toLowerCase(baseChar); - } - return baseChar; - } - - /** - * Table mapping most combined Latin, Greek, and Cyrillic characters - * to their base characters. If c is in range, BASE_CHARS[c] == c - * if c is not a combined character, or the base character if it - * is combined. - * - * cf. native/jni/src/utils/char_utils.cpp - */ - private static final char BASE_CHARS[] = { - /* U+0000 */ 0x0000, 0x0001, 0x0002, 0x0003, 0x0004, 0x0005, 0x0006, 0x0007, - /* U+0008 */ 0x0008, 0x0009, 0x000A, 0x000B, 0x000C, 0x000D, 0x000E, 0x000F, - /* U+0010 */ 0x0010, 0x0011, 0x0012, 0x0013, 0x0014, 0x0015, 0x0016, 0x0017, - /* U+0018 */ 0x0018, 0x0019, 0x001A, 0x001B, 0x001C, 0x001D, 0x001E, 0x001F, - /* U+0020 */ 0x0020, 0x0021, 0x0022, 0x0023, 0x0024, 0x0025, 0x0026, 0x0027, - /* U+0028 */ 0x0028, 0x0029, 0x002A, 0x002B, 0x002C, 0x002D, 0x002E, 0x002F, - /* U+0030 */ 0x0030, 0x0031, 0x0032, 0x0033, 0x0034, 0x0035, 0x0036, 0x0037, - /* U+0038 */ 0x0038, 0x0039, 0x003A, 0x003B, 0x003C, 0x003D, 0x003E, 0x003F, - /* U+0040 */ 0x0040, 0x0041, 0x0042, 0x0043, 0x0044, 0x0045, 0x0046, 0x0047, - /* U+0048 */ 0x0048, 0x0049, 0x004A, 0x004B, 0x004C, 0x004D, 0x004E, 0x004F, - /* U+0050 */ 0x0050, 0x0051, 0x0052, 0x0053, 0x0054, 0x0055, 0x0056, 0x0057, - /* U+0058 */ 0x0058, 0x0059, 0x005A, 0x005B, 0x005C, 0x005D, 0x005E, 0x005F, - /* U+0060 */ 0x0060, 0x0061, 0x0062, 0x0063, 0x0064, 0x0065, 0x0066, 0x0067, - /* U+0068 */ 0x0068, 0x0069, 0x006A, 0x006B, 0x006C, 0x006D, 0x006E, 0x006F, - /* U+0070 */ 0x0070, 0x0071, 0x0072, 0x0073, 0x0074, 0x0075, 0x0076, 0x0077, - /* U+0078 */ 0x0078, 0x0079, 0x007A, 0x007B, 0x007C, 0x007D, 0x007E, 0x007F, - /* U+0080 */ 0x0080, 0x0081, 0x0082, 0x0083, 0x0084, 0x0085, 0x0086, 0x0087, - /* U+0088 */ 0x0088, 0x0089, 0x008A, 0x008B, 0x008C, 0x008D, 0x008E, 0x008F, - /* U+0090 */ 0x0090, 0x0091, 0x0092, 0x0093, 0x0094, 0x0095, 0x0096, 0x0097, - /* U+0098 */ 0x0098, 0x0099, 0x009A, 0x009B, 0x009C, 0x009D, 0x009E, 0x009F, - /* U+00A0 */ 0x0020, 0x00A1, 0x00A2, 0x00A3, 0x00A4, 0x00A5, 0x00A6, 0x00A7, - /* U+00A8 */ 0x0020, 0x00A9, 0x0061, 0x00AB, 0x00AC, 0x00AD, 0x00AE, 0x0020, - /* U+00B0 */ 0x00B0, 0x00B1, 0x0032, 0x0033, 0x0020, 0x03BC, 0x00B6, 0x00B7, - /* U+00B8 */ 0x0020, 0x0031, 0x006F, 0x00BB, 0x0031, 0x0031, 0x0033, 0x00BF, - /* U+00C0 */ 0x0041, 0x0041, 0x0041, 0x0041, 0x0041, 0x0041, 0x00C6, 0x0043, - /* U+00C8 */ 0x0045, 0x0045, 0x0045, 0x0045, 0x0049, 0x0049, 0x0049, 0x0049, - /* U+00D0 */ 0x00D0, 0x004E, 0x004F, 0x004F, 0x004F, 0x004F, 0x004F, 0x00D7, - /* U+00D8 */ 0x004F, 0x0055, 0x0055, 0x0055, 0x0055, 0x0059, 0x00DE, 0x0073, - // U+00D8: Manually changed from 00D8 to 004F - // TODO: Check if it's really acceptable to consider Ø a diacritical variant of O - // U+00DF: Manually changed from 00DF to 0073 - /* U+00E0 */ 0x0061, 0x0061, 0x0061, 0x0061, 0x0061, 0x0061, 0x00E6, 0x0063, - /* U+00E8 */ 0x0065, 0x0065, 0x0065, 0x0065, 0x0069, 0x0069, 0x0069, 0x0069, - /* U+00F0 */ 0x00F0, 0x006E, 0x006F, 0x006F, 0x006F, 0x006F, 0x006F, 0x00F7, - /* U+00F8 */ 0x006F, 0x0075, 0x0075, 0x0075, 0x0075, 0x0079, 0x00FE, 0x0079, - // U+00F8: Manually changed from 00F8 to 006F - // TODO: Check if it's really acceptable to consider ø a diacritical variant of o - /* U+0100 */ 0x0041, 0x0061, 0x0041, 0x0061, 0x0041, 0x0061, 0x0043, 0x0063, - /* U+0108 */ 0x0043, 0x0063, 0x0043, 0x0063, 0x0043, 0x0063, 0x0044, 0x0064, - /* U+0110 */ 0x0110, 0x0111, 0x0045, 0x0065, 0x0045, 0x0065, 0x0045, 0x0065, - /* U+0118 */ 0x0045, 0x0065, 0x0045, 0x0065, 0x0047, 0x0067, 0x0047, 0x0067, - /* U+0120 */ 0x0047, 0x0067, 0x0047, 0x0067, 0x0048, 0x0068, 0x0126, 0x0127, - /* U+0128 */ 0x0049, 0x0069, 0x0049, 0x0069, 0x0049, 0x0069, 0x0049, 0x0069, - /* U+0130 */ 0x0049, 0x0131, 0x0049, 0x0069, 0x004A, 0x006A, 0x004B, 0x006B, - /* U+0138 */ 0x0138, 0x004C, 0x006C, 0x004C, 0x006C, 0x004C, 0x006C, 0x004C, - /* U+0140 */ 0x006C, 0x004C, 0x006C, 0x004E, 0x006E, 0x004E, 0x006E, 0x004E, - // U+0141: Manually changed from 0141 to 004C - // U+0142: Manually changed from 0142 to 006C - /* U+0148 */ 0x006E, 0x02BC, 0x014A, 0x014B, 0x004F, 0x006F, 0x004F, 0x006F, - /* U+0150 */ 0x004F, 0x006F, 0x0152, 0x0153, 0x0052, 0x0072, 0x0052, 0x0072, - /* U+0158 */ 0x0052, 0x0072, 0x0053, 0x0073, 0x0053, 0x0073, 0x0053, 0x0073, - /* U+0160 */ 0x0053, 0x0073, 0x0054, 0x0074, 0x0054, 0x0074, 0x0166, 0x0167, - /* U+0168 */ 0x0055, 0x0075, 0x0055, 0x0075, 0x0055, 0x0075, 0x0055, 0x0075, - /* U+0170 */ 0x0055, 0x0075, 0x0055, 0x0075, 0x0057, 0x0077, 0x0059, 0x0079, - /* U+0178 */ 0x0059, 0x005A, 0x007A, 0x005A, 0x007A, 0x005A, 0x007A, 0x0073, - /* U+0180 */ 0x0180, 0x0181, 0x0182, 0x0183, 0x0184, 0x0185, 0x0186, 0x0187, - /* U+0188 */ 0x0188, 0x0189, 0x018A, 0x018B, 0x018C, 0x018D, 0x018E, 0x018F, - /* U+0190 */ 0x0190, 0x0191, 0x0192, 0x0193, 0x0194, 0x0195, 0x0196, 0x0197, - /* U+0198 */ 0x0198, 0x0199, 0x019A, 0x019B, 0x019C, 0x019D, 0x019E, 0x019F, - /* U+01A0 */ 0x004F, 0x006F, 0x01A2, 0x01A3, 0x01A4, 0x01A5, 0x01A6, 0x01A7, - /* U+01A8 */ 0x01A8, 0x01A9, 0x01AA, 0x01AB, 0x01AC, 0x01AD, 0x01AE, 0x0055, - /* U+01B0 */ 0x0075, 0x01B1, 0x01B2, 0x01B3, 0x01B4, 0x01B5, 0x01B6, 0x01B7, - /* U+01B8 */ 0x01B8, 0x01B9, 0x01BA, 0x01BB, 0x01BC, 0x01BD, 0x01BE, 0x01BF, - /* U+01C0 */ 0x01C0, 0x01C1, 0x01C2, 0x01C3, 0x0044, 0x0044, 0x0064, 0x004C, - /* U+01C8 */ 0x004C, 0x006C, 0x004E, 0x004E, 0x006E, 0x0041, 0x0061, 0x0049, - /* U+01D0 */ 0x0069, 0x004F, 0x006F, 0x0055, 0x0075, 0x0055, 0x0075, 0x0055, - // U+01D5: Manually changed from 00DC to 0055 - // U+01D6: Manually changed from 00FC to 0075 - // U+01D7: Manually changed from 00DC to 0055 - /* U+01D8 */ 0x0075, 0x0055, 0x0075, 0x0055, 0x0075, 0x01DD, 0x0041, 0x0061, - // U+01D8: Manually changed from 00FC to 0075 - // U+01D9: Manually changed from 00DC to 0055 - // U+01DA: Manually changed from 00FC to 0075 - // U+01DB: Manually changed from 00DC to 0055 - // U+01DC: Manually changed from 00FC to 0075 - // U+01DE: Manually changed from 00C4 to 0041 - // U+01DF: Manually changed from 00E4 to 0061 - /* U+01E0 */ 0x0041, 0x0061, 0x00C6, 0x00E6, 0x01E4, 0x01E5, 0x0047, 0x0067, - // U+01E0: Manually changed from 0226 to 0041 - // U+01E1: Manually changed from 0227 to 0061 - /* U+01E8 */ 0x004B, 0x006B, 0x004F, 0x006F, 0x004F, 0x006F, 0x01B7, 0x0292, - // U+01EC: Manually changed from 01EA to 004F - // U+01ED: Manually changed from 01EB to 006F - /* U+01F0 */ 0x006A, 0x0044, 0x0044, 0x0064, 0x0047, 0x0067, 0x01F6, 0x01F7, - /* U+01F8 */ 0x004E, 0x006E, 0x0041, 0x0061, 0x00C6, 0x00E6, 0x004F, 0x006F, - // U+01FA: Manually changed from 00C5 to 0041 - // U+01FB: Manually changed from 00E5 to 0061 - // U+01FE: Manually changed from 00D8 to 004F - // TODO: Check if it's really acceptable to consider Ø a diacritical variant of O - // U+01FF: Manually changed from 00F8 to 006F - // TODO: Check if it's really acceptable to consider ø a diacritical variant of o - /* U+0200 */ 0x0041, 0x0061, 0x0041, 0x0061, 0x0045, 0x0065, 0x0045, 0x0065, - /* U+0208 */ 0x0049, 0x0069, 0x0049, 0x0069, 0x004F, 0x006F, 0x004F, 0x006F, - /* U+0210 */ 0x0052, 0x0072, 0x0052, 0x0072, 0x0055, 0x0075, 0x0055, 0x0075, - /* U+0218 */ 0x0053, 0x0073, 0x0054, 0x0074, 0x021C, 0x021D, 0x0048, 0x0068, - /* U+0220 */ 0x0220, 0x0221, 0x0222, 0x0223, 0x0224, 0x0225, 0x0041, 0x0061, - /* U+0228 */ 0x0045, 0x0065, 0x004F, 0x006F, 0x004F, 0x006F, 0x004F, 0x006F, - // U+022A: Manually changed from 00D6 to 004F - // U+022B: Manually changed from 00F6 to 006F - // U+022C: Manually changed from 00D5 to 004F - // U+022D: Manually changed from 00F5 to 006F - /* U+0230 */ 0x004F, 0x006F, 0x0059, 0x0079, 0x0234, 0x0235, 0x0236, 0x0237, - // U+0230: Manually changed from 022E to 004F - // U+0231: Manually changed from 022F to 006F - /* U+0238 */ 0x0238, 0x0239, 0x023A, 0x023B, 0x023C, 0x023D, 0x023E, 0x023F, - /* U+0240 */ 0x0240, 0x0241, 0x0242, 0x0243, 0x0244, 0x0245, 0x0246, 0x0247, - /* U+0248 */ 0x0248, 0x0249, 0x024A, 0x024B, 0x024C, 0x024D, 0x024E, 0x024F, - /* U+0250 */ 0x0250, 0x0251, 0x0252, 0x0253, 0x0254, 0x0255, 0x0256, 0x0257, - /* U+0258 */ 0x0258, 0x0259, 0x025A, 0x025B, 0x025C, 0x025D, 0x025E, 0x025F, - /* U+0260 */ 0x0260, 0x0261, 0x0262, 0x0263, 0x0264, 0x0265, 0x0266, 0x0267, - /* U+0268 */ 0x0268, 0x0269, 0x026A, 0x026B, 0x026C, 0x026D, 0x026E, 0x026F, - /* U+0270 */ 0x0270, 0x0271, 0x0272, 0x0273, 0x0274, 0x0275, 0x0276, 0x0277, - /* U+0278 */ 0x0278, 0x0279, 0x027A, 0x027B, 0x027C, 0x027D, 0x027E, 0x027F, - /* U+0280 */ 0x0280, 0x0281, 0x0282, 0x0283, 0x0284, 0x0285, 0x0286, 0x0287, - /* U+0288 */ 0x0288, 0x0289, 0x028A, 0x028B, 0x028C, 0x028D, 0x028E, 0x028F, - /* U+0290 */ 0x0290, 0x0291, 0x0292, 0x0293, 0x0294, 0x0295, 0x0296, 0x0297, - /* U+0298 */ 0x0298, 0x0299, 0x029A, 0x029B, 0x029C, 0x029D, 0x029E, 0x029F, - /* U+02A0 */ 0x02A0, 0x02A1, 0x02A2, 0x02A3, 0x02A4, 0x02A5, 0x02A6, 0x02A7, - /* U+02A8 */ 0x02A8, 0x02A9, 0x02AA, 0x02AB, 0x02AC, 0x02AD, 0x02AE, 0x02AF, - /* U+02B0 */ 0x0068, 0x0266, 0x006A, 0x0072, 0x0279, 0x027B, 0x0281, 0x0077, - /* U+02B8 */ 0x0079, 0x02B9, 0x02BA, 0x02BB, 0x02BC, 0x02BD, 0x02BE, 0x02BF, - /* U+02C0 */ 0x02C0, 0x02C1, 0x02C2, 0x02C3, 0x02C4, 0x02C5, 0x02C6, 0x02C7, - /* U+02C8 */ 0x02C8, 0x02C9, 0x02CA, 0x02CB, 0x02CC, 0x02CD, 0x02CE, 0x02CF, - /* U+02D0 */ 0x02D0, 0x02D1, 0x02D2, 0x02D3, 0x02D4, 0x02D5, 0x02D6, 0x02D7, - /* U+02D8 */ 0x0020, 0x0020, 0x0020, 0x0020, 0x0020, 0x0020, 0x02DE, 0x02DF, - /* U+02E0 */ 0x0263, 0x006C, 0x0073, 0x0078, 0x0295, 0x02E5, 0x02E6, 0x02E7, - /* U+02E8 */ 0x02E8, 0x02E9, 0x02EA, 0x02EB, 0x02EC, 0x02ED, 0x02EE, 0x02EF, - /* U+02F0 */ 0x02F0, 0x02F1, 0x02F2, 0x02F3, 0x02F4, 0x02F5, 0x02F6, 0x02F7, - /* U+02F8 */ 0x02F8, 0x02F9, 0x02FA, 0x02FB, 0x02FC, 0x02FD, 0x02FE, 0x02FF, - /* U+0300 */ 0x0300, 0x0301, 0x0302, 0x0303, 0x0304, 0x0305, 0x0306, 0x0307, - /* U+0308 */ 0x0308, 0x0309, 0x030A, 0x030B, 0x030C, 0x030D, 0x030E, 0x030F, - /* U+0310 */ 0x0310, 0x0311, 0x0312, 0x0313, 0x0314, 0x0315, 0x0316, 0x0317, - /* U+0318 */ 0x0318, 0x0319, 0x031A, 0x031B, 0x031C, 0x031D, 0x031E, 0x031F, - /* U+0320 */ 0x0320, 0x0321, 0x0322, 0x0323, 0x0324, 0x0325, 0x0326, 0x0327, - /* U+0328 */ 0x0328, 0x0329, 0x032A, 0x032B, 0x032C, 0x032D, 0x032E, 0x032F, - /* U+0330 */ 0x0330, 0x0331, 0x0332, 0x0333, 0x0334, 0x0335, 0x0336, 0x0337, - /* U+0338 */ 0x0338, 0x0339, 0x033A, 0x033B, 0x033C, 0x033D, 0x033E, 0x033F, - /* U+0340 */ 0x0300, 0x0301, 0x0342, 0x0313, 0x0308, 0x0345, 0x0346, 0x0347, - /* U+0348 */ 0x0348, 0x0349, 0x034A, 0x034B, 0x034C, 0x034D, 0x034E, 0x034F, - /* U+0350 */ 0x0350, 0x0351, 0x0352, 0x0353, 0x0354, 0x0355, 0x0356, 0x0357, - /* U+0358 */ 0x0358, 0x0359, 0x035A, 0x035B, 0x035C, 0x035D, 0x035E, 0x035F, - /* U+0360 */ 0x0360, 0x0361, 0x0362, 0x0363, 0x0364, 0x0365, 0x0366, 0x0367, - /* U+0368 */ 0x0368, 0x0369, 0x036A, 0x036B, 0x036C, 0x036D, 0x036E, 0x036F, - /* U+0370 */ 0x0370, 0x0371, 0x0372, 0x0373, 0x02B9, 0x0375, 0x0376, 0x0377, - /* U+0378 */ 0x0378, 0x0379, 0x0020, 0x037B, 0x037C, 0x037D, 0x003B, 0x037F, - /* U+0380 */ 0x0380, 0x0381, 0x0382, 0x0383, 0x0020, 0x00A8, 0x0391, 0x00B7, - /* U+0388 */ 0x0395, 0x0397, 0x0399, 0x038B, 0x039F, 0x038D, 0x03A5, 0x03A9, - /* U+0390 */ 0x03CA, 0x0391, 0x0392, 0x0393, 0x0394, 0x0395, 0x0396, 0x0397, - /* U+0398 */ 0x0398, 0x0399, 0x039A, 0x039B, 0x039C, 0x039D, 0x039E, 0x039F, - /* U+03A0 */ 0x03A0, 0x03A1, 0x03A2, 0x03A3, 0x03A4, 0x03A5, 0x03A6, 0x03A7, - /* U+03A8 */ 0x03A8, 0x03A9, 0x0399, 0x03A5, 0x03B1, 0x03B5, 0x03B7, 0x03B9, - /* U+03B0 */ 0x03CB, 0x03B1, 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7, - /* U+03B8 */ 0x03B8, 0x03B9, 0x03BA, 0x03BB, 0x03BC, 0x03BD, 0x03BE, 0x03BF, - /* U+03C0 */ 0x03C0, 0x03C1, 0x03C2, 0x03C3, 0x03C4, 0x03C5, 0x03C6, 0x03C7, - /* U+03C8 */ 0x03C8, 0x03C9, 0x03B9, 0x03C5, 0x03BF, 0x03C5, 0x03C9, 0x03CF, - /* U+03D0 */ 0x03B2, 0x03B8, 0x03A5, 0x03D2, 0x03D2, 0x03C6, 0x03C0, 0x03D7, - /* U+03D8 */ 0x03D8, 0x03D9, 0x03DA, 0x03DB, 0x03DC, 0x03DD, 0x03DE, 0x03DF, - /* U+03E0 */ 0x03E0, 0x03E1, 0x03E2, 0x03E3, 0x03E4, 0x03E5, 0x03E6, 0x03E7, - /* U+03E8 */ 0x03E8, 0x03E9, 0x03EA, 0x03EB, 0x03EC, 0x03ED, 0x03EE, 0x03EF, - /* U+03F0 */ 0x03BA, 0x03C1, 0x03C2, 0x03F3, 0x0398, 0x03B5, 0x03F6, 0x03F7, - /* U+03F8 */ 0x03F8, 0x03A3, 0x03FA, 0x03FB, 0x03FC, 0x03FD, 0x03FE, 0x03FF, - /* U+0400 */ 0x0415, 0x0415, 0x0402, 0x0413, 0x0404, 0x0405, 0x0406, 0x0406, - /* U+0408 */ 0x0408, 0x0409, 0x040A, 0x040B, 0x041A, 0x0418, 0x0423, 0x040F, - /* U+0410 */ 0x0410, 0x0411, 0x0412, 0x0413, 0x0414, 0x0415, 0x0416, 0x0417, - /* U+0418 */ 0x0418, 0x0419, 0x041A, 0x041B, 0x041C, 0x041D, 0x041E, 0x041F, - // U+0419: Manually changed from 0418 to 0419 - /* U+0420 */ 0x0420, 0x0421, 0x0422, 0x0423, 0x0424, 0x0425, 0x0426, 0x0427, - /* U+0428 */ 0x0428, 0x0429, 0x042C, 0x042B, 0x042C, 0x042D, 0x042E, 0x042F, - // U+042A: Manually changed from 042A to 042C - /* U+0430 */ 0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437, - /* U+0438 */ 0x0438, 0x0439, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, 0x043F, - // U+0439: Manually changed from 0438 to 0439 - /* U+0440 */ 0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447, - /* U+0448 */ 0x0448, 0x0449, 0x044C, 0x044B, 0x044C, 0x044D, 0x044E, 0x044F, - // U+044A: Manually changed from 044A to 044C - /* U+0450 */ 0x0435, 0x0435, 0x0452, 0x0433, 0x0454, 0x0455, 0x0456, 0x0456, - /* U+0458 */ 0x0458, 0x0459, 0x045A, 0x045B, 0x043A, 0x0438, 0x0443, 0x045F, - /* U+0460 */ 0x0460, 0x0461, 0x0462, 0x0463, 0x0464, 0x0465, 0x0466, 0x0467, - /* U+0468 */ 0x0468, 0x0469, 0x046A, 0x046B, 0x046C, 0x046D, 0x046E, 0x046F, - /* U+0470 */ 0x0470, 0x0471, 0x0472, 0x0473, 0x0474, 0x0475, 0x0474, 0x0475, - /* U+0478 */ 0x0478, 0x0479, 0x047A, 0x047B, 0x047C, 0x047D, 0x047E, 0x047F, - /* U+0480 */ 0x0480, 0x0481, 0x0482, 0x0483, 0x0484, 0x0485, 0x0486, 0x0487, - /* U+0488 */ 0x0488, 0x0489, 0x048A, 0x048B, 0x048C, 0x048D, 0x048E, 0x048F, - /* U+0490 */ 0x0490, 0x0491, 0x0492, 0x0493, 0x0494, 0x0495, 0x0496, 0x0497, - /* U+0498 */ 0x0498, 0x0499, 0x049A, 0x049B, 0x049C, 0x049D, 0x049E, 0x049F, - /* U+04A0 */ 0x04A0, 0x04A1, 0x04A2, 0x04A3, 0x04A4, 0x04A5, 0x04A6, 0x04A7, - /* U+04A8 */ 0x04A8, 0x04A9, 0x04AA, 0x04AB, 0x04AC, 0x04AD, 0x04AE, 0x04AF, - /* U+04B0 */ 0x04B0, 0x04B1, 0x04B2, 0x04B3, 0x04B4, 0x04B5, 0x04B6, 0x04B7, - /* U+04B8 */ 0x04B8, 0x04B9, 0x04BA, 0x04BB, 0x04BC, 0x04BD, 0x04BE, 0x04BF, - /* U+04C0 */ 0x04C0, 0x0416, 0x0436, 0x04C3, 0x04C4, 0x04C5, 0x04C6, 0x04C7, - /* U+04C8 */ 0x04C8, 0x04C9, 0x04CA, 0x04CB, 0x04CC, 0x04CD, 0x04CE, 0x04CF, - /* U+04D0 */ 0x0410, 0x0430, 0x0410, 0x0430, 0x04D4, 0x04D5, 0x0415, 0x0435, - /* U+04D8 */ 0x04D8, 0x04D9, 0x04D8, 0x04D9, 0x0416, 0x0436, 0x0417, 0x0437, - /* U+04E0 */ 0x04E0, 0x04E1, 0x0418, 0x0438, 0x0418, 0x0438, 0x041E, 0x043E, - /* U+04E8 */ 0x04E8, 0x04E9, 0x04E8, 0x04E9, 0x042D, 0x044D, 0x0423, 0x0443, - /* U+04F0 */ 0x0423, 0x0443, 0x0423, 0x0443, 0x0427, 0x0447, 0x04F6, 0x04F7, - /* U+04F8 */ 0x042B, 0x044B, 0x04FA, 0x04FB, 0x04FC, 0x04FD, 0x04FE, 0x04FF, - }; -} diff --git a/java/src/com/android/inputmethod/latin/InputAttributes.java b/java/src/com/android/inputmethod/latin/InputAttributes.java index 8caf6f17f..fcf043031 100644 --- a/java/src/com/android/inputmethod/latin/InputAttributes.java +++ b/java/src/com/android/inputmethod/latin/InputAttributes.java @@ -52,8 +52,7 @@ public final class InputAttributes { } else if (inputClass == 0) { // TODO: is this check still necessary? Log.w(TAG, String.format("Unexpected input class: inputType=0x%08x" - + " imeOptions=0x%08x", - inputType, editorInfo.imeOptions)); + + " imeOptions=0x%08x", inputType, editorInfo.imeOptions)); } mIsSettingsSuggestionStripOn = false; mInputTypeNoAutoCorrect = false; @@ -204,8 +203,7 @@ public final class InputAttributes { public static boolean inPrivateImeOptions(String packageName, String key, EditorInfo editorInfo) { if (editorInfo == null) return false; - final String findingKey = (packageName != null) ? packageName + "." + key - : key; + final String findingKey = (packageName != null) ? packageName + "." + key : key; return StringUtils.containsInCommaSplittableText(findingKey, editorInfo.privateImeOptions); } } diff --git a/java/src/com/android/inputmethod/latin/InputView.java b/java/src/com/android/inputmethod/latin/InputView.java index 81ccf83d8..76b0912f6 100644 --- a/java/src/com/android/inputmethod/latin/InputView.java +++ b/java/src/com/android/inputmethod/latin/InputView.java @@ -23,87 +23,203 @@ import android.view.MotionEvent; import android.view.View; import android.widget.LinearLayout; -public final class InputView extends LinearLayout { - private View mSuggestionStripView; - private View mKeyboardView; - private int mKeyboardTopPadding; +import com.android.inputmethod.keyboard.MainKeyboardView; +import com.android.inputmethod.latin.suggestions.MoreSuggestionsView; +import com.android.inputmethod.latin.suggestions.SuggestionStripView; - private boolean mIsForwardingEvent; +public final class InputView extends LinearLayout { private final Rect mInputViewRect = new Rect(); - private final Rect mEventForwardingRect = new Rect(); - private final Rect mEventReceivingRect = new Rect(); + private KeyboardTopPaddingForwarder mKeyboardTopPaddingForwarder; + private MoreSuggestionsViewCanceler mMoreSuggestionsViewCanceler; public InputView(final Context context, final AttributeSet attrs) { super(context, attrs, 0); } - public void setKeyboardGeometry(final int keyboardTopPadding) { - mKeyboardTopPadding = keyboardTopPadding; - } - @Override protected void onFinishInflate() { - mSuggestionStripView = findViewById(R.id.suggestion_strip_view); - mKeyboardView = findViewById(R.id.keyboard_view); + final SuggestionStripView suggestionStripView = + (SuggestionStripView)findViewById(R.id.suggestion_strip_view); + final MainKeyboardView mainKeyboardView = + (MainKeyboardView)findViewById(R.id.keyboard_view); + mKeyboardTopPaddingForwarder = new KeyboardTopPaddingForwarder( + mainKeyboardView, suggestionStripView); + mMoreSuggestionsViewCanceler = new MoreSuggestionsViewCanceler( + mainKeyboardView, suggestionStripView); + } + + public void setKeyboardTopPadding(final int keyboardTopPadding) { + mKeyboardTopPaddingForwarder.setKeyboardTopPadding(keyboardTopPadding); } @Override public boolean dispatchTouchEvent(final MotionEvent me) { - if (mSuggestionStripView.getVisibility() != VISIBLE - || mKeyboardView.getVisibility() != VISIBLE) { - return super.dispatchTouchEvent(me); - } - - // The touch events that hit the top padding of keyboard should be forwarded to - // {@link SuggestionStripView}. final Rect rect = mInputViewRect; - this.getGlobalVisibleRect(rect); + getGlobalVisibleRect(rect); final int x = (int)me.getX() + rect.left; final int y = (int)me.getY() + rect.top; - final Rect forwardingRect = mEventForwardingRect; - mKeyboardView.getGlobalVisibleRect(forwardingRect); - if (!mIsForwardingEvent && !forwardingRect.contains(x, y)) { - return super.dispatchTouchEvent(me); + // The touch events that hit the top padding of keyboard should be + // forwarded to {@link SuggestionStripView}. + if (mKeyboardTopPaddingForwarder.dispatchTouchEvent(x, y, me)) { + return true; + } + // To cancel {@link MoreSuggestionsView}, we should intercept a touch event to + // {@link MainKeyboardView} and dismiss the {@link MoreSuggestionsView}. + if (mMoreSuggestionsViewCanceler.dispatchTouchEvent(x, y, me)) { + return true; + } + return super.dispatchTouchEvent(me); + } + + /** + * This class forwards series of {@link MotionEvent}s from <code>Forwarder</code> view to + * <code>Receiver</code> view. + * + * @param <Sender> a {@link View} that may send a {@link MotionEvent} to <Receiver>. + * @param <Receiver> a {@link View} that receives forwarded {@link MotionEvent} from + * <Forwarder>. + */ + private static abstract class MotionEventForwarder<Sender extends View, Receiver extends View> { + protected final Sender mSenderView; + protected final Receiver mReceiverView; + + private boolean mIsForwardingEvent; + protected final Rect mEventSendingRect = new Rect(); + protected final Rect mEventReceivingRect = new Rect(); + + public MotionEventForwarder(final Sender senderView, final Receiver receiverView) { + mSenderView = senderView; + mReceiverView = receiverView; } - final int forwardingLimitY = forwardingRect.top + mKeyboardTopPadding; - boolean sendToTarget = false; + // Return true if a touch event of global coordinate x, y needs to be forwarded. + protected abstract boolean needsToForward(final int x, final int y); - switch (me.getAction()) { - case MotionEvent.ACTION_DOWN: - if (y < forwardingLimitY) { - // This down event and further move and up events should be forwarded to the target. - mIsForwardingEvent = true; - sendToTarget = true; + // Translate global x-coordinate to <code>Receiver</code> local coordinate. + protected int translateX(final int x) { + return x - mEventReceivingRect.left; + } + + // Translate global y-coordinate to <code>Receiver</code> local coordinate. + protected int translateY(final int y) { + return y - mEventReceivingRect.top; + } + + // Callback when a {@link MotionEvent} is forwarded. + protected void onForwardingEvent(final MotionEvent me) {} + + // Dispatches a {@link MotioneEvent} to <code>Receiver</code> if needed and returns true. + // Otherwise returns false. + public boolean dispatchTouchEvent(final int x, final int y, final MotionEvent me) { + // Forwards a {link MotionEvent} only if both <code>Sender</code> and + // <code>Receiver</code> are visible. + if (mSenderView.getVisibility() != View.VISIBLE || + mReceiverView.getVisibility() != View.VISIBLE) { + return false; + } + final Rect sendingRect = mEventSendingRect; + mSenderView.getGlobalVisibleRect(sendingRect); + if (!mIsForwardingEvent && !sendingRect.contains(x, y)) { + return false; } - break; - case MotionEvent.ACTION_MOVE: - sendToTarget = mIsForwardingEvent; - break; - case MotionEvent.ACTION_UP: - case MotionEvent.ACTION_CANCEL: - sendToTarget = mIsForwardingEvent; - mIsForwardingEvent = false; - break; + + boolean shouldForwardToReceiver = false; + + switch (me.getActionMasked()) { + case MotionEvent.ACTION_DOWN: + // If the down event happens in the forwarding area, successive {@link MotionEvent}s + // should be forwarded. + if (needsToForward(x, y)) { + mIsForwardingEvent = true; + shouldForwardToReceiver = true; + } + break; + case MotionEvent.ACTION_MOVE: + shouldForwardToReceiver = mIsForwardingEvent; + break; + case MotionEvent.ACTION_UP: + case MotionEvent.ACTION_CANCEL: + shouldForwardToReceiver = mIsForwardingEvent; + mIsForwardingEvent = false; + break; + } + + if (!shouldForwardToReceiver) { + return false; + } + + final Rect receivingRect = mEventReceivingRect; + mReceiverView.getGlobalVisibleRect(receivingRect); + // Translate global coordinates to <code>Receiver</code> local coordinates. + me.setLocation(translateX(x), translateY(y)); + mReceiverView.dispatchTouchEvent(me); + onForwardingEvent(me); + return true; } + } + + /** + * This class forwards {@link MotionEvent}s happened in the top padding of + * {@link MainKeyboardView} to {@link SuggestionStripView}. + */ + private static class KeyboardTopPaddingForwarder + extends MotionEventForwarder<MainKeyboardView, SuggestionStripView> { + private int mKeyboardTopPadding; - if (!sendToTarget) { - return super.dispatchTouchEvent(me); + public KeyboardTopPaddingForwarder(final MainKeyboardView mainKeyboardView, + final SuggestionStripView suggestionStripView) { + super(mainKeyboardView, suggestionStripView); } - final Rect receivingRect = mEventReceivingRect; - mSuggestionStripView.getGlobalVisibleRect(receivingRect); - final int translatedX = x - receivingRect.left; - final int translatedY; - if (y < forwardingLimitY) { - // The forwarded event should have coordinates that are inside of the target. - translatedY = Math.min(y - receivingRect.top, receivingRect.height() - 1); - } else { - translatedY = y - receivingRect.top; + public void setKeyboardTopPadding(final int keyboardTopPadding) { + mKeyboardTopPadding = keyboardTopPadding; + } + + private boolean isInKeyboardTopPadding(final int y) { + return y < mEventSendingRect.top + mKeyboardTopPadding; + } + + @Override + protected boolean needsToForward(final int x, final int y) { + return isInKeyboardTopPadding(y); + } + + @Override + protected int translateY(final int y) { + final int translatedY = super.translateY(y); + if (isInKeyboardTopPadding(y)) { + // The forwarded event should have coordinates that are inside of + // the target. + return Math.min(translatedY, mEventReceivingRect.height() - 1); + } + return translatedY; + } + } + + /** + * This class forwards {@link MotionEvent}s happened in the {@link MainKeyboardView} to + * {@link SuggestionStripView} when the {@link MoreSuggestionsView} is showing. + * {@link SuggestionStripView} dismisses {@link MoreSuggestionsView} when it receives those + * events. + */ + private static class MoreSuggestionsViewCanceler + extends MotionEventForwarder<MainKeyboardView, SuggestionStripView> { + public MoreSuggestionsViewCanceler(final MainKeyboardView mainKeyboardView, + final SuggestionStripView suggestionStripView) { + super(mainKeyboardView, suggestionStripView); + } + + @Override + protected boolean needsToForward(final int x, final int y) { + return mReceiverView.isShowingMoreSuggestionPanel() && mEventSendingRect.contains(x, y); + } + + @Override + protected void onForwardingEvent(final MotionEvent me) { + if (me.getActionMasked() == MotionEvent.ACTION_DOWN) { + mReceiverView.dismissMoreSuggestionsPanel(); + } } - me.setLocation(translatedX, translatedY); - mSuggestionStripView.dispatchTouchEvent(me); - return true; } } diff --git a/java/src/com/android/inputmethod/latin/LatinIME.java b/java/src/com/android/inputmethod/latin/LatinIME.java index 77d07019f..67b570277 100644 --- a/java/src/com/android/inputmethod/latin/LatinIME.java +++ b/java/src/com/android/inputmethod/latin/LatinIME.java @@ -68,7 +68,6 @@ import com.android.inputmethod.compat.InputMethodServiceCompatUtils; import com.android.inputmethod.compat.SuggestionSpanUtils; import com.android.inputmethod.dictionarypack.DictionaryPackConstants; import com.android.inputmethod.event.EventInterpreter; -import com.android.inputmethod.keyboard.KeyDetector; import com.android.inputmethod.keyboard.Keyboard; import com.android.inputmethod.keyboard.KeyboardActionListener; import com.android.inputmethod.keyboard.KeyboardId; @@ -78,10 +77,7 @@ import com.android.inputmethod.latin.Suggest.OnGetSuggestedWordsCallback; import com.android.inputmethod.latin.SuggestedWords.SuggestedWordInfo; import com.android.inputmethod.latin.define.ProductionFlag; import com.android.inputmethod.latin.personalization.DictionaryDecayBroadcastReciever; -import com.android.inputmethod.latin.personalization.PersonalizationDictionary; import com.android.inputmethod.latin.personalization.PersonalizationDictionarySessionRegister; -import com.android.inputmethod.latin.personalization.PersonalizationHelper; -import com.android.inputmethod.latin.personalization.PersonalizationPredictionDictionary; import com.android.inputmethod.latin.personalization.UserHistoryDictionary; import com.android.inputmethod.latin.settings.Settings; import com.android.inputmethod.latin.settings.SettingsActivity; @@ -97,12 +93,11 @@ import com.android.inputmethod.latin.utils.InputTypeUtils; import com.android.inputmethod.latin.utils.IntentUtils; import com.android.inputmethod.latin.utils.JniUtils; import com.android.inputmethod.latin.utils.LatinImeLoggerUtils; +import com.android.inputmethod.latin.utils.LeakGuardHandlerWrapper; import com.android.inputmethod.latin.utils.RecapitalizeStatus; -import com.android.inputmethod.latin.utils.StaticInnerHandlerWrapper; import com.android.inputmethod.latin.utils.StringUtils; import com.android.inputmethod.latin.utils.TargetPackageInfoGetterTask; import com.android.inputmethod.latin.utils.TextRange; -import com.android.inputmethod.latin.utils.UserHistoryForgettingCurveUtils; import com.android.inputmethod.research.ResearchLogger; import java.io.FileDescriptor; @@ -110,6 +105,7 @@ import java.io.PrintWriter; import java.util.ArrayList; import java.util.Locale; import java.util.TreeSet; +import java.util.concurrent.TimeUnit; /** * Input method implementation for Qwerty'ish keyboard. @@ -181,9 +177,6 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen private boolean mIsMainDictionaryAvailable; private UserBinaryDictionary mUserDictionary; - private UserHistoryDictionary mUserHistoryDictionary; - private PersonalizationPredictionDictionary mPersonalizationPredictionDictionary; - private PersonalizationDictionary mPersonalizationDictionary; private boolean mIsUserDictionaryAvailable; private LastComposedWord mLastComposedWord = LastComposedWord.NOT_A_COMPOSED_WORD; @@ -196,9 +189,6 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen private int mLastSelectionStart = NOT_A_CURSOR_POSITION; private int mLastSelectionEnd = NOT_A_CURSOR_POSITION; - // Whether we are expecting an onUpdateSelection event to fire. If it does when we don't - // "expect" it, it means the user actually moved the cursor. - private boolean mExpectingUpdateSelection; private int mDeleteCount; private long mLastKeyTime; private final TreeSet<Long> mCurrentlyPressedHardwareKeys = CollectionUtils.newTreeSet(); @@ -227,7 +217,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen public final UIHandler mHandler = new UIHandler(this); private InputUpdater mInputUpdater; - public static final class UIHandler extends StaticInnerHandlerWrapper<LatinIME> { + public static final class UIHandler extends LeakGuardHandlerWrapper<LatinIME> { private static final int MSG_UPDATE_SHIFT_STATE = 0; private static final int MSG_PENDING_IMS_CALLBACK = 1; private static final int MSG_UPDATE_SUGGESTION_STRIP = 2; @@ -236,6 +226,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen private static final int MSG_REOPEN_DICTIONARIES = 5; private static final int MSG_ON_END_BATCH_INPUT = 6; private static final int MSG_RESET_CACHES = 7; + // Update this when adding new messages + private static final int MSG_LAST = MSG_RESET_CACHES; private static final int ARG1_NOT_GESTURE_INPUT = 0; private static final int ARG1_DISMISS_GESTURE_FLOATING_PREVIEW_TEXT = 1; @@ -248,12 +240,12 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen private long mDoubleSpacePeriodTimeout; private long mDoubleSpacePeriodTimerStart; - public UIHandler(final LatinIME outerInstance) { - super(outerInstance); + public UIHandler(final LatinIME ownerInstance) { + super(ownerInstance); } public void onCreate() { - final Resources res = getOuterInstance().getResources(); + final Resources res = getOwnerInstance().getResources(); mDelayUpdateSuggestions = res.getInteger(R.integer.config_delay_update_suggestions); mDelayUpdateShiftState = @@ -264,7 +256,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen @Override public void handleMessage(final Message msg) { - final LatinIME latinIme = getOuterInstance(); + final LatinIME latinIme = getOwnerInstance(); final KeyboardSwitcher switcher = latinIme.mKeyboardSwitcher; switch (msg.what) { case MSG_UPDATE_SUGGESTION_STRIP: @@ -347,6 +339,13 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen removeMessages(MSG_UPDATE_SHIFT_STATE); } + @UsedForTesting + public void removeAllMessages() { + for (int i = 0; i <= MSG_LAST; ++i) { + removeMessages(i); + } + } + public void showGesturePreviewAndSuggestionStrip(final SuggestedWords suggestedWords, final boolean dismissGestureFloatingPreviewText) { removeMessages(MSG_SHOW_GESTURE_PREVIEW_AND_SUGGESTION_STRIP); @@ -401,7 +400,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen removeMessages(MSG_PENDING_IMS_CALLBACK); resetPendingImsCallback(); mIsOrientationChanging = true; - final LatinIME latinIme = getOuterInstance(); + final LatinIME latinIme = getOwnerInstance(); if (latinIme.isInputViewShown()) { latinIme.mKeyboardSwitcher.saveKeyboardState(); } @@ -434,7 +433,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen mIsOrientationChanging = false; mPendingSuccessiveImsCallback = true; } - final LatinIME latinIme = getOuterInstance(); + final LatinIME latinIme = getOwnerInstance(); executePendingImsCallback(latinIme, editorInfo, restarting); latinIme.onStartInputInternal(editorInfo, restarting); } @@ -453,7 +452,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen sendMessageDelayed(obtainMessage(MSG_PENDING_IMS_CALLBACK), PENDING_IMS_CALLBACK_DURATION); } - final LatinIME latinIme = getOuterInstance(); + final LatinIME latinIme = getOwnerInstance(); executePendingImsCallback(latinIme, editorInfo, restarting); latinIme.onStartInputViewInternal(editorInfo, restarting); mAppliedEditorInfo = editorInfo; @@ -465,7 +464,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // Typically this is the first onFinishInputView after orientation changed. mHasPendingFinishInputView = true; } else { - final LatinIME latinIme = getOuterInstance(); + final LatinIME latinIme = getOwnerInstance(); latinIme.onFinishInputViewInternal(finishingInput); mAppliedEditorInfo = null; } @@ -476,7 +475,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // Typically this is the first onFinishInput after orientation changed. mHasPendingFinishInput = true; } else { - final LatinIME latinIme = getOuterInstance(); + final LatinIME latinIme = getOwnerInstance(); executePendingImsCallback(latinIme, null, false); latinIme.onFinishInputInternal(); } @@ -589,9 +588,9 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // the layout; at this time, we need to skip resetting the contacts dictionary. It will // be done later inside {@see #initSuggest()} when the reopenDictionaries message is // processed. - if (!mHandler.hasPendingReopenDictionaries()) { - // May need to reset the contacts dictionary depending on the user settings. - resetContactsDictionary(null == mSuggest ? null : mSuggest.getContactsDictionary()); + if (!mHandler.hasPendingReopenDictionaries() && mSuggest != null) { + // May need to reset dictionaries depending on the user settings. + mSuggest.setAdditionalDictionaries(mSuggest /* oldSuggest */, mSettings.getCurrent()); } } @@ -609,7 +608,6 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen final Locale switcherSubtypeLocale = mSubtypeSwitcher.getCurrentSubtypeLocale(); final String switcherLocaleStr = switcherSubtypeLocale.toString(); final Locale subtypeLocale; - final String localeStr; if (TextUtils.isEmpty(switcherLocaleStr)) { // This happens in very rare corner cases - for example, immediately after a switch // to LatinIME has been requested, about a frame later another switch happens. In this @@ -619,10 +617,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // of knowing anyway. Log.e(TAG, "System is reporting no current subtype."); subtypeLocale = getResources().getConfiguration().locale; - localeStr = subtypeLocale.toString(); } else { subtypeLocale = switcherSubtypeLocale; - localeStr = switcherLocaleStr; } final Suggest newSuggest = new Suggest(this /* Context */, subtypeLocale, @@ -637,71 +633,15 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen ResearchLogger.getInstance().initSuggest(newSuggest); } - mUserDictionary = new UserBinaryDictionary(this, localeStr); + mUserDictionary = new UserBinaryDictionary(this, subtypeLocale); mIsUserDictionaryAvailable = mUserDictionary.isEnabled(); newSuggest.setUserDictionary(mUserDictionary); - - final SharedPreferences prefs = PreferenceManager.getDefaultSharedPreferences(this); - - mUserHistoryDictionary = PersonalizationHelper.getUserHistoryDictionary( - this, localeStr, prefs); - newSuggest.setUserHistoryDictionary(mUserHistoryDictionary); - mPersonalizationDictionary = PersonalizationHelper - .getPersonalizationDictionary(this, localeStr, prefs); - newSuggest.setPersonalizationDictionary(mPersonalizationDictionary); - mPersonalizationPredictionDictionary = PersonalizationHelper - .getPersonalizationPredictionDictionary(this, localeStr, prefs); - newSuggest.setPersonalizationPredictionDictionary(mPersonalizationPredictionDictionary); - + newSuggest.setAdditionalDictionaries(mSuggest /* oldSuggest */, mSettings.getCurrent()); final Suggest oldSuggest = mSuggest; - resetContactsDictionary(null != oldSuggest ? oldSuggest.getContactsDictionary() : null); mSuggest = newSuggest; if (oldSuggest != null) oldSuggest.close(); } - /** - * Resets the contacts dictionary in mSuggest according to the user settings. - * - * This method takes an optional contacts dictionary to use when the locale hasn't changed - * since the contacts dictionary can be opened or closed as necessary depending on the settings. - * - * @param oldContactsDictionary an optional dictionary to use, or null - */ - private void resetContactsDictionary(final ContactsBinaryDictionary oldContactsDictionary) { - final Suggest suggest = mSuggest; - final boolean shouldSetDictionary = - (null != suggest && mSettings.getCurrent().mUseContactsDict); - - final ContactsBinaryDictionary dictionaryToUse; - if (!shouldSetDictionary) { - // Make sure the dictionary is closed. If it is already closed, this is a no-op, - // so it's safe to call it anyways. - if (null != oldContactsDictionary) oldContactsDictionary.close(); - dictionaryToUse = null; - } else { - final Locale locale = mSubtypeSwitcher.getCurrentSubtypeLocale(); - if (null != oldContactsDictionary) { - if (!oldContactsDictionary.mLocale.equals(locale)) { - // If the locale has changed then recreate the contacts dictionary. This - // allows locale dependent rules for handling bigram name predictions. - oldContactsDictionary.close(); - dictionaryToUse = new ContactsBinaryDictionary(this, locale); - } else { - // Make sure the old contacts dictionary is opened. If it is already open, - // this is a no-op, so it's safe to call it anyways. - oldContactsDictionary.reopen(this); - dictionaryToUse = oldContactsDictionary; - } - } else { - dictionaryToUse = new ContactsBinaryDictionary(this, locale); - } - } - - if (null != suggest) { - suggest.setContactsDictionary(dictionaryToUse); - } - } - /* package private */ void resetSuggestMainDict() { final Locale subtypeLocale = mSubtypeSwitcher.getCurrentSubtypeLocale(); mSuggest.resetMainDict(this, subtypeLocale, this /* SuggestInitializationListener */); @@ -760,8 +700,9 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen .findViewById(android.R.id.extractArea); mKeyPreviewBackingView = view.findViewById(R.id.key_preview_backing); mSuggestionStripView = (SuggestionStripView)view.findViewById(R.id.suggestion_strip_view); - if (mSuggestionStripView != null) + if (mSuggestionStripView != null) { mSuggestionStripView.setListener(this, view); + } if (LatinImeLogger.sVISUALDEBUG) { mKeyPreviewBackingView.setBackgroundColor(0x10FF0000); } @@ -907,7 +848,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // refresh later. final boolean canReachInputConnection; if (!mConnection.resetCachesUponCursorMoveAndReturnSuccess(editorInfo.initialSelStart, - false /* shouldFinishComposition */)) { + editorInfo.initialSelEnd, false /* shouldFinishComposition */)) { // We try resetting the caches up to 5 times before giving up. mHandler.postResetCaches(isDifferentTextField, 5 /* remainingTries */); // mLastSelection{Start,End} are reset later in this method, don't need to do it here @@ -1022,11 +963,6 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // Only for debug mBoostPersonalizationDictionaryForDebug = currentSettingsValues.mBoostPersonalizationDictionaryForDebug; - if (mBoostPersonalizationDictionaryForDebug) { - UserHistoryForgettingCurveUtils.boostMaxFreqForDebug(); - } else { - UserHistoryForgettingCurveUtils.resetMaxFreqForDebug(); - } } } @@ -1088,16 +1024,9 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen + ", ce=" + composingSpanEnd); } if (ProductionFlag.USES_DEVELOPMENT_ONLY_DIAGNOSTICS) { - final boolean expectingUpdateSelectionFromLogger = - ResearchLogger.getAndClearLatinIMEExpectingUpdateSelection(); ResearchLogger.latinIME_onUpdateSelection(mLastSelectionStart, mLastSelectionEnd, oldSelStart, oldSelEnd, newSelStart, newSelEnd, composingSpanStart, - composingSpanEnd, mExpectingUpdateSelection, - expectingUpdateSelectionFromLogger, mConnection); - if (expectingUpdateSelectionFromLogger) { - // TODO: Investigate. Quitting now sounds wrong - we won't do the resetting work - return; - } + composingSpanEnd, mConnection); } final boolean selectionChanged = mLastSelectionStart != newSelStart @@ -1116,14 +1045,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // TODO: revisit this when LatinIME supports hardware keyboards. // NOTE: the test harness subclasses LatinIME and overrides isInputViewShown(). // TODO: find a better way to simulate actual execution. - if (isInputViewShown() && !mExpectingUpdateSelection - && !mConnection.isBelatedExpectedUpdate(oldSelStart, newSelStart)) { - // TAKE CARE: there is a race condition when we enter this test even when the user - // did not explicitly move the cursor. This happens when typing fast, where two keys - // turn this flag on in succession and both onUpdateSelection() calls arrive after - // the second one - the first call successfully avoids this test, but the second one - // enters. For the moment we rely on noComposingSpan to further reduce the impact. - + if (isInputViewShown() && !mConnection.isBelatedExpectedUpdate(oldSelStart, newSelStart, + oldSelEnd, newSelEnd)) { // TODO: the following is probably better done in resetEntireInputState(). // it should only happen when the cursor moved, and the very purpose of the // test below is to narrow down whether this happened or not. Likewise with @@ -1149,13 +1072,13 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // Another option would be to send suggestions each time we set the composing // text, but that is probably too expensive to do, so we decided to leave things // as is. - resetEntireInputState(newSelStart); + resetEntireInputState(newSelStart, newSelEnd); } else { - // resetEntireInputState calls resetCachesUponCursorMove, but with the second - // argument as true. But in all cases where we don't reset the entire input state, - // we still want to tell the rich input connection about the new cursor position so - // that it can update its caches. - mConnection.resetCachesUponCursorMoveAndReturnSuccess(newSelStart, + // resetEntireInputState calls resetCachesUponCursorMove, but forcing the + // composition to end. But in all cases where we don't reset the entire input + // state, we still want to tell the rich input connection about the new cursor + // position so that it can update its caches. + mConnection.resetCachesUponCursorMoveAndReturnSuccess(newSelStart, newSelEnd, false /* shouldFinishComposition */); } @@ -1168,7 +1091,6 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen mRecapitalizeStatus.deactivate(); mKeyboardSwitcher.updateShiftState(); } - mExpectingUpdateSelection = false; // Make a note of the cursor position mLastSelectionStart = newSelStart; @@ -1354,8 +1276,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen @Override public boolean onEvaluateFullscreenMode() { // Reread resource value here, because this method is called by framework anytime as needed. - final boolean isFullscreenModeAllowed = - Settings.readUseFullscreenMode(getResources()); + final boolean isFullscreenModeAllowed = Settings.readUseFullscreenMode(getResources()); if (super.onEvaluateFullscreenMode() && isFullscreenModeAllowed) { // TODO: Remove this hack. Actually we should not really assume NO_EXTRACT_UI // implies NO_FULLSCREEN. However, the framework mistakenly does. i.e. NO_EXTRACT_UI @@ -1380,7 +1301,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // This will reset the whole input state to the starting state. It will clear // the composing word, reset the last composed word, tell the inputconnection about it. - private void resetEntireInputState(final int newCursorPosition) { + private void resetEntireInputState(final int newSelStart, final int newSelEnd) { final boolean shouldFinishComposition = mWordComposer.isComposingWord(); resetComposingState(true /* alsoResetLastComposedWord */); final SettingsValues settingsValues = mSettings.getCurrent(); @@ -1389,14 +1310,15 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } else { setSuggestedWords(settingsValues.mSuggestPuncList, false); } - mConnection.resetCachesUponCursorMoveAndReturnSuccess(newCursorPosition, + mConnection.resetCachesUponCursorMoveAndReturnSuccess(newSelStart, newSelEnd, shouldFinishComposition); } private void resetComposingState(final boolean alsoResetLastComposedWord) { mWordComposer.reset(); - if (alsoResetLastComposedWord) + if (alsoResetLastComposedWord) { mLastComposedWord = LastComposedWord.NOT_A_COMPOSED_WORD; + } } private void commitTyped(final String separatorString) { @@ -1443,15 +1365,16 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen if (0 != (auto & TextUtils.CAP_MODE_CHARACTERS)) { return WordComposer.CAPS_MODE_AUTO_SHIFT_LOCKED; } - if (0 != auto) return WordComposer.CAPS_MODE_AUTO_SHIFTED; + if (0 != auto) { + return WordComposer.CAPS_MODE_AUTO_SHIFTED; + } return WordComposer.CAPS_MODE_OFF; } private void swapSwapperAndSpace() { final CharSequence lastTwo = mConnection.getTextBeforeCursor(2, 0); // It is guaranteed lastTwo.charAt(1) is a swapper - else this method is not called. - if (lastTwo != null && lastTwo.length() == 2 - && lastTwo.charAt(0) == Constants.CODE_SPACE) { + if (lastTwo != null && lastTwo.length() == 2 && lastTwo.charAt(0) == Constants.CODE_SPACE) { mConnection.deleteSurroundingText(2, 0); final String text = lastTwo.charAt(1) + " "; mConnection.commitText(text, 1); @@ -1491,6 +1414,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen ResearchLogger.latinIME_maybeDoubleSpacePeriod(textToInsert, false /* isBatchMode */); } + mWordComposer.discardPreviousWordForSuggestion(); mKeyboardSwitcher.updateShiftState(); return true; } @@ -1730,7 +1654,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen if (mWordComposer.isCursorFrontOrMiddleOfComposingWord()) { // If we are in the middle of a recorrection, we need to commit the recorrection // first so that we can insert the character at the current cursor position. - resetEntireInputState(mLastSelectionStart); + resetEntireInputState(mLastSelectionStart, mLastSelectionEnd); } else { commitTyped(LastComposedWord.NOT_A_SEPARATOR); } @@ -1746,7 +1670,6 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } handleCharacter(primaryCode, keyX, keyY, spaceState); } - mExpectingUpdateSelection = true; return didAutoCorrect; } @@ -1786,9 +1709,9 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen mInputUpdater.onStartBatchInput(); mHandler.cancelUpdateSuggestionStrip(); mConnection.beginBatchEdit(); - final SettingsValues settingsValues = mSettings.getCurrent(); + final SettingsValues currentSettingsValues = mSettings.getCurrent(); if (mWordComposer.isComposingWord()) { - if (settingsValues.mIsInternal) { + if (currentSettingsValues.mIsInternal) { if (mWordComposer.isBatchMode()) { LatinImeLoggerUtils.onAutoCorrection( "", mWordComposer.getTypedWord(), " ", mWordComposer); @@ -1799,7 +1722,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen if (mWordComposer.isCursorFrontOrMiddleOfComposingWord()) { // If we are in the middle of a recorrection, we need to commit the recorrection // first so that we can insert the batch input at the current cursor position. - resetEntireInputState(mLastSelectionStart); + resetEntireInputState(mLastSelectionStart, mLastSelectionEnd); } else if (wordComposerSize <= 1) { // We auto-correct the previous (typed, not gestured) string iff it's one character // long. The reason for this is, even in the middle of gesture typing, you'll still @@ -1812,15 +1735,26 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } else { commitTyped(LastComposedWord.NOT_A_SEPARATOR); } - mExpectingUpdateSelection = true; } final int codePointBeforeCursor = mConnection.getCodePointBeforeCursor(); if (Character.isLetterOrDigit(codePointBeforeCursor) - || settingsValues.isUsuallyFollowedBySpace(codePointBeforeCursor)) { + || currentSettingsValues.isUsuallyFollowedBySpace(codePointBeforeCursor)) { + final boolean autoShiftHasBeenOverriden = mKeyboardSwitcher.getKeyboardShiftMode() != + getCurrentAutoCapsState(); mSpaceState = SPACE_STATE_PHANTOM; + if (!autoShiftHasBeenOverriden) { + // When we change the space state, we need to update the shift state of the + // keyboard unless it has been overridden manually. This is happening for example + // after typing some letters and a period, then gesturing; the keyboard is not in + // caps mode yet, but since a gesture is starting, it should go in caps mode, + // unless the user explictly said it should not. + mKeyboardSwitcher.updateShiftState(); + } } mConnection.endBatchEdit(); - mWordComposer.setCapitalizedModeAtStartComposingTime(getActualCapsMode()); + mWordComposer.setCapitalizedModeAndPreviousWordAtStartComposingTime(getActualCapsMode(), + // Prev word is 1st word before cursor + getNthPreviousWordForSuggestion(currentSettingsValues, 1 /* nthPreviousWord */)); } static final class InputUpdater implements Handler.Callback { @@ -1993,7 +1927,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen mConnection.commitText(commitParts[0], 0); mSpaceState = SPACE_STATE_PHANTOM; mKeyboardSwitcher.updateShiftState(); - mWordComposer.setCapitalizedModeAtStartComposingTime(getActualCapsMode()); + mWordComposer.setCapitalizedModeAndPreviousWordAtStartComposingTime( + getActualCapsMode(), commitParts[0]); ++mAutoCommitSequenceNumber; } } @@ -2003,8 +1938,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // This method must run in UI Thread. public void onEndBatchInputAsyncInternal(final SuggestedWords suggestedWords) { - final String batchInputText = suggestedWords.isEmpty() - ? null : suggestedWords.getWord(0); + final String batchInputText = suggestedWords.isEmpty() ? null : suggestedWords.getWord(0); if (TextUtils.isEmpty(batchInputText)) { return; } @@ -2026,7 +1960,6 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen mWordComposer.setBatchInputWord(batchInputText); mConnection.setComposingText(batchInputText, 1); } - mExpectingUpdateSelection = true; mConnection.endBatchEdit(); if (ProductionFlag.USES_DEVELOPMENT_ONLY_DIAGNOSTICS) { ResearchLogger.latinIME_onEndBatchInput(batchInputText, 0, suggestedWords); @@ -2080,9 +2013,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } private void handleBackspace(final int spaceState) { - // We revert these in this method if the deletion doesn't happen. mDeleteCount++; - mExpectingUpdateSelection = true; // In many cases, we may have to put the keyboard in auto-shift state again. However // we want to wait a few milliseconds before doing it to avoid the keyboard flashing @@ -2092,7 +2023,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen if (mWordComposer.isCursorFrontOrMiddleOfComposingWord()) { // If we are in the middle of a recorrection, we need to commit the recorrection // first so that we can remove the character at the current cursor position. - resetEntireInputState(mLastSelectionStart); + resetEntireInputState(mLastSelectionStart, mLastSelectionEnd); // When we exit this if-clause, mWordComposer.isComposingWord() will return false. } if (mWordComposer.isComposingWord()) { @@ -2163,6 +2094,10 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // later (typically, in a subsequent press on backspace). mLastSelectionEnd = mLastSelectionStart; mConnection.deleteSurroundingText(numCharsDeleted, 0); + if (ProductionFlag.USES_DEVELOPMENT_ONLY_DIAGNOSTICS) { + ResearchLogger.latinIME_handleBackspace(numCharsDeleted, + false /* shouldUncommitLogUnit */); + } } else { // There is no selection, just delete one character. if (NOT_A_CURSOR_POSITION == mLastSelectionEnd) { @@ -2185,22 +2120,27 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } else { final int codePointBeforeCursor = mConnection.getCodePointBeforeCursor(); if (codePointBeforeCursor == Constants.NOT_A_CODE) { - // Nothing to delete before the cursor. We have to revert the deletion - // states that were updated at the beginning of this method. - mDeleteCount--; - mExpectingUpdateSelection = false; + // Nothing to delete before the cursor. return; } final int lengthToDelete = Character.isSupplementaryCodePoint(codePointBeforeCursor) ? 2 : 1; mConnection.deleteSurroundingText(lengthToDelete, 0); + if (ProductionFlag.USES_DEVELOPMENT_ONLY_DIAGNOSTICS) { + ResearchLogger.latinIME_handleBackspace(lengthToDelete, + true /* shouldUncommitLogUnit */); + } if (mDeleteCount > DELETE_ACCELERATE_AT) { final int codePointBeforeCursorToDeleteAgain = mConnection.getCodePointBeforeCursor(); if (codePointBeforeCursorToDeleteAgain != Constants.NOT_A_CODE) { final int lengthToDeleteAgain = Character.isSupplementaryCodePoint( - codePointBeforeCursorToDeleteAgain) ? 2 : 1; + codePointBeforeCursorToDeleteAgain) ? 2 : 1; mConnection.deleteSurroundingText(lengthToDeleteAgain, 0); + if (ProductionFlag.USES_DEVELOPMENT_ONLY_DIAGNOSTICS) { + ResearchLogger.latinIME_handleBackspace(lengthToDeleteAgain, + true /* shouldUncommitLogUnit */); + } } } } @@ -2217,8 +2157,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen /* * Strip a trailing space if necessary and returns whether it's a swap weak space situation. */ - private boolean maybeStripSpace(final int code, - final int spaceState, final boolean isFromSuggestionStrip) { + private boolean maybeStripSpace(final int code, final int spaceState, + final boolean isFromSuggestionStrip) { if (Constants.CODE_ENTER == code && SPACE_STATE_SWAP_PUNCTUATION == spaceState) { mConnection.removeTrailingSpace(); return false; @@ -2233,8 +2173,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen return false; } - private void handleCharacter(final int primaryCode, final int x, - final int y, final int spaceState) { + private void handleCharacter(final int primaryCode, final int x, final int y, + final int spaceState) { // TODO: refactor this method to stop flipping isComposingWord around all the time, and // make it shorter (possibly cut into several pieces). Also factor handleNonSpecialCharacter // which has the same name as other handle* methods but is not the same. @@ -2254,7 +2194,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen if (mWordComposer.isCursorFrontOrMiddleOfComposingWord()) { // If we are in the middle of a recorrection, we need to commit the recorrection // first so that we can insert the character at the current cursor position. - resetEntireInputState(mLastSelectionStart); + resetEntireInputState(mLastSelectionStart, mLastSelectionEnd); isComposingWord = false; } // We want to find out whether to start composing a new word with this character. If so, @@ -2284,25 +2224,24 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen resetComposingState(false /* alsoResetLastComposedWord */); } if (isComposingWord) { - final int keyX, keyY; - if (Constants.isValidCoordinate(x) && Constants.isValidCoordinate(y)) { - final KeyDetector keyDetector = - mKeyboardSwitcher.getMainKeyboardView().getKeyDetector(); - keyX = keyDetector.getTouchX(x); - keyY = keyDetector.getTouchY(y); - } else { - keyX = x; - keyY = y; - } + final MainKeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); + // TODO: We should reconsider which coordinate system should be used to represent + // keyboard event. + final int keyX = mainKeyboardView.getKeyX(x); + final int keyY = mainKeyboardView.getKeyY(y); mWordComposer.add(primaryCode, keyX, keyY); // If it's the first letter, make note of auto-caps state if (mWordComposer.size() == 1) { - mWordComposer.setCapitalizedModeAtStartComposingTime(getActualCapsMode()); + // We pass 1 to getPreviousWordForSuggestion because we were not composing a word + // yet, so the word we want is the 1st word before the cursor. + mWordComposer.setCapitalizedModeAndPreviousWordAtStartComposingTime( + getActualCapsMode(), + getNthPreviousWordForSuggestion(currentSettings, 1 /* nthPreviousWord */)); } mConnection.setComposingText(getTextWithUnderline(mWordComposer.getTypedWord()), 1); } else { - final boolean swapWeakSpace = maybeStripSpace(primaryCode, - spaceState, Constants.SUGGESTION_STRIP_COORDINATE == x); + final boolean swapWeakSpace = maybeStripSpace(primaryCode, spaceState, + Constants.SUGGESTION_STRIP_COORDINATE == x); sendKeyCodePoint(primaryCode); @@ -2366,7 +2305,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen if (mWordComposer.isCursorFrontOrMiddleOfComposingWord()) { // If we are in the middle of a recorrection, we need to commit the recorrection // first so that we can insert the separator at the current cursor position. - resetEntireInputState(mLastSelectionStart); + resetEntireInputState(mLastSelectionStart, mLastSelectionEnd); } if (mWordComposer.isComposingWord()) { // May have changed since we stored wasComposing if (currentSettings.mCorrectionEnabled) { @@ -2539,6 +2478,24 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } } + /** + * Get the nth previous word before the cursor as context for the suggestion process. + * @param currentSettings the current settings values. + * @param nthPreviousWord reverse index of the word to get (1-indexed) + * @return the nth previous word before the cursor. + */ + private String getNthPreviousWordForSuggestion(final SettingsValues currentSettings, + final int nthPreviousWord) { + if (currentSettings.mCurrentLanguageHasSpaces) { + // If we are typing in a language with spaces we can just look up the previous + // word from textview. + return mConnection.getNthPreviousWord(currentSettings, nthPreviousWord); + } else { + return LastComposedWord.NOT_A_COMPOSED_WORD == mLastComposedWord ? null + : mLastComposedWord.mCommittedWord; + } + } + private void getSuggestedWords(final int sessionId, final int sequenceNumber, final OnGetSuggestedWordsCallback callback) { final Keyboard keyboard = mKeyboardSwitcher.getKeyboard(); @@ -2552,17 +2509,25 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // should just skip whitespace if any, so 1. final SettingsValues currentSettings = mSettings.getCurrent(); final int[] additionalFeaturesOptions = currentSettings.mAdditionalFeaturesSettingValues; - final String prevWord; - if (currentSettings.mCurrentLanguageHasSpaces) { - // If we are typing in a language with spaces we can just look up the previous - // word from textview. - prevWord = mConnection.getNthPreviousWord(currentSettings.mWordSeparators, - mWordComposer.isComposingWord() ? 2 : 1); - } else { - prevWord = LastComposedWord.NOT_A_COMPOSED_WORD == mLastComposedWord ? null - : mLastComposedWord.mCommittedWord; + + if (DEBUG) { + if (mWordComposer.isComposingWord() || mWordComposer.isBatchMode()) { + final String previousWord = mWordComposer.getPreviousWordForSuggestion(); + // TODO: this is for checking consistency with older versions. Remove this when + // we are confident this is stable. + // We're checking the previous word in the text field against the memorized previous + // word. If we are composing a word we should have the second word before the cursor + // memorized, otherwise we should have the first. + final String rereadPrevWord = getNthPreviousWordForSuggestion(currentSettings, + mWordComposer.isComposingWord() ? 2 : 1); + if (!TextUtils.equals(previousWord, rereadPrevWord)) { + throw new RuntimeException("Unexpected previous word: " + + previousWord + " <> " + rereadPrevWord); + } + } } - suggest.getSuggestedWords(mWordComposer, prevWord, keyboard.getProximityInfo(), + suggest.getSuggestedWords(mWordComposer, mWordComposer.getPreviousWordForSuggestion(), + keyboard.getProximityInfo(), currentSettings.mBlockPotentiallyOffensive, currentSettings.mCorrectionEnabled, additionalFeaturesOptions, sessionId, sequenceNumber, callback); } @@ -2679,7 +2644,6 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen ResearchLogger.latinIme_commitCurrentAutoCorrection(typedWord, autoCorrection, separator, mWordComposer.isBatchMode(), suggestedWords); } - mExpectingUpdateSelection = true; commitChosenWord(autoCorrection, LastComposedWord.COMMIT_TYPE_DECIDED_WORD, separator); if (!typedWord.equals(autoCorrection)) { @@ -2750,7 +2714,6 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // typed word. final String replacedWord = mWordComposer.getTypedWord(); LatinImeLogger.logOnManualSuggestion(replacedWord, suggestion, index, suggestedWords); - mExpectingUpdateSelection = true; commitChosenWord(suggestion, LastComposedWord.COMMIT_TYPE_MANUAL_PICK, LastComposedWord.NOT_A_SEPARATOR); if (ProductionFlag.USES_DEVELOPMENT_ONLY_DIAGNOSTICS) { @@ -2806,6 +2769,19 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // strings. mLastComposedWord = mWordComposer.commitWord(commitType, chosenWord, separatorString, prevWord); + final boolean shouldDiscardPreviousWordForSuggestion; + if (0 == StringUtils.codePointCount(separatorString)) { + // Separator is 0-length. Discard the word only if the current language has spaces. + shouldDiscardPreviousWordForSuggestion = + mSettings.getCurrent().mCurrentLanguageHasSpaces; + } else { + // Otherwise, we discard if the separator contains any non-whitespace. + shouldDiscardPreviousWordForSuggestion = + !StringUtils.containsOnlyWhitespace(separatorString); + } + if (shouldDiscardPreviousWordForSuggestion) { + mWordComposer.discardPreviousWordForSuggestion(); + } } private void setPunctuationSuggestions() { @@ -2830,10 +2806,10 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen final SettingsValues currentSettings = mSettings.getCurrent(); if (!currentSettings.mCorrectionEnabled) return null; - final UserHistoryDictionary userHistoryDictionary = mUserHistoryDictionary; + final UserHistoryDictionary userHistoryDictionary = suggest.getUserHistoryDictionary(); if (userHistoryDictionary == null) return null; - final String prevWord = mConnection.getNthPreviousWord(currentSettings.mWordSeparators, 2); + final String prevWord = mConnection.getNthPreviousWord(currentSettings, 2); final String secondWord; if (mWordComposer.wasAutoCapitalized() && !mWordComposer.isMostlyCaps()) { secondWord = suggestion.toLowerCase(mSubtypeSwitcher.getCurrentSubtypeLocale()); @@ -2845,7 +2821,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen final int maxFreq = AutoCorrectionUtils.getMaxFrequency( suggest.getUnigramDictionaries(), suggestion); if (maxFreq == 0) return null; - userHistoryDictionary.addToDictionary(prevWord, secondWord, maxFreq > 0); + userHistoryDictionary.addToDictionary(prevWord, secondWord, maxFreq > 0, + (int)TimeUnit.MILLISECONDS.toSeconds((System.currentTimeMillis()))); return prevWord; } @@ -2901,7 +2878,13 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } } } - mWordComposer.setComposingWord(typedWord, mKeyboardSwitcher.getKeyboard()); + mWordComposer.setComposingWord(typedWord, + getNthPreviousWordForSuggestion(currentSettings, + // We want the previous word for suggestion. If we have chars in the word + // before the cursor, then we want the word before that, hence 2; otherwise, + // we want the word immediately before the cursor, hence 1. + 0 == numberOfCharsInWordBeforeCursor ? 1 : 2), + mKeyboardSwitcher.getKeyboard()); mWordComposer.setCursorPositionWithinWord( typedWord.codePointCount(0, numberOfCharsInWordBeforeCursor)); mConnection.setComposingRegion( @@ -2979,7 +2962,11 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } private void restartSuggestionsOnWordBeforeCursor(final String word) { - mWordComposer.setComposingWord(word, mKeyboardSwitcher.getKeyboard()); + mWordComposer.setComposingWord(word, + // Previous word is the 2nd word before cursor because we are restarting on the + // 1st word before cursor. + getNthPreviousWordForSuggestion(mSettings.getCurrent(), 2 /* nthPreviousWord */), + mKeyboardSwitcher.getKeyboard()); final int length = word.length(); mConnection.deleteSurroundingText(length, 0); mConnection.setComposingText(word, 1); @@ -2997,7 +2984,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen * @param remainingTries How many times we may try again before giving up. */ private void retryResetCaches(final boolean tryResumeSuggestions, final int remainingTries) { - if (!mConnection.resetCachesUponCursorMoveAndReturnSuccess(mLastSelectionStart, false)) { + if (!mConnection.resetCachesUponCursorMoveAndReturnSuccess(mLastSelectionStart, + mLastSelectionEnd, false)) { if (0 < remainingTries) { mHandler.postResetCaches(tryResumeSuggestions, remainingTries - 1); return; @@ -3024,8 +3012,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen throw new RuntimeException("revertCommit, but we are composing a word"); } final CharSequence wordBeforeCursor = - mConnection.getTextBeforeCursor(deleteLength, 0) - .subSequence(0, cancelLength); + mConnection.getTextBeforeCursor(deleteLength, 0).subSequence(0, cancelLength); if (!TextUtils.equals(committedWord, wordBeforeCursor)) { throw new RuntimeException("revertCommit check failed: we thought we were " + "reverting \"" + committedWord @@ -3034,7 +3021,9 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } mConnection.deleteSurroundingText(deleteLength, 0); if (!TextUtils.isEmpty(previousWord) && !TextUtils.isEmpty(committedWord)) { - mUserHistoryDictionary.cancelAddingUserHistory(previousWord, committedWord); + if (mSuggest != null) { + mSuggest.cancelAddingUserHistory(previousWord, committedWord); + } } final String stringToCommit = originallyTypedWord + mLastComposedWord.mSeparatorString; if (mSettings.getCurrent().mCurrentLanguageHasSpaces) { @@ -3045,7 +3034,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } else { // For languages without spaces, we revert the typed string but the cursor is flush // with the typed word, so we need to resume suggestions right away. - mWordComposer.setComposingWord(stringToCommit, mKeyboardSwitcher.getKeyboard()); + mWordComposer.setComposingWord(stringToCommit, previousWord, + mKeyboardSwitcher.getKeyboard()); mConnection.setComposingText(stringToCommit, 1); } if (mSettings.isInternal()) { @@ -3095,8 +3085,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen private void hapticAndAudioFeedback(final int code, final int repeatCount) { final MainKeyboardView keyboardView = mKeyboardSwitcher.getMainKeyboardView(); - if (keyboardView != null && keyboardView.isInSlidingKeyInput()) { - // No need to feedback while sliding input. + if (keyboardView != null && keyboardView.isInDraggingFinger()) { + // No need to feedback while finger is dragging. return; } if (repeatCount > 0) { @@ -3226,8 +3216,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen final Intent intent = IntentUtils.getInputLanguageSelectionIntent( mRichImm.getInputMethodIdOfThisIme(), Intent.FLAG_ACTIVITY_NEW_TASK - | Intent.FLAG_ACTIVITY_RESET_TASK_IF_NEEDED - | Intent.FLAG_ACTIVITY_CLEAR_TOP); + | Intent.FLAG_ACTIVITY_RESET_TASK_IF_NEEDED + | Intent.FLAG_ACTIVITY_CLEAR_TOP); startActivity(intent); break; case 1: @@ -3236,9 +3226,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } } }; - final AlertDialog.Builder builder = new AlertDialog.Builder(this) - .setItems(items, listener) - .setTitle(title); + final AlertDialog.Builder builder = + new AlertDialog.Builder(this).setItems(items, listener).setTitle(title); showOptionDialog(builder.create()); } @@ -3299,11 +3288,13 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen final Printer p = new PrintWriterPrinter(fout); p.println("LatinIME state :"); + p.println(" VersionCode = " + ApplicationUtils.getVersionCode(this)); + p.println(" VersionName = " + ApplicationUtils.getVersionName(this)); final Keyboard keyboard = mKeyboardSwitcher.getKeyboard(); final int keyboardMode = keyboard != null ? keyboard.mId.mMode : -1; p.println(" Keyboard mode = " + keyboardMode); final SettingsValues settingsValues = mSettings.getCurrent(); - p.println(" mIsSuggestionsSuggestionsRequested = " + p.println(" mIsSuggestionsRequested = " + settingsValues.isSuggestionsRequested(mDisplayOrientation)); p.println(" mCorrectionEnabled=" + settingsValues.mCorrectionEnabled); p.println(" isComposingWord=" + mWordComposer.isComposingWord()); diff --git a/java/src/com/android/inputmethod/latin/RichInputConnection.java b/java/src/com/android/inputmethod/latin/RichInputConnection.java index 673d1b4c2..37311acf2 100644 --- a/java/src/com/android/inputmethod/latin/RichInputConnection.java +++ b/java/src/com/android/inputmethod/latin/RichInputConnection.java @@ -57,14 +57,19 @@ public final class RichInputConnection { private static final int INVALID_CURSOR_POSITION = -1; /** - * This variable contains an expected value for the cursor position. This is where the - * cursor may end up after all the keyboard-triggered updates have passed. We keep this to - * compare it to the actual cursor position to guess whether the move was caused by a - * keyboard command or not. - * It's not really the cursor position: the cursor may not be there yet, and it's also expected - * there be cases where it never actually comes to be there. + * This variable contains an expected value for the selection start position. This is where the + * cursor or selection start may end up after all the keyboard-triggered updates have passed. We + * keep this to compare it to the actual selection start to guess whether the move was caused by + * a keyboard command or not. + * It's not really the selection start position: the selection start may not be there yet, and + * in some cases, it may never arrive there. */ - private int mExpectedCursorPosition = INVALID_CURSOR_POSITION; // in chars, not code points + private int mExpectedSelStart = INVALID_CURSOR_POSITION; // in chars, not code points + /** + * The expected selection end. Only differs from mExpectedSelStart if a non-empty selection is + * expected. The same caveats as mExpectedSelStart apply. + */ + private int mExpectedSelEnd = INVALID_CURSOR_POSITION; // in chars, not code points /** * This contains the committed text immediately preceding the cursor and the composing * text if any. It is refreshed when the cursor moves by calling upon the TextView. @@ -103,16 +108,16 @@ public final class RichInputConnection { final String reference = (beforeCursor.length() <= actualLength) ? beforeCursor.toString() : beforeCursor.subSequence(beforeCursor.length() - actualLength, beforeCursor.length()).toString(); - if (et.selectionStart != mExpectedCursorPosition + if (et.selectionStart != mExpectedSelStart || !(reference.equals(internal.toString()))) { - final String context = "Expected cursor position = " + mExpectedCursorPosition - + "\nActual cursor position = " + et.selectionStart + final String context = "Expected selection start = " + mExpectedSelStart + + "\nActual selection start = " + et.selectionStart + "\nExpected text = " + internal.length() + " " + internal + "\nActual text = " + reference.length() + " " + reference; ((LatinIME)mParent).debugDumpStateAndCrashWithException(context); } else { Log.e(TAG, DebugLogUtils.getStackTrace(2)); - Log.e(TAG, "Exp <> Actual : " + mExpectedCursorPosition + " <> " + et.selectionStart); + Log.e(TAG, "Exp <> Actual : " + mExpectedSelStart + " <> " + et.selectionStart); } } @@ -150,16 +155,50 @@ public final class RichInputConnection { * data, so we empty the cache and note that we don't know the new cursor position, and we * return false so that the caller knows about this and can retry later. * - * @param newCursorPosition The new position of the cursor, as received from the system. - * @param shouldFinishComposition Whether we should finish the composition in progress. + * @param newSelStart the new position of the selection start, as received from the system. + * @param newSelEnd the new position of the selection end, as received from the system. + * @param shouldFinishComposition whether we should finish the composition in progress. * @return true if we were able to connect to the editor successfully, false otherwise. When * this method returns false, the caches could not be correctly refreshed so they were only * reset: the caller should try again later to return to normal operation. */ - public boolean resetCachesUponCursorMoveAndReturnSuccess(final int newCursorPosition, - final boolean shouldFinishComposition) { - mExpectedCursorPosition = newCursorPosition; + public boolean resetCachesUponCursorMoveAndReturnSuccess(final int newSelStart, + final int newSelEnd, final boolean shouldFinishComposition) { + mExpectedSelStart = newSelStart; + mExpectedSelEnd = newSelEnd; mComposingText.setLength(0); + final boolean didReloadTextSuccessfully = reloadTextCache(); + if (!didReloadTextSuccessfully) { + Log.d(TAG, "Will try to retrieve text later."); + return false; + } + final int lengthOfTextBeforeCursor = mCommittedTextBeforeComposingText.length(); + if (lengthOfTextBeforeCursor > newSelStart + || (lengthOfTextBeforeCursor < Constants.EDITOR_CONTENTS_CACHE_SIZE + && newSelStart < Constants.EDITOR_CONTENTS_CACHE_SIZE)) { + // newSelStart and newSelEnd may be lying -- when rotating the device (probably a + // framework bug). If we have less chars than we asked for, then we know how many chars + // we have, and if we got more than newSelStart says, then we know it was lying. In both + // cases the length is more reliable. Note that we only have to check newSelStart (not + // newSelEnd) since if newSelEnd is wrong, the newSelStart will be wrong as well. + mExpectedSelStart = lengthOfTextBeforeCursor; + mExpectedSelEnd = lengthOfTextBeforeCursor; + } + if (null != mIC && shouldFinishComposition) { + mIC.finishComposingText(); + if (ProductionFlag.USES_DEVELOPMENT_ONLY_DIAGNOSTICS) { + ResearchLogger.richInputConnection_finishComposingText(); + } + } + return true; + } + + /** + * Reload the cached text from the InputConnection. + * + * @return true if successful + */ + private boolean reloadTextCache() { mCommittedTextBeforeComposingText.setLength(0); mIC = mParent.getCurrentInputConnection(); // Call upon the inputconnection directly since our own method is using the cache, and @@ -169,27 +208,12 @@ public final class RichInputConnection { if (null == textBeforeCursor) { // For some reason the app thinks we are not connected to it. This looks like a // framework bug... Fall back to ground state and return false. - mExpectedCursorPosition = INVALID_CURSOR_POSITION; - Log.e(TAG, "Unable to connect to the editor to retrieve text... will retry later"); + mExpectedSelStart = INVALID_CURSOR_POSITION; + mExpectedSelEnd = INVALID_CURSOR_POSITION; + Log.e(TAG, "Unable to connect to the editor to retrieve text."); return false; } mCommittedTextBeforeComposingText.append(textBeforeCursor); - final int lengthOfTextBeforeCursor = textBeforeCursor.length(); - if (lengthOfTextBeforeCursor > newCursorPosition - || (lengthOfTextBeforeCursor < Constants.EDITOR_CONTENTS_CACHE_SIZE - && newCursorPosition < Constants.EDITOR_CONTENTS_CACHE_SIZE)) { - // newCursorPosition may be lying -- when rotating the device (probably a framework - // bug). If we have less chars than we asked for, then we know how many chars we have, - // and if we got more than newCursorPosition says, then we know it was lying. In both - // cases the length is more reliable - mExpectedCursorPosition = lengthOfTextBeforeCursor; - } - if (null != mIC && shouldFinishComposition) { - mIC.finishComposingText(); - if (ProductionFlag.USES_DEVELOPMENT_ONLY_DIAGNOSTICS) { - ResearchLogger.richInputConnection_finishComposingText(); - } - } return true; } @@ -218,7 +242,8 @@ public final class RichInputConnection { if (DEBUG_BATCH_NESTING) checkBatchEdit(); if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); mCommittedTextBeforeComposingText.append(text); - mExpectedCursorPosition += text.length() - mComposingText.length(); + mExpectedSelStart += text.length() - mComposingText.length(); + mExpectedSelEnd = mExpectedSelStart; mComposingText.setLength(0); if (null != mIC) { mIC.commitText(text, i); @@ -231,7 +256,7 @@ public final class RichInputConnection { } public boolean canDeleteCharacters() { - return mExpectedCursorPosition > 0; + return mExpectedSelStart > 0; } /** @@ -268,11 +293,10 @@ public final class RichInputConnection { // heavy pressing of delete, for example DEFAULT_TEXT_CACHE_SIZE - 5 times or so. // getCapsMode should be updated to be able to return a "not enough info" result so that // we can get more context only when needed. - if (TextUtils.isEmpty(mCommittedTextBeforeComposingText) && 0 != mExpectedCursorPosition) { - final CharSequence textBeforeCursor = getTextBeforeCursor( - Constants.EDITOR_CONTENTS_CACHE_SIZE, 0); - if (!TextUtils.isEmpty(textBeforeCursor)) { - mCommittedTextBeforeComposingText.append(textBeforeCursor); + if (TextUtils.isEmpty(mCommittedTextBeforeComposingText) && 0 != mExpectedSelStart) { + if (!reloadTextCache()) { + Log.w(TAG, "Unable to connect to the editor. " + + "Setting caps mode without knowing text."); } } // This never calls InputConnection#getCapsMode - in fact, it's a static method that @@ -295,8 +319,8 @@ public final class RichInputConnection { // However, if we don't have an expected cursor position, then we should always // go fetch the cache again (as it happens, INVALID_CURSOR_POSITION < 0, so we need to // test for this explicitly) - if (INVALID_CURSOR_POSITION != mExpectedCursorPosition - && (cachedLength >= n || cachedLength >= mExpectedCursorPosition)) { + if (INVALID_CURSOR_POSITION != mExpectedSelStart + && (cachedLength >= n || cachedLength >= mExpectedSelStart)) { final StringBuilder s = new StringBuilder(mCommittedTextBeforeComposingText); // We call #toString() here to create a temporary object. // In some situations, this method is called on a worker thread, and it's possible @@ -336,10 +360,14 @@ public final class RichInputConnection { + remainingChars, 0); mCommittedTextBeforeComposingText.setLength(len); } - if (mExpectedCursorPosition > beforeLength) { - mExpectedCursorPosition -= beforeLength; + if (mExpectedSelStart > beforeLength) { + mExpectedSelStart -= beforeLength; + mExpectedSelEnd -= beforeLength; } else { - mExpectedCursorPosition = 0; + // There are fewer characters before the cursor in the buffer than we are being asked to + // delete. Only delete what is there. + mExpectedSelStart = 0; + mExpectedSelEnd -= mExpectedSelStart; } if (null != mIC) { mIC.deleteSurroundingText(beforeLength, afterLength); @@ -373,7 +401,8 @@ public final class RichInputConnection { switch (keyEvent.getKeyCode()) { case KeyEvent.KEYCODE_ENTER: mCommittedTextBeforeComposingText.append("\n"); - mExpectedCursorPosition += 1; + mExpectedSelStart += 1; + mExpectedSelEnd = mExpectedSelStart; break; case KeyEvent.KEYCODE_DEL: if (0 == mComposingText.length()) { @@ -385,18 +414,24 @@ public final class RichInputConnection { } else { mComposingText.delete(mComposingText.length() - 1, mComposingText.length()); } - if (mExpectedCursorPosition > 0) mExpectedCursorPosition -= 1; + if (mExpectedSelStart > 0 && mExpectedSelStart == mExpectedSelEnd) { + // TODO: Handle surrogate pairs. + mExpectedSelStart -= 1; + } + mExpectedSelEnd = mExpectedSelStart; break; case KeyEvent.KEYCODE_UNKNOWN: if (null != keyEvent.getCharacters()) { mCommittedTextBeforeComposingText.append(keyEvent.getCharacters()); - mExpectedCursorPosition += keyEvent.getCharacters().length(); + mExpectedSelStart += keyEvent.getCharacters().length(); + mExpectedSelEnd = mExpectedSelStart; } break; default: - final String text = new String(new int[] { keyEvent.getUnicodeChar() }, 0, 1); + final String text = StringUtils.newSingleCodePointString(keyEvent.getUnicodeChar()); mCommittedTextBeforeComposingText.append(text); - mExpectedCursorPosition += text.length(); + mExpectedSelStart += text.length(); + mExpectedSelEnd = mExpectedSelStart; break; } } @@ -430,10 +465,12 @@ public final class RichInputConnection { public void setComposingText(final CharSequence text, final int newCursorPosition) { if (DEBUG_BATCH_NESTING) checkBatchEdit(); if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); - mExpectedCursorPosition += text.length() - mComposingText.length(); + mExpectedSelStart += text.length() - mComposingText.length(); + mExpectedSelEnd = mExpectedSelStart; mComposingText.setLength(0); mComposingText.append(text); - // TODO: support values of i != 1. At this time, this is never called with i != 1. + // TODO: support values of newCursorPosition != 1. At this time, this is never called with + // newCursorPosition != 1. if (null != mIC) { mIC.setComposingText(text, newCursorPosition); if (ProductionFlag.USES_DEVELOPMENT_ONLY_DIAGNOSTICS) { @@ -443,19 +480,31 @@ public final class RichInputConnection { if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); } - public void setSelection(final int start, final int end) { + /** + * Set the selection of the text editor. + * + * Calls through to {@link InputConnection#setSelection(int, int)}. + * + * @param start the character index where the selection should start. + * @param end the character index where the selection should end. + * @return Returns true on success, false if the input connection is no longer valid either when + * setting the selection or when retrieving the text cache at that point. + */ + public boolean setSelection(final int start, final int end) { if (DEBUG_BATCH_NESTING) checkBatchEdit(); if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); + mExpectedSelStart = start; + mExpectedSelEnd = end; if (null != mIC) { - mIC.setSelection(start, end); + final boolean isIcValid = mIC.setSelection(start, end); + if (!isIcValid) { + return false; + } if (ProductionFlag.USES_DEVELOPMENT_ONLY_DIAGNOSTICS) { ResearchLogger.richInputConnection_setSelection(start, end); } } - mExpectedCursorPosition = start; - mCommittedTextBeforeComposingText.setLength(0); - mCommittedTextBeforeComposingText.append( - getTextBeforeCursor(Constants.EDITOR_CONTENTS_CACHE_SIZE, 0)); + return reloadTextCache(); } public void commitCorrection(final CorrectionInfo correctionInfo) { @@ -476,7 +525,8 @@ public final class RichInputConnection { // text should never be null, but just in case, it's better to insert nothing than to crash if (null == text) text = ""; mCommittedTextBeforeComposingText.append(text); - mExpectedCursorPosition += text.length() - mComposingText.length(); + mExpectedSelStart += text.length() - mComposingText.length(); + mExpectedSelEnd = mExpectedSelStart; mComposingText.setLength(0); if (null != mIC) { mIC.commitCompletion(completionInfo); @@ -488,7 +538,7 @@ public final class RichInputConnection { } @SuppressWarnings("unused") - public String getNthPreviousWord(final String sentenceSeperators, final int n) { + public String getNthPreviousWord(final SettingsValues currentSettingsValues, final int n) { mIC = mParent.getCurrentInputConnection(); if (null == mIC) return null; final CharSequence prev = getTextBeforeCursor(LOOKBACK_CHARACTER_NUM, 0); @@ -507,7 +557,7 @@ public final class RichInputConnection { } } } - return getNthPreviousWord(prev, sentenceSeperators, n); + return getNthPreviousWord(prev, currentSettingsValues, n); } private static boolean isSeparator(int code, String sep) { @@ -531,7 +581,7 @@ public final class RichInputConnection { // (n = 2) "abc |" -> null // (n = 2) "abc. def|" -> null public static String getNthPreviousWord(final CharSequence prev, - final String sentenceSeperators, final int n) { + final SettingsValues currentSettingsValues, final int n) { if (prev == null) return null; final String[] w = spaceRegex.split(prev); @@ -543,7 +593,8 @@ public final class RichInputConnection { // If ends in a separator, return null final char lastChar = nthPrevWord.charAt(length - 1); - if (sentenceSeperators.contains(String.valueOf(lastChar))) return null; + if (currentSettingsValues.isWordSeparator(lastChar) + || currentSettingsValues.isWordConnector(lastChar)) return null; return nthPrevWord; } @@ -758,20 +809,30 @@ public final class RichInputConnection { * this update and not the ones in-between. This is almost impossible to achieve even trying * very very hard. * - * @param oldSelStart The value of the old cursor position in the update. - * @param newSelStart The value of the new cursor position in the update. + * @param oldSelStart The value of the old selection in the update. + * @param newSelStart The value of the new selection in the update. + * @param oldSelEnd The value of the old selection end in the update. + * @param newSelEnd The value of the new selection end in the update. * @return whether this is a belated expected update or not. */ - public boolean isBelatedExpectedUpdate(final int oldSelStart, final int newSelStart) { - // If this is an update that arrives at our expected position, it's a belated update. - if (newSelStart == mExpectedCursorPosition) return true; - // If this is an update that moves the cursor from our expected position, it must be - // an explicit move. - if (oldSelStart == mExpectedCursorPosition) return false; - // The following returns true if newSelStart is between oldSelStart and - // mCurrentCursorPosition. We assume that if the updated position is between the old - // position and the expected position, then it must be a belated update. - return (newSelStart - oldSelStart) * (mExpectedCursorPosition - newSelStart) >= 0; + public boolean isBelatedExpectedUpdate(final int oldSelStart, final int newSelStart, + final int oldSelEnd, final int newSelEnd) { + // This update is "belated" if we are expecting it. That is, mExpectedSelStart and + // mExpectedSelEnd match the new values that the TextView is updating TO. + if (mExpectedSelStart == newSelStart && mExpectedSelEnd == newSelEnd) return true; + // This update is not belated if mExpectedSelStart and mExpeectedSelend match the old + // values, and one of newSelStart or newSelEnd is updated to a different value. In this + // case, there is likely something other than the IME that has moved the selection endpoint + // to the new value. + if (mExpectedSelStart == oldSelStart && mExpectedSelEnd == oldSelEnd + && (oldSelStart != newSelStart || oldSelEnd != newSelEnd)) return false; + // If nether of the above two cases holds, then the system may be having trouble keeping up + // with updates. If 1) the selection is a cursor, 2) newSelStart is between oldSelStart + // and mExpectedSelStart, and 3) newSelEnd is between oldSelEnd and mExpectedSelEnd, then + // assume a belated update. + return (newSelStart == newSelEnd) + && (newSelStart - oldSelStart) * (mExpectedSelStart - newSelStart) >= 0 + && (newSelEnd - oldSelEnd) * (mExpectedSelEnd - newSelEnd) >= 0; } /** diff --git a/java/src/com/android/inputmethod/latin/SubtypeSwitcher.java b/java/src/com/android/inputmethod/latin/SubtypeSwitcher.java index cd9c89f04..379eaaa3d 100644 --- a/java/src/com/android/inputmethod/latin/SubtypeSwitcher.java +++ b/java/src/com/android/inputmethod/latin/SubtypeSwitcher.java @@ -32,7 +32,9 @@ import android.view.inputmethod.InputMethodManager; import android.view.inputmethod.InputMethodSubtype; import com.android.inputmethod.annotations.UsedForTesting; +import com.android.inputmethod.compat.InputMethodSubtypeCompatUtils; import com.android.inputmethod.keyboard.KeyboardSwitcher; +import com.android.inputmethod.latin.utils.LocaleUtils; import com.android.inputmethod.latin.utils.SubtypeLocaleUtils; import java.util.List; @@ -56,23 +58,34 @@ public final class SubtypeSwitcher { private InputMethodSubtype mEmojiSubtype; private boolean mIsNetworkConnected; + private static final String KEYBOARD_MODE = "keyboard"; // Dummy no language QWERTY subtype. See {@link R.xml.method}. - private static final InputMethodSubtype DUMMY_NO_LANGUAGE_SUBTYPE = new InputMethodSubtype( - R.string.subtype_no_language_qwerty, R.drawable.ic_ime_switcher_dark, - SubtypeLocaleUtils.NO_LANGUAGE, "keyboard", "KeyboardLayoutSet=" - + SubtypeLocaleUtils.QWERTY - + "," + Constants.Subtype.ExtraValue.ASCII_CAPABLE - + ",EnabledWhenDefaultIsNotAsciiCapable," - + Constants.Subtype.ExtraValue.EMOJI_CAPABLE, - false /* isAuxiliary */, false /* overridesImplicitlyEnabledSubtype */); + private static final int SUBTYPE_ID_OF_DUMMY_NO_LANGUAGE_SUBTYPE = 0xdde0bfd3; + private static final String EXTRA_VALUE_OF_DUMMY_NO_LANGUAGE_SUBTYPE = + "KeyboardLayoutSet=" + SubtypeLocaleUtils.QWERTY + + "," + Constants.Subtype.ExtraValue.ASCII_CAPABLE + + "," + Constants.Subtype.ExtraValue.ENABLED_WHEN_DEFAULT_IS_NOT_ASCII_CAPABLE + + "," + Constants.Subtype.ExtraValue.EMOJI_CAPABLE; + private static final InputMethodSubtype DUMMY_NO_LANGUAGE_SUBTYPE = + InputMethodSubtypeCompatUtils.newInputMethodSubtype( + R.string.subtype_no_language_qwerty, R.drawable.ic_ime_switcher_dark, + SubtypeLocaleUtils.NO_LANGUAGE, KEYBOARD_MODE, + EXTRA_VALUE_OF_DUMMY_NO_LANGUAGE_SUBTYPE, + false /* isAuxiliary */, false /* overridesImplicitlyEnabledSubtype */, + SUBTYPE_ID_OF_DUMMY_NO_LANGUAGE_SUBTYPE); // Caveat: We probably should remove this when we add an Emoji subtype in {@link R.xml.method}. // Dummy Emoji subtype. See {@link R.xml.method}. - private static final InputMethodSubtype DUMMY_EMOJI_SUBTYPE = new InputMethodSubtype( - R.string.subtype_emoji, R.drawable.ic_ime_switcher_dark, - SubtypeLocaleUtils.NO_LANGUAGE, "keyboard", "KeyboardLayoutSet=" - + SubtypeLocaleUtils.EMOJI + "," - + Constants.Subtype.ExtraValue.EMOJI_CAPABLE, - false /* isAuxiliary */, false /* overridesImplicitlyEnabledSubtype */); + private static final int SUBTYPE_ID_OF_DUMMY_EMOJI_SUBTYPE = 0xd78b2ed0; + private static final String EXTRA_VALUE_OF_DUMMY_EMOJI_SUBTYPE = + "KeyboardLayoutSet=" + SubtypeLocaleUtils.EMOJI + + "," + Constants.Subtype.ExtraValue.EMOJI_CAPABLE; + private static final InputMethodSubtype DUMMY_EMOJI_SUBTYPE = + InputMethodSubtypeCompatUtils.newInputMethodSubtype( + R.string.subtype_emoji, R.drawable.ic_ime_switcher_dark, + SubtypeLocaleUtils.NO_LANGUAGE, KEYBOARD_MODE, + EXTRA_VALUE_OF_DUMMY_EMOJI_SUBTYPE, + false /* isAuxiliary */, false /* overridesImplicitlyEnabledSubtype */, + SUBTYPE_ID_OF_DUMMY_EMOJI_SUBTYPE); static final class NeedsToDisplayLanguage { private int mEnabledSubtypeCount; @@ -256,18 +269,23 @@ public final class SubtypeSwitcher { return mNeedsToDisplayLanguage.getValue(); } - private static Locale sForcedLocaleForTesting = null; + private static InputMethodSubtype sForcedSubtypeForTesting = null; @UsedForTesting - void forceLocale(final Locale locale) { - sForcedLocaleForTesting = locale; + void forceSubtype(final InputMethodSubtype subtype) { + sForcedSubtypeForTesting = subtype; } public Locale getCurrentSubtypeLocale() { - if (null != sForcedLocaleForTesting) return sForcedLocaleForTesting; + if (null != sForcedSubtypeForTesting) { + return LocaleUtils.constructLocaleFromString(sForcedSubtypeForTesting.getLocale()); + } return SubtypeLocaleUtils.getSubtypeLocale(getCurrentSubtype()); } public InputMethodSubtype getCurrentSubtype() { + if (null != sForcedSubtypeForTesting) { + return sForcedSubtypeForTesting; + } return mRichImm.getCurrentInputMethodSubtype(getNoLanguageSubtype()); } @@ -279,8 +297,8 @@ public final class SubtypeSwitcher { if (mNoLanguageSubtype != null) { return mNoLanguageSubtype; } - Log.w(TAG, "Can't find no lanugage with QWERTY subtype"); - Log.w(TAG, "No input method subtype found; return dummy subtype: " + Log.w(TAG, "Can't find any language with QWERTY subtype"); + Log.w(TAG, "No input method subtype found; returning dummy subtype: " + DUMMY_NO_LANGUAGE_SUBTYPE); return DUMMY_NO_LANGUAGE_SUBTYPE; } @@ -293,8 +311,9 @@ public final class SubtypeSwitcher { if (mEmojiSubtype != null) { return mEmojiSubtype; } - Log.w(TAG, "Can't find Emoji subtype"); - Log.w(TAG, "No input method subtype found; return dummy subtype: " + DUMMY_EMOJI_SUBTYPE); + Log.w(TAG, "Can't find emoji subtype"); + Log.w(TAG, "No input method subtype found; returning dummy subtype: " + + DUMMY_EMOJI_SUBTYPE); return DUMMY_EMOJI_SUBTYPE; } } diff --git a/java/src/com/android/inputmethod/latin/Suggest.java b/java/src/com/android/inputmethod/latin/Suggest.java index 0a4c7a55d..32ab1f3df 100644 --- a/java/src/com/android/inputmethod/latin/Suggest.java +++ b/java/src/com/android/inputmethod/latin/Suggest.java @@ -25,9 +25,10 @@ import com.android.inputmethod.annotations.UsedForTesting; import com.android.inputmethod.keyboard.ProximityInfo; import com.android.inputmethod.latin.SuggestedWords.SuggestedWordInfo; import com.android.inputmethod.latin.personalization.PersonalizationDictionary; -import com.android.inputmethod.latin.personalization.PersonalizationPredictionDictionary; +import com.android.inputmethod.latin.personalization.PersonalizationHelper; import com.android.inputmethod.latin.personalization.UserHistoryDictionary; import com.android.inputmethod.latin.settings.Settings; +import com.android.inputmethod.latin.settings.SettingsValues; import com.android.inputmethod.latin.utils.AutoCorrectionUtils; import com.android.inputmethod.latin.utils.BoundedTreeSet; import com.android.inputmethod.latin.utils.CollectionUtils; @@ -72,7 +73,9 @@ public final class Suggest { CollectionUtils.newConcurrentHashMap(); private HashSet<String> mOnlyDictionarySetForDebug = null; private Dictionary mMainDictionary; - private ContactsBinaryDictionary mContactsDict; + private ContactsBinaryDictionary mContactsDictionary; + private UserHistoryDictionary mUserHistoryDictionary; + private PersonalizationDictionary mPersonalizationDictionary; @UsedForTesting private boolean mIsCurrentlyWaitingForMainDictionary = false; @@ -81,24 +84,28 @@ public final class Suggest { // Locale used for upper- and title-casing words public final Locale mLocale; + private final Context mContext; + public Suggest(final Context context, final Locale locale, final SuggestInitializationListener listener) { initAsynchronously(context, locale, listener); mLocale = locale; + mContext = context; + // TODO: Use SettingsValues instead of Settings. // initialize a debug flag for the personalization if (Settings.readUseOnlyPersonalizationDictionaryForDebug( PreferenceManager.getDefaultSharedPreferences(context))) { mOnlyDictionarySetForDebug = new HashSet<String>(); mOnlyDictionarySetForDebug.add(Dictionary.TYPE_PERSONALIZATION); - mOnlyDictionarySetForDebug.add(Dictionary.TYPE_PERSONALIZATION_PREDICTION_IN_JAVA); } } @UsedForTesting - Suggest(final AssetFileAddress[] dictionaryList, final Locale locale) { + Suggest(final Context context, final AssetFileAddress[] dictionaryList, final Locale locale) { final Dictionary mainDict = DictionaryFactory.createDictionaryForTest(dictionaryList, false /* useFullEditDistance */, locale); mLocale = locale; + mContext = context; mMainDictionary = mainDict; addOrReplaceDictionaryInternal(Dictionary.TYPE_MAIN, mainDict); } @@ -165,7 +172,15 @@ public final class Suggest { } public ContactsBinaryDictionary getContactsDictionary() { - return mContactsDict; + return mContactsDictionary; + } + + public UserHistoryDictionary getUserHistoryDictionary() { + return mUserHistoryDictionary; + } + + public PersonalizationDictionary getPersonalizationDictionary() { + return mPersonalizationDictionary; } public ConcurrentHashMap<String, Dictionary> getUnigramDictionaries() { @@ -186,24 +201,120 @@ public final class Suggest { * won't be used. */ public void setContactsDictionary(final ContactsBinaryDictionary contactsDictionary) { - mContactsDict = contactsDictionary; + mContactsDictionary = contactsDictionary; addOrReplaceDictionaryInternal(Dictionary.TYPE_CONTACTS, contactsDictionary); } public void setUserHistoryDictionary(final UserHistoryDictionary userHistoryDictionary) { + mUserHistoryDictionary = userHistoryDictionary; addOrReplaceDictionaryInternal(Dictionary.TYPE_USER_HISTORY, userHistoryDictionary); } - public void setPersonalizationPredictionDictionary( - final PersonalizationPredictionDictionary personalizationPredictionDictionary) { - addOrReplaceDictionaryInternal(Dictionary.TYPE_PERSONALIZATION_PREDICTION_IN_JAVA, - personalizationPredictionDictionary); - } - public void setPersonalizationDictionary( final PersonalizationDictionary personalizationDictionary) { - addOrReplaceDictionaryInternal(Dictionary.TYPE_PERSONALIZATION, - personalizationDictionary); + mPersonalizationDictionary = personalizationDictionary; + addOrReplaceDictionaryInternal(Dictionary.TYPE_PERSONALIZATION, personalizationDictionary); + } + + /** + * Set dictionaries that can be turned off according to the user settings. + * + * @param oldSuggest the instance having old dictionaries + * @param settingsValues current SettingsValues + */ + public void setAdditionalDictionaries(final Suggest oldSuggest, + final SettingsValues settingsValues) { + // Contacts dictionary + resetContactsDictionary(null != oldSuggest ? oldSuggest.getContactsDictionary() : null, + settingsValues); + // User history dictionary & Personalization dictionary + resetPersonalizedDictionaries(oldSuggest, settingsValues); + } + + /** + * Set the user history dictionary and personalization dictionary according to the user + * settings. + * + * @param oldSuggest the instance that has been used + * @param settingsValues current settingsValues + */ + // TODO: Consolidate resetPersonalizedDictionaries() and resetContactsDictionary(). Call up the + // new method for each dictionary. + private void resetPersonalizedDictionaries(final Suggest oldSuggest, + final SettingsValues settingsValues) { + final boolean shouldSetDictionaries = settingsValues.mUsePersonalizedDicts; + + final UserHistoryDictionary oldUserHistoryDictionary = (null == oldSuggest) ? null : + oldSuggest.getUserHistoryDictionary(); + final PersonalizationDictionary oldPersonalizationDictionary = (null == oldSuggest) ? null : + oldSuggest.getPersonalizationDictionary(); + final UserHistoryDictionary userHistoryDictionaryToUse; + final PersonalizationDictionary personalizationDictionaryToUse; + if (!shouldSetDictionaries) { + userHistoryDictionaryToUse = null; + personalizationDictionaryToUse = null; + } else { + if (null != oldUserHistoryDictionary + && oldUserHistoryDictionary.mLocale.equals(mLocale)) { + userHistoryDictionaryToUse = oldUserHistoryDictionary; + } else { + userHistoryDictionaryToUse = + PersonalizationHelper.getUserHistoryDictionary(mContext, mLocale); + } + if (null != oldPersonalizationDictionary + && oldPersonalizationDictionary.mLocale.equals(mLocale)) { + personalizationDictionaryToUse = oldPersonalizationDictionary; + } else { + personalizationDictionaryToUse = + PersonalizationHelper.getPersonalizationDictionary(mContext, mLocale); + } + } + setUserHistoryDictionary(userHistoryDictionaryToUse); + setPersonalizationDictionary(personalizationDictionaryToUse); + } + + /** + * Set the contacts dictionary according to the user settings. + * + * This method takes an optional contacts dictionary to use when the locale hasn't changed + * since the contacts dictionary can be opened or closed as necessary depending on the settings. + * + * @param oldContactsDictionary an optional dictionary to use, or null + * @param settingsValues current settingsValues + */ + private void resetContactsDictionary(final ContactsBinaryDictionary oldContactsDictionary, + final SettingsValues settingsValues) { + final boolean shouldSetDictionary = settingsValues.mUseContactsDict; + final ContactsBinaryDictionary dictionaryToUse; + if (!shouldSetDictionary) { + // Make sure the dictionary is closed. If it is already closed, this is a no-op, + // so it's safe to call it anyways. + if (null != oldContactsDictionary) oldContactsDictionary.close(); + dictionaryToUse = null; + } else { + if (null != oldContactsDictionary) { + if (!oldContactsDictionary.mLocale.equals(mLocale)) { + // If the locale has changed then recreate the contacts dictionary. This + // allows locale dependent rules for handling bigram name predictions. + oldContactsDictionary.close(); + dictionaryToUse = new ContactsBinaryDictionary(mContext, mLocale); + } else { + // Make sure the old contacts dictionary is opened. If it is already open, + // this is a no-op, so it's safe to call it anyways. + oldContactsDictionary.reopen(mContext); + dictionaryToUse = oldContactsDictionary; + } + } else { + dictionaryToUse = new ContactsBinaryDictionary(mContext, mLocale); + } + } + setContactsDictionary(dictionaryToUse); + } + + public void cancelAddingUserHistory(final String previousWord, final String committedWord) { + if (mUserHistoryDictionary != null) { + mUserHistoryDictionary.cancelAddingUserHistory(previousWord, committedWord); + } } public void setAutoCorrectionThreshold(float threshold) { @@ -432,7 +543,8 @@ public final class Suggest { final String scoreInfoString; if (normalizedScore > 0) { scoreInfoString = String.format( - Locale.ROOT, "%d (%4.2f)", cur.mScore, normalizedScore); + Locale.ROOT, "%d (%4.2f), %s", cur.mScore, normalizedScore, + cur.mSourceDict.mDictType); } else { scoreInfoString = Integer.toString(cur.mScore); } diff --git a/java/src/com/android/inputmethod/latin/SynchronouslyLoadedUserBinaryDictionary.java b/java/src/com/android/inputmethod/latin/SynchronouslyLoadedUserBinaryDictionary.java index 6405b5e46..9cb2f5bc4 100644 --- a/java/src/com/android/inputmethod/latin/SynchronouslyLoadedUserBinaryDictionary.java +++ b/java/src/com/android/inputmethod/latin/SynchronouslyLoadedUserBinaryDictionary.java @@ -22,14 +22,15 @@ import com.android.inputmethod.keyboard.ProximityInfo; import com.android.inputmethod.latin.SuggestedWords.SuggestedWordInfo; import java.util.ArrayList; +import java.util.Locale; public final class SynchronouslyLoadedUserBinaryDictionary extends UserBinaryDictionary { - public SynchronouslyLoadedUserBinaryDictionary(final Context context, final String locale) { + public SynchronouslyLoadedUserBinaryDictionary(final Context context, final Locale locale) { this(context, locale, false); } - public SynchronouslyLoadedUserBinaryDictionary(final Context context, final String locale, + public SynchronouslyLoadedUserBinaryDictionary(final Context context, final Locale locale, final boolean alsoUseMoreRestrictiveLocales) { super(context, locale, alsoUseMoreRestrictiveLocales); } diff --git a/java/src/com/android/inputmethod/latin/UserBinaryDictionary.java b/java/src/com/android/inputmethod/latin/UserBinaryDictionary.java index 15b3d8d02..6fc98220f 100644 --- a/java/src/com/android/inputmethod/latin/UserBinaryDictionary.java +++ b/java/src/com/android/inputmethod/latin/UserBinaryDictionary.java @@ -18,7 +18,6 @@ package com.android.inputmethod.latin; import android.content.ContentProviderClient; import android.content.ContentResolver; -import android.content.ContentUris; import android.content.Context; import android.database.ContentObserver; import android.database.Cursor; @@ -33,6 +32,7 @@ import com.android.inputmethod.compat.UserDictionaryCompatUtils; import com.android.inputmethod.latin.utils.LocaleUtils; import com.android.inputmethod.latin.utils.SubtypeLocaleUtils; +import java.io.File; import java.util.Arrays; import java.util.Locale; @@ -75,20 +75,26 @@ public class UserBinaryDictionary extends ExpandableBinaryDictionary { final private String mLocale; final private boolean mAlsoUseMoreRestrictiveLocales; - public UserBinaryDictionary(final Context context, final String locale) { + public UserBinaryDictionary(final Context context, final Locale locale) { this(context, locale, false); } - public UserBinaryDictionary(final Context context, final String locale, + // Dummy constructor for tests. + public UserBinaryDictionary(final Context context, final Locale locale, final File file) { + this(context, locale); + } + + public UserBinaryDictionary(final Context context, final Locale locale, final boolean alsoUseMoreRestrictiveLocales) { - super(context, getFilenameWithLocale(NAME, locale), Dictionary.TYPE_USER, + super(context, getDictNameWithLocale(NAME, locale), locale, Dictionary.TYPE_USER, false /* isUpdatable */); if (null == locale) throw new NullPointerException(); // Catch the error earlier - if (SubtypeLocaleUtils.NO_LANGUAGE.equals(locale)) { + final String localeStr = locale.toString(); + if (SubtypeLocaleUtils.NO_LANGUAGE.equals(localeStr)) { // If we don't have a locale, insert into the "all locales" user dictionary. mLocale = USER_DICTIONARY_ALL_LANGUAGES; } else { - mLocale = locale; + mLocale = localeStr; } mAlsoUseMoreRestrictiveLocales = alsoUseMoreRestrictiveLocales; // Perform a managed query. The Activity will handle closing and re-querying the cursor diff --git a/java/src/com/android/inputmethod/latin/WordComposer.java b/java/src/com/android/inputmethod/latin/WordComposer.java index 039dadc66..7da97e57a 100644 --- a/java/src/com/android/inputmethod/latin/WordComposer.java +++ b/java/src/com/android/inputmethod/latin/WordComposer.java @@ -48,6 +48,11 @@ public final class WordComposer { // at any given time. However this is not limited in size, while mPrimaryKeyCodes is limited // to MAX_WORD_LENGTH code points. private final StringBuilder mTypedWord; + // The previous word (before the composing word). Used as context for suggestions. May be null + // after resetting and before starting a new composing word, or when there is no context like + // at the start of text for example. It can also be set to null externally when the user + // enters a separator that does not let bigrams across, like a period or a comma. + private String mPreviousWordForSuggestion; private String mAutoCorrection; private boolean mIsResumed; private boolean mIsBatchMode; @@ -85,6 +90,7 @@ public final class WordComposer { mIsBatchMode = false; mCursorPositionWithinWord = 0; mRejectedBatchModeSuggestion = null; + mPreviousWordForSuggestion = null; refreshSize(); } @@ -101,6 +107,7 @@ public final class WordComposer { mIsBatchMode = source.mIsBatchMode; mCursorPositionWithinWord = source.mCursorPositionWithinWord; mRejectedBatchModeSuggestion = source.mRejectedBatchModeSuggestion; + mPreviousWordForSuggestion = source.mPreviousWordForSuggestion; refreshSize(); } @@ -118,6 +125,7 @@ public final class WordComposer { mIsBatchMode = false; mCursorPositionWithinWord = 0; mRejectedBatchModeSuggestion = null; + mPreviousWordForSuggestion = null; refreshSize(); } @@ -284,8 +292,13 @@ public final class WordComposer { /** * Set the currently composing word to the one passed as an argument. * This will register NOT_A_COORDINATE for X and Ys, and use the passed keyboard for proximity. + * @param word the char sequence to set as the composing word. + * @param previousWord the previous word, to use as context for suggestions. Can be null if + * the context is nil (typically, at start of text). + * @param keyboard the keyboard this is typed on, for coordinate info/proximity. */ - public void setComposingWord(final CharSequence word, final Keyboard keyboard) { + public void setComposingWord(final CharSequence word, final String previousWord, + final Keyboard keyboard) { reset(); final int length = word.length(); for (int i = 0; i < length; i = Character.offsetByCodePoints(word, i, 1)) { @@ -293,6 +306,7 @@ public final class WordComposer { addKeyInfo(codePoint, keyboard); } mIsResumed = true; + mPreviousWordForSuggestion = previousWord; } /** @@ -343,6 +357,10 @@ public final class WordComposer { return mTypedWord.toString(); } + public String getPreviousWordForSuggestion() { + return mPreviousWordForSuggestion; + } + /** * Whether or not the user typed a capital letter as the first letter in the word * @return capitalization preference @@ -388,18 +406,21 @@ public final class WordComposer { } /** - * Saves the caps mode at the start of composing. + * Saves the caps mode and the previous word at the start of composing. * - * WordComposer needs to know about this for several reasons. The first is, we need to know - * after the fact what the reason was, to register the correct form into the user history - * dictionary: if the word was automatically capitalized, we should insert it in all-lower - * case but if it's a manual pressing of shift, then it should be inserted as is. + * WordComposer needs to know about the caps mode for several reasons. The first is, we need + * to know after the fact what the reason was, to register the correct form into the user + * history dictionary: if the word was automatically capitalized, we should insert it in + * all-lower case but if it's a manual pressing of shift, then it should be inserted as is. * Also, batch input needs to know about the current caps mode to display correctly * capitalized suggestions. * @param mode the mode at the time of start + * @param previousWord the previous word as context for suggestions. May be null if none. */ - public void setCapitalizedModeAtStartComposingTime(final int mode) { + public void setCapitalizedModeAndPreviousWordAtStartComposingTime(final int mode, + final String previousWord) { mCapitalizedMode = mode; + mPreviousWordForSuggestion = previousWord; } /** @@ -451,6 +472,7 @@ public final class WordComposer { mCapsCount = 0; mDigitsCount = 0; mIsBatchMode = false; + mPreviousWordForSuggestion = mTypedWord.toString(); mTypedWord.setLength(0); mCodePointSize = 0; mTrailingSingleQuotesCount = 0; @@ -464,7 +486,15 @@ public final class WordComposer { return lastComposedWord; } - public void resumeSuggestionOnLastComposedWord(final LastComposedWord lastComposedWord) { + // Call this when the recorded previous word should be discarded. This is typically called + // when the user inputs a separator that's not whitespace (including the case of the + // double-space-to-period feature). + public void discardPreviousWordForSuggestion() { + mPreviousWordForSuggestion = null; + } + + public void resumeSuggestionOnLastComposedWord(final LastComposedWord lastComposedWord, + final String previousWord) { mPrimaryKeyCodes = lastComposedWord.mPrimaryKeyCodes; mInputPointers.set(lastComposedWord.mInputPointers); mTypedWord.setLength(0); @@ -475,6 +505,7 @@ public final class WordComposer { mCursorPositionWithinWord = mCodePointSize; mRejectedBatchModeSuggestion = null; mIsResumed = true; + mPreviousWordForSuggestion = previousWord; } public boolean isBatchMode() { diff --git a/java/src/com/android/inputmethod/latin/makedict/AbstractDictDecoder.java b/java/src/com/android/inputmethod/latin/makedict/AbstractDictDecoder.java index fda97dafc..f8fa68f45 100644 --- a/java/src/com/android/inputmethod/latin/makedict/AbstractDictDecoder.java +++ b/java/src/com/android/inputmethod/latin/makedict/AbstractDictDecoder.java @@ -32,36 +32,35 @@ import java.util.TreeMap; * A base class of the binary dictionary decoder. */ public abstract class AbstractDictDecoder implements DictDecoder { - protected FileHeader readHeader(final DictBuffer dictBuffer) + private static final int SUCCESS = 0; + private static final int ERROR_CANNOT_READ = 1; + private static final int ERROR_WRONG_FORMAT = 2; + + protected FileHeader readHeader(final DictBuffer headerBuffer) throws IOException, UnsupportedFormatException { - if (dictBuffer == null) { + if (headerBuffer == null) { openDictBuffer(); } - final int version = HeaderReader.readVersion(dictBuffer); + final int version = HeaderReader.readVersion(headerBuffer); if (version < FormatSpec.MINIMUM_SUPPORTED_VERSION || version > FormatSpec.MAXIMUM_SUPPORTED_VERSION) { throw new UnsupportedFormatException("Unsupported version : " + version); } // TODO: Remove this field. - final int optionsFlags = HeaderReader.readOptionFlags(dictBuffer); - - final int headerSize = HeaderReader.readHeaderSize(dictBuffer); - + final int optionsFlags = HeaderReader.readOptionFlags(headerBuffer); + final int headerSize = HeaderReader.readHeaderSize(headerBuffer); if (headerSize < 0) { throw new UnsupportedFormatException("header size can't be negative."); } - final HashMap<String, String> attributes = HeaderReader.readAttributes(dictBuffer, + final HashMap<String, String> attributes = HeaderReader.readAttributes(headerBuffer, headerSize); final FileHeader header = new FileHeader(headerSize, - new FusionDictionary.DictionaryOptions(attributes, - 0 != (optionsFlags & FormatSpec.GERMAN_UMLAUT_PROCESSING_FLAG), - 0 != (optionsFlags & FormatSpec.FRENCH_LIGATURE_PROCESSING_FLAG)), - new FormatOptions(version, - 0 != (optionsFlags & FormatSpec.SUPPORTS_DYNAMIC_UPDATE), - 0 != (optionsFlags & FormatSpec.CONTAINS_TIMESTAMP_FLAG))); + new FusionDictionary.DictionaryOptions(attributes), + new FormatOptions(version, + 0 != (optionsFlags & FormatSpec.CONTAINS_TIMESTAMP_FLAG))); return header; } @@ -204,4 +203,25 @@ public abstract class AbstractDictDecoder implements DictDecoder { return readLength; } } + + /** + * Check whether the header contains the expected information. This is a no-error method, + * that will return an error code and never throw a checked exception. + * @return an error code, either ERROR_* or SUCCESS. + */ + private int checkHeader() { + try { + readHeader(); + } catch (IOException e) { + return ERROR_CANNOT_READ; + } catch (UnsupportedFormatException e) { + return ERROR_WRONG_FORMAT; + } + return SUCCESS; + } + + @Override + public boolean hasValidRawBinaryDictionary() { + return checkHeader() == SUCCESS; + } } diff --git a/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java b/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java index 216492b4d..2dbb5eb93 100644 --- a/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java +++ b/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java @@ -24,12 +24,9 @@ import com.android.inputmethod.latin.makedict.FusionDictionary.PtNodeArray; import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; import java.io.File; -import java.io.FileInputStream; -import java.io.FileNotFoundException; import java.io.IOException; import java.io.OutputStream; import java.nio.ByteBuffer; -import java.nio.channels.FileChannel; import java.util.ArrayList; import java.util.Map; import java.util.TreeMap; @@ -61,6 +58,7 @@ public final class BinaryDictDecoderUtils { public int readInt(); public int position(); public void position(int newPosition); + @UsedForTesting public void put(final byte b); public int limit(); @UsedForTesting @@ -169,6 +167,15 @@ public final class BinaryDictDecoderUtils { return size; } + @UsedForTesting + static int getCharArraySize(final int[] chars, final int start, final int end) { + int size = 0; + for (int i = start; i < end; ++i) { + size += getCharSize(chars[i]); + } + return size; + } + /** * Writes a char array to a byte buffer. * @@ -200,8 +207,7 @@ public final class BinaryDictDecoderUtils { * @param word the string to write. * @return the size written, in bytes. */ - static int writeString(final byte[] buffer, final int origin, - final String word) { + static int writeString(final byte[] buffer, final int origin, final String word) { final int length = word.length(); int index = origin; for (int i = 0; i < length; i = word.offsetByCodePoints(i, 1)) { @@ -223,22 +229,63 @@ public final class BinaryDictDecoderUtils { * * This will also write the terminator byte. * - * @param buffer the OutputStream to write to. + * @param stream the OutputStream to write to. * @param word the string to write. + * @return the size written, in bytes. */ - static void writeString(final OutputStream buffer, final String word) throws IOException { + static int writeString(final OutputStream stream, final String word) throws IOException { final int length = word.length(); + int written = 0; for (int i = 0; i < length; i = word.offsetByCodePoints(i, 1)) { final int codePoint = word.codePointAt(i); - if (1 == getCharSize(codePoint)) { - buffer.write((byte) codePoint); + final int charSize = getCharSize(codePoint); + if (1 == charSize) { + stream.write((byte) codePoint); + } else { + stream.write((byte) (0xFF & (codePoint >> 16))); + stream.write((byte) (0xFF & (codePoint >> 8))); + stream.write((byte) (0xFF & codePoint)); + } + written += charSize; + } + stream.write(FormatSpec.PTNODE_CHARACTERS_TERMINATOR); + written += FormatSpec.PTNODE_TERMINATOR_SIZE; + return written; + } + + /** + * Writes an array of code points with our character format to an OutputStream. + * + * This will also write the terminator byte. + * + * @param stream the OutputStream to write to. + * @param codePoints the array of code points + * @return the size written, in bytes. + */ + // TODO: Merge this method with writeCharArray and rename the various write* methods to + // make the difference clear. + @UsedForTesting + static int writeCodePoints(final OutputStream stream, final int[] codePoints, + final int startIndex, final int endIndex) + throws IOException { + int written = 0; + for (int i = startIndex; i < endIndex; ++i) { + final int codePoint = codePoints[i]; + final int charSize = getCharSize(codePoint); + if (1 == charSize) { + stream.write((byte) codePoint); } else { - buffer.write((byte) (0xFF & (codePoint >> 16))); - buffer.write((byte) (0xFF & (codePoint >> 8))); - buffer.write((byte) (0xFF & codePoint)); + stream.write((byte) (0xFF & (codePoint >> 16))); + stream.write((byte) (0xFF & (codePoint >> 8))); + stream.write((byte) (0xFF & codePoint)); } + written += charSize; + } + if (endIndex - startIndex > 1) { + stream.write(FormatSpec.PTNODE_CHARACTERS_TERMINATOR); + written += FormatSpec.PTNODE_TERMINATOR_SIZE; } - buffer.write(FormatSpec.PTNODE_CHARACTERS_TERMINATOR); + return written; } /** @@ -286,7 +333,7 @@ public final class BinaryDictDecoderUtils { static int readChildrenAddress(final DictBuffer dictBuffer, final int optionFlags, final FormatOptions options) { - if (options.mSupportsDynamicUpdate) { + if (options.supportsDynamicUpdate()) { final int address = dictBuffer.readUnsignedInt24(); if (address == 0) return FormatSpec.NO_CHILDREN_ADDRESS; if ((address & FormatSpec.MSB24) != 0) { @@ -392,7 +439,7 @@ public final class BinaryDictDecoderUtils { final FormatOptions options) { dictDecoder.setPosition(headerSize); final int count = dictDecoder.readPtNodeCount(); - int groupPos = headerSize + BinaryDictIOUtils.getPtNodeCountSize(count); + int groupPos = dictDecoder.getPosition(); final StringBuilder builder = new StringBuilder(); WeightedString result = null; @@ -454,9 +501,9 @@ public final class BinaryDictDecoderUtils { do { // Scan the linked-list node. final int nodeArrayHeadPos = dictDecoder.getPosition(); final int count = dictDecoder.readPtNodeCount(); - int groupOffsetPos = nodeArrayHeadPos + BinaryDictIOUtils.getPtNodeCountSize(count); + int groupPos = dictDecoder.getPosition(); for (int i = count; i > 0; --i) { // Scan the array of PtNode. - PtNodeInfo info = dictDecoder.readPtNode(groupOffsetPos, options); + PtNodeInfo info = dictDecoder.readPtNode(groupPos, options); if (BinaryDictIOUtils.isMovedPtNode(info.mFlags, options)) continue; ArrayList<WeightedString> shortcutTargets = info.mShortcutTargets; ArrayList<WeightedString> bigrams = null; @@ -492,15 +539,15 @@ public final class BinaryDictDecoderUtils { 0 != (info.mFlags & FormatSpec.FLAG_IS_NOT_A_WORD), 0 != (info.mFlags & FormatSpec.FLAG_IS_BLACKLISTED))); } - groupOffsetPos = info.mEndAddress; + groupPos = info.mEndAddress; } // reach the end of the array. - if (options.mSupportsDynamicUpdate) { + if (options.supportsDynamicUpdate()) { final boolean hasValidForwardLink = dictDecoder.readAndFollowForwardLink(); if (!hasValidForwardLink) break; } - } while (options.mSupportsDynamicUpdate && dictDecoder.hasNextPtNodeArray()); + } while (options.supportsDynamicUpdate() && dictDecoder.hasNextPtNodeArray()); final PtNodeArray nodeArray = new PtNodeArray(nodeArrayContents); nodeArray.mCachedAddressBeforeUpdate = nodeArrayOriginPos; @@ -556,7 +603,7 @@ public final class BinaryDictDecoderUtils { Map<Integer, PtNodeArray> reverseNodeArrayMapping = new TreeMap<Integer, PtNodeArray>(); Map<Integer, PtNode> reversePtNodeMapping = new TreeMap<Integer, PtNode>(); - final PtNodeArray root = readNodeArray(dictDecoder, fileHeader.mHeaderSize, + final PtNodeArray root = readNodeArray(dictDecoder, fileHeader.mBodyOffset, reverseNodeArrayMapping, reversePtNodeMapping, fileHeader.mFormatOptions); FusionDictionary newDict = new FusionDictionary(root, fileHeader.mDictionaryOptions); @@ -592,32 +639,10 @@ public final class BinaryDictDecoderUtils { /** * Basic test to find out whether the file is a binary dictionary or not. * - * Concretely this only tests the magic number. - * * @param file The file to test. * @return true if it's a binary dictionary, false otherwise */ public static boolean isBinaryDictionary(final File file) { - FileInputStream inStream = null; - try { - inStream = new FileInputStream(file); - final ByteBuffer buffer = inStream.getChannel().map( - FileChannel.MapMode.READ_ONLY, 0, file.length()); - final int version = getFormatVersion(new ByteBufferDictBuffer(buffer)); - return (version >= FormatSpec.MINIMUM_SUPPORTED_VERSION - && version <= FormatSpec.MAXIMUM_SUPPORTED_VERSION); - } catch (FileNotFoundException e) { - return false; - } catch (IOException e) { - return false; - } finally { - if (inStream != null) { - try { - inStream.close(); - } catch (IOException e) { - // do nothing - } - } - } + return FormatSpec.getDictDecoder(file).hasValidRawBinaryDictionary(); } } diff --git a/java/src/com/android/inputmethod/latin/makedict/BinaryDictEncoderUtils.java b/java/src/com/android/inputmethod/latin/makedict/BinaryDictEncoderUtils.java index f761829de..bb40e0dd5 100644 --- a/java/src/com/android/inputmethod/latin/makedict/BinaryDictEncoderUtils.java +++ b/java/src/com/android/inputmethod/latin/makedict/BinaryDictEncoderUtils.java @@ -16,10 +16,11 @@ package com.android.inputmethod.latin.makedict; +import com.android.inputmethod.annotations.UsedForTesting; import com.android.inputmethod.latin.makedict.BinaryDictDecoderUtils.CharEncoding; +import com.android.inputmethod.latin.makedict.BinaryDictDecoderUtils.DictBuffer; import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions; import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode; -import com.android.inputmethod.latin.makedict.FusionDictionary.DictionaryOptions; import com.android.inputmethod.latin.makedict.FusionDictionary.PtNodeArray; import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; @@ -160,7 +161,7 @@ public class BinaryDictEncoderUtils { node.mCachedSize = nodeSize; size += nodeSize; } - if (options.mSupportsDynamicUpdate) { + if (options.supportsDynamicUpdate()) { size += FormatSpec.FORWARD_LINK_ADDRESS_SIZE; } ptNodeArray.mCachedSize = size; @@ -245,6 +246,27 @@ public class BinaryDictEncoderUtils { } } + @UsedForTesting + static void writeUIntToDictBuffer(final DictBuffer dictBuffer, final int value, + final int size) { + switch(size) { + case 4: + dictBuffer.put((byte) ((value >> 24) & 0xFF)); + /* fall through */ + case 3: + dictBuffer.put((byte) ((value >> 16) & 0xFF)); + /* fall through */ + case 2: + dictBuffer.put((byte) ((value >> 8) & 0xFF)); + /* fall through */ + case 1: + dictBuffer.put((byte) (value & 0xFF)); + break; + default: + /* nop */ + } + } + // End utility methods // This method is responsible for finding a nice ordering of the nodes that favors run-time @@ -377,7 +399,7 @@ public class BinaryDictEncoderUtils { nodeSize += FormatSpec.PTNODE_FREQUENCY_SIZE; } } - if (formatOptions.mSupportsDynamicUpdate) { + if (formatOptions.supportsDynamicUpdate()) { nodeSize += FormatSpec.SIGNED_CHILDREN_ADDRESS_SIZE; } else if (null != ptNode.mChildren) { nodeSize += getByteSize(getOffsetToTargetNodeArrayDuringUpdate(ptNodeArray, @@ -397,7 +419,7 @@ public class BinaryDictEncoderUtils { ptNode.mCachedSize = nodeSize; size += nodeSize; } - if (formatOptions.mSupportsDynamicUpdate) { + if (formatOptions.supportsDynamicUpdate()) { size += FormatSpec.FORWARD_LINK_ADDRESS_SIZE; } if (ptNodeArray.mCachedSize != size) { @@ -513,7 +535,7 @@ public class BinaryDictEncoderUtils { if (passes > MAX_PASSES) throw new RuntimeException("Too many passes - probably a bug"); } while (changesDone); - if (formatOptions.mSupportsDynamicUpdate) { + if (formatOptions.supportsDynamicUpdate()) { computeParentAddresses(flatNodes); } final PtNodeArray lastPtNodeArray = flatNodes.get(flatNodes.size() - 1); @@ -622,7 +644,7 @@ public class BinaryDictEncoderUtils { byte flags = 0; if (hasMultipleChars) flags |= FormatSpec.FLAG_HAS_MULTIPLE_CHARS; if (isTerminal) flags |= FormatSpec.FLAG_IS_TERMINAL; - if (formatOptions.mSupportsDynamicUpdate) { + if (formatOptions.supportsDynamicUpdate()) { flags |= FormatSpec.FLAG_IS_NOT_MOVED; } else if (true) { switch (childrenAddressSize) { @@ -690,6 +712,13 @@ public class BinaryDictEncoderUtils { + word + " is " + unigramFrequency); bigramFrequency = unigramFrequency; } + bigramFlags += getBigramFrequencyDiff(unigramFrequency, bigramFrequency) + & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY; + return bigramFlags; + } + + public static int getBigramFrequencyDiff(final int unigramFrequency, + final int bigramFrequency) { // We compute the difference between 255 (which means probability = 1) and the // unigram score. We split this into a number of discrete steps. // Now, the steps are numbered 0~15; 0 represents an increase of 1 step while 15 @@ -723,22 +752,15 @@ public class BinaryDictEncoderUtils { // include this bigram in the dictionary. For now, register as 0, and live with the // small over-estimation that we get in this case. TODO: actually remove this bigram // if discretizedFrequency < 0. - final int finalBigramFrequency = discretizedFrequency > 0 ? discretizedFrequency : 0; - bigramFlags += finalBigramFrequency & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY; - return bigramFlags; + return discretizedFrequency > 0 ? discretizedFrequency : 0; } /** - * Makes the 2-byte value for options flags. + * Makes the 2-byte value for options flags. Unused at the moment, and always 0. */ - private static final int makeOptionsValue(final FusionDictionary dictionary, - final FormatOptions formatOptions) { - final DictionaryOptions options = dictionary.mOptions; - final boolean hasBigrams = dictionary.hasBigrams(); - return (options.mFrenchLigatureProcessing ? FormatSpec.FRENCH_LIGATURE_PROCESSING_FLAG : 0) - + (options.mGermanUmlautProcessing ? FormatSpec.GERMAN_UMLAUT_PROCESSING_FLAG : 0) - + (hasBigrams ? FormatSpec.CONTAINS_BIGRAMS_FLAG : 0) - + (formatOptions.mSupportsDynamicUpdate ? FormatSpec.SUPPORTS_DYNAMIC_UPDATE : 0); + private static final int makeOptionsValue(final FormatOptions formatOptions) { + // TODO: why doesn't this handle CONTAINS_TIMESTAMP_FLAG? + return 0; } /** @@ -826,7 +848,7 @@ public class BinaryDictEncoderUtils { } dictEncoder.writePtNode(ptNode, parentPosition, formatOptions, dict); } - if (formatOptions.mSupportsDynamicUpdate) { + if (formatOptions.supportsDynamicUpdate()) { dictEncoder.writeForwardLinkAddress(FormatSpec.NO_FORWARD_LINK_ADDRESS); } if (dictEncoder.getPosition() != ptNodeArray.mCachedAddressAfterUpdate @@ -927,7 +949,7 @@ public class BinaryDictEncoderUtils { headerBuffer.write((byte) (0xFF & version)); // Options flags - final int options = makeOptionsValue(dict, formatOptions); + final int options = makeOptionsValue(formatOptions); headerBuffer.write((byte) (0xFF & (options >> 8))); headerBuffer.write((byte) (0xFF & options)); final int headerSizeOffset = headerBuffer.size(); diff --git a/java/src/com/android/inputmethod/latin/makedict/BinaryDictIOUtils.java b/java/src/com/android/inputmethod/latin/makedict/BinaryDictIOUtils.java index d5516ef46..07ba777c7 100644 --- a/java/src/com/android/inputmethod/latin/makedict/BinaryDictIOUtils.java +++ b/java/src/com/android/inputmethod/latin/makedict/BinaryDictIOUtils.java @@ -23,7 +23,6 @@ import com.android.inputmethod.latin.makedict.BinaryDictDecoderUtils.DictBuffer; import com.android.inputmethod.latin.makedict.FormatSpec.FileHeader; import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions; import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode; -import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; import com.android.inputmethod.latin.utils.ByteArrayDictBuffer; import java.io.File; @@ -32,7 +31,6 @@ import java.io.FileNotFoundException; import java.io.IOException; import java.io.OutputStream; import java.util.ArrayList; -import java.util.Iterator; import java.util.Map; import java.util.Stack; @@ -62,7 +60,7 @@ public final class BinaryDictIOUtils { * Retrieves all node arrays without recursive call. */ private static void readUnigramsAndBigramsBinaryInner(final DictDecoder dictDecoder, - final int headerSize, final Map<Integer, String> words, + final int bodyOffset, final Map<Integer, String> words, final Map<Integer, Integer> frequencies, final Map<Integer, ArrayList<PendingAttribute>> bigrams, final FormatOptions formatOptions) { @@ -71,7 +69,7 @@ public final class BinaryDictIOUtils { Stack<Position> stack = new Stack<Position>(); int index = 0; - Position initPos = new Position(headerSize, 0); + Position initPos = new Position(bodyOffset, 0); stack.push(initPos); while (!stack.empty()) { @@ -87,7 +85,7 @@ public final class BinaryDictIOUtils { if (p.mNumOfPtNode == Position.NOT_READ_PTNODE_COUNT) { p.mNumOfPtNode = dictDecoder.readPtNodeCount(); - p.mAddress += getPtNodeCountSize(p.mNumOfPtNode); + p.mAddress = dictDecoder.getPosition(); p.mPosition = 0; } if (p.mNumOfPtNode == 0) { @@ -112,7 +110,7 @@ public final class BinaryDictIOUtils { } if (p.mPosition == p.mNumOfPtNode) { - if (formatOptions.mSupportsDynamicUpdate) { + if (formatOptions.supportsDynamicUpdate()) { final boolean hasValidForwardLinkAddress = dictDecoder.readAndFollowForwardLink(); if (hasValidForwardLinkAddress && dictDecoder.hasNextPtNodeArray()) { @@ -154,7 +152,7 @@ public final class BinaryDictIOUtils { UnsupportedFormatException { // Read header final FileHeader header = dictDecoder.readHeader(); - readUnigramsAndBigramsBinaryInner(dictDecoder, header.mHeaderSize, words, + readUnigramsAndBigramsBinaryInner(dictDecoder, header.mBodyOffset, words, frequencies, bigrams, header.mFormatOptions); } @@ -228,7 +226,7 @@ public final class BinaryDictIOUtils { // a forward link address that we need to consult and possibly resume // search on the next node array in the linked list. if (foundNextPtNode) break; - if (!header.mFormatOptions.mSupportsDynamicUpdate) { + if (!header.mFormatOptions.supportsDynamicUpdate()) { return FormatSpec.NOT_VALID_WORD; } @@ -245,8 +243,8 @@ public final class BinaryDictIOUtils { /** * @return the size written, in bytes. Always 3 bytes. */ - static int writeSInt24ToBuffer(final DictBuffer dictBuffer, - final int value) { + @UsedForTesting + static int writeSInt24ToBuffer(final DictBuffer dictBuffer, final int value) { final int absValue = Math.abs(value); dictBuffer.put((byte)(((value < 0 ? 0x80 : 0) | (absValue >> 16)) & 0xFF)); dictBuffer.put((byte)((absValue >> 8) & 0xFF)); @@ -257,6 +255,7 @@ public final class BinaryDictIOUtils { /** * @return the size written, in bytes. Always 3 bytes. */ + @UsedForTesting static int writeSInt24ToStream(final OutputStream destination, final int value) throws IOException { final int absValue = Math.abs(value); @@ -266,28 +265,7 @@ public final class BinaryDictIOUtils { return 3; } - /** - * @return the size written, in bytes. 1, 2, or 3 bytes. - */ - private static int writeVariableAddress(final OutputStream destination, final int value) - throws IOException { - switch (BinaryDictEncoderUtils.getByteSize(value)) { - case 1: - destination.write((byte)value); - break; - case 2: - destination.write((byte)(0xFF & (value >> 8))); - destination.write((byte)(0xFF & value)); - break; - case 3: - destination.write((byte)(0xFF & (value >> 16))); - destination.write((byte)(0xFF & (value >> 8))); - destination.write((byte)(0xFF & value)); - break; - } - return BinaryDictEncoderUtils.getByteSize(value); - } - + @UsedForTesting static void skipString(final DictBuffer dictBuffer, final boolean hasMultipleChars) { if (hasMultipleChars) { @@ -301,176 +279,25 @@ public final class BinaryDictIOUtils { } /** - * Write a string to a stream. - * - * @param destination the stream to write. - * @param word the string to be written. - * @return the size written, in bytes. - * @throws IOException - */ - private static int writeString(final OutputStream destination, final String word) - throws IOException { - int size = 0; - final int length = word.length(); - for (int i = 0; i < length; i = word.offsetByCodePoints(i, 1)) { - final int codePoint = word.codePointAt(i); - if (CharEncoding.getCharSize(codePoint) == 1) { - destination.write((byte)codePoint); - size++; - } else { - destination.write((byte)(0xFF & (codePoint >> 16))); - destination.write((byte)(0xFF & (codePoint >> 8))); - destination.write((byte)(0xFF & codePoint)); - size += 3; - } - } - destination.write((byte)FormatSpec.PTNODE_CHARACTERS_TERMINATOR); - size += FormatSpec.PTNODE_TERMINATOR_SIZE; - return size; - } - - /** - * Write a PtNode to an output stream from a PtNodeInfo. - * A PtNode is an in-memory representation of a node in the patricia trie. - * A PtNode info is a container for low-level information about how the - * PtNode is stored in the binary format. + * Writes a PtNodeCount to the stream. * * @param destination the stream to write. - * @param info the PtNode info to be written. - * @return the size written, in bytes. + * @param ptNodeCount the count. + * @return the size written in bytes. */ - private static int writePtNode(final OutputStream destination, final PtNodeInfo info) - throws IOException { - int size = FormatSpec.PTNODE_FLAGS_SIZE; - destination.write((byte)info.mFlags); - final int parentOffset = info.mParentAddress == FormatSpec.NO_PARENT_ADDRESS ? - FormatSpec.NO_PARENT_ADDRESS : info.mParentAddress - info.mOriginalAddress; - size += writeSInt24ToStream(destination, parentOffset); - - for (int i = 0; i < info.mCharacters.length; ++i) { - if (CharEncoding.getCharSize(info.mCharacters[i]) == 1) { - destination.write((byte)info.mCharacters[i]); - size++; - } else { - size += writeSInt24ToStream(destination, info.mCharacters[i]); - } - } - if (info.mCharacters.length > 1) { - destination.write((byte)FormatSpec.PTNODE_CHARACTERS_TERMINATOR); - size++; - } - - if ((info.mFlags & FormatSpec.FLAG_IS_TERMINAL) != 0) { - destination.write((byte)info.mFrequency); - size++; - } - - if (DBG) { - MakedictLog.d("writePtNode origin=" + info.mOriginalAddress + ", size=" + size - + ", child=" + info.mChildrenAddress + ", characters =" - + new String(info.mCharacters, 0, info.mCharacters.length)); - } - final int childrenOffset = info.mChildrenAddress == FormatSpec.NO_CHILDREN_ADDRESS ? - 0 : info.mChildrenAddress - (info.mOriginalAddress + size); - writeSInt24ToStream(destination, childrenOffset); - size += FormatSpec.SIGNED_CHILDREN_ADDRESS_SIZE; - - if (info.mShortcutTargets != null && info.mShortcutTargets.size() > 0) { - final int shortcutListSize = - BinaryDictEncoderUtils.getShortcutListSize(info.mShortcutTargets); - destination.write((byte)(shortcutListSize >> 8)); - destination.write((byte)(shortcutListSize & 0xFF)); - size += 2; - final Iterator<WeightedString> shortcutIterator = info.mShortcutTargets.iterator(); - while (shortcutIterator.hasNext()) { - final WeightedString target = shortcutIterator.next(); - destination.write((byte)BinaryDictEncoderUtils.makeShortcutFlags( - shortcutIterator.hasNext(), target.mFrequency)); - size++; - size += writeString(destination, target.mWord); - } - } - - if (info.mBigrams != null) { - // TODO: Consolidate this code with the code that computes the size of the bigram list - // in BinaryDictEncoderUtils#computeActualNodeArraySize - for (int i = 0; i < info.mBigrams.size(); ++i) { - - final int bigramFrequency = info.mBigrams.get(i).mFrequency; - int bigramFlags = (i < info.mBigrams.size() - 1) - ? FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_HAS_NEXT : 0; - size++; - final int bigramOffset = info.mBigrams.get(i).mAddress - (info.mOriginalAddress - + size); - bigramFlags |= (bigramOffset < 0) ? FormatSpec.FLAG_BIGRAM_ATTR_OFFSET_NEGATIVE : 0; - switch (BinaryDictEncoderUtils.getByteSize(bigramOffset)) { - case 1: - bigramFlags |= FormatSpec.FLAG_BIGRAM_ATTR_ADDRESS_TYPE_ONEBYTE; - break; - case 2: - bigramFlags |= FormatSpec.FLAG_BIGRAM_ATTR_ADDRESS_TYPE_TWOBYTES; - break; - case 3: - bigramFlags |= FormatSpec.FLAG_BIGRAM_ATTR_ADDRESS_TYPE_THREEBYTES; - break; - } - bigramFlags |= bigramFrequency & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY; - destination.write((byte)bigramFlags); - size += writeVariableAddress(destination, Math.abs(bigramOffset)); - } - } - return size; - } - - /** - * Compute the size of the PtNode. - */ - static int computePtNodeSize(final PtNodeInfo info, final FormatOptions formatOptions) { - int size = FormatSpec.PTNODE_FLAGS_SIZE + FormatSpec.PARENT_ADDRESS_SIZE - + BinaryDictEncoderUtils.getPtNodeCharactersSize(info.mCharacters) - + getChildrenAddressSize(info.mFlags, formatOptions); - if ((info.mFlags & FormatSpec.FLAG_IS_TERMINAL) != 0) { - size += FormatSpec.PTNODE_FREQUENCY_SIZE; - } - if (info.mShortcutTargets != null && !info.mShortcutTargets.isEmpty()) { - size += BinaryDictEncoderUtils.getShortcutListSize(info.mShortcutTargets); - } - if (info.mBigrams != null) { - for (final PendingAttribute attr : info.mBigrams) { - size += FormatSpec.PTNODE_FLAGS_SIZE; - size += BinaryDictEncoderUtils.getByteSize(attr.mAddress); - } - } - return size; - } - - /** - * Write a node array to the stream. - * - * @param destination the stream to write. - * @param infos an array of PtNodeInfo to be written. - * @return the size written, in bytes. - * @throws IOException - */ - static int writeNodes(final OutputStream destination, final PtNodeInfo[] infos) + @UsedForTesting + static int writePtNodeCount(final OutputStream destination, final int ptNodeCount) throws IOException { - int size = getPtNodeCountSize(infos.length); - switch (getPtNodeCountSize(infos.length)) { - case 1: - destination.write((byte)infos.length); - break; - case 2: - final int encodedPtNodeCount = - infos.length | FormatSpec.LARGE_PTNODE_ARRAY_SIZE_FIELD_SIZE_FLAG; - destination.write((byte)(encodedPtNodeCount >> 8)); - destination.write((byte)(encodedPtNodeCount & 0xFF)); - break; - default: - throw new RuntimeException("Invalid node count size."); + final int countSize = BinaryDictIOUtils.getPtNodeCountSize(ptNodeCount); + // the count must fit on one byte or two bytes. + // Please see comments in FormatSpec. + if (countSize != 1 && countSize != 2) { + throw new RuntimeException("Strange size from getPtNodeCountSize : " + countSize); } - for (final PtNodeInfo info : infos) size += writePtNode(destination, info); - writeSInt24ToStream(destination, FormatSpec.NO_FORWARD_LINK_ADDRESS); - return size + FormatSpec.FORWARD_LINK_ADDRESS_SIZE; + final int encodedPtNodeCount = (countSize == 2) ? + (ptNodeCount | FormatSpec.LARGE_PTNODE_ARRAY_SIZE_FIELD_SIZE_FLAG) : ptNodeCount; + BinaryDictEncoderUtils.writeUIntToStream(destination, encodedPtNodeCount, countSize); + return countSize; } private static final int HEADER_READING_BUFFER_SIZE = 16384; @@ -529,7 +356,7 @@ public final class BinaryDictIOUtils { * Helper method to check whether the node is moved. */ public static boolean isMovedPtNode(final int flags, final FormatOptions options) { - return options.mSupportsDynamicUpdate + return options.supportsDynamicUpdate() && ((flags & FormatSpec.MASK_CHILDREN_ADDRESS_TYPE) == FormatSpec.FLAG_IS_MOVED); } @@ -538,14 +365,14 @@ public final class BinaryDictIOUtils { */ public static boolean supportsDynamicUpdate(final FormatOptions options) { return options.mVersion >= FormatSpec.FIRST_VERSION_WITH_DYNAMIC_UPDATE - && options.mSupportsDynamicUpdate; + && options.supportsDynamicUpdate(); } /** * Helper method to check whether the node is deleted. */ public static boolean isDeletedPtNode(final int flags, final FormatOptions formatOptions) { - return formatOptions.mSupportsDynamicUpdate + return formatOptions.supportsDynamicUpdate() && ((flags & FormatSpec.MASK_CHILDREN_ADDRESS_TYPE) == FormatSpec.FLAG_IS_DELETED); } @@ -568,7 +395,7 @@ public final class BinaryDictIOUtils { static int getChildrenAddressSize(final int optionFlags, final FormatOptions formatOptions) { - if (formatOptions.mSupportsDynamicUpdate) return FormatSpec.SIGNED_CHILDREN_ADDRESS_SIZE; + if (formatOptions.supportsDynamicUpdate()) return FormatSpec.SIGNED_CHILDREN_ADDRESS_SIZE; switch (optionFlags & FormatSpec.MASK_CHILDREN_ADDRESS_TYPE) { case FormatSpec.FLAG_CHILDREN_ADDRESS_TYPE_ONEBYTE: return 1; diff --git a/java/src/com/android/inputmethod/latin/makedict/DictDecoder.java b/java/src/com/android/inputmethod/latin/makedict/DictDecoder.java index 3dbeee099..b4838f00f 100644 --- a/java/src/com/android/inputmethod/latin/makedict/DictDecoder.java +++ b/java/src/com/android/inputmethod/latin/makedict/DictDecoder.java @@ -35,6 +35,7 @@ import java.util.TreeMap; /** * An interface of binary dictionary decoders. */ +// TODO: Straighten out responsibility for the buffer's file pointer. public interface DictDecoder { /** @@ -43,7 +44,7 @@ public interface DictDecoder { public FileHeader readHeader() throws IOException, UnsupportedFormatException; /** - * Reads PtNode from nodeAddress. + * Reads PtNode from ptNodePos. * @param ptNodePos the position of PtNode. * @param formatOptions the format options. * @return PtNodeInfo. @@ -127,7 +128,8 @@ public interface DictDecoder { * Opens the dictionary file and makes DictBuffer. */ @UsedForTesting - public void openDictBuffer() throws FileNotFoundException, IOException; + public void openDictBuffer() throws FileNotFoundException, IOException, + UnsupportedFormatException; @UsedForTesting public boolean isDictBufferOpen(); @@ -228,4 +230,9 @@ public interface DictDecoder { } public void skipPtNode(final FormatOptions formatOptions); + + /** + * @return whether this decoder has a valid binary dictionary that it can decode. + */ + public boolean hasValidRawBinaryDictionary(); } diff --git a/java/src/com/android/inputmethod/latin/makedict/DictUpdater.java b/java/src/com/android/inputmethod/latin/makedict/DictUpdater.java deleted file mode 100644 index c4f7ec91f..000000000 --- a/java/src/com/android/inputmethod/latin/makedict/DictUpdater.java +++ /dev/null @@ -1,54 +0,0 @@ -/* - * Copyright (C) 2013 The Android Open Source Project - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -package com.android.inputmethod.latin.makedict; - -import com.android.inputmethod.annotations.UsedForTesting; -import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; - -import java.io.IOException; -import java.util.ArrayList; - -/** - * An interface of a binary dictionary updater. - */ -@UsedForTesting -public interface DictUpdater extends DictDecoder { - - /** - * Deletes the word from the binary dictionary. - * - * @param word the word to be deleted. - */ - @UsedForTesting - public void deleteWord(final String word) throws IOException, UnsupportedFormatException; - - /** - * Inserts a word into a binary dictionary. - * - * @param word the word to be inserted. - * @param frequency the frequency of the new word. - * @param bigramStrings bigram list, or null if none. - * @param shortcuts shortcut list, or null if none. - * @param isBlackListEntry whether this should be a blacklist entry. - */ - // TODO: Support batch insertion. - @UsedForTesting - public void insertWord(final String word, final int frequency, - final ArrayList<WeightedString> bigramStrings, - final ArrayList<WeightedString> shortcuts, final boolean isNotAWord, - final boolean isBlackListEntry) throws IOException, UnsupportedFormatException; -} diff --git a/java/src/com/android/inputmethod/latin/makedict/DynamicBinaryDictIOUtils.java b/java/src/com/android/inputmethod/latin/makedict/DynamicBinaryDictIOUtils.java deleted file mode 100644 index 28da9ffdd..000000000 --- a/java/src/com/android/inputmethod/latin/makedict/DynamicBinaryDictIOUtils.java +++ /dev/null @@ -1,492 +0,0 @@ -/* - * Copyright (C) 2013 The Android Open Source Project - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -package com.android.inputmethod.latin.makedict; - -import com.android.inputmethod.annotations.UsedForTesting; -import com.android.inputmethod.latin.Constants; -import com.android.inputmethod.latin.makedict.BinaryDictDecoderUtils.DictBuffer; -import com.android.inputmethod.latin.makedict.FormatSpec.FileHeader; -import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions; -import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; -import com.android.inputmethod.latin.utils.CollectionUtils; - -import java.io.IOException; -import java.io.OutputStream; -import java.util.ArrayList; -import java.util.Arrays; - -/** - * The utility class to help dynamic updates on the binary dictionary. - * - * All the methods in this class are static. - */ -@UsedForTesting -public final class DynamicBinaryDictIOUtils { - private static final boolean DBG = false; - private static final int MAX_JUMPS = 10000; - - private DynamicBinaryDictIOUtils() { - // This utility class is not publicly instantiable. - } - - /* package */ static int markAsDeleted(final int flags) { - return (flags & (~FormatSpec.MASK_CHILDREN_ADDRESS_TYPE)) | FormatSpec.FLAG_IS_DELETED; - } - - /** - * Update a parent address in a PtNode that is referred to by ptNodeOriginAddress. - * - * @param dictUpdater the DictUpdater to write. - * @param ptNodeOriginAddress the address of the PtNode. - * @param newParentAddress the absolute address of the parent. - * @param formatOptions file format options. - */ - private static void updateParentAddress(final Ver3DictUpdater dictUpdater, - final int ptNodeOriginAddress, final int newParentAddress, - final FormatOptions formatOptions) { - final DictBuffer dictBuffer = dictUpdater.getDictBuffer(); - final int originalPosition = dictBuffer.position(); - dictBuffer.position(ptNodeOriginAddress); - if (!formatOptions.mSupportsDynamicUpdate) { - throw new RuntimeException("this file format does not support parent addresses"); - } - final int flags = dictBuffer.readUnsignedByte(); - if (BinaryDictIOUtils.isMovedPtNode(flags, formatOptions)) { - // If the node is moved, the parent address is stored in the destination node. - // We are guaranteed to process the destination node later, so there is no need to - // update anything here. - dictBuffer.position(originalPosition); - return; - } - if (DBG) { - MakedictLog.d("update parent address flags=" + flags + ", " + ptNodeOriginAddress); - } - final int parentOffset = newParentAddress - ptNodeOriginAddress; - BinaryDictIOUtils.writeSInt24ToBuffer(dictBuffer, parentOffset); - dictBuffer.position(originalPosition); - } - - /** - * Update parent addresses in a node array stored at ptNodeOriginAddress. - * - * @param dictUpdater the DictUpdater to be modified. - * @param ptNodeOriginAddress the address of the node array to update. - * @param newParentAddress the address to be written. - * @param formatOptions file format options. - */ - private static void updateParentAddresses(final Ver3DictUpdater dictUpdater, - final int ptNodeOriginAddress, final int newParentAddress, - final FormatOptions formatOptions) { - final int originalPosition = dictUpdater.getPosition(); - dictUpdater.setPosition(ptNodeOriginAddress); - do { - final int count = dictUpdater.readPtNodeCount(); - for (int i = 0; i < count; ++i) { - updateParentAddress(dictUpdater, dictUpdater.getPosition(), newParentAddress, - formatOptions); - dictUpdater.skipPtNode(formatOptions); - } - if (!dictUpdater.readAndFollowForwardLink()) break; - if (dictUpdater.getPosition() == FormatSpec.NO_FORWARD_LINK_ADDRESS) break; - } while (formatOptions.mSupportsDynamicUpdate); - dictUpdater.setPosition(originalPosition); - } - - /** - * Update a children address in a PtNode that is addressed by ptNodeOriginAddress. - * - * @param dictUpdater the DictUpdater to write. - * @param ptNodeOriginAddress the address of the PtNode. - * @param newChildrenAddress the absolute address of the child. - * @param formatOptions file format options. - */ - private static void updateChildrenAddress(final Ver3DictUpdater dictUpdater, - final int ptNodeOriginAddress, final int newChildrenAddress, - final FormatOptions formatOptions) { - final DictBuffer dictBuffer = dictUpdater.getDictBuffer(); - final int originalPosition = dictBuffer.position(); - dictBuffer.position(ptNodeOriginAddress); - final int flags = dictBuffer.readUnsignedByte(); - BinaryDictDecoderUtils.readParentAddress(dictBuffer, formatOptions); - BinaryDictIOUtils.skipString(dictBuffer, (flags & FormatSpec.FLAG_HAS_MULTIPLE_CHARS) != 0); - if ((flags & FormatSpec.FLAG_IS_TERMINAL) != 0) dictBuffer.readUnsignedByte(); - final int childrenOffset = newChildrenAddress == FormatSpec.NO_CHILDREN_ADDRESS - ? FormatSpec.NO_CHILDREN_ADDRESS : newChildrenAddress - dictBuffer.position(); - BinaryDictIOUtils.writeSInt24ToBuffer(dictBuffer, childrenOffset); - dictBuffer.position(originalPosition); - } - - /** - * Helper method to move a PtNode to the tail of the file. - */ - private static int movePtNode(final OutputStream destination, - final Ver3DictUpdater dictUpdater, final PtNodeInfo info, - final int nodeArrayOriginAddress, final int oldNodeAddress, - final FormatOptions formatOptions) throws IOException { - final DictBuffer dictBuffer = dictUpdater.getDictBuffer(); - updateParentAddress(dictUpdater, oldNodeAddress, dictBuffer.limit() + 1, formatOptions); - dictBuffer.position(oldNodeAddress); - final int currentFlags = dictBuffer.readUnsignedByte(); - dictBuffer.position(oldNodeAddress); - dictBuffer.put((byte)(FormatSpec.FLAG_IS_MOVED | (currentFlags - & (~FormatSpec.MASK_MOVE_AND_DELETE_FLAG)))); - int size = FormatSpec.PTNODE_FLAGS_SIZE; - updateForwardLink(dictUpdater, nodeArrayOriginAddress, dictBuffer.limit(), formatOptions); - size += BinaryDictIOUtils.writeNodes(destination, new PtNodeInfo[] { info }); - return size; - } - - @SuppressWarnings("unused") - private static void updateForwardLink(final Ver3DictUpdater dictUpdater, - final int nodeArrayOriginAddress, final int newNodeArrayAddress, - final FormatOptions formatOptions) { - final DictBuffer dictBuffer = dictUpdater.getDictBuffer(); - dictUpdater.setPosition(nodeArrayOriginAddress); - int jumpCount = 0; - while (jumpCount++ < MAX_JUMPS) { - final int count = dictUpdater.readPtNodeCount(); - for (int i = 0; i < count; ++i) { - dictUpdater.readPtNode(dictUpdater.getPosition(), formatOptions); - } - final int forwardLinkAddress = dictBuffer.readUnsignedInt24(); - if (forwardLinkAddress == FormatSpec.NO_FORWARD_LINK_ADDRESS) { - dictBuffer.position(dictBuffer.position() - FormatSpec.FORWARD_LINK_ADDRESS_SIZE); - BinaryDictIOUtils.writeSInt24ToBuffer(dictBuffer, newNodeArrayAddress); - return; - } - dictBuffer.position(forwardLinkAddress); - } - if (DBG && jumpCount >= MAX_JUMPS) { - throw new RuntimeException("too many jumps, probably a bug."); - } - } - - /** - * Move a PtNode that is referred to by oldPtNodeOrigin to the tail of the file, and set the - * children address to the byte after the PtNode. - * - * @param fileEndAddress the address of the tail of the file. - * @param codePoints the characters to put inside the PtNode. - * @param length how many code points to read from codePoints. - * @param flags the flags for this PtNode. - * @param frequency the frequency of this terminal. - * @param parentAddress the address of the parent PtNode of this PtNode. - * @param shortcutTargets the shortcut targets for this PtNode. - * @param bigrams the bigrams for this PtNode. - * @param destination the stream representing the tail of the file. - * @param dictUpdater the DictUpdater. - * @param oldPtNodeArrayOrigin the origin of the old PtNode array this PtNode was a part of. - * @param oldPtNodeOrigin the old origin where this PtNode used to be stored. - * @param formatOptions format options for this dictionary. - * @return the size written, in bytes. - * @throws IOException if the file can't be accessed - */ - private static int movePtNode(final int fileEndAddress, final int[] codePoints, - final int length, final int flags, final int frequency, final int parentAddress, - final ArrayList<WeightedString> shortcutTargets, - final ArrayList<PendingAttribute> bigrams, final OutputStream destination, - final Ver3DictUpdater dictUpdater, final int oldPtNodeArrayOrigin, - final int oldPtNodeOrigin, final FormatOptions formatOptions) throws IOException { - int size = 0; - final int newPtNodeOrigin = fileEndAddress + 1; - final int[] writtenCharacters = Arrays.copyOfRange(codePoints, 0, length); - final PtNodeInfo tmpInfo = new PtNodeInfo(newPtNodeOrigin, -1 /* endAddress */, - flags, writtenCharacters, frequency, parentAddress, FormatSpec.NO_CHILDREN_ADDRESS, - shortcutTargets, bigrams); - size = BinaryDictIOUtils.computePtNodeSize(tmpInfo, formatOptions); - final PtNodeInfo newInfo = new PtNodeInfo(newPtNodeOrigin, newPtNodeOrigin + size, - flags, writtenCharacters, frequency, parentAddress, - fileEndAddress + 1 + size + FormatSpec.FORWARD_LINK_ADDRESS_SIZE, shortcutTargets, - bigrams); - movePtNode(destination, dictUpdater, newInfo, oldPtNodeArrayOrigin, oldPtNodeOrigin, - formatOptions); - return 1 + size + FormatSpec.FORWARD_LINK_ADDRESS_SIZE; - } - - /** - * Converts a list of WeightedString to a list of PendingAttribute. - */ - public static ArrayList<PendingAttribute> resolveBigramPositions(final DictUpdater dictUpdater, - final ArrayList<WeightedString> bigramStrings) - throws IOException, UnsupportedFormatException { - if (bigramStrings == null) return CollectionUtils.newArrayList(); - final ArrayList<PendingAttribute> bigrams = CollectionUtils.newArrayList(); - for (final WeightedString bigram : bigramStrings) { - final int pos = dictUpdater.getTerminalPosition(bigram.mWord); - if (pos == FormatSpec.NOT_VALID_WORD) { - // TODO: figure out what is the correct thing to do here. - } else { - bigrams.add(new PendingAttribute(bigram.mFrequency, pos)); - } - } - return bigrams; - } - - /** - * Insert a word into a binary dictionary. - * - * @param dictUpdater the dict updater. - * @param destination a stream to the underlying file, with the pointer at the end of the file. - * @param word the word to insert. - * @param frequency the frequency of the new word. - * @param bigramStrings bigram list, or null if none. - * @param shortcuts shortcut list, or null if none. - * @param isBlackListEntry whether this should be a blacklist entry. - * @throws IOException if the file can't be accessed. - * @throws UnsupportedFormatException if the existing dictionary is in an unexpected format. - */ - // TODO: Support batch insertion. - // TODO: Remove @UsedForTesting once UserHistoryDictionary is implemented by BinaryDictionary. - @UsedForTesting - public static void insertWord(final Ver3DictUpdater dictUpdater, - final OutputStream destination, final String word, final int frequency, - final ArrayList<WeightedString> bigramStrings, - final ArrayList<WeightedString> shortcuts, final boolean isNotAWord, - final boolean isBlackListEntry) - throws IOException, UnsupportedFormatException { - final ArrayList<PendingAttribute> bigrams = resolveBigramPositions(dictUpdater, - bigramStrings); - final DictBuffer dictBuffer = dictUpdater.getDictBuffer(); - - final boolean isTerminal = true; - final boolean hasBigrams = !bigrams.isEmpty(); - final boolean hasShortcuts = shortcuts != null && !shortcuts.isEmpty(); - - // find the insert position of the word. - if (dictBuffer.position() != 0) dictBuffer.position(0); - final FileHeader fileHeader = dictUpdater.readHeader(); - - int wordPos = 0, address = dictBuffer.position(), nodeOriginAddress = dictBuffer.position(); - final int[] codePoints = FusionDictionary.getCodePoints(word); - final int wordLen = codePoints.length; - - for (int depth = 0; depth < Constants.DICTIONARY_MAX_WORD_LENGTH; ++depth) { - if (wordPos >= wordLen) break; - nodeOriginAddress = dictBuffer.position(); - int nodeParentAddress = -1; - final int ptNodeCount = BinaryDictDecoderUtils.readPtNodeCount(dictBuffer); - boolean foundNextNode = false; - - for (int i = 0; i < ptNodeCount; ++i) { - address = dictBuffer.position(); - final PtNodeInfo currentInfo = dictUpdater.readPtNode(address, - fileHeader.mFormatOptions); - final boolean isMovedNode = BinaryDictIOUtils.isMovedPtNode(currentInfo.mFlags, - fileHeader.mFormatOptions); - if (isMovedNode) continue; - nodeParentAddress = (currentInfo.mParentAddress == FormatSpec.NO_PARENT_ADDRESS) - ? FormatSpec.NO_PARENT_ADDRESS : currentInfo.mParentAddress + address; - boolean matched = true; - for (int p = 0; p < currentInfo.mCharacters.length; ++p) { - if (wordPos + p >= wordLen) { - /* - * splitting - * before - * abcd - ef - * - * insert "abc" - * - * after - * abc - d - ef - */ - final int newNodeAddress = dictBuffer.limit(); - final int flags = BinaryDictEncoderUtils.makePtNodeFlags(p > 1, - isTerminal, 0, hasShortcuts, hasBigrams, false /* isNotAWord */, - false /* isBlackListEntry */, fileHeader.mFormatOptions); - int written = movePtNode(newNodeAddress, currentInfo.mCharacters, p, flags, - frequency, nodeParentAddress, shortcuts, bigrams, destination, - dictUpdater, nodeOriginAddress, address, fileHeader.mFormatOptions); - - final int[] characters2 = Arrays.copyOfRange(currentInfo.mCharacters, p, - currentInfo.mCharacters.length); - if (currentInfo.mChildrenAddress != FormatSpec.NO_CHILDREN_ADDRESS) { - updateParentAddresses(dictUpdater, currentInfo.mChildrenAddress, - newNodeAddress + written + 1, fileHeader.mFormatOptions); - } - final PtNodeInfo newInfo2 = new PtNodeInfo( - newNodeAddress + written + 1, -1 /* endAddress */, - currentInfo.mFlags, characters2, currentInfo.mFrequency, - newNodeAddress + 1, currentInfo.mChildrenAddress, - currentInfo.mShortcutTargets, currentInfo.mBigrams); - BinaryDictIOUtils.writeNodes(destination, new PtNodeInfo[] { newInfo2 }); - return; - } else if (codePoints[wordPos + p] != currentInfo.mCharacters[p]) { - if (p > 0) { - /* - * splitting - * before - * ab - cd - * - * insert "ac" - * - * after - * a - b - cd - * | - * - c - */ - - final int newNodeAddress = dictBuffer.limit(); - final int childrenAddress = currentInfo.mChildrenAddress; - - // move prefix - final int prefixFlags = BinaryDictEncoderUtils.makePtNodeFlags(p > 1, - false /* isTerminal */, 0 /* childrenAddressSize*/, - false /* hasShortcut */, false /* hasBigrams */, - false /* isNotAWord */, false /* isBlackListEntry */, - fileHeader.mFormatOptions); - int written = movePtNode(newNodeAddress, currentInfo.mCharacters, p, - prefixFlags, -1 /* frequency */, nodeParentAddress, null, null, - destination, dictUpdater, nodeOriginAddress, address, - fileHeader.mFormatOptions); - - final int[] suffixCharacters = Arrays.copyOfRange( - currentInfo.mCharacters, p, currentInfo.mCharacters.length); - if (currentInfo.mChildrenAddress != FormatSpec.NO_CHILDREN_ADDRESS) { - updateParentAddresses(dictUpdater, currentInfo.mChildrenAddress, - newNodeAddress + written + 1, fileHeader.mFormatOptions); - } - final int suffixFlags = BinaryDictEncoderUtils.makePtNodeFlags( - suffixCharacters.length > 1, - (currentInfo.mFlags & FormatSpec.FLAG_IS_TERMINAL) != 0, - 0 /* childrenAddressSize */, - (currentInfo.mFlags & FormatSpec.FLAG_HAS_SHORTCUT_TARGETS) - != 0, - (currentInfo.mFlags & FormatSpec.FLAG_HAS_BIGRAMS) != 0, - isNotAWord, isBlackListEntry, fileHeader.mFormatOptions); - final PtNodeInfo suffixInfo = new PtNodeInfo( - newNodeAddress + written + 1, -1 /* endAddress */, suffixFlags, - suffixCharacters, currentInfo.mFrequency, newNodeAddress + 1, - currentInfo.mChildrenAddress, currentInfo.mShortcutTargets, - currentInfo.mBigrams); - written += BinaryDictIOUtils.computePtNodeSize(suffixInfo, - fileHeader.mFormatOptions) + 1; - - final int[] newCharacters = Arrays.copyOfRange(codePoints, wordPos + p, - codePoints.length); - final int flags = BinaryDictEncoderUtils.makePtNodeFlags( - newCharacters.length > 1, isTerminal, - 0 /* childrenAddressSize */, hasShortcuts, hasBigrams, - isNotAWord, isBlackListEntry, fileHeader.mFormatOptions); - final PtNodeInfo newInfo = new PtNodeInfo( - newNodeAddress + written, -1 /* endAddress */, flags, - newCharacters, frequency, newNodeAddress + 1, - FormatSpec.NO_CHILDREN_ADDRESS, shortcuts, bigrams); - BinaryDictIOUtils.writeNodes(destination, - new PtNodeInfo[] { suffixInfo, newInfo }); - return; - } - matched = false; - break; - } - } - - if (matched) { - if (wordPos + currentInfo.mCharacters.length == wordLen) { - // the word exists in the dictionary. - // only update the PtNode. - final int newNodeAddress = dictBuffer.limit(); - final boolean hasMultipleChars = currentInfo.mCharacters.length > 1; - final int flags = BinaryDictEncoderUtils.makePtNodeFlags(hasMultipleChars, - isTerminal, 0 /* childrenAddressSize */, hasShortcuts, hasBigrams, - isNotAWord, isBlackListEntry, fileHeader.mFormatOptions); - final PtNodeInfo newInfo = new PtNodeInfo(newNodeAddress + 1, - -1 /* endAddress */, flags, currentInfo.mCharacters, frequency, - nodeParentAddress, currentInfo.mChildrenAddress, shortcuts, - bigrams); - movePtNode(destination, dictUpdater, newInfo, nodeOriginAddress, address, - fileHeader.mFormatOptions); - return; - } - wordPos += currentInfo.mCharacters.length; - if (currentInfo.mChildrenAddress == FormatSpec.NO_CHILDREN_ADDRESS) { - /* - * found the prefix of the word. - * make new PtNode and link to the PtNode from this PtNode. - * - * before - * ab - cd - * - * insert "abcde" - * - * after - * ab - cd - e - */ - final int newNodeArrayAddress = dictBuffer.limit(); - updateChildrenAddress(dictUpdater, address, newNodeArrayAddress, - fileHeader.mFormatOptions); - final int newNodeAddress = newNodeArrayAddress + 1; - final boolean hasMultipleChars = (wordLen - wordPos) > 1; - final int flags = BinaryDictEncoderUtils.makePtNodeFlags(hasMultipleChars, - isTerminal, 0 /* childrenAddressSize */, hasShortcuts, hasBigrams, - isNotAWord, isBlackListEntry, fileHeader.mFormatOptions); - final int[] characters = Arrays.copyOfRange(codePoints, wordPos, wordLen); - final PtNodeInfo newInfo = new PtNodeInfo(newNodeAddress, -1, flags, - characters, frequency, address, FormatSpec.NO_CHILDREN_ADDRESS, - shortcuts, bigrams); - BinaryDictIOUtils.writeNodes(destination, new PtNodeInfo[] { newInfo }); - return; - } - dictBuffer.position(currentInfo.mChildrenAddress); - foundNextNode = true; - break; - } - } - - if (foundNextNode) continue; - - // reached the end of the array. - final int linkAddressPosition = dictBuffer.position(); - int nextLink = dictBuffer.readUnsignedInt24(); - if ((nextLink & FormatSpec.MSB24) != 0) { - nextLink = -(nextLink & FormatSpec.SINT24_MAX); - } - if (nextLink == FormatSpec.NO_FORWARD_LINK_ADDRESS) { - /* - * expand this node. - * - * before - * ab - cd - * - * insert "abef" - * - * after - * ab - cd - * | - * - ef - */ - - // change the forward link address. - final int newNodeAddress = dictBuffer.limit(); - dictBuffer.position(linkAddressPosition); - BinaryDictIOUtils.writeSInt24ToBuffer(dictBuffer, newNodeAddress); - - final int[] characters = Arrays.copyOfRange(codePoints, wordPos, wordLen); - final int flags = BinaryDictEncoderUtils.makePtNodeFlags(characters.length > 1, - isTerminal, 0 /* childrenAddressSize */, hasShortcuts, hasBigrams, - isNotAWord, isBlackListEntry, fileHeader.mFormatOptions); - final PtNodeInfo newInfo = new PtNodeInfo(newNodeAddress + 1, - -1 /* endAddress */, flags, characters, frequency, nodeParentAddress, - FormatSpec.NO_CHILDREN_ADDRESS, shortcuts, bigrams); - BinaryDictIOUtils.writeNodes(destination, new PtNodeInfo[]{ newInfo }); - return; - } else { - depth--; - dictBuffer.position(nextLink); - } - } - } -} diff --git a/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java b/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java index b56234f6d..437fa942b 100644 --- a/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java +++ b/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java @@ -40,12 +40,8 @@ public final class FormatSpec { * p | not used 3 bits * t | each unigram and bigram entry has a time stamp? * i | 1 bit, 1 = yes, 0 = no : CONTAINS_TIMESTAMP_FLAG - * o | has bigrams ? 1 bit, 1 = yes, 0 = no : CONTAINS_BIGRAMS_FLAG - * n | FRENCH_LIGATURE_PROCESSING_FLAG - * f | supports dynamic updates ? 1 bit, 1 = yes, 0 = no : SUPPORTS_DYNAMIC_UPDATE - * l | GERMAN_UMLAUT_PROCESSING_FLAG - * a | - * gs + * o | + * nflags * * h | * e | size of the file header, 4bytes @@ -82,45 +78,36 @@ public final class FormatSpec { * s * * f | - * o | IF SUPPORTS_DYNAMIC_UPDATE (defined in the file header) - * r | forward link address, 3byte - * w | 1 byte = bbbbbbbb match - * a | case 1xxxxxxx => -((xxxxxxx << 16) + (next byte << 8) + next byte) - * r | otherwise => (xxxxxxx << 16) + (next byte << 8) + next byte - * d | - * linkaddress + * o | forward link address, 3byte + * r | 1 byte = bbbbbbbb match + * w | case 1xxxxxxx => -((xxxxxxx << 16) + (next byte << 8) + next byte) + * a | otherwise => (xxxxxxx << 16) + (next byte << 8) + next byte + * r | + * dlinkaddress */ /* Node (FusionDictionary.PtNode) layout is as follows: - * | IF !SUPPORTS_DYNAMIC_UPDATE - * | addressType xx : mask with MASK_CHILDREN_ADDRESS_TYPE - * | 2 bits, 00 = no children : FLAG_CHILDREN_ADDRESS_TYPE_NOADDRESS - * f | 01 = 1 byte : FLAG_CHILDREN_ADDRESS_TYPE_ONEBYTE - * l | 10 = 2 bytes : FLAG_CHILDREN_ADDRESS_TYPE_TWOBYTES - * a | 11 = 3 bytes : FLAG_CHILDREN_ADDRESS_TYPE_THREEBYTES - * g | ELSE - * s | is moved ? 2 bits, 11 = no : FLAG_IS_NOT_MOVED - * | This must be the same as FLAG_CHILDREN_ADDRESS_TYPE_THREEBYTES - * | 01 = yes : FLAG_IS_MOVED - * | the new address is stored in the same place as the parent address - * | is deleted? 10 = yes : FLAG_IS_DELETED - * | has several chars ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_MULTIPLE_CHARS - * | has a terminal ? 1 bit, 1 = yes, 0 = no : FLAG_IS_TERMINAL - * | has shortcut targets ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_SHORTCUT_TARGETS + * | is moved ? 2 bits, 11 = no : FLAG_IS_NOT_MOVED + * | This must be the same as FLAG_CHILDREN_ADDRESS_TYPE_THREEBYTES + * | 01 = yes : FLAG_IS_MOVED + * f | the new address is stored in the same place as the parent address + * l | is deleted? 10 = yes : FLAG_IS_DELETED + * a | has several chars ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_MULTIPLE_CHARS + * g | has a terminal ? 1 bit, 1 = yes, 0 = no : FLAG_IS_TERMINAL + * s | has shortcut targets ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_SHORTCUT_TARGETS * | has bigrams ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_BIGRAMS * | is not a word ? 1 bit, 1 = yes, 0 = no : FLAG_IS_NOT_A_WORD * | is blacklisted ? 1 bit, 1 = yes, 0 = no : FLAG_IS_BLACKLISTED * * p | - * a | IF SUPPORTS_DYNAMIC_UPDATE (defined in the file header) - * r | parent address, 3byte - * e | 1 byte = bbbbbbbb match - * n | case 1xxxxxxx => -((0xxxxxxx << 16) + (next byte << 8) + next byte) - * t | otherwise => (bbbbbbbb << 16) + (next byte << 8) + next byte - * a | This address is relative to the head of the PtNode. - * d | If the node doesn't have a parent, this field is set to 0. + * a | parent address, 3byte + * r | 1 byte = bbbbbbbb match + * e | case 1xxxxxxx => -((0xxxxxxx << 16) + (next byte << 8) + next byte) + * n | otherwise => (bbbbbbbb << 16) + (next byte << 8) + next byte + * t | This address is relative to the head of the PtNode. + * a | If the node doesn't have a parent, this field is set to 0. * d | - * ress + * dress * * c | IF FLAG_HAS_MULTIPLE_CHARS * h | char, char, char, char n * (1 or 3 bytes) : use PtNodeInfo for i/o helpers @@ -134,23 +121,16 @@ public final class FormatSpec { * e | frequency 1 byte * q | * - * c | IF SUPPORTS_DYNAMIC_UPDATE - * h | children address, 3 bytes - * i | 1 byte = bbbbbbbb match - * l | case 1xxxxxxx => -((0xxxxxxx << 16) + (next byte << 8) + next byte) - * d | otherwise => (bbbbbbbb<<16) + (next byte << 8) + next byte - * r | if this node doesn't have children, this field is set to 0. - * e | (see BinaryDictEncoderUtils#writeVariableSignedAddress) - * n | ELSIF 00 = FLAG_CHILDREN_ADDRESS_TYPE_NOADDRESS == addressType - * a | // nothing - * d | ELSIF 01 = FLAG_CHILDREN_ADDRESS_TYPE_ONEBYTE == addressType - * d | children address, 1 byte - * r | ELSIF 10 = FLAG_CHILDREN_ADDRESS_TYPE_TWOBYTES == addressType - * e | children address, 2 bytes - * s | ELSE // 11 = FLAG_CHILDREN_ADDRESS_TYPE_THREEBYTES = addressType - * s | children address, 3 bytes - * | END - * | This address is relative to the position of this field. + * c | + * h | children address, 3 bytes + * i | 1 byte = bbbbbbbb match + * l | case 1xxxxxxx => -((0xxxxxxx << 16) + (next byte << 8) + next byte) + * d | otherwise => (bbbbbbbb<<16) + (next byte << 8) + next byte + * r | if this node doesn't have children, this field is set to 0. + * e | (see BinaryDictEncoderUtils#writeVariableSignedAddress) + * n | This address is relative to the position of this field. + * a | + * ddress * * | IF FLAG_IS_TERMINAL && FLAG_HAS_SHORTCUT_TARGETS * | shortcut string list @@ -199,20 +179,21 @@ public final class FormatSpec { */ public static final int MAGIC_NUMBER = 0x9BC13AFE; - static final int MINIMUM_SUPPORTED_VERSION = 2; - static final int MAXIMUM_SUPPORTED_VERSION = 4; static final int NOT_A_VERSION_NUMBER = -1; static final int FIRST_VERSION_WITH_DYNAMIC_UPDATE = 3; static final int FIRST_VERSION_WITH_TERMINAL_ID = 4; - static final int VERSION3 = 3; - static final int VERSION4 = 4; + + // These MUST have the same values as the relevant constants in format_utils.h. + // From version 4 on, we use version * 100 + revision as a version number. That allows + // us to change the format during development while having testing devices remove + // older files with each upgrade, while still having a readable versioning scheme. + public static final int VERSION2 = 2; + public static final int VERSION4 = 401; + static final int MINIMUM_SUPPORTED_VERSION = VERSION2; + static final int MAXIMUM_SUPPORTED_VERSION = VERSION4; // These options need to be the same numeric values as the one in the native reading code. - static final int GERMAN_UMLAUT_PROCESSING_FLAG = 0x1; // TODO: Make the native reading code read this variable. - static final int SUPPORTS_DYNAMIC_UPDATE = 0x2; - static final int FRENCH_LIGATURE_PROCESSING_FLAG = 0x4; - static final int CONTAINS_BIGRAMS_FLAG = 0x8; static final int CONTAINS_TIMESTAMP_FLAG = 0x10; // TODO: Make this value adaptative to content data, store it in the header, and @@ -263,8 +244,10 @@ public final class FormatSpec { static final int PTNODE_ATTRIBUTE_MAX_ADDRESS_SIZE = 3; static final int PTNODE_SHORTCUT_LIST_SIZE_SIZE = 2; - // These values are used only by version 4 or later. + // These values are used only by version 4 or later. They MUST match the definitions in + // ver4_dict_constants.cpp. static final String TRIE_FILE_EXTENSION = ".trie"; + public static final String HEADER_FILE_EXTENSION = ".header"; static final String FREQ_FILE_EXTENSION = ".freq"; static final String UNIGRAM_TIMESTAMP_FILE_EXTENSION = ".timestamp"; // tat = Terminal Address Table @@ -278,9 +261,9 @@ public final class FormatSpec { static final int UNIGRAM_TIMESTAMP_SIZE = 4; // With the English main dictionary as of October 2013, the size of bigram address table is - // is 584KB with the block size being 4. - // This is 91% of that of full address table. - static final int BIGRAM_ADDRESS_TABLE_BLOCK_SIZE = 4; + // is 345KB with the block size being 16. + // This is 54% of that of full address table. + static final int BIGRAM_ADDRESS_TABLE_BLOCK_SIZE = 16; static final int BIGRAM_CONTENT_COUNT = 2; static final int BIGRAM_FREQ_CONTENT_INDEX = 0; static final int BIGRAM_TIMESTAMP_CONTENT_INDEX = 1; @@ -293,7 +276,7 @@ public final class FormatSpec { static final int SHORTCUT_CONTENT_COUNT = 1; static final int SHORTCUT_CONTENT_INDEX = 0; // With the English main dictionary as of October 2013, the size of shortcut address table is - // 29KB with the block size being 64. + // 26KB with the block size being 64. // This is only 4.4% of that of full address table. static final int SHORTCUT_ADDRESS_TABLE_BLOCK_SIZE = 64; static final String SHORTCUT_CONTENT_ID = "_shortcut"; @@ -331,43 +314,36 @@ public final class FormatSpec { */ public static final class FormatOptions { public final int mVersion; - public final boolean mSupportsDynamicUpdate; public final boolean mHasTerminalId; public final boolean mHasTimestamp; - @UsedForTesting - public FormatOptions(final int version) { - this(version, false); - } @UsedForTesting - public FormatOptions(final int version, final boolean supportsDynamicUpdate) { - this(version, supportsDynamicUpdate, false /* hasTimestamp */); + public FormatOptions(final int version) { + this(version, false /* hasTimestamp */); } - public FormatOptions(final int version, final boolean supportsDynamicUpdate, - final boolean hasTimestamp) { + public FormatOptions(final int version, final boolean hasTimestamp) { mVersion = version; - if (version < FIRST_VERSION_WITH_DYNAMIC_UPDATE && supportsDynamicUpdate) { - throw new RuntimeException("Dynamic updates are only supported with versions " - + FIRST_VERSION_WITH_DYNAMIC_UPDATE + " and ulterior."); - } - mSupportsDynamicUpdate = supportsDynamicUpdate; mHasTerminalId = (version >= FIRST_VERSION_WITH_TERMINAL_ID); mHasTimestamp = hasTimestamp; } + + public boolean supportsDynamicUpdate() { + return mVersion >= FIRST_VERSION_WITH_DYNAMIC_UPDATE; + } } /** * Class representing file header. */ public static final class FileHeader { - public final int mHeaderSize; + public final int mBodyOffset; public final DictionaryOptions mDictionaryOptions; public final FormatOptions mFormatOptions; // Note that these are corresponding definitions in native code in latinime::HeaderPolicy // and latinime::HeaderReadWriteUtils. - public static final String SUPPORTS_DYNAMIC_UPDATE_ATTRIBUTE = "SUPPORTS_DYNAMIC_UPDATE"; public static final String USES_FORGETTING_CURVE_ATTRIBUTE = "USES_FORGETTING_CURVE"; + public static final String HAS_HISTORICAL_INFO_ATTRIBUTE = "HAS_HISTORICAL_INFO"; public static final String ATTRIBUTE_VALUE_TRUE = "1"; public static final String DICTIONARY_VERSION_ATTRIBUTE = "version"; @@ -375,10 +351,20 @@ public final class FormatSpec { public static final String DICTIONARY_ID_ATTRIBUTE = "dictionary"; private static final String DICTIONARY_DESCRIPTION_ATTRIBUTE = "description"; public FileHeader(final int headerSize, final DictionaryOptions dictionaryOptions, - final FormatOptions formatOptions) { - mHeaderSize = headerSize; + final FormatOptions formatOptions) throws UnsupportedFormatException { mDictionaryOptions = dictionaryOptions; mFormatOptions = formatOptions; + mBodyOffset = formatOptions.mVersion < VERSION4 ? headerSize : 0; + if (null == getLocaleString()) { + throw new UnsupportedFormatException("Cannot create a FileHeader without a locale"); + } + if (null == getVersion()) { + throw new UnsupportedFormatException( + "Cannot create a FileHeader without a version"); + } + if (null == getId()) { + throw new UnsupportedFormatException("Cannot create a FileHeader without an ID"); + } } // Helper method to get the locale as a String @@ -415,7 +401,7 @@ public final class FormatSpec { if (dictFile.isDirectory()) { return new Ver4DictDecoder(dictFile, bufferType); } else if (dictFile.isFile()) { - return new Ver3DictDecoder(dictFile, bufferType); + return new Ver2DictDecoder(dictFile, bufferType); } return null; } @@ -425,7 +411,7 @@ public final class FormatSpec { if (dictFile.isDirectory()) { return new Ver4DictDecoder(dictFile, factory); } else if (dictFile.isFile()) { - return new Ver3DictDecoder(dictFile, factory); + return new Ver2DictDecoder(dictFile, factory); } return null; } diff --git a/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java b/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java index 3bb218bea..fdf2ae7b5 100644 --- a/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java +++ b/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java @@ -303,14 +303,9 @@ public final class FusionDictionary implements Iterable<Word> { * Options global to the dictionary. */ public static final class DictionaryOptions { - public final boolean mGermanUmlautProcessing; - public final boolean mFrenchLigatureProcessing; public final HashMap<String, String> mAttributes; - public DictionaryOptions(final HashMap<String, String> attributes, - final boolean germanUmlautProcessing, final boolean frenchLigatureProcessing) { + public DictionaryOptions(final HashMap<String, String> attributes) { mAttributes = attributes; - mGermanUmlautProcessing = germanUmlautProcessing; - mFrenchLigatureProcessing = frenchLigatureProcessing; } @Override public String toString() { // Convenience method @@ -339,14 +334,6 @@ public final class FusionDictionary implements Iterable<Word> { } s.append("\n"); } - if (mGermanUmlautProcessing) { - s.append(indent); - s.append("Needs German umlaut processing\n"); - } - if (mFrenchLigatureProcessing) { - s.append(indent); - s.append("Needs French ligature processing\n"); - } return s.toString(); } } @@ -701,138 +688,6 @@ public final class FusionDictionary implements Iterable<Word> { } /** - * Recursively count the number of nodes in a given branch of the trie. - * - * @param nodeArray the node array to count. - * @return the number of nodes in this branch. - */ - public static int countNodeArrays(final PtNodeArray nodeArray) { - int size = 1; - for (int i = nodeArray.mData.size() - 1; i >= 0; --i) { - PtNode ptNode = nodeArray.mData.get(i); - if (null != ptNode.mChildren) - size += countNodeArrays(ptNode.mChildren); - } - return size; - } - - // Recursively find out whether there are any bigrams. - // This can be pretty expensive especially if there aren't any (we return as soon - // as we find one, so it's much cheaper if there are bigrams) - private static boolean hasBigramsInternal(final PtNodeArray nodeArray) { - if (null == nodeArray) return false; - for (int i = nodeArray.mData.size() - 1; i >= 0; --i) { - PtNode ptNode = nodeArray.mData.get(i); - if (null != ptNode.mBigrams) return true; - if (hasBigramsInternal(ptNode.mChildren)) return true; - } - return false; - } - - /** - * Finds out whether there are any bigrams in this dictionary. - * - * @return true if there is any bigram, false otherwise. - */ - // TODO: this is expensive especially for large dictionaries without any bigram. - // The up side is, this is always accurate and correct and uses no memory. We should - // find a more efficient way of doing this, without compromising too much on memory - // and ease of use. - public boolean hasBigrams() { - return hasBigramsInternal(mRootNodeArray); - } - - // Historically, the tails of the words were going to be merged to save space. - // However, that would prevent the code to search for a specific address in log(n) - // time so this was abandoned. - // The code is still of interest as it does add some compression to any dictionary - // that has no need for attributes. Implementations that does not read attributes should be - // able to read a dictionary with merged tails. - // Also, the following code does support frequencies, as in, it will only merges - // tails that share the same frequency. Though it would result in the above loss of - // performance while searching by address, it is still technically possible to merge - // tails that contain attributes, but this code does not take that into account - it does - // not compare attributes and will merge terminals with different attributes regardless. - public void mergeTails() { - MakedictLog.i("Do not merge tails"); - return; - -// MakedictLog.i("Merging PtNodes. Number of PtNodes : " + countPtNodes(root)); -// MakedictLog.i("Number of PtNodes : " + countPtNodes(root)); -// -// final HashMap<String, ArrayList<PtNodeArray>> repository = -// new HashMap<String, ArrayList<PtNodeArray>>(); -// mergeTailsInner(repository, root); -// -// MakedictLog.i("Number of different pseudohashes : " + repository.size()); -// int size = 0; -// for (ArrayList<PtNodeArray> a : repository.values()) { -// size += a.size(); -// } -// MakedictLog.i("Number of nodes after merge : " + (1 + size)); -// MakedictLog.i("Recursively seen nodes : " + countNodes(root)); - } - - // The following methods are used by the deactivated mergeTails() -// private static boolean isEqual(PtNodeArray a, PtNodeArray b) { -// if (null == a && null == b) return true; -// if (null == a || null == b) return false; -// if (a.data.size() != b.data.size()) return false; -// final int size = a.data.size(); -// for (int i = size - 1; i >= 0; --i) { -// PtNode aPtNode = a.data.get(i); -// PtNode bPtNode = b.data.get(i); -// if (aPtNode.frequency != bPtNode.frequency) return false; -// if (aPtNode.alternates == null && bPtNode.alternates != null) return false; -// if (aPtNode.alternates != null && !aPtNode.equals(bPtNode.alternates)) return false; -// if (!Arrays.equals(aPtNode.chars, bPtNode.chars)) return false; -// if (!isEqual(aPtNode.children, bPtNode.children)) return false; -// } -// return true; -// } - -// static private HashMap<String, ArrayList<PtNodeArray>> mergeTailsInner( -// final HashMap<String, ArrayList<PtNodeArray>> map, final PtNodeArray nodeArray) { -// final ArrayList<PtNode> branches = nodeArray.data; -// final int nodeSize = branches.size(); -// for (int i = 0; i < nodeSize; ++i) { -// PtNode ptNode = branches.get(i); -// if (null != ptNode.children) { -// String pseudoHash = getPseudoHash(ptNode.children); -// ArrayList<PtNodeArray> similarList = map.get(pseudoHash); -// if (null == similarList) { -// similarList = new ArrayList<PtNodeArray>(); -// map.put(pseudoHash, similarList); -// } -// boolean merged = false; -// for (PtNodeArray similar : similarList) { -// if (isEqual(ptNode.children, similar)) { -// ptNode.children = similar; -// merged = true; -// break; -// } -// } -// if (!merged) { -// similarList.add(ptNode.children); -// } -// mergeTailsInner(map, ptNode.children); -// } -// } -// return map; -// } - -// private static String getPseudoHash(final PtNodeArray nodeArray) { -// StringBuilder s = new StringBuilder(); -// for (PtNode ptNode : nodeArray.data) { -// s.append(ptNode.frequency); -// for (int ch : ptNode.chars) { -// s.append(Character.toChars(ch)); -// } -// } -// return s.toString(); -// } - - /** * Iterator to walk through a dictionary. * * This is purely for convenience. diff --git a/java/src/com/android/inputmethod/latin/makedict/SparseTableContentReader.java b/java/src/com/android/inputmethod/latin/makedict/SparseTableContentReader.java new file mode 100644 index 000000000..63e1f56f5 --- /dev/null +++ b/java/src/com/android/inputmethod/latin/makedict/SparseTableContentReader.java @@ -0,0 +1,126 @@ +/* + * Copyright (C) 2013 The Android Open Source Project + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package com.android.inputmethod.latin.makedict; + +import com.android.inputmethod.latin.makedict.BinaryDictDecoderUtils.DictBuffer; +import com.android.inputmethod.latin.makedict.DictDecoder.DictionaryBufferFactory; + +import java.io.File; +import java.io.FileNotFoundException; +import java.io.IOException; + +/** + * An auxiliary class for reading SparseTable and data written by SparseTableContentWriter. + */ +public class SparseTableContentReader { + + /** + * An interface of a function which is passed to SparseTableContentReader.read. + */ + public interface SparseTableContentReaderInterface { + /** + * Reads data. + * + * @param buffer the DictBuffer. The position of the buffer is set to the head of data. + */ + public void read(final DictBuffer buffer); + } + + protected final int mContentCount; + protected final int mBlockSize; + protected final File mBaseDir; + protected final File mLookupTableFile; + protected final File[] mAddressTableFiles; + protected final File[] mContentFiles; + protected DictBuffer mLookupTableBuffer; + protected final DictBuffer[] mAddressTableBuffers; + private final DictBuffer[] mContentBuffers; + protected final DictionaryBufferFactory mFactory; + + /** + * Sole constructor of SparseTableContentReader. + * + * @param name the name of SparseTable. + * @param blockSize the block size of the content table. + * @param baseDir the directory which contains the files of the content table. + * @param contentFilenames the file names of content files. + * @param contentSuffixes the ids of contents. These ids are used for a suffix of a name of + * address files and content files. + * @param factory the DictionaryBufferFactory which is used for opening the files. + */ + public SparseTableContentReader(final String name, final int blockSize, final File baseDir, + final String[] contentFilenames, final String[] contentSuffixes, + final DictionaryBufferFactory factory) { + if (contentFilenames.length != contentSuffixes.length) { + throw new RuntimeException("The length of contentFilenames and the length of" + + " contentSuffixes are different " + contentFilenames.length + ", " + + contentSuffixes.length); + } + mBlockSize = blockSize; + mBaseDir = baseDir; + mFactory = factory; + mContentCount = contentFilenames.length; + mLookupTableFile = new File(baseDir, name + FormatSpec.LOOKUP_TABLE_FILE_SUFFIX); + mAddressTableFiles = new File[mContentCount]; + mContentFiles = new File[mContentCount]; + for (int i = 0; i < mContentCount; ++i) { + mAddressTableFiles[i] = new File(mBaseDir, + name + FormatSpec.CONTENT_TABLE_FILE_SUFFIX + contentSuffixes[i]); + mContentFiles[i] = new File(mBaseDir, contentFilenames[i] + contentSuffixes[i]); + } + mAddressTableBuffers = new DictBuffer[mContentCount]; + mContentBuffers = new DictBuffer[mContentCount]; + } + + public void openBuffers() throws FileNotFoundException, IOException { + mLookupTableBuffer = mFactory.getDictionaryBuffer(mLookupTableFile); + for (int i = 0; i < mContentCount; ++i) { + mAddressTableBuffers[i] = mFactory.getDictionaryBuffer(mAddressTableFiles[i]); + mContentBuffers[i] = mFactory.getDictionaryBuffer(mContentFiles[i]); + } + } + + /** + * Calls the read() callback of the reader with the appropriate buffer appropriately positioned. + * @param contentNumber the index in the original contentFilenames[] array. + * @param terminalId the terminal ID to read. + * @param reader the reader on which to call the callback. + */ + protected void read(final int contentNumber, final int terminalId, + final SparseTableContentReaderInterface reader) { + if (terminalId < 0 || (terminalId / mBlockSize) * SparseTable.SIZE_OF_INT_IN_BYTES + >= mLookupTableBuffer.limit()) { + return; + } + + mLookupTableBuffer.position((terminalId / mBlockSize) * SparseTable.SIZE_OF_INT_IN_BYTES); + final int indexInAddressTable = mLookupTableBuffer.readInt(); + if (indexInAddressTable == SparseTable.NOT_EXIST) { + return; + } + + mAddressTableBuffers[contentNumber].position(SparseTable.SIZE_OF_INT_IN_BYTES + * ((indexInAddressTable * mBlockSize) + (terminalId % mBlockSize))); + final int address = mAddressTableBuffers[contentNumber].readInt(); + if (address == SparseTable.NOT_EXIST) { + return; + } + + mContentBuffers[contentNumber].position(address); + reader.read(mContentBuffers[contentNumber]); + } +}
\ No newline at end of file diff --git a/java/src/com/android/inputmethod/latin/makedict/SparseTableContentWriter.java b/java/src/com/android/inputmethod/latin/makedict/SparseTableContentWriter.java new file mode 100644 index 000000000..49f0fd624 --- /dev/null +++ b/java/src/com/android/inputmethod/latin/makedict/SparseTableContentWriter.java @@ -0,0 +1,93 @@ +/* + * Copyright (C) 2013 The Android Open Source Project + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package com.android.inputmethod.latin.makedict; + +import java.io.File; +import java.io.FileNotFoundException; +import java.io.FileOutputStream; +import java.io.IOException; +import java.io.OutputStream; + +/** + * An auxiliary class for writing data associated with SparseTable to files. + */ +public class SparseTableContentWriter { + public interface SparseTableContentWriterInterface { + public void write(final OutputStream outStream) throws IOException; + } + + private final int mContentCount; + private final SparseTable mSparseTable; + private final File mLookupTableFile; + protected final File mBaseDir; + private final File[] mAddressTableFiles; + private final File[] mContentFiles; + protected final OutputStream[] mContentOutStreams; + + /** + * Sole constructor of SparseTableContentWriter. + * + * @param name the name of SparseTable. + * @param initialCapacity the initial capacity of SparseTable. + * @param blockSize the block size of the content table. + * @param baseDir the directory which contains the files of the content table. + * @param contentFilenames the file names of content files. + * @param contentIds the ids of contents. These ids are used for a suffix of a name of address + * files and content files. + */ + public SparseTableContentWriter(final String name, final int initialCapacity, + final int blockSize, final File baseDir, final String[] contentFilenames, + final String[] contentIds) { + if (contentFilenames.length != contentIds.length) { + throw new RuntimeException("The length of contentFilenames and the length of" + + " contentIds are different " + contentFilenames.length + ", " + + contentIds.length); + } + mContentCount = contentFilenames.length; + mSparseTable = new SparseTable(initialCapacity, blockSize, mContentCount); + mLookupTableFile = new File(baseDir, name + FormatSpec.LOOKUP_TABLE_FILE_SUFFIX); + mAddressTableFiles = new File[mContentCount]; + mContentFiles = new File[mContentCount]; + mBaseDir = baseDir; + for (int i = 0; i < mContentCount; ++i) { + mAddressTableFiles[i] = new File(mBaseDir, + name + FormatSpec.CONTENT_TABLE_FILE_SUFFIX + contentIds[i]); + mContentFiles[i] = new File(mBaseDir, contentFilenames[i] + contentIds[i]); + } + mContentOutStreams = new OutputStream[mContentCount]; + } + + public void openStreams() throws FileNotFoundException { + for (int i = 0; i < mContentCount; ++i) { + mContentOutStreams[i] = new FileOutputStream(mContentFiles[i]); + } + } + + protected void write(final int contentIndex, final int index, + final SparseTableContentWriterInterface writer) throws IOException { + mSparseTable.set(contentIndex, index, (int) mContentFiles[contentIndex].length()); + writer.write(mContentOutStreams[contentIndex]); + mContentOutStreams[contentIndex].flush(); + } + + public void closeStreams() throws IOException { + mSparseTable.writeToFiles(mLookupTableFile, mAddressTableFiles); + for (int i = 0; i < mContentCount; ++i) { + mContentOutStreams[i].close(); + } + } +}
\ No newline at end of file diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver3DictDecoder.java b/java/src/com/android/inputmethod/latin/makedict/Ver2DictDecoder.java index acab4f8a5..ea0a2c6c2 100644 --- a/java/src/com/android/inputmethod/latin/makedict/Ver3DictDecoder.java +++ b/java/src/com/android/inputmethod/latin/makedict/Ver2DictDecoder.java @@ -23,7 +23,6 @@ import com.android.inputmethod.latin.makedict.FormatSpec.FileHeader; import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions; import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode; import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; -import com.android.inputmethod.latin.utils.JniUtils; import android.util.Log; @@ -34,18 +33,11 @@ import java.util.ArrayList; import java.util.Arrays; /** - * An implementation of DictDecoder for version 3 binary dictionary. + * An implementation of DictDecoder for version 2 binary dictionary. */ @UsedForTesting -public class Ver3DictDecoder extends AbstractDictDecoder { - private static final String TAG = Ver3DictDecoder.class.getSimpleName(); - - static { - JniUtils.loadNativeLibrary(); - } - - // TODO: implement something sensical instead of just a phony method - private static native int doNothing(); +public class Ver2DictDecoder extends AbstractDictDecoder { + private static final String TAG = Ver2DictDecoder.class.getSimpleName(); protected static class PtNodeReader extends AbstractDictDecoder.PtNodeReader { private static int readFrequency(final DictBuffer dictBuffer) { @@ -57,7 +49,7 @@ public class Ver3DictDecoder extends AbstractDictDecoder { private final DictionaryBufferFactory mBufferFactory; protected DictBuffer mDictBuffer; - /* package */ Ver3DictDecoder(final File file, final int factoryFlag) { + /* package */ Ver2DictDecoder(final File file, final int factoryFlag) { mDictionaryBinaryFile = file; mDictBuffer = null; @@ -72,7 +64,7 @@ public class Ver3DictDecoder extends AbstractDictDecoder { } } - /* package */ Ver3DictDecoder(final File file, final DictionaryBufferFactory factory) { + /* package */ Ver2DictDecoder(final File file, final DictionaryBufferFactory factory) { mDictionaryBinaryFile = file; mBufferFactory = factory; } @@ -166,7 +158,7 @@ public class Ver3DictDecoder extends AbstractDictDecoder { final ArrayList<PendingAttribute> bigrams; if (0 != (flags & FormatSpec.FLAG_HAS_BIGRAMS)) { bigrams = new ArrayList<PendingAttribute>(); - addressPointer += PtNodeReader.readBigramAddresses(mDictBuffer, bigrams, + addressPointer += PtNodeReader.readBigramAddresses(mDictBuffer, bigrams, addressPointer); if (bigrams.size() >= FormatSpec.MAX_BIGRAMS_IN_A_PTNODE) { throw new RuntimeException("Too many bigrams in a PtNode (" + bigrams.size() diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver3DictEncoder.java b/java/src/com/android/inputmethod/latin/makedict/Ver2DictEncoder.java index 5da34534e..e5430423d 100644 --- a/java/src/com/android/inputmethod/latin/makedict/Ver3DictEncoder.java +++ b/java/src/com/android/inputmethod/latin/makedict/Ver2DictEncoder.java @@ -31,16 +31,16 @@ import java.util.ArrayList; import java.util.Iterator; /** - * An implementation of DictEncoder for version 3 binary dictionary. + * An implementation of DictEncoder for version 2 binary dictionary. */ -public class Ver3DictEncoder implements DictEncoder { +public class Ver2DictEncoder implements DictEncoder { private final File mDictFile; private OutputStream mOutStream; private byte[] mBuffer; private int mPosition; - public Ver3DictEncoder(final File dictFile) { + public Ver2DictEncoder(final File dictFile) { mDictFile = dictFile; mOutStream = null; mBuffer = null; @@ -49,7 +49,7 @@ public class Ver3DictEncoder implements DictEncoder { // This constructor is used only by BinaryDictOffdeviceUtilsTests. // If you want to use this in the production code, you should consider keeping consistency of // the interface of Ver3DictDecoder by using factory. - public Ver3DictEncoder(final OutputStream outStream) { + public Ver2DictEncoder(final OutputStream outStream) { mDictFile = null; mOutStream = outStream; } @@ -68,7 +68,7 @@ public class Ver3DictEncoder implements DictEncoder { @Override public void writeDictionary(final FusionDictionary dict, final FormatOptions formatOptions) throws IOException, UnsupportedFormatException { - if (formatOptions.mVersion > FormatSpec.VERSION3) { + if (formatOptions.mVersion > FormatSpec.VERSION2) { throw new UnsupportedFormatException( "The given format options has wrong version number : " + formatOptions.mVersion); @@ -169,7 +169,7 @@ public class Ver3DictEncoder implements DictEncoder { private void writeChildrenPosition(final PtNode ptNode, final FormatOptions formatOptions) { final int childrenPos = BinaryDictEncoderUtils.getChildrenPosition(ptNode, formatOptions); - if (formatOptions.mSupportsDynamicUpdate) { + if (formatOptions.supportsDynamicUpdate()) { mPosition += BinaryDictEncoderUtils.writeSignedChildrenPosition(mBuffer, mPosition, childrenPos); } else { diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver3DictUpdater.java b/java/src/com/android/inputmethod/latin/makedict/Ver3DictUpdater.java deleted file mode 100644 index 07adda625..000000000 --- a/java/src/com/android/inputmethod/latin/makedict/Ver3DictUpdater.java +++ /dev/null @@ -1,82 +0,0 @@ -/* - * Copyright (C) 2013 The Android Open Source Project - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -package com.android.inputmethod.latin.makedict; - -import com.android.inputmethod.annotations.UsedForTesting; -import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; - -import java.io.File; -import java.io.FileNotFoundException; -import java.io.FileOutputStream; -import java.io.IOException; -import java.io.OutputStream; -import java.util.ArrayList; - -/** - * An implementation of DictUpdater for version 3 binary dictionary. - */ -@UsedForTesting -public class Ver3DictUpdater extends Ver3DictDecoder implements DictUpdater { - private OutputStream mOutStream; - - @UsedForTesting - public Ver3DictUpdater(final File dictFile, final int factoryType) { - // DictUpdater must have an updatable DictBuffer. - super(dictFile, ((factoryType & MASK_DICTBUFFER) == USE_BYTEARRAY) - ? USE_BYTEARRAY : USE_WRITABLE_BYTEBUFFER); - mOutStream = null; - } - - private void openStreamAndBuffer() throws FileNotFoundException, IOException { - super.openDictBuffer(); - mOutStream = new FileOutputStream(mDictionaryBinaryFile, true /* append */); - } - - private void close() throws IOException { - if (mOutStream != null) { - mOutStream.close(); - mOutStream = null; - } - } - - @Override @UsedForTesting - public void deleteWord(final String word) throws IOException, UnsupportedFormatException { - if (mOutStream == null) openStreamAndBuffer(); - mDictBuffer.position(0); - readHeader(); - final int wordPos = getTerminalPosition(word); - if (wordPos != FormatSpec.NOT_VALID_WORD) { - mDictBuffer.position(wordPos); - final int flags = mDictBuffer.readUnsignedByte(); - mDictBuffer.position(wordPos); - mDictBuffer.put((byte) DynamicBinaryDictIOUtils.markAsDeleted(flags)); - } - close(); - } - - @Override @UsedForTesting - public void insertWord(final String word, final int frequency, - final ArrayList<WeightedString> bigramStrings, - final ArrayList<WeightedString> shortcuts, - final boolean isNotAWord, final boolean isBlackListEntry) - throws IOException, UnsupportedFormatException { - if (mOutStream == null) openStreamAndBuffer(); - DynamicBinaryDictIOUtils.insertWord(this, mOutStream, word, frequency, bigramStrings, - shortcuts, isNotAWord, isBlackListEntry); - close(); - } -} diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver4DictDecoder.java b/java/src/com/android/inputmethod/latin/makedict/Ver4DictDecoder.java index 734223ec2..7071893d2 100644 --- a/java/src/com/android/inputmethod/latin/makedict/Ver4DictDecoder.java +++ b/java/src/com/android/inputmethod/latin/makedict/Ver4DictDecoder.java @@ -40,26 +40,52 @@ import java.util.Arrays; public class Ver4DictDecoder extends AbstractDictDecoder { private static final String TAG = Ver4DictDecoder.class.getSimpleName(); - private static final int FILETYPE_TRIE = 1; - private static final int FILETYPE_FREQUENCY = 2; - private static final int FILETYPE_TERMINAL_ADDRESS_TABLE = 3; - private static final int FILETYPE_BIGRAM_FREQ = 4; - private static final int FILETYPE_SHORTCUT = 5; - - private final File mDictDirectory; - private final DictionaryBufferFactory mBufferFactory; + protected static final int FILETYPE_TRIE = 1; + protected static final int FILETYPE_FREQUENCY = 2; + protected static final int FILETYPE_TERMINAL_ADDRESS_TABLE = 3; + protected static final int FILETYPE_BIGRAM_FREQ = 4; + protected static final int FILETYPE_SHORTCUT = 5; + protected static final int FILETYPE_HEADER = 6; + + protected final File mDictDirectory; + protected final DictionaryBufferFactory mBufferFactory; protected DictBuffer mDictBuffer; - private DictBuffer mFrequencyBuffer; - private DictBuffer mTerminalAddressTableBuffer; - private DictBuffer mBigramBuffer; - private DictBuffer mShortcutBuffer; - private SparseTable mBigramAddressTable; - private SparseTable mShortcutAddressTable; + protected DictBuffer mHeaderBuffer; + protected DictBuffer mFrequencyBuffer; + protected DictBuffer mTerminalAddressTableBuffer; + private BigramContentReader mBigramReader; + private ShortcutContentReader mShortcutReader; + + /** + * Raw PtNode info straight out of a trie file in version 4 dictionary. + */ + protected static final class Ver4PtNodeInfo { + public final int mFlags; + public final int[] mCharacters; + public final int mTerminalId; + public final int mChildrenPos; + public final int mParentPos; + public final int mNodeSize; + public int mStartIndexOfCharacters; + public int mEndIndexOfCharacters; // exclusive + + public Ver4PtNodeInfo(final int flags, final int[] characters, final int terminalId, + final int childrenPos, final int parentPos, final int nodeSize) { + mFlags = flags; + mCharacters = characters; + mTerminalId = terminalId; + mChildrenPos = childrenPos; + mParentPos = parentPos; + mNodeSize = nodeSize; + mStartIndexOfCharacters = 0; + mEndIndexOfCharacters = characters.length; + } + } @UsedForTesting /* package */ Ver4DictDecoder(final File dictDirectory, final int factoryFlag) { mDictDirectory = dictDirectory; - mDictBuffer = mFrequencyBuffer = null; + mDictBuffer = mHeaderBuffer = mFrequencyBuffer = null; if ((factoryFlag & MASK_DICTBUFFER) == USE_READONLY_BYTEBUFFER) { mBufferFactory = new DictionaryBufferFromReadOnlyByteBufferFactory(); @@ -76,13 +102,16 @@ public class Ver4DictDecoder extends AbstractDictDecoder { /* package */ Ver4DictDecoder(final File dictDirectory, final DictionaryBufferFactory factory) { mDictDirectory = dictDirectory; mBufferFactory = factory; - mDictBuffer = mFrequencyBuffer = null; + mDictBuffer = mHeaderBuffer = mFrequencyBuffer = null; } - private File getFile(final int fileType) { + protected File getFile(final int fileType) throws UnsupportedFormatException { if (fileType == FILETYPE_TRIE) { return new File(mDictDirectory, mDictDirectory.getName() + FormatSpec.TRIE_FILE_EXTENSION); + } else if (fileType == FILETYPE_HEADER) { + return new File(mDictDirectory, + mDictDirectory.getName() + FormatSpec.HEADER_FILE_EXTENSION); } else if (fileType == FILETYPE_FREQUENCY) { return new File(mDictDirectory, mDictDirectory.getName() + FormatSpec.FREQ_FILE_EXTENSION); @@ -98,20 +127,27 @@ public class Ver4DictDecoder extends AbstractDictDecoder { mDictDirectory.getName() + FormatSpec.SHORTCUT_FILE_EXTENSION + FormatSpec.SHORTCUT_CONTENT_ID); } else { - throw new RuntimeException("Unsupported kind of file : " + fileType); + throw new UnsupportedFormatException("Unsupported kind of file : " + fileType); } } @Override - public void openDictBuffer() throws FileNotFoundException, IOException { + public void openDictBuffer() throws FileNotFoundException, IOException, + UnsupportedFormatException { + if (!mDictDirectory.isDirectory()) { + throw new UnsupportedFormatException("Format 4 dictionary needs a directory"); + } + mHeaderBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_HEADER)); mDictBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_TRIE)); mFrequencyBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_FREQUENCY)); mTerminalAddressTableBuffer = mBufferFactory.getDictionaryBuffer( getFile(FILETYPE_TERMINAL_ADDRESS_TABLE)); - mBigramBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_BIGRAM_FREQ)); - loadBigramAddressSparseTable(); - mShortcutBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_SHORTCUT)); - loadShortcutAddressSparseTable(); + mBigramReader = new BigramContentReader(mDictDirectory.getName(), + mDictDirectory, mBufferFactory, false); + mBigramReader.openBuffers(); + mShortcutReader = new ShortcutContentReader(mDictDirectory.getName(), mDictDirectory, + mBufferFactory); + mShortcutReader.openBuffers(); } @Override @@ -119,46 +155,134 @@ public class Ver4DictDecoder extends AbstractDictDecoder { return mDictBuffer != null; } + @UsedForTesting + /* package */ DictBuffer getHeaderBuffer() { + return mHeaderBuffer; + } + + @UsedForTesting /* package */ DictBuffer getDictBuffer() { return mDictBuffer; } @Override public FileHeader readHeader() throws IOException, UnsupportedFormatException { - if (mDictBuffer == null) { + if (mHeaderBuffer == null) { openDictBuffer(); } - final FileHeader header = super.readHeader(mDictBuffer); + mHeaderBuffer.position(0); + final FileHeader header = super.readHeader(mHeaderBuffer); final int version = header.mFormatOptions.mVersion; - if (version != 4) { + if (version != FormatSpec.VERSION4) { throw new UnsupportedFormatException("File header has a wrong version : " + version); } return header; } - private void loadBigramAddressSparseTable() throws IOException { - final File lookupIndexFile = new File(mDictDirectory, mDictDirectory.getName() - + FormatSpec.BIGRAM_FILE_EXTENSION + FormatSpec.LOOKUP_TABLE_FILE_SUFFIX); - final File freqsFile = new File(mDictDirectory, mDictDirectory.getName() - + FormatSpec.BIGRAM_FILE_EXTENSION + FormatSpec.CONTENT_TABLE_FILE_SUFFIX - + FormatSpec.BIGRAM_FREQ_CONTENT_ID); - mBigramAddressTable = SparseTable.readFromFiles(lookupIndexFile, new File[] { freqsFile }, - FormatSpec.BIGRAM_ADDRESS_TABLE_BLOCK_SIZE); + /** + * An auxiliary class for reading bigrams. + */ + protected static class BigramContentReader extends SparseTableContentReader { + public BigramContentReader(final String name, final File baseDir, + final DictionaryBufferFactory factory, final boolean hasTimestamp) { + super(name + FormatSpec.BIGRAM_FILE_EXTENSION, + FormatSpec.BIGRAM_ADDRESS_TABLE_BLOCK_SIZE, baseDir, + getContentFilenames(name, hasTimestamp), getContentIds(hasTimestamp), factory); + } + + // TODO: Consolidate this method and BigramContentWriter.getContentFilenames. + protected static String[] getContentFilenames(final String name, + final boolean hasTimestamp) { + final String[] contentFilenames; + if (hasTimestamp) { + contentFilenames = new String[] { name + FormatSpec.BIGRAM_FILE_EXTENSION, + name + FormatSpec.BIGRAM_FILE_EXTENSION }; + } else { + contentFilenames = new String[] { name + FormatSpec.BIGRAM_FILE_EXTENSION }; + } + return contentFilenames; + } + + // TODO: Consolidate this method and BigramContentWriter.getContentIds. + protected static String[] getContentIds(final boolean hasTimestamp) { + final String[] contentIds; + if (hasTimestamp) { + contentIds = new String[] { FormatSpec.BIGRAM_FREQ_CONTENT_ID, + FormatSpec.BIGRAM_TIMESTAMP_CONTENT_ID }; + } else { + contentIds = new String[] { FormatSpec.BIGRAM_FREQ_CONTENT_ID }; + } + return contentIds; + } + + public ArrayList<PendingAttribute> readTargetsAndFrequencies(final int terminalId, + final DictBuffer terminalAddressTableBuffer) { + final ArrayList<PendingAttribute> bigrams = CollectionUtils.newArrayList(); + read(FormatSpec.BIGRAM_FREQ_CONTENT_INDEX, terminalId, + new SparseTableContentReaderInterface() { + @Override + public void read(final DictBuffer buffer) { + while (bigrams.size() < FormatSpec.MAX_BIGRAMS_IN_A_PTNODE) { + // If bigrams.size() reaches FormatSpec.MAX_BIGRAMS_IN_A_PTNODE, + // remaining bigram entries are ignored. + final int bigramFlags = buffer.readUnsignedByte(); + final int targetTerminalId = buffer.readUnsignedInt24(); + terminalAddressTableBuffer.position(targetTerminalId + * FormatSpec.TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE); + final int targetAddress = + terminalAddressTableBuffer.readUnsignedInt24(); + bigrams.add(new PendingAttribute(bigramFlags + & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY, + targetAddress)); + if (0 == (bigramFlags + & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_HAS_NEXT)) { + break; + } + } + if (bigrams.size() >= FormatSpec.MAX_BIGRAMS_IN_A_PTNODE) { + throw new RuntimeException("Too many bigrams in a PtNode (" + + bigrams.size() + " but max is " + + FormatSpec.MAX_BIGRAMS_IN_A_PTNODE + ")"); + } + } + }); + if (bigrams.isEmpty()) return null; + return bigrams; + } } - // TODO: Let's have something like SparseTableContentsReader in this class. - private void loadShortcutAddressSparseTable() throws IOException { - final File lookupIndexFile = new File(mDictDirectory, mDictDirectory.getName() - + FormatSpec.SHORTCUT_FILE_EXTENSION + FormatSpec.LOOKUP_TABLE_FILE_SUFFIX); - final File contentFile = new File(mDictDirectory, mDictDirectory.getName() - + FormatSpec.SHORTCUT_FILE_EXTENSION + FormatSpec.CONTENT_TABLE_FILE_SUFFIX - + FormatSpec.SHORTCUT_CONTENT_ID); - final File timestampsFile = new File(mDictDirectory, mDictDirectory.getName() - + FormatSpec.SHORTCUT_FILE_EXTENSION + FormatSpec.CONTENT_TABLE_FILE_SUFFIX - + FormatSpec.SHORTCUT_CONTENT_ID); - mShortcutAddressTable = SparseTable.readFromFiles(lookupIndexFile, - new File[] { contentFile, timestampsFile }, - FormatSpec.SHORTCUT_ADDRESS_TABLE_BLOCK_SIZE); + /** + * An auxiliary class for reading shortcuts. + */ + protected static class ShortcutContentReader extends SparseTableContentReader { + public ShortcutContentReader(final String name, final File baseDir, + final DictionaryBufferFactory factory) { + super(name + FormatSpec.SHORTCUT_FILE_EXTENSION, + FormatSpec.SHORTCUT_ADDRESS_TABLE_BLOCK_SIZE, baseDir, + new String[] { name + FormatSpec.SHORTCUT_FILE_EXTENSION }, + new String[] { FormatSpec.SHORTCUT_CONTENT_ID }, factory); + } + + public ArrayList<WeightedString> readShortcuts(final int terminalId) { + final ArrayList<WeightedString> shortcuts = CollectionUtils.newArrayList(); + read(FormatSpec.SHORTCUT_CONTENT_INDEX, terminalId, + new SparseTableContentReaderInterface() { + @Override + public void read(final DictBuffer buffer) { + while (true) { + final int flags = buffer.readUnsignedByte(); + final String word = CharEncoding.readString(buffer); + shortcuts.add(new WeightedString(word, + flags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY)); + if (0 == (flags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_HAS_NEXT)) { + break; + } + } + } + }); + if (shortcuts.isEmpty()) return null; + return shortcuts; + } } protected static class PtNodeReader extends AbstractDictDecoder.PtNodeReader { @@ -172,102 +296,82 @@ public class Ver4DictDecoder extends AbstractDictDecoder { } } - private ArrayList<WeightedString> readShortcuts(final int terminalId) { - if (mShortcutAddressTable.get(0, terminalId) == SparseTable.NOT_EXIST) return null; - - final ArrayList<WeightedString> ret = CollectionUtils.newArrayList(); - final int posOfShortcuts = mShortcutAddressTable.get(FormatSpec.SHORTCUT_CONTENT_INDEX, - terminalId); - mShortcutBuffer.position(posOfShortcuts); - while (true) { - final int flags = mShortcutBuffer.readUnsignedByte(); - final String word = CharEncoding.readString(mShortcutBuffer); - ret.add(new WeightedString(word, - flags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY)); - if (0 == (flags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_HAS_NEXT)) break; - } - return ret; - } + private final int[] mCharacterBufferForReadingVer4PtNodeInfo + = new int[FormatSpec.MAX_WORD_LENGTH]; + /** + * Reads PtNode from ptNodePos in the trie file and returns Ver4PtNodeInfo. + * + * @param ptNodePos the position of PtNode. + * @param options the format options. + * @return Ver4PtNodeInfo. + */ // TODO: Make this buffer thread safe. // TODO: Support words longer than FormatSpec.MAX_WORD_LENGTH. - private final int[] mCharacterBuffer = new int[FormatSpec.MAX_WORD_LENGTH]; - @Override - public PtNodeInfo readPtNode(int ptNodePos, FormatOptions options) { - int addressPointer = ptNodePos; + protected Ver4PtNodeInfo readVer4PtNodeInfo(final int ptNodePos, final FormatOptions options) { + int readingPos = ptNodePos; final int flags = PtNodeReader.readPtNodeOptionFlags(mDictBuffer); - addressPointer += FormatSpec.PTNODE_FLAGS_SIZE; + readingPos += FormatSpec.PTNODE_FLAGS_SIZE; - final int parentAddress = PtNodeReader.readParentAddress(mDictBuffer, options); + final int parentPos = PtNodeReader.readParentAddress(mDictBuffer, options); if (BinaryDictIOUtils.supportsDynamicUpdate(options)) { - addressPointer += FormatSpec.PARENT_ADDRESS_SIZE; + readingPos += FormatSpec.PARENT_ADDRESS_SIZE; } final int characters[]; if (0 != (flags & FormatSpec.FLAG_HAS_MULTIPLE_CHARS)) { int index = 0; int character = CharEncoding.readChar(mDictBuffer); - addressPointer += CharEncoding.getCharSize(character); + readingPos += CharEncoding.getCharSize(character); while (FormatSpec.INVALID_CHARACTER != character && index < FormatSpec.MAX_WORD_LENGTH) { - mCharacterBuffer[index++] = character; + mCharacterBufferForReadingVer4PtNodeInfo[index++] = character; character = CharEncoding.readChar(mDictBuffer); - addressPointer += CharEncoding.getCharSize(character); + readingPos += CharEncoding.getCharSize(character); } - characters = Arrays.copyOfRange(mCharacterBuffer, 0, index); + characters = Arrays.copyOfRange(mCharacterBufferForReadingVer4PtNodeInfo, 0, index); } else { final int character = CharEncoding.readChar(mDictBuffer); - addressPointer += CharEncoding.getCharSize(character); + readingPos += CharEncoding.getCharSize(character); characters = new int[] { character }; } final int terminalId; if (0 != (FormatSpec.FLAG_IS_TERMINAL & flags)) { terminalId = PtNodeReader.readTerminalId(mDictBuffer); - addressPointer += FormatSpec.PTNODE_TERMINAL_ID_SIZE; + readingPos += FormatSpec.PTNODE_TERMINAL_ID_SIZE; } else { terminalId = PtNode.NOT_A_TERMINAL; } + int childrenPos = PtNodeReader.readChildrenAddress(mDictBuffer, flags, options); + if (childrenPos != FormatSpec.NO_CHILDREN_ADDRESS) { + childrenPos += readingPos; + } + readingPos += BinaryDictIOUtils.getChildrenAddressSize(flags, options); + + return new Ver4PtNodeInfo(flags, characters, terminalId, childrenPos, parentPos, + readingPos - ptNodePos); + } + + @Override + public PtNodeInfo readPtNode(int ptNodePos, FormatOptions options) { + final Ver4PtNodeInfo nodeInfo = readVer4PtNodeInfo(ptNodePos, options); + final int frequency; - if (0 != (FormatSpec.FLAG_IS_TERMINAL & flags)) { - frequency = PtNodeReader.readFrequency(mFrequencyBuffer, terminalId); + if (0 != (FormatSpec.FLAG_IS_TERMINAL & nodeInfo.mFlags)) { + frequency = PtNodeReader.readFrequency(mFrequencyBuffer, nodeInfo.mTerminalId); } else { frequency = PtNode.NOT_A_TERMINAL; } - int childrenAddress = PtNodeReader.readChildrenAddress(mDictBuffer, flags, options); - if (childrenAddress != FormatSpec.NO_CHILDREN_ADDRESS) { - childrenAddress += addressPointer; - } - addressPointer += BinaryDictIOUtils.getChildrenAddressSize(flags, options); - final ArrayList<WeightedString> shortcutTargets = readShortcuts(terminalId); - - final ArrayList<PendingAttribute> bigrams; - if (0 != (flags & FormatSpec.FLAG_HAS_BIGRAMS)) { - bigrams = new ArrayList<PendingAttribute>(); - final int posOfBigrams = mBigramAddressTable.get(0 /* contentTableIndex */, terminalId); - mBigramBuffer.position(posOfBigrams); - while (bigrams.size() < FormatSpec.MAX_BIGRAMS_IN_A_PTNODE) { - // If bigrams.size() reaches FormatSpec.MAX_BIGRAMS_IN_A_PTNODE, - // remaining bigram entries are ignored. - final int bigramFlags = mBigramBuffer.readUnsignedByte(); - final int targetTerminalId = mBigramBuffer.readUnsignedInt24(); - mTerminalAddressTableBuffer.position( - targetTerminalId * FormatSpec.TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE); - final int targetAddress = mTerminalAddressTableBuffer.readUnsignedInt24(); - bigrams.add(new PendingAttribute( - bigramFlags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY, - targetAddress)); - if (0 == (bigramFlags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_HAS_NEXT)) break; - } - if (bigrams.size() >= FormatSpec.MAX_BIGRAMS_IN_A_PTNODE) { - throw new RuntimeException("Too many bigrams in a PtNode (" + bigrams.size() - + " but max is " + FormatSpec.MAX_BIGRAMS_IN_A_PTNODE + ")"); - } - } else { - bigrams = null; - } - return new PtNodeInfo(ptNodePos, addressPointer, flags, characters, frequency, - parentAddress, childrenAddress, shortcutTargets, bigrams); + + final ArrayList<WeightedString> shortcutTargets = mShortcutReader.readShortcuts( + nodeInfo.mTerminalId); + final ArrayList<PendingAttribute> bigrams = mBigramReader.readTargetsAndFrequencies( + nodeInfo.mTerminalId, mTerminalAddressTableBuffer); + + return new PtNodeInfo(ptNodePos, ptNodePos + nodeInfo.mNodeSize, nodeInfo.mFlags, + nodeInfo.mCharacters, frequency, nodeInfo.mParentPos, nodeInfo.mChildrenPos, + shortcutTargets, bigrams); } private void deleteDictFiles() { @@ -318,10 +422,14 @@ public class Ver4DictDecoder extends AbstractDictDecoder { @Override public boolean readAndFollowForwardLink() { - final int nextAddress = mDictBuffer.readUnsignedInt24(); - if (nextAddress >= 0 && nextAddress < mDictBuffer.limit()) { - mDictBuffer.position(nextAddress); - return true; + final int forwardLinkPos = mDictBuffer.position(); + int nextRelativePos = BinaryDictDecoderUtils.readSInt24(mDictBuffer); + if (nextRelativePos != FormatSpec.NO_FORWARD_LINK_ADDRESS) { + final int nextPos = forwardLinkPos + nextRelativePos; + if (nextPos >= 0 && nextPos < mDictBuffer.limit()) { + mDictBuffer.position(nextPos); + return true; + } } return false; } @@ -332,6 +440,7 @@ public class Ver4DictDecoder extends AbstractDictDecoder { } @Override + @UsedForTesting public void skipPtNode(final FormatOptions formatOptions) { final int flags = PtNodeReader.readPtNodeOptionFlags(mDictBuffer); PtNodeReader.readParentAddress(mDictBuffer, formatOptions); diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver4DictEncoder.java b/java/src/com/android/inputmethod/latin/makedict/Ver4DictEncoder.java index 8d5b48a9b..d34aa171e 100644 --- a/java/src/com/android/inputmethod/latin/makedict/Ver4DictEncoder.java +++ b/java/src/com/android/inputmethod/latin/makedict/Ver4DictEncoder.java @@ -1,5 +1,4 @@ /* -/* * Copyright (C) 2013 The Android Open Source Project * * Licensed under the Apache License, Version 2.0 (the "License"); @@ -18,21 +17,15 @@ package com.android.inputmethod.latin.makedict; import com.android.inputmethod.annotations.UsedForTesting; -import com.android.inputmethod.latin.makedict.BinaryDictDecoderUtils.CharEncoding; -import com.android.inputmethod.latin.makedict.FormatSpec.FileHeader; +import com.android.inputmethod.latin.BinaryDictionary; +import com.android.inputmethod.latin.Dictionary; import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions; -import com.android.inputmethod.latin.makedict.FusionDictionary.DictionaryOptions; import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode; -import com.android.inputmethod.latin.makedict.FusionDictionary.PtNodeArray; import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; +import com.android.inputmethod.latin.utils.LocaleUtils; import java.io.File; -import java.io.FileNotFoundException; -import java.io.FileOutputStream; import java.io.IOException; -import java.io.OutputStream; -import java.util.ArrayList; -import java.util.Iterator; /** * An implementation of DictEncoder for version 4 binary dictionary. @@ -40,244 +33,19 @@ import java.util.Iterator; @UsedForTesting public class Ver4DictEncoder implements DictEncoder { private final File mDictPlacedDir; - private byte[] mTrieBuf; - private int mTriePos; - private int mHeaderSize; - private OutputStream mTrieOutStream; - private OutputStream mFreqOutStream; - private OutputStream mUnigramTimestampOutStream; - private OutputStream mTerminalAddressTableOutStream; - private File mDictDir; - private String mBaseFilename; - private BigramContentWriter mBigramWriter; - private ShortcutContentWriter mShortcutWriter; @UsedForTesting public Ver4DictEncoder(final File dictPlacedDir) { mDictPlacedDir = dictPlacedDir; } - private interface SparseTableContentWriterInterface { - public void write(final OutputStream outStream) throws IOException; - } - - private static class SparseTableContentWriter { - private final int mContentCount; - private final SparseTable mSparseTable; - private final File mLookupTableFile; - protected final File mBaseDir; - private final File[] mAddressTableFiles; - private final File[] mContentFiles; - protected final OutputStream[] mContentOutStreams; - - public SparseTableContentWriter(final String name, final int initialCapacity, - final int blockSize, final File baseDir, final String[] contentFilenames, - final String[] contentIds) { - if (contentFilenames.length != contentIds.length) { - throw new RuntimeException("The length of contentFilenames and the length of" - + " contentIds are different " + contentFilenames.length + ", " - + contentIds.length); - } - mContentCount = contentFilenames.length; - mSparseTable = new SparseTable(initialCapacity, blockSize, mContentCount); - mLookupTableFile = new File(baseDir, name + FormatSpec.LOOKUP_TABLE_FILE_SUFFIX); - mAddressTableFiles = new File[mContentCount]; - mContentFiles = new File[mContentCount]; - mBaseDir = baseDir; - for (int i = 0; i < mContentCount; ++i) { - mAddressTableFiles[i] = new File(mBaseDir, - name + FormatSpec.CONTENT_TABLE_FILE_SUFFIX + contentIds[i]); - mContentFiles[i] = new File(mBaseDir, contentFilenames[i] + contentIds[i]); - } - mContentOutStreams = new OutputStream[mContentCount]; - } - - public void openStreams() throws FileNotFoundException { - for (int i = 0; i < mContentCount; ++i) { - mContentOutStreams[i] = new FileOutputStream(mContentFiles[i]); - } - } - - protected void write(final int contentIndex, final int index, - final SparseTableContentWriterInterface writer) throws IOException { - mSparseTable.set(contentIndex, index, (int) mContentFiles[contentIndex].length()); - writer.write(mContentOutStreams[contentIndex]); - mContentOutStreams[contentIndex].flush(); - } - - public void closeStreams() throws IOException { - mSparseTable.writeToFiles(mLookupTableFile, mAddressTableFiles); - for (int i = 0; i < mContentCount; ++i) { - mContentOutStreams[i].close(); - } - } - } - - private static class BigramContentWriter extends SparseTableContentWriter { - private final boolean mWriteTimestamp; - - public BigramContentWriter(final String name, final int initialCapacity, - final File baseDir, final boolean writeTimestamp) { - super(name + FormatSpec.BIGRAM_FILE_EXTENSION, initialCapacity, - FormatSpec.BIGRAM_ADDRESS_TABLE_BLOCK_SIZE, baseDir, - getContentFilenames(name, writeTimestamp), getContentIds(writeTimestamp)); - mWriteTimestamp = writeTimestamp; - } - - private static String[] getContentFilenames(final String name, - final boolean writeTimestamp) { - final String[] contentFilenames; - if (writeTimestamp) { - contentFilenames = new String[] { name + FormatSpec.BIGRAM_FILE_EXTENSION, - name + FormatSpec.BIGRAM_FILE_EXTENSION }; - } else { - contentFilenames = new String[] { name + FormatSpec.BIGRAM_FILE_EXTENSION }; - } - return contentFilenames; - } - - private static String[] getContentIds(final boolean writeTimestamp) { - final String[] contentIds; - if (writeTimestamp) { - contentIds = new String[] { FormatSpec.BIGRAM_FREQ_CONTENT_ID, - FormatSpec.BIGRAM_TIMESTAMP_CONTENT_ID }; - } else { - contentIds = new String[] { FormatSpec.BIGRAM_FREQ_CONTENT_ID }; - } - return contentIds; - } - - public void writeBigramsForOneWord(final int terminalId, final int bigramCount, - final Iterator<WeightedString> bigramIterator, final FusionDictionary dict) - throws IOException { - write(FormatSpec.BIGRAM_FREQ_CONTENT_INDEX, terminalId, - new SparseTableContentWriterInterface() { - @Override - public void write(final OutputStream outStream) throws IOException { - writeBigramsForOneWordInternal(outStream, bigramIterator, dict); - }}); - if (mWriteTimestamp) { - write(FormatSpec.BIGRAM_TIMESTAMP_CONTENT_INDEX, terminalId, - new SparseTableContentWriterInterface() { - @Override - public void write(final OutputStream outStream) throws IOException { - initBigramTimestampsCountersAndLevelsForOneWordInternal(outStream, - bigramCount); - }}); - } - } - - private void writeBigramsForOneWordInternal(final OutputStream outStream, - final Iterator<WeightedString> bigramIterator, final FusionDictionary dict) - throws IOException { - while (bigramIterator.hasNext()) { - final WeightedString bigram = bigramIterator.next(); - final PtNode target = - FusionDictionary.findWordInTree(dict.mRootNodeArray, bigram.mWord); - final int unigramFrequencyForThisWord = target.mFrequency; - final int bigramFlags = BinaryDictEncoderUtils.makeBigramFlags( - bigramIterator.hasNext(), 0, bigram.mFrequency, - unigramFrequencyForThisWord, bigram.mWord); - BinaryDictEncoderUtils.writeUIntToStream(outStream, bigramFlags, - FormatSpec.PTNODE_ATTRIBUTE_FLAGS_SIZE); - BinaryDictEncoderUtils.writeUIntToStream(outStream, target.mTerminalId, - FormatSpec.PTNODE_ATTRIBUTE_MAX_ADDRESS_SIZE); - } - } - - private void initBigramTimestampsCountersAndLevelsForOneWordInternal( - final OutputStream outStream, final int bigramCount) throws IOException { - for (int i = 0; i < bigramCount; ++i) { - // TODO: Figure out what initial values should be. - BinaryDictEncoderUtils.writeUIntToStream(outStream, 0 /* value */, - FormatSpec.BIGRAM_TIMESTAMP_SIZE); - BinaryDictEncoderUtils.writeUIntToStream(outStream, 0 /* value */, - FormatSpec.BIGRAM_COUNTER_SIZE); - BinaryDictEncoderUtils.writeUIntToStream(outStream, 0 /* value */, - FormatSpec.BIGRAM_LEVEL_SIZE); - } - } - } - - private static class ShortcutContentWriter extends SparseTableContentWriter { - public ShortcutContentWriter(final String name, final int initialCapacity, - final File baseDir) { - super(name + FormatSpec.SHORTCUT_FILE_EXTENSION, initialCapacity, - FormatSpec.SHORTCUT_ADDRESS_TABLE_BLOCK_SIZE, baseDir, - new String[] { name + FormatSpec.SHORTCUT_FILE_EXTENSION }, - new String[] { FormatSpec.SHORTCUT_CONTENT_ID }); - } - - public void writeShortcutForOneWord(final int terminalId, - final Iterator<WeightedString> shortcutIterator) throws IOException { - write(FormatSpec.SHORTCUT_CONTENT_INDEX, terminalId, - new SparseTableContentWriterInterface() { - @Override - public void write(final OutputStream outStream) throws IOException { - writeShortcutForOneWordInternal(outStream, shortcutIterator); - } - }); - } - - private void writeShortcutForOneWordInternal(final OutputStream outStream, - final Iterator<WeightedString> shortcutIterator) throws IOException { - while (shortcutIterator.hasNext()) { - final WeightedString target = shortcutIterator.next(); - final int shortcutFlags = BinaryDictEncoderUtils.makeShortcutFlags( - shortcutIterator.hasNext(), target.mFrequency); - BinaryDictEncoderUtils.writeUIntToStream(outStream, shortcutFlags, - FormatSpec.PTNODE_ATTRIBUTE_FLAGS_SIZE); - CharEncoding.writeString(outStream, target.mWord); - } - } - } - - private void openStreams(final FormatOptions formatOptions, final DictionaryOptions dictOptions) - throws FileNotFoundException, IOException { - final FileHeader header = new FileHeader(0, dictOptions, formatOptions); - mBaseFilename = header.getId() + "." + header.getVersion(); - mDictDir = new File(mDictPlacedDir, mBaseFilename); - final File trieFile = new File(mDictDir, mBaseFilename + FormatSpec.TRIE_FILE_EXTENSION); - final File freqFile = new File(mDictDir, mBaseFilename + FormatSpec.FREQ_FILE_EXTENSION); - final File timestampFile = new File(mDictDir, - mBaseFilename + FormatSpec.UNIGRAM_TIMESTAMP_FILE_EXTENSION); - final File terminalAddressTableFile = new File(mDictDir, - mBaseFilename + FormatSpec.TERMINAL_ADDRESS_TABLE_FILE_EXTENSION); - if (!mDictDir.isDirectory()) { - if (mDictDir.exists()) mDictDir.delete(); - mDictDir.mkdirs(); - } - mTrieOutStream = new FileOutputStream(trieFile); - mFreqOutStream = new FileOutputStream(freqFile); - mTerminalAddressTableOutStream = new FileOutputStream(terminalAddressTableFile); - if (formatOptions.mHasTimestamp) { - mUnigramTimestampOutStream = new FileOutputStream(timestampFile); - } - } - - private void close() throws IOException { - try { - if (mTrieOutStream != null) { - mTrieOutStream.close(); - } - if (mFreqOutStream != null) { - mFreqOutStream.close(); - } - if (mTerminalAddressTableOutStream != null) { - mTerminalAddressTableOutStream.close(); - } - if (mUnigramTimestampOutStream != null) { - mUnigramTimestampOutStream.close(); - } - } finally { - mTrieOutStream = null; - mFreqOutStream = null; - mTerminalAddressTableOutStream = null; - } - } - + // TODO: This builds a FusionDictionary first and iterates it to add words to the binary + // dictionary. However, it is possible to just add words directly to the binary dictionary + // instead. + // In the long run, when we stop supporting version 2, FusionDictionary will become deprecated + // and we can remove it. Then we'll be able to just call BinaryDictionary directly. @Override - public void writeDictionary(final FusionDictionary dict, final FormatOptions formatOptions) + public void writeDictionary(FusionDictionary dict, FormatOptions formatOptions) throws IOException, UnsupportedFormatException { if (formatOptions.mVersion != FormatSpec.VERSION4) { throw new UnsupportedFormatException("File header has a wrong version number : " @@ -286,190 +54,71 @@ public class Ver4DictEncoder implements DictEncoder { if (!mDictPlacedDir.isDirectory()) { throw new UnsupportedFormatException("Given path is not a directory."); } - - if (mTrieOutStream == null) { - openStreams(formatOptions, dict.mOptions); - } - - mHeaderSize = BinaryDictEncoderUtils.writeDictionaryHeader(mTrieOutStream, dict, - formatOptions); - - MakedictLog.i("Flattening the tree..."); - ArrayList<PtNodeArray> flatNodes = BinaryDictEncoderUtils.flattenTree(dict.mRootNodeArray); - int terminalCount = 0; - for (final PtNodeArray array : flatNodes) { - for (final PtNode node : array.mData) { - if (node.isTerminal()) node.mTerminalId = terminalCount++; + if (!BinaryDictionary.createEmptyDictFile(mDictPlacedDir.getAbsolutePath(), + FormatSpec.VERSION4, dict.mOptions.mAttributes)) { + throw new IOException("Cannot create dictionary file : " + + mDictPlacedDir.getAbsolutePath()); + } + final BinaryDictionary binaryDict = new BinaryDictionary(mDictPlacedDir.getAbsolutePath(), + 0l, mDictPlacedDir.length(), true /* useFullEditDistance */, + LocaleUtils.constructLocaleFromString(dict.mOptions.mAttributes.get( + FormatSpec.FileHeader.DICTIONARY_LOCALE_ATTRIBUTE)), + Dictionary.TYPE_USER /* Dictionary type. Does not matter for us */, + true /* isUpdatable */); + if (!binaryDict.isValidDictionary()) { + // Somehow createEmptyDictFile returned true, but the file was not created correctly + throw new IOException("Cannot create dictionary file"); + } + for (final Word word : dict) { + // TODO: switch to addMultipleDictionaryEntries when they support shortcuts + if (null == word.mShortcutTargets || word.mShortcutTargets.isEmpty()) { + binaryDict.addUnigramWord(word.mWord, word.mFrequency, + null /* shortcutTarget */, 0 /* shortcutProbability */, + word.mIsNotAWord, word.mIsBlacklistEntry, 0 /* timestamp */); + } else { + for (final WeightedString shortcutTarget : word.mShortcutTargets) { + binaryDict.addUnigramWord(word.mWord, word.mFrequency, + shortcutTarget.mWord, shortcutTarget.mFrequency, + word.mIsNotAWord, word.mIsBlacklistEntry, 0 /* timestamp */); + } + } + if (binaryDict.needsToRunGC(true /* mindsBlockByGC */)) { + binaryDict.flushWithGC(); } } - - MakedictLog.i("Computing addresses..."); - BinaryDictEncoderUtils.computeAddresses(dict, flatNodes, formatOptions); - if (MakedictLog.DBG) BinaryDictEncoderUtils.checkFlatPtNodeArrayList(flatNodes); - - writeTerminalData(flatNodes, terminalCount); - if (formatOptions.mHasTimestamp) { - initUnigramTimestamps(terminalCount); - } - mBigramWriter = new BigramContentWriter(mBaseFilename, terminalCount, mDictDir, - formatOptions.mHasTimestamp); - writeBigrams(flatNodes, dict); - mShortcutWriter = new ShortcutContentWriter(mBaseFilename, terminalCount, mDictDir); - writeShortcuts(flatNodes); - - final PtNodeArray lastNodeArray = flatNodes.get(flatNodes.size() - 1); - final int bufferSize = lastNodeArray.mCachedAddressAfterUpdate + lastNodeArray.mCachedSize; - mTrieBuf = new byte[bufferSize]; - - MakedictLog.i("Writing file..."); - for (PtNodeArray nodeArray : flatNodes) { - BinaryDictEncoderUtils.writePlacedPtNodeArray(dict, this, nodeArray, formatOptions); - } - if (MakedictLog.DBG) { - BinaryDictEncoderUtils.showStatistics(flatNodes); - MakedictLog.i("has " + terminalCount + " terminals."); + for (final Word word0 : dict) { + if (null == word0.mBigrams) continue; + for (final WeightedString word1 : word0.mBigrams) { + binaryDict.addBigramWords(word0.mWord, word1.mWord, word1.mFrequency, + 0 /* timestamp */); + if (binaryDict.needsToRunGC(true /* mindsBlockByGC */)) { + binaryDict.flushWithGC(); + } + } } - mTrieOutStream.write(mTrieBuf); - - MakedictLog.i("Done"); - close(); + binaryDict.flushWithGC(); + binaryDict.close(); } @Override public void setPosition(int position) { - if (mTrieBuf == null || position < 0 || position >- mTrieBuf.length) return; - mTriePos = position; } @Override public int getPosition() { - return mTriePos; + return 0; } @Override public void writePtNodeCount(int ptNodeCount) { - final int countSize = BinaryDictIOUtils.getPtNodeCountSize(ptNodeCount); - // ptNodeCount must fit on one byte or two bytes. - // Please see comments in FormatSpec - if (countSize != 1 && countSize != 2) { - throw new RuntimeException("Strange size from getPtNodeCountSize : " + countSize); - } - final int encodedPtNodeCount = (countSize == 2) ? - (ptNodeCount | FormatSpec.LARGE_PTNODE_ARRAY_SIZE_FIELD_SIZE_FLAG) : ptNodeCount; - mTriePos = BinaryDictEncoderUtils.writeUIntToBuffer(mTrieBuf, mTriePos, encodedPtNodeCount, - countSize); - } - - private void writePtNodeFlags(final PtNode ptNode, final FormatOptions formatOptions) { - final int childrenPos = BinaryDictEncoderUtils.getChildrenPosition(ptNode, formatOptions); - mTriePos = BinaryDictEncoderUtils.writeUIntToBuffer(mTrieBuf, mTriePos, - BinaryDictEncoderUtils.makePtNodeFlags(ptNode, childrenPos, formatOptions), - FormatSpec.PTNODE_FLAGS_SIZE); - } - - private void writeParentPosition(int parentPos, final PtNode ptNode, - final FormatOptions formatOptions) { - if (parentPos != FormatSpec.NO_PARENT_ADDRESS) { - parentPos -= ptNode.mCachedAddressAfterUpdate; - } - mTriePos = BinaryDictEncoderUtils.writeParentAddress(mTrieBuf, mTriePos, parentPos, - formatOptions); - } - - private void writeCharacters(final int[] characters, final boolean hasSeveralChars) { - mTriePos = CharEncoding.writeCharArray(characters, mTrieBuf, mTriePos); - if (hasSeveralChars) { - mTrieBuf[mTriePos++] = FormatSpec.PTNODE_CHARACTERS_TERMINATOR; - } - } - - private void writeTerminalId(final int terminalId) { - mTriePos = BinaryDictEncoderUtils.writeUIntToBuffer(mTrieBuf, mTriePos, terminalId, - FormatSpec.PTNODE_TERMINAL_ID_SIZE); - } - - private void writeChildrenPosition(PtNode ptNode, FormatOptions formatOptions) { - final int childrenPos = BinaryDictEncoderUtils.getChildrenPosition(ptNode, formatOptions); - if (formatOptions.mSupportsDynamicUpdate) { - mTriePos += BinaryDictEncoderUtils.writeSignedChildrenPosition(mTrieBuf, - mTriePos, childrenPos); - } else { - mTriePos += BinaryDictEncoderUtils.writeChildrenPosition(mTrieBuf, - mTriePos, childrenPos); - } - } - - private void writeBigrams(final ArrayList<PtNodeArray> flatNodes, final FusionDictionary dict) - throws IOException { - mBigramWriter.openStreams(); - for (final PtNodeArray nodeArray : flatNodes) { - for (final PtNode ptNode : nodeArray.mData) { - if (ptNode.mBigrams != null) { - mBigramWriter.writeBigramsForOneWord(ptNode.mTerminalId, ptNode.mBigrams.size(), - ptNode.mBigrams.iterator(), dict); - } - } - } - mBigramWriter.closeStreams(); - } - - private void writeShortcuts(final ArrayList<PtNodeArray> flatNodes) throws IOException { - mShortcutWriter.openStreams(); - for (final PtNodeArray nodeArray : flatNodes) { - for (final PtNode ptNode : nodeArray.mData) { - if (ptNode.mShortcutTargets != null && !ptNode.mShortcutTargets.isEmpty()) { - mShortcutWriter.writeShortcutForOneWord(ptNode.mTerminalId, - ptNode.mShortcutTargets.iterator()); - } - } - } - mShortcutWriter.closeStreams(); } @Override public void writeForwardLinkAddress(int forwardLinkAddress) { - mTriePos = BinaryDictEncoderUtils.writeUIntToBuffer(mTrieBuf, mTriePos, - forwardLinkAddress, FormatSpec.FORWARD_LINK_ADDRESS_SIZE); } @Override - public void writePtNode(final PtNode ptNode, final int parentPosition, - final FormatOptions formatOptions, final FusionDictionary dict) { - writePtNodeFlags(ptNode, formatOptions); - writeParentPosition(parentPosition, ptNode, formatOptions); - writeCharacters(ptNode.mChars, ptNode.hasSeveralChars()); - if (ptNode.isTerminal()) { - writeTerminalId(ptNode.mTerminalId); - } - writeChildrenPosition(ptNode, formatOptions); - } - - private void writeTerminalData(final ArrayList<PtNodeArray> flatNodes, - final int terminalCount) throws IOException { - final byte[] freqBuf = new byte[terminalCount * FormatSpec.FREQUENCY_AND_FLAGS_SIZE]; - final byte[] terminalAddressTableBuf = - new byte[terminalCount * FormatSpec.TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE]; - for (final PtNodeArray nodeArray : flatNodes) { - for (final PtNode ptNode : nodeArray.mData) { - if (ptNode.isTerminal()) { - BinaryDictEncoderUtils.writeUIntToBuffer(freqBuf, - ptNode.mTerminalId * FormatSpec.FREQUENCY_AND_FLAGS_SIZE, - ptNode.mFrequency, FormatSpec.FREQUENCY_AND_FLAGS_SIZE); - BinaryDictEncoderUtils.writeUIntToBuffer(terminalAddressTableBuf, - ptNode.mTerminalId * FormatSpec.TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE, - ptNode.mCachedAddressAfterUpdate + mHeaderSize, - FormatSpec.TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE); - } - } - } - mFreqOutStream.write(freqBuf); - mTerminalAddressTableOutStream.write(terminalAddressTableBuf); - } - - private void initUnigramTimestamps(final int terminalCount) throws IOException { - // Initial value of time stamps for each word is 0. - final byte[] unigramTimestampBuf = - new byte[terminalCount * FormatSpec.UNIGRAM_TIMESTAMP_SIZE]; - mUnigramTimestampOutStream.write(unigramTimestampBuf); + public void writePtNode( + PtNode ptNode, int parentPosition, FormatOptions formatOptions, FusionDictionary dict) { } } diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver4DictUpdater.java b/java/src/com/android/inputmethod/latin/makedict/Ver4DictUpdater.java deleted file mode 100644 index 3d8f186ba..000000000 --- a/java/src/com/android/inputmethod/latin/makedict/Ver4DictUpdater.java +++ /dev/null @@ -1,59 +0,0 @@ -/* - * Copyright (C) 2013 The Android Open Source Project - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -package com.android.inputmethod.latin.makedict; - -import com.android.inputmethod.annotations.UsedForTesting; -import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; - -import java.io.File; -import java.io.IOException; -import java.util.ArrayList; - -/** - * An implementation of DictUpdater for version 4 binary dictionary. - */ -@UsedForTesting -public class Ver4DictUpdater extends Ver4DictDecoder implements DictUpdater { - - @UsedForTesting - public Ver4DictUpdater(final File dictDirectory, final int factoryType) { - // DictUpdater must have an updatable DictBuffer. - super(dictDirectory, ((factoryType & MASK_DICTBUFFER) == USE_BYTEARRAY) - ? USE_BYTEARRAY : USE_WRITABLE_BYTEBUFFER); - } - - @Override - public void deleteWord(final String word) throws IOException, UnsupportedFormatException { - if (mDictBuffer == null) openDictBuffer(); - readHeader(); - final int wordPos = getTerminalPosition(word); - if (wordPos != FormatSpec.NOT_VALID_WORD) { - mDictBuffer.position(wordPos); - final int flags = PtNodeReader.readPtNodeOptionFlags(mDictBuffer); - mDictBuffer.position(wordPos); - mDictBuffer.put((byte) DynamicBinaryDictIOUtils.markAsDeleted(flags)); - } - } - - @Override - public void insertWord(final String word, final int frequency, - final ArrayList<WeightedString> bigramStrings, final ArrayList<WeightedString> shortcuts, - final boolean isNotAWord, final boolean isBlackListEntry) - throws IOException, UnsupportedFormatException { - // TODO: Implement this method. - } -} diff --git a/java/src/com/android/inputmethod/latin/personalization/DecayingExpandableBinaryDictionaryBase.java b/java/src/com/android/inputmethod/latin/personalization/DecayingExpandableBinaryDictionaryBase.java index 1de15a333..701c29023 100644 --- a/java/src/com/android/inputmethod/latin/personalization/DecayingExpandableBinaryDictionaryBase.java +++ b/java/src/com/android/inputmethod/latin/personalization/DecayingExpandableBinaryDictionaryBase.java @@ -17,18 +17,16 @@ package com.android.inputmethod.latin.personalization; import android.content.Context; -import android.content.SharedPreferences; import android.util.Log; import com.android.inputmethod.annotations.UsedForTesting; +import com.android.inputmethod.latin.BinaryDictionary.LanguageModelParam; import com.android.inputmethod.latin.Constants; import com.android.inputmethod.latin.Dictionary; import com.android.inputmethod.latin.ExpandableBinaryDictionary; -import com.android.inputmethod.latin.LatinImeLogger; import com.android.inputmethod.latin.makedict.DictDecoder; import com.android.inputmethod.latin.makedict.FormatSpec; -import com.android.inputmethod.latin.settings.Settings; -import com.android.inputmethod.latin.utils.CollectionUtils; +import com.android.inputmethod.latin.makedict.UnsupportedFormatException; import com.android.inputmethod.latin.utils.UserHistoryDictIOUtils; import com.android.inputmethod.latin.utils.UserHistoryDictIOUtils.OnAddWordListener; @@ -36,7 +34,9 @@ import java.io.File; import java.io.IOException; import java.util.ArrayList; import java.util.HashMap; +import java.util.Locale; import java.util.Map; +import java.util.concurrent.TimeUnit; /** * This class is a base class of a dictionary that supports decaying for the personalized language @@ -45,8 +45,7 @@ import java.util.Map; public abstract class DecayingExpandableBinaryDictionaryBase extends ExpandableBinaryDictionary { private static final String TAG = DecayingExpandableBinaryDictionaryBase.class.getSimpleName(); public static final boolean DBG_SAVE_RESTORE = false; - private static final boolean DBG_STRESS_TEST = false; - private static final boolean PROFILE_SAVE_RESTORE = LatinImeLogger.sDBG; + private static final boolean DBG_DUMP_ON_CLOSE = false; /** Any pair being typed or picked */ public static final int FREQUENCY_FOR_TYPED = 2; @@ -54,52 +53,56 @@ public abstract class DecayingExpandableBinaryDictionaryBase extends ExpandableB public static final int FREQUENCY_FOR_WORDS_IN_DICTS = FREQUENCY_FOR_TYPED; public static final int FREQUENCY_FOR_WORDS_NOT_IN_DICTS = Dictionary.NOT_A_PROBABILITY; - /** Locale for which this user history dictionary is storing words */ - private final String mLocale; + public static final int REQUIRED_BINARY_DICTIONARY_VERSION = FormatSpec.VERSION4; - private final String mFileName; + /** The locale for this dictionary. */ + public final Locale mLocale; - private final SharedPreferences mPrefs; + private final String mDictName; - private final ArrayList<PersonalizationDictionaryUpdateSession> mSessions = - CollectionUtils.newArrayList(); - - // Should always be false except when we use this class for test - @UsedForTesting boolean mIsTest = false; + /* package */ DecayingExpandableBinaryDictionaryBase(final Context context, + final Locale locale, final String dictionaryType, final String dictName) { + super(context, dictName, locale, dictionaryType, true); + mLocale = locale; + mDictName = dictName; + if (mLocale != null && mLocale.toString().length() > 1) { + reloadDictionaryIfRequired(); + } + } + // Creates an instance that uses a given dictionary file for testing. + @UsedForTesting /* package */ DecayingExpandableBinaryDictionaryBase(final Context context, - final String locale, final SharedPreferences sp, final String dictionaryType, - final String fileName) { - super(context, fileName, dictionaryType, true); + final Locale locale, final String dictionaryType, final String dictName, + final File dictFile) { + super(context, dictName, locale, dictionaryType, true, dictFile); mLocale = locale; - mFileName = fileName; - mPrefs = sp; - if (mLocale != null && mLocale.length() > 1) { - asyncLoadDictionaryToMemory(); + mDictName = dictName; + if (mLocale != null && mLocale.toString().length() > 1) { reloadDictionaryIfRequired(); } } @Override public void close() { - if (!ExpandableBinaryDictionary.ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE) { - closeBinaryDictionary(); + if (DBG_DUMP_ON_CLOSE) { + dumpAllWordsForDebug(); } // Flush pending writes. - // TODO: Remove after this class become to use a dynamic binary dictionary. - asyncFlashAllBinaryDictionary(); - Settings.writeLastUserHistoryWriteTime(mPrefs, mLocale); + asyncFlushBinaryDictionary(); } @Override protected Map<String, String> getHeaderAttributeMap() { HashMap<String, String> attributeMap = new HashMap<String, String>(); - attributeMap.put(FormatSpec.FileHeader.SUPPORTS_DYNAMIC_UPDATE_ATTRIBUTE, - FormatSpec.FileHeader.ATTRIBUTE_VALUE_TRUE); attributeMap.put(FormatSpec.FileHeader.USES_FORGETTING_CURVE_ATTRIBUTE, FormatSpec.FileHeader.ATTRIBUTE_VALUE_TRUE); - attributeMap.put(FormatSpec.FileHeader.DICTIONARY_ID_ATTRIBUTE, mFileName); - attributeMap.put(FormatSpec.FileHeader.DICTIONARY_LOCALE_ATTRIBUTE, mLocale); + attributeMap.put(FormatSpec.FileHeader.HAS_HISTORICAL_INFO_ATTRIBUTE, + FormatSpec.FileHeader.ATTRIBUTE_VALUE_TRUE); + attributeMap.put(FormatSpec.FileHeader.DICTIONARY_ID_ATTRIBUTE, mDictName); + attributeMap.put(FormatSpec.FileHeader.DICTIONARY_LOCALE_ATTRIBUTE, mLocale.toString()); + attributeMap.put(FormatSpec.FileHeader.DICTIONARY_VERSION_ATTRIBUTE, + String.valueOf(TimeUnit.MILLISECONDS.toSeconds(System.currentTimeMillis()))); return attributeMap; } @@ -113,6 +116,25 @@ public abstract class DecayingExpandableBinaryDictionaryBase extends ExpandableB return false; } + @Override + protected boolean matchesExpectedBinaryDictFormatVersionForThisType(final int formatVersion) { + // This class is using format 4 because it's used by all version 4 dictionaries. + // TODO: remove this when all dynamically generated dicts use version 4. + return formatVersion == REQUIRED_BINARY_DICTIONARY_VERSION; + } + + public void addMultipleDictionaryEntriesToDictionary( + final ArrayList<LanguageModelParam> languageModelParams, + final ExpandableBinaryDictionary.AddMultipleDictionaryEntriesCallback callback) { + if (languageModelParams == null || languageModelParams.isEmpty()) { + if (callback != null) { + callback.onFinished(); + } + return; + } + addMultipleDictionaryEntriesDynamically(languageModelParams, callback); + } + /** * Pair will be added to the decaying dictionary. * @@ -121,70 +143,63 @@ public abstract class DecayingExpandableBinaryDictionaryBase extends ExpandableB * context, as in beginning of a sentence for example. * The second word may not be null (a NullPointerException would be thrown). */ - public void addToDictionary(final String word0, final String word1, final boolean isValid) { + public void addToDictionary(final String word0, final String word1, final boolean isValid, + final int timestamp) { if (word1.length() >= Constants.DICTIONARY_MAX_WORD_LENGTH || (word0 != null && word0.length() >= Constants.DICTIONARY_MAX_WORD_LENGTH)) { return; } - final int frequency = ENABLE_BINARY_DICTIONARY_DYNAMIC_UPDATE ? - (isValid ? FREQUENCY_FOR_WORDS_IN_DICTS : FREQUENCY_FOR_WORDS_NOT_IN_DICTS) : - FREQUENCY_FOR_TYPED; - addWordDynamically(word1, null /* shortcutTarget */, frequency, 0 /* shortcutFreq */, - false /* isNotAWord */); + final int frequency = isValid ? + FREQUENCY_FOR_WORDS_IN_DICTS : FREQUENCY_FOR_WORDS_NOT_IN_DICTS; + addWordDynamically(word1, frequency, null /* shortcutTarget */, 0 /* shortcutFreq */, + false /* isNotAWord */, false /* isBlacklisted */, timestamp); // Do not insert a word as a bigram of itself if (word1.equals(word0)) { return; } if (null != word0) { - addBigramDynamically(word0, word1, frequency, isValid); + addBigramDynamically(word0, word1, frequency, timestamp); } } - public void cancelAddingUserHistory(final String word0, final String word1) { - removeBigramDynamically(word0, word1); - } - @Override protected void loadDictionaryAsync() { - final int[] profTotalCount = { 0 }; - final String locale = getLocale(); - if (DBG_STRESS_TEST) { - try { - Log.w(TAG, "Start stress in loading: " + locale); - Thread.sleep(15000); - Log.w(TAG, "End stress in loading"); - } catch (InterruptedException e) { + // Never loaded to memory in Java side. + } + + @UsedForTesting + public void dumpAllWordsForDebug() { + runAfterGcForDebug(new Runnable() { + @Override + public void run() { + dumpAllWordsForDebugLocked(); } - } - final long last = Settings.readLastUserHistoryWriteTime(mPrefs, locale); - final long now = System.currentTimeMillis(); - final ExpandableBinaryDictionary dictionary = this; + }); + } + + private void dumpAllWordsForDebugLocked() { + Log.d(TAG, "dumpAllWordsForDebug started."); final OnAddWordListener listener = new OnAddWordListener() { @Override public void setUnigram(final String word, final String shortcutTarget, final int frequency, final int shortcutFreq) { - if (DBG_SAVE_RESTORE) { - Log.d(TAG, "load unigram: " + word + "," + frequency); - } - addWord(word, shortcutTarget, frequency, shortcutFreq, false /* isNotAWord */); - ++profTotalCount[0]; + Log.d(TAG, "load unigram: " + word + "," + frequency); } @Override public void setBigram(final String word0, final String word1, final int frequency) { if (word0.length() < Constants.DICTIONARY_MAX_WORD_LENGTH && word1.length() < Constants.DICTIONARY_MAX_WORD_LENGTH) { - if (DBG_SAVE_RESTORE) { - Log.d(TAG, "load bigram: " + word0 + "," + word1 + "," + frequency); - } - ++profTotalCount[0]; - addBigram(word0, word1, frequency, last); + Log.d(TAG, "load bigram: " + word0 + "," + word1 + "," + frequency); + } else { + Log.d(TAG, "Skip inserting a too long bigram: " + word0 + "," + word1 + "," + + frequency); } } }; // Load the dictionary from binary file - final File dictFile = new File(mContext.getFilesDir(), mFileName); + final File dictFile = new File(mContext.getFilesDir(), mDictName); final DictDecoder dictDecoder = FormatSpec.getDictDecoder(dictFile, DictDecoder.USE_BYTEARRAY); if (dictDecoder == null) { @@ -198,35 +213,17 @@ public abstract class DecayingExpandableBinaryDictionaryBase extends ExpandableB UserHistoryDictIOUtils.readDictionaryBinary(dictDecoder, listener); } catch (IOException e) { Log.d(TAG, "IOException on opening a bytebuffer", e); - } finally { - if (PROFILE_SAVE_RESTORE) { - final long diff = System.currentTimeMillis() - now; - Log.d(TAG, "PROF: Load UserHistoryDictionary: " - + locale + ", " + diff + "ms. load " + profTotalCount[0] + "entries."); - } + } catch (UnsupportedFormatException e) { + Log.d(TAG, "Unsupported format, can't read the dictionary", e); } } - protected String getLocale() { - return mLocale; - } - - public void registerUpdateSession(PersonalizationDictionaryUpdateSession session) { - session.setPredictionDictionary(this); - mSessions.add(session); - session.onDictionaryReady(); - } - - public void unRegisterUpdateSession(PersonalizationDictionaryUpdateSession session) { - mSessions.remove(session); - } - @UsedForTesting public void clearAndFlushDictionary() { // Clear the node structure on memory clear(); // Then flush the cleared state of the dictionary on disk. - asyncFlashAllBinaryDictionary(); + asyncFlushBinaryDictionary(); } /* package */ void decayIfNeeded() { diff --git a/java/src/com/android/inputmethod/latin/personalization/DynamicPersonalizationDictionaryWriter.java b/java/src/com/android/inputmethod/latin/personalization/DynamicPersonalizationDictionaryWriter.java deleted file mode 100644 index 6f152bb91..000000000 --- a/java/src/com/android/inputmethod/latin/personalization/DynamicPersonalizationDictionaryWriter.java +++ /dev/null @@ -1,190 +0,0 @@ -/* - * Copyright (C) 2013 The Android Open Source Project - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -package com.android.inputmethod.latin.personalization; - -import android.content.Context; - -import com.android.inputmethod.annotations.UsedForTesting; -import com.android.inputmethod.compat.ActivityManagerCompatUtils; -import com.android.inputmethod.keyboard.ProximityInfo; -import com.android.inputmethod.latin.AbstractDictionaryWriter; -import com.android.inputmethod.latin.ExpandableDictionary; -import com.android.inputmethod.latin.WordComposer; -import com.android.inputmethod.latin.ExpandableDictionary.NextWord; -import com.android.inputmethod.latin.SuggestedWords.SuggestedWordInfo; -import com.android.inputmethod.latin.makedict.DictEncoder; -import com.android.inputmethod.latin.makedict.FormatSpec; -import com.android.inputmethod.latin.makedict.UnsupportedFormatException; -import com.android.inputmethod.latin.utils.UserHistoryDictIOUtils; -import com.android.inputmethod.latin.utils.UserHistoryDictIOUtils.BigramDictionaryInterface; -import com.android.inputmethod.latin.utils.UserHistoryForgettingCurveUtils; -import com.android.inputmethod.latin.utils.UserHistoryForgettingCurveUtils.ForgettingCurveParams; - -import java.io.IOException; -import java.util.ArrayList; -import java.util.Map; - -// Currently this class is used to implement dynamic prodiction dictionary. -// TODO: Move to native code. -public class DynamicPersonalizationDictionaryWriter extends AbstractDictionaryWriter { - private static final String TAG = DynamicPersonalizationDictionaryWriter.class.getSimpleName(); - /** Maximum number of pairs. Pruning will start when databases goes above this number. */ - public static final int DEFAULT_MAX_HISTORY_BIGRAMS = 10000; - public static final int LOW_MEMORY_MAX_HISTORY_BIGRAMS = 2000; - - /** Any pair being typed or picked */ - private static final int FREQUENCY_FOR_TYPED = 2; - - private static final int BINARY_DICT_VERSION = 3; - private static final FormatSpec.FormatOptions FORMAT_OPTIONS = - new FormatSpec.FormatOptions(BINARY_DICT_VERSION, true /* supportsDynamicUpdate */); - - private final UserHistoryDictionaryBigramList mBigramList = - new UserHistoryDictionaryBigramList(); - private final ExpandableDictionary mExpandableDictionary; - private final int mMaxHistoryBigrams; - - public DynamicPersonalizationDictionaryWriter(final Context context, final String dictType) { - super(context, dictType); - mExpandableDictionary = new ExpandableDictionary(dictType); - final boolean isLowRamDevice = ActivityManagerCompatUtils.isLowRamDevice(context); - mMaxHistoryBigrams = isLowRamDevice ? - LOW_MEMORY_MAX_HISTORY_BIGRAMS : DEFAULT_MAX_HISTORY_BIGRAMS; - } - - @Override - public void clear() { - mBigramList.evictAll(); - mExpandableDictionary.clearDictionary(); - } - - /** - * Adds a word unigram to the fusion dictionary. Call updateBinaryDictionary when all changes - * are done to update the binary dictionary. - * @param word The word to add. - * @param shortcutTarget A shortcut target for this word, or null if none. - * @param frequency The frequency for this unigram. - * @param shortcutFreq The frequency of the shortcut (0~15, with 15 = whitelist). Ignored - * if shortcutTarget is null. - * @param isNotAWord true if this is not a word, i.e. shortcut only. - */ - @Override - public void addUnigramWord(final String word, final String shortcutTarget, final int frequency, - final int shortcutFreq, final boolean isNotAWord) { - if (mBigramList.size() > mMaxHistoryBigrams * 2) { - // Too many entries: just stop adding new vocabulary and wait next refresh. - return; - } - mExpandableDictionary.addWord(word, shortcutTarget, frequency, shortcutFreq); - mBigramList.addBigram(null, word, (byte)frequency); - } - - @Override - public void addBigramWords(final String word0, final String word1, final int frequency, - final boolean isValid, final long lastModifiedTime) { - if (mBigramList.size() > mMaxHistoryBigrams * 2) { - // Too many entries: just stop adding new vocabulary and wait next refresh. - return; - } - if (lastModifiedTime > 0) { - mExpandableDictionary.setBigramAndGetFrequency(word0, word1, - new ForgettingCurveParams(frequency, System.currentTimeMillis(), - lastModifiedTime)); - mBigramList.addBigram(word0, word1, (byte)frequency); - } else { - mExpandableDictionary.setBigramAndGetFrequency(word0, word1, - new ForgettingCurveParams(isValid)); - mBigramList.addBigram(word0, word1, (byte)frequency); - } - } - - @Override - public void removeBigramWords(final String word0, final String word1) { - if (mBigramList.removeBigram(word0, word1)) { - mExpandableDictionary.removeBigram(word0, word1); - } - } - - @Override - protected void writeDictionary(final DictEncoder dictEncoder, - final Map<String, String> attributeMap) throws IOException, UnsupportedFormatException { - UserHistoryDictIOUtils.writeDictionary(dictEncoder, - new FrequencyProvider(mBigramList, mExpandableDictionary, mMaxHistoryBigrams), - mBigramList, FORMAT_OPTIONS); - } - - private static class FrequencyProvider implements BigramDictionaryInterface { - private final UserHistoryDictionaryBigramList mBigramList; - private final ExpandableDictionary mExpandableDictionary; - private final int mMaxHistoryBigrams; - - public FrequencyProvider(final UserHistoryDictionaryBigramList bigramList, - final ExpandableDictionary expandableDictionary, final int maxHistoryBigrams) { - mBigramList = bigramList; - mExpandableDictionary = expandableDictionary; - mMaxHistoryBigrams = maxHistoryBigrams; - } - - @Override - public int getFrequency(final String word0, final String word1) { - final int freq; - if (word0 == null) { // unigram - freq = FREQUENCY_FOR_TYPED; - } else { // bigram - final NextWord nw = mExpandableDictionary.getBigramWord(word0, word1); - if (nw != null) { - final ForgettingCurveParams forgettingCurveParams = nw.getFcParams(); - final byte prevFc = mBigramList.getBigrams(word0).get(word1); - final byte fc = forgettingCurveParams.getFc(); - final boolean isValid = forgettingCurveParams.isValid(); - if (prevFc > 0 && prevFc == fc) { - freq = fc & 0xFF; - } else if (UserHistoryForgettingCurveUtils. - needsToSave(fc, isValid, mBigramList.size() <= mMaxHistoryBigrams)) { - freq = fc & 0xFF; - } else { - // Delete this entry - freq = -1; - } - } else { - // Delete this entry - freq = -1; - } - } - return freq; - } - } - - @Override - public ArrayList<SuggestedWordInfo> getSuggestions(final WordComposer composer, - final String prevWord, final ProximityInfo proximityInfo, - boolean blockOffensiveWords, final int[] additionalFeaturesOptions) { - return mExpandableDictionary.getSuggestions(composer, prevWord, proximityInfo, - blockOffensiveWords, additionalFeaturesOptions); - } - - @Override - public boolean isValidWord(final String word) { - return mExpandableDictionary.isValidWord(word); - } - - @UsedForTesting - public boolean isInBigramListForTests(final String word) { - // TODO: Use native method to determine whether the word is in dictionary or not - return mBigramList.containsKey(word) || mBigramList.getBigrams(null).containsKey(word); - } -} diff --git a/java/src/com/android/inputmethod/latin/personalization/PersonalizationDictionary.java b/java/src/com/android/inputmethod/latin/personalization/PersonalizationDictionary.java index f257165cb..596562f1d 100644 --- a/java/src/com/android/inputmethod/latin/personalization/PersonalizationDictionary.java +++ b/java/src/com/android/inputmethod/latin/personalization/PersonalizationDictionary.java @@ -16,53 +16,37 @@ package com.android.inputmethod.latin.personalization; +import com.android.inputmethod.annotations.UsedForTesting; import com.android.inputmethod.latin.Dictionary; -import com.android.inputmethod.latin.ExpandableBinaryDictionary; import com.android.inputmethod.latin.utils.CollectionUtils; +import java.io.File; +import java.util.ArrayList; +import java.util.Locale; + import android.content.Context; -import android.content.SharedPreferences; -import java.util.ArrayList; +public class PersonalizationDictionary extends DecayingExpandableBinaryDictionaryBase { + private static final String NAME = PersonalizationDictionary.class.getSimpleName(); -/** - * This class is a dictionary for the personalized language model that uses binary dictionary. - */ -public class PersonalizationDictionary extends ExpandableBinaryDictionary { - private static final String NAME = "personalization"; private final ArrayList<PersonalizationDictionaryUpdateSession> mSessions = CollectionUtils.newArrayList(); - /** Locale for which this user history dictionary is storing words */ - private final String mLocale; - - public PersonalizationDictionary(final Context context, final String locale, - final SharedPreferences prefs) { - // TODO: Make isUpdatable true. - super(context, getFilenameWithLocale(NAME, locale), Dictionary.TYPE_PERSONALIZATION, - false /* isUpdatable */); - mLocale = locale; - // TODO: Restore last updated time - loadDictionary(); - } - - @Override - protected void loadDictionaryAsync() { - // TODO: Implement - } - - @Override - protected boolean hasContentChanged() { - return false; + /* package */ PersonalizationDictionary(final Context context, final Locale locale) { + super(context, locale, Dictionary.TYPE_PERSONALIZATION, + getDictNameWithLocale(NAME, locale)); } - @Override - protected boolean needsToReloadBeforeWriting() { - return false; + // Creates an instance that uses a given dictionary file for testing. + @UsedForTesting + public PersonalizationDictionary(final Context context, final Locale locale, + final File dictFile) { + super(context, locale, Dictionary.TYPE_PERSONALIZATION, getDictNameWithLocale(NAME, locale), + dictFile); } public void registerUpdateSession(PersonalizationDictionaryUpdateSession session) { - session.setDictionary(this); + session.setPredictionDictionary(this); mSessions.add(session); session.onDictionaryReady(); } diff --git a/java/src/com/android/inputmethod/latin/personalization/PersonalizationDictionarySessionRegister.java b/java/src/com/android/inputmethod/latin/personalization/PersonalizationDictionarySessionRegister.java index c1833ff14..542bda621 100644 --- a/java/src/com/android/inputmethod/latin/personalization/PersonalizationDictionarySessionRegister.java +++ b/java/src/com/android/inputmethod/latin/personalization/PersonalizationDictionarySessionRegister.java @@ -20,18 +20,18 @@ import android.content.Context; import android.content.res.Configuration; public class PersonalizationDictionarySessionRegister { - public static void init(Context context) { + public static void init(final Context context) { } public static void onConfigurationChanged(final Context context, final Configuration conf) { } - public static void onUpdateData(Context context, String type) { + public static void onUpdateData(final Context context, final String type) { } - public static void onRemoveData(Context context, String type) { + public static void onRemoveData(final Context context, final String type) { } - public static void onDestroy(Context context) { + public static void onDestroy(final Context context) { } } diff --git a/java/src/com/android/inputmethod/latin/personalization/PersonalizationDictionaryUpdateSession.java b/java/src/com/android/inputmethod/latin/personalization/PersonalizationDictionaryUpdateSession.java index a86f6e584..61354762b 100644 --- a/java/src/com/android/inputmethod/latin/personalization/PersonalizationDictionaryUpdateSession.java +++ b/java/src/com/android/inputmethod/latin/personalization/PersonalizationDictionaryUpdateSession.java @@ -18,61 +18,37 @@ package com.android.inputmethod.latin.personalization; import android.content.Context; +import com.android.inputmethod.latin.BinaryDictionary.LanguageModelParam; +import com.android.inputmethod.latin.ExpandableBinaryDictionary; + import java.lang.ref.WeakReference; import java.util.ArrayList; +import java.util.Locale; /** * This class is a session where a data provider can communicate with a personalization * dictionary. */ public abstract class PersonalizationDictionaryUpdateSession { - /** - * This class is a parameter for a new unigram or bigram word which will be added - * to the personalization dictionary. - */ - public static class PersonalizationLanguageModelParam { - public final String mWord0; - public final String mWord1; - public final boolean mIsValid; - public final int mFrequency; - public PersonalizationLanguageModelParam(String word0, String word1, boolean isValid, - int frequency) { - mWord0 = word0; - mWord1 = word1; - mIsValid = isValid; - mFrequency = frequency; - } - } - - // TODO: Use a dynamic binary dictionary instead public WeakReference<PersonalizationDictionary> mDictionary; - public WeakReference<DecayingExpandableBinaryDictionaryBase> mPredictionDictionary; - public final String mSystemLocale; - public PersonalizationDictionaryUpdateSession(String locale) { + public final Locale mSystemLocale; + + public PersonalizationDictionaryUpdateSession(final Locale locale) { mSystemLocale = locale; } public abstract void onDictionaryReady(); - public abstract void onDictionaryClosed(Context context); + public abstract void onDictionaryClosed(final Context context); - public void setDictionary(PersonalizationDictionary dictionary) { + public void setPredictionDictionary(final PersonalizationDictionary dictionary) { mDictionary = new WeakReference<PersonalizationDictionary>(dictionary); } - public void setPredictionDictionary(DecayingExpandableBinaryDictionaryBase dictionary) { - mPredictionDictionary = - new WeakReference<DecayingExpandableBinaryDictionaryBase>(dictionary); - } - protected PersonalizationDictionary getDictionary() { return mDictionary == null ? null : mDictionary.get(); } - protected DecayingExpandableBinaryDictionaryBase getPredictionDictionary() { - return mPredictionDictionary == null ? null : mPredictionDictionary.get(); - } - private void unsetDictionary() { final PersonalizationDictionary dictionary = getDictionary(); if (dictionary == null) { @@ -81,48 +57,30 @@ public abstract class PersonalizationDictionaryUpdateSession { dictionary.unRegisterUpdateSession(this); } - private void unsetPredictionDictionary() { - final DecayingExpandableBinaryDictionaryBase dictionary = getPredictionDictionary(); - if (dictionary == null) { - return; - } - dictionary.unRegisterUpdateSession(this); - } - - public void clearAndFlushPredictionDictionary(Context context) { - final DecayingExpandableBinaryDictionaryBase dictionary = getPredictionDictionary(); + public void clearAndFlushDictionary(final Context context) { + final PersonalizationDictionary dictionary = getDictionary(); if (dictionary == null) { return; } dictionary.clearAndFlushDictionary(); } - public void closeSession(Context context) { + public void closeSession(final Context context) { unsetDictionary(); - unsetPredictionDictionary(); onDictionaryClosed(context); } - // TODO: Support multi locale to add bigram - public void addBigramToPersonalizationDictionary(String word0, String word1, boolean isValid, - int frequency) { - final DecayingExpandableBinaryDictionaryBase dictionary = getPredictionDictionary(); - if (dictionary == null) { - return; - } - dictionary.addToDictionary(word0, word1, isValid); - } - - // Bulk import - // TODO: Support multi locale to add bigram - public void addBigramsToPersonalizationDictionary( - final ArrayList<PersonalizationLanguageModelParam> lmParams) { - final DecayingExpandableBinaryDictionaryBase dictionary = getPredictionDictionary(); + // TODO: Support multi locale. + public void addMultipleDictionaryEntriesToDictionary( + final ArrayList<LanguageModelParam> languageModelParams, + final ExpandableBinaryDictionary.AddMultipleDictionaryEntriesCallback callback) { + final PersonalizationDictionary dictionary = getDictionary(); if (dictionary == null) { + if (callback != null) { + callback.onFinished(); + } return; } - for (final PersonalizationLanguageModelParam lmParam : lmParams) { - dictionary.addToDictionary(lmParam.mWord0, lmParam.mWord1, lmParam.mIsValid); - } + dictionary.addMultipleDictionaryEntriesToDictionary(languageModelParams, callback); } } diff --git a/java/src/com/android/inputmethod/latin/personalization/PersonalizationHelper.java b/java/src/com/android/inputmethod/latin/personalization/PersonalizationHelper.java index 221ddeeba..d55cae132 100644 --- a/java/src/com/android/inputmethod/latin/personalization/PersonalizationHelper.java +++ b/java/src/com/android/inputmethod/latin/personalization/PersonalizationHelper.java @@ -19,11 +19,10 @@ package com.android.inputmethod.latin.personalization; import com.android.inputmethod.latin.utils.CollectionUtils; import android.content.Context; -import android.content.SharedPreferences; -import android.preference.PreferenceManager; import android.util.Log; import java.lang.ref.SoftReference; +import java.util.Locale; import java.util.concurrent.ConcurrentHashMap; public class PersonalizationHelper { @@ -31,21 +30,16 @@ public class PersonalizationHelper { private static final boolean DEBUG = false; private static final ConcurrentHashMap<String, SoftReference<UserHistoryDictionary>> sLangUserHistoryDictCache = CollectionUtils.newConcurrentHashMap(); - private static final ConcurrentHashMap<String, SoftReference<PersonalizationDictionary>> sLangPersonalizationDictCache = CollectionUtils.newConcurrentHashMap(); - private static final ConcurrentHashMap<String, - SoftReference<PersonalizationPredictionDictionary>> - sLangPersonalizationPredictionDictCache = - CollectionUtils.newConcurrentHashMap(); - public static UserHistoryDictionary getUserHistoryDictionary( - final Context context, final String locale, final SharedPreferences sp) { + final Context context, final Locale locale) { + final String localeStr = locale.toString(); synchronized (sLangUserHistoryDictCache) { - if (sLangUserHistoryDictCache.containsKey(locale)) { + if (sLangUserHistoryDictCache.containsKey(localeStr)) { final SoftReference<UserHistoryDictionary> ref = - sLangUserHistoryDictCache.get(locale); + sLangUserHistoryDictCache.get(localeStr); final UserHistoryDictionary dict = ref == null ? null : ref.get(); if (dict != null) { if (DEBUG) { @@ -55,8 +49,9 @@ public class PersonalizationHelper { return dict; } } - final UserHistoryDictionary dict = new UserHistoryDictionary(context, locale, sp); - sLangUserHistoryDictCache.put(locale, new SoftReference<UserHistoryDictionary>(dict)); + final UserHistoryDictionary dict = new UserHistoryDictionary(context, locale); + sLangUserHistoryDictCache.put(localeStr, + new SoftReference<UserHistoryDictionary>(dict)); return dict; } } @@ -74,57 +69,30 @@ public class PersonalizationHelper { } public static void registerPersonalizationDictionaryUpdateSession(final Context context, - final PersonalizationDictionaryUpdateSession session, String locale) { - final PersonalizationPredictionDictionary predictionDictionary = - getPersonalizationPredictionDictionary(context, locale, - PreferenceManager.getDefaultSharedPreferences(context)); - predictionDictionary.registerUpdateSession(session); - final PersonalizationDictionary dictionary = - getPersonalizationDictionary(context, locale, - PreferenceManager.getDefaultSharedPreferences(context)); - dictionary.registerUpdateSession(session); + final PersonalizationDictionaryUpdateSession session, final Locale locale) { + final PersonalizationDictionary personalizationDictionary = + getPersonalizationDictionary(context, locale); + personalizationDictionary.registerUpdateSession(session); } public static PersonalizationDictionary getPersonalizationDictionary( - final Context context, final String locale, final SharedPreferences sp) { + final Context context, final Locale locale) { + final String localeStr = locale.toString(); synchronized (sLangPersonalizationDictCache) { - if (sLangPersonalizationDictCache.containsKey(locale)) { + if (sLangPersonalizationDictCache.containsKey(localeStr)) { final SoftReference<PersonalizationDictionary> ref = - sLangPersonalizationDictCache.get(locale); + sLangPersonalizationDictCache.get(localeStr); final PersonalizationDictionary dict = ref == null ? null : ref.get(); if (dict != null) { if (DEBUG) { - Log.w(TAG, "Use cached PersonalizationDictCache for " + locale); + Log.w(TAG, "Use cached PersonalizationDictionary for " + locale); } return dict; } } - final PersonalizationDictionary dict = - new PersonalizationDictionary(context, locale, sp); + final PersonalizationDictionary dict = new PersonalizationDictionary(context, locale); sLangPersonalizationDictCache.put( - locale, new SoftReference<PersonalizationDictionary>(dict)); - return dict; - } - } - - public static PersonalizationPredictionDictionary getPersonalizationPredictionDictionary( - final Context context, final String locale, final SharedPreferences sp) { - synchronized (sLangPersonalizationPredictionDictCache) { - if (sLangPersonalizationPredictionDictCache.containsKey(locale)) { - final SoftReference<PersonalizationPredictionDictionary> ref = - sLangPersonalizationPredictionDictCache.get(locale); - final PersonalizationPredictionDictionary dict = ref == null ? null : ref.get(); - if (dict != null) { - if (DEBUG) { - Log.w(TAG, "Use cached PersonalizationPredictionDictionary for " + locale); - } - return dict; - } - } - final PersonalizationPredictionDictionary dict = - new PersonalizationPredictionDictionary(context, locale, sp); - sLangPersonalizationPredictionDictCache.put( - locale, new SoftReference<PersonalizationPredictionDictionary>(dict)); + localeStr, new SoftReference<PersonalizationDictionary>(dict)); return dict; } } diff --git a/java/src/com/android/inputmethod/latin/personalization/PersonalizationPredictionDictionary.java b/java/src/com/android/inputmethod/latin/personalization/PersonalizationPredictionDictionary.java deleted file mode 100644 index 432954453..000000000 --- a/java/src/com/android/inputmethod/latin/personalization/PersonalizationPredictionDictionary.java +++ /dev/null @@ -1,37 +0,0 @@ -/* - * Copyright (C) 2013 The Android Open Source Project - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -package com.android.inputmethod.latin.personalization; - -import com.android.inputmethod.latin.Dictionary; -import com.android.inputmethod.latin.ExpandableBinaryDictionary; - -import android.content.Context; -import android.content.SharedPreferences; - -public class PersonalizationPredictionDictionary extends DecayingExpandableBinaryDictionaryBase { - private static final String NAME = PersonalizationPredictionDictionary.class.getSimpleName(); - - /* package */ PersonalizationPredictionDictionary(final Context context, final String locale, - final SharedPreferences sp) { - super(context, locale, sp, Dictionary.TYPE_PERSONALIZATION_PREDICTION_IN_JAVA, - getDictionaryFileName(locale)); - } - - private static String getDictionaryFileName(final String locale) { - return NAME + "." + locale + ExpandableBinaryDictionary.DICT_FILE_EXTENSION; - } -} diff --git a/java/src/com/android/inputmethod/latin/personalization/UserHistoryDictionary.java b/java/src/com/android/inputmethod/latin/personalization/UserHistoryDictionary.java index a60226d7e..868f21cbc 100644 --- a/java/src/com/android/inputmethod/latin/personalization/UserHistoryDictionary.java +++ b/java/src/com/android/inputmethod/latin/personalization/UserHistoryDictionary.java @@ -16,11 +16,13 @@ package com.android.inputmethod.latin.personalization; +import com.android.inputmethod.annotations.UsedForTesting; import com.android.inputmethod.latin.Dictionary; -import com.android.inputmethod.latin.ExpandableBinaryDictionary; + +import java.io.File; +import java.util.Locale; import android.content.Context; -import android.content.SharedPreferences; /** * Locally gathers stats about the words user types and various other signals like auto-correction @@ -29,12 +31,19 @@ import android.content.SharedPreferences; public class UserHistoryDictionary extends DecayingExpandableBinaryDictionaryBase { /* package for tests */ static final String NAME = UserHistoryDictionary.class.getSimpleName(); - /* package */ UserHistoryDictionary(final Context context, final String locale, - final SharedPreferences sp) { - super(context, locale, sp, Dictionary.TYPE_USER_HISTORY, getDictionaryFileName(locale)); + /* package */ UserHistoryDictionary(final Context context, final Locale locale) { + super(context, locale, Dictionary.TYPE_USER_HISTORY, getDictNameWithLocale(NAME, locale)); + } + + // Creates an instance that uses a given dictionary file for testing. + @UsedForTesting + public UserHistoryDictionary(final Context context, final Locale locale, + final File dictFile) { + super(context, locale, Dictionary.TYPE_USER_HISTORY, getDictNameWithLocale(NAME, locale), + dictFile); } - private static String getDictionaryFileName(final String locale) { - return NAME + "." + locale + ExpandableBinaryDictionary.DICT_FILE_EXTENSION; + public void cancelAddingUserHistory(final String word0, final String word1) { + removeBigramDynamically(word0, word1); } } diff --git a/java/src/com/android/inputmethod/latin/settings/DebugSettings.java b/java/src/com/android/inputmethod/latin/settings/DebugSettings.java index da1fb73fe..d060485bd 100644 --- a/java/src/com/android/inputmethod/latin/settings/DebugSettings.java +++ b/java/src/com/android/inputmethod/latin/settings/DebugSettings.java @@ -112,8 +112,7 @@ public final class DebugSettings extends PreferenceFragment updateDebugMode(); mServiceNeedsRestart = true; } - } else if (key.equals(PREF_FORCE_NON_DISTINCT_MULTITOUCH) - || key.equals(PREF_USE_ONLY_PERSONALIZATION_DICTIONARY_FOR_DEBUG)) { + } else if (key.equals(PREF_FORCE_NON_DISTINCT_MULTITOUCH)) { mServiceNeedsRestart = true; } } diff --git a/java/src/com/android/inputmethod/latin/settings/Settings.java b/java/src/com/android/inputmethod/latin/settings/Settings.java index df2c6907f..94f145813 100644 --- a/java/src/com/android/inputmethod/latin/settings/Settings.java +++ b/java/src/com/android/inputmethod/latin/settings/Settings.java @@ -27,12 +27,10 @@ import com.android.inputmethod.latin.AudioAndHapticFeedbackManager; import com.android.inputmethod.latin.InputAttributes; import com.android.inputmethod.latin.R; import com.android.inputmethod.latin.utils.AdditionalSubtypeUtils; -import com.android.inputmethod.latin.utils.LocaleUtils; import com.android.inputmethod.latin.utils.ResourceUtils; import com.android.inputmethod.latin.utils.RunInLocale; import com.android.inputmethod.latin.utils.StringUtils; -import java.util.HashMap; import java.util.Locale; import java.util.concurrent.locks.ReentrantLock; @@ -53,10 +51,9 @@ public final class Settings implements SharedPreferences.OnSharedPreferenceChang public static final String PREF_AUTO_CORRECTION_THRESHOLD = "auto_correction_threshold"; public static final String PREF_SHOW_SUGGESTIONS_SETTING = "show_suggestions_setting"; public static final String PREF_MISC_SETTINGS = "misc_settings"; - public static final String PREF_LAST_USER_DICTIONARY_WRITE_TIME = - "last_user_dictionary_write_time"; public static final String PREF_ADVANCED_SETTINGS = "pref_advanced_settings"; public static final String PREF_KEY_USE_CONTACTS_DICT = "pref_key_use_contacts_dict"; + public static final String PREF_KEY_USE_PERSONALIZED_DICTS = "pref_key_use_personalized_dicts"; public static final String PREF_KEY_USE_DOUBLE_SPACE_PERIOD = "pref_key_use_double_space_period"; public static final String PREF_BLOCK_POTENTIALLY_OFFENSIVE = @@ -229,16 +226,15 @@ public final class Settings implements SharedPreferences.OnSharedPreferenceChang res.getBoolean(R.bool.config_default_phrase_gesture_enabled)); } - public static boolean readFromBuildConfigIfToShowKeyPreviewPopupSettingsOption( - final Resources res) { - return res.getBoolean(R.bool.config_enable_show_option_of_key_preview_popup); + public static boolean readFromBuildConfigIfToShowKeyPreviewPopupOption(final Resources res) { + return res.getBoolean(R.bool.config_enable_show_key_preview_popup_option); } public static boolean readKeyPreviewPopupEnabled(final SharedPreferences prefs, final Resources res) { final boolean defaultKeyPreviewPopup = res.getBoolean( R.bool.config_default_key_preview_popup); - if (!readFromBuildConfigIfToShowKeyPreviewPopupSettingsOption(res)) { + if (!readFromBuildConfigIfToShowKeyPreviewPopupOption(res)) { return defaultKeyPreviewPopup; } return prefs.getBoolean(PREF_POPUP_ON, defaultKeyPreviewPopup); @@ -333,25 +329,6 @@ public final class Settings implements SharedPreferences.OnSharedPreferenceChang return prefs.getBoolean(DebugSettings.PREF_USABILITY_STUDY_MODE, true); } - public static long readLastUserHistoryWriteTime(final SharedPreferences prefs, - final String locale) { - final String str = prefs.getString(PREF_LAST_USER_DICTIONARY_WRITE_TIME, ""); - final HashMap<String, Long> map = LocaleUtils.localeAndTimeStrToHashMap(str); - if (map.containsKey(locale)) { - return map.get(locale); - } - return 0; - } - - public static void writeLastUserHistoryWriteTime(final SharedPreferences prefs, - final String locale) { - final String oldStr = prefs.getString(PREF_LAST_USER_DICTIONARY_WRITE_TIME, ""); - final HashMap<String, Long> map = LocaleUtils.localeAndTimeStrToHashMap(oldStr); - map.put(locale, System.currentTimeMillis()); - final String newStr = LocaleUtils.localeAndTimeHashMapToStr(map); - prefs.edit().putString(PREF_LAST_USER_DICTIONARY_WRITE_TIME, newStr).apply(); - } - public static boolean readUseFullscreenMode(final Resources res) { return res.getBoolean(R.bool.config_use_fullscreen_mode); } diff --git a/java/src/com/android/inputmethod/latin/settings/SettingsFragment.java b/java/src/com/android/inputmethod/latin/settings/SettingsFragment.java index 5c60a7350..d7a3e95b3 100644 --- a/java/src/com/android/inputmethod/latin/settings/SettingsFragment.java +++ b/java/src/com/android/inputmethod/latin/settings/SettingsFragment.java @@ -48,7 +48,6 @@ import com.android.inputmethod.latin.utils.AdditionalSubtypeUtils; import com.android.inputmethod.latin.utils.ApplicationUtils; import com.android.inputmethod.latin.utils.FeedbackUtils; import com.android.inputmethod.latin.utils.SubtypeLocaleUtils; -import com.android.inputmethod.research.ResearchLogger; import com.android.inputmethodcommon.InputMethodSettingsFragment; import java.util.TreeSet; @@ -61,13 +60,6 @@ public final class SettingsFragment extends InputMethodSettingsFragment DBG_USE_INTERNAL_PERSONAL_DICTIONARY_SETTINGS || Build.VERSION.SDK_INT <= 18 /* Build.VERSION.JELLY_BEAN_MR2 */; - private CheckBoxPreference mVoiceInputKeyPreference; - private ListPreference mShowCorrectionSuggestionsPreference; - private ListPreference mAutoCorrectionThresholdPreference; - private ListPreference mKeyPreviewPopupDismissDelay; - // Use bigrams to predict the next word when there is no input for it yet - private CheckBoxPreference mBigramPrediction; - private void setPreferenceEnabled(final String preferenceKey, final boolean enabled) { final Preference preference = findPreference(preferenceKey); if (preference != null) { @@ -75,6 +67,18 @@ public final class SettingsFragment extends InputMethodSettingsFragment } } + private void updateListPreferenceSummaryToCurrentValue(final String prefKey) { + // Because the "%s" summary trick of {@link ListPreference} doesn't work properly before + // KitKat, we need to update the summary programmatically. + final ListPreference listPreference = (ListPreference)findPreference(prefKey); + if (listPreference == null) { + return; + } + final CharSequence entries[] = listPreference.getEntries(); + final int entryIndex = listPreference.findIndexOfValue(listPreference.getValue()); + listPreference.setSummary(entryIndex < 0 ? null : entries[entryIndex]); + } + private static void removePreference(final String preferenceKey, final PreferenceGroup parent) { if (parent == null) { return; @@ -107,16 +111,9 @@ public final class SettingsFragment extends InputMethodSettingsFragment SubtypeLocaleUtils.init(context); AudioAndHapticFeedbackManager.init(context); - mVoiceInputKeyPreference = - (CheckBoxPreference) findPreference(Settings.PREF_VOICE_INPUT_KEY); - mShowCorrectionSuggestionsPreference = - (ListPreference) findPreference(Settings.PREF_SHOW_SUGGESTIONS_SETTING); final SharedPreferences prefs = getPreferenceManager().getSharedPreferences(); prefs.registerOnSharedPreferenceChangeListener(this); - mAutoCorrectionThresholdPreference = - (ListPreference) findPreference(Settings.PREF_AUTO_CORRECTION_THRESHOLD); - mBigramPrediction = (CheckBoxPreference) findPreference(Settings.PREF_BIGRAM_PREDICTIONS); ensureConsistencyOfAutoCorrectionSettings(); final PreferenceGroup generalSettings = @@ -143,12 +140,7 @@ public final class SettingsFragment extends InputMethodSettingsFragment feedbackSettings.setOnPreferenceClickListener(new OnPreferenceClickListener() { @Override public boolean onPreferenceClick(final Preference pref) { - if (ProductionFlag.USES_DEVELOPMENT_ONLY_DIAGNOSTICS) { - // Use development-only feedback mechanism - ResearchLogger.getInstance().presentFeedbackDialogFromSettings(); - } else { - FeedbackUtils.showFeedbackForm(getActivity()); - } + FeedbackUtils.showFeedbackForm(getActivity()); return true; } }); @@ -167,7 +159,7 @@ public final class SettingsFragment extends InputMethodSettingsFragment final boolean showVoiceKeyOption = res.getBoolean( R.bool.config_enable_show_voice_key_option); if (!showVoiceKeyOption) { - generalSettings.removePreference(mVoiceInputKeyPreference); + removePreference(Settings.PREF_VOICE_INPUT_KEY, generalSettings); } final PreferenceGroup advancedSettings = @@ -177,26 +169,27 @@ public final class SettingsFragment extends InputMethodSettingsFragment removePreference(Settings.PREF_VIBRATION_DURATION_SETTINGS, advancedSettings); } - mKeyPreviewPopupDismissDelay = - (ListPreference) findPreference(Settings.PREF_KEY_PREVIEW_POPUP_DISMISS_DELAY); - if (!Settings.readFromBuildConfigIfToShowKeyPreviewPopupSettingsOption(res)) { + if (!Settings.readFromBuildConfigIfToShowKeyPreviewPopupOption(res)) { removePreference(Settings.PREF_POPUP_ON, generalSettings); removePreference(Settings.PREF_KEY_PREVIEW_POPUP_DISMISS_DELAY, advancedSettings); } else { + // TODO: Cleanup this setup. + final ListPreference keyPreviewPopupDismissDelay = + (ListPreference) findPreference(Settings.PREF_KEY_PREVIEW_POPUP_DISMISS_DELAY); final String popupDismissDelayDefaultValue = Integer.toString(res.getInteger( R.integer.config_key_preview_linger_timeout)); - mKeyPreviewPopupDismissDelay.setEntries(new String[] { + keyPreviewPopupDismissDelay.setEntries(new String[] { res.getString(R.string.key_preview_popup_dismiss_no_delay), res.getString(R.string.key_preview_popup_dismiss_default_delay), }); - mKeyPreviewPopupDismissDelay.setEntryValues(new String[] { + keyPreviewPopupDismissDelay.setEntryValues(new String[] { "0", popupDismissDelayDefaultValue }); - if (null == mKeyPreviewPopupDismissDelay.getValue()) { - mKeyPreviewPopupDismissDelay.setValue(popupDismissDelayDefaultValue); + if (null == keyPreviewPopupDismissDelay.getValue()) { + keyPreviewPopupDismissDelay.setValue(popupDismissDelayDefaultValue); } - mKeyPreviewPopupDismissDelay.setEnabled( + keyPreviewPopupDismissDelay.setEnabled( Settings.readKeyPreviewPopupEnabled(prefs, res)); } @@ -243,20 +236,19 @@ public final class SettingsFragment extends InputMethodSettingsFragment @Override public void onResume() { super.onResume(); - final boolean isShortcutImeEnabled = SubtypeSwitcher.getInstance().isShortcutImeEnabled(); - if (!isShortcutImeEnabled) { - getPreferenceScreen().removePreference(mVoiceInputKeyPreference); - } final SharedPreferences prefs = getPreferenceManager().getSharedPreferences(); + final Resources res = getResources(); + final boolean isShortcutImeEnabled = SubtypeSwitcher.getInstance().isShortcutImeEnabled(); + setPreferenceEnabled(Settings.PREF_VOICE_INPUT_KEY, isShortcutImeEnabled); final CheckBoxPreference showSetupWizardIcon = (CheckBoxPreference)findPreference(Settings.PREF_SHOW_SETUP_WIZARD_ICON); if (showSetupWizardIcon != null) { showSetupWizardIcon.setChecked(Settings.readShowSetupWizardIcon(prefs, getActivity())); } - updateShowCorrectionSuggestionsSummary(); - updateKeyPreviewPopupDelaySummary(); - updateColorSchemeSummary(prefs, getResources()); - updateCustomInputStylesSummary(); + updateListPreferenceSummaryToCurrentValue(Settings.PREF_SHOW_SUGGESTIONS_SETTING); + updateListPreferenceSummaryToCurrentValue(Settings.PREF_KEY_PREVIEW_POPUP_DISMISS_DELAY); + updateListPreferenceSummaryToCurrentValue(Settings.PREF_KEYBOARD_LAYOUT); + updateCustomInputStylesSummary(prefs, res); } @Override @@ -287,50 +279,26 @@ public final class SettingsFragment extends InputMethodSettingsFragment LauncherIconVisibilityManager.updateSetupWizardIconVisibility(getActivity()); } ensureConsistencyOfAutoCorrectionSettings(); - updateShowCorrectionSuggestionsSummary(); - updateKeyPreviewPopupDelaySummary(); - updateColorSchemeSummary(prefs, res); + updateListPreferenceSummaryToCurrentValue(Settings.PREF_SHOW_SUGGESTIONS_SETTING); + updateListPreferenceSummaryToCurrentValue(Settings.PREF_KEY_PREVIEW_POPUP_DISMISS_DELAY); + updateListPreferenceSummaryToCurrentValue(Settings.PREF_KEYBOARD_LAYOUT); refreshEnablingsOfKeypressSoundAndVibrationSettings(prefs, getResources()); } private void ensureConsistencyOfAutoCorrectionSettings() { final String autoCorrectionOff = getResources().getString( R.string.auto_correction_threshold_mode_index_off); - final String currentSetting = mAutoCorrectionThresholdPreference.getValue(); - mBigramPrediction.setEnabled(!currentSetting.equals(autoCorrectionOff)); - } - - private void updateShowCorrectionSuggestionsSummary() { - mShowCorrectionSuggestionsPreference.setSummary( - getResources().getStringArray(R.array.prefs_suggestion_visibilities) - [mShowCorrectionSuggestionsPreference.findIndexOfValue( - mShowCorrectionSuggestionsPreference.getValue())]); + final ListPreference autoCorrectionThresholdPref = (ListPreference)findPreference( + Settings.PREF_AUTO_CORRECTION_THRESHOLD); + final String currentSetting = autoCorrectionThresholdPref.getValue(); + setPreferenceEnabled( + Settings.PREF_BIGRAM_PREDICTIONS, !currentSetting.equals(autoCorrectionOff)); } - private void updateColorSchemeSummary(final SharedPreferences prefs, final Resources res) { - // Because the "%s" summary trick of {@link ListPreference} doesn't work properly before - // KitKat, we need to update the summary by code. - final Preference preference = findPreference(Settings.PREF_KEYBOARD_LAYOUT); - if (!(preference instanceof ListPreference)) { - Log.w(TAG, "Can't find Keyboard Color Scheme preference"); - return; - } - final ListPreference colorSchemePreference = (ListPreference)preference; - final int themeIndex = Settings.readKeyboardThemeIndex(prefs, res); - int entryIndex = colorSchemePreference.findIndexOfValue(Integer.toString(themeIndex)); - if (entryIndex < 0) { - final int defaultThemeIndex = Settings.resetAndGetDefaultKeyboardThemeIndex(prefs, res); - entryIndex = colorSchemePreference.findIndexOfValue( - Integer.toString(defaultThemeIndex)); - } - colorSchemePreference.setSummary(colorSchemePreference.getEntries()[entryIndex]); - } - - private void updateCustomInputStylesSummary() { + private void updateCustomInputStylesSummary(final SharedPreferences prefs, + final Resources res) { final PreferenceScreen customInputStyles = (PreferenceScreen)findPreference(Settings.PREF_CUSTOM_INPUT_STYLES); - final SharedPreferences prefs = getPreferenceManager().getSharedPreferences(); - final Resources res = getResources(); final String prefSubtype = Settings.readPrefAdditionalSubtypes(prefs, res); final InputMethodSubtype[] subtypes = AdditionalSubtypeUtils.createAdditionalSubtypesArray(prefSubtype); @@ -342,13 +310,6 @@ public final class SettingsFragment extends InputMethodSettingsFragment customInputStyles.setSummary(styles); } - private void updateKeyPreviewPopupDelaySummary() { - final ListPreference lp = mKeyPreviewPopupDismissDelay; - final CharSequence[] entries = lp.getEntries(); - if (entries == null || entries.length <= 0) return; - lp.setSummary(entries[lp.findIndexOfValue(lp.getValue())]); - } - private void refreshEnablingsOfKeypressSoundAndVibrationSettings( final SharedPreferences sp, final Resources res) { setPreferenceEnabled(Settings.PREF_VIBRATION_DURATION_SETTINGS, diff --git a/java/src/com/android/inputmethod/latin/settings/SettingsValues.java b/java/src/com/android/inputmethod/latin/settings/SettingsValues.java index f331c78e5..d98e547fc 100644 --- a/java/src/com/android/inputmethod/latin/settings/SettingsValues.java +++ b/java/src/com/android/inputmethod/latin/settings/SettingsValues.java @@ -29,11 +29,9 @@ import com.android.inputmethod.latin.Dictionary; import com.android.inputmethod.latin.InputAttributes; import com.android.inputmethod.latin.R; import com.android.inputmethod.latin.RichInputMethodManager; -import com.android.inputmethod.latin.SubtypeSwitcher; import com.android.inputmethod.latin.SuggestedWords; import com.android.inputmethod.latin.SuggestedWords.SuggestedWordInfo; import com.android.inputmethod.latin.utils.CollectionUtils; -import com.android.inputmethod.latin.utils.InputTypeUtils; import com.android.inputmethod.latin.utils.StringUtils; import java.util.ArrayList; @@ -67,10 +65,11 @@ public final class SettingsValues { public final boolean mVibrateOn; public final boolean mSoundOn; public final boolean mKeyPreviewPopupOn; - private final boolean mShowsVoiceInputKey; + public final boolean mShowsVoiceInputKey; public final boolean mIncludesOtherImesInLanguageSwitchList; public final boolean mShowsLanguageSwitchKey; public final boolean mUseContactsDict; + public final boolean mUsePersonalizedDicts; public final boolean mUseDoubleSpacePeriod; public final boolean mBlockPotentiallyOffensive; // Use bigrams to predict the next word when there is no input for it yet @@ -148,6 +147,7 @@ public final class SettingsValues { Settings.PREF_INCLUDE_OTHER_IMES_IN_LANGUAGE_SWITCH_LIST, false); mShowsLanguageSwitchKey = Settings.readShowsLanguageSwitchKey(prefs); mUseContactsDict = prefs.getBoolean(Settings.PREF_KEY_USE_CONTACTS_DICT, true); + mUsePersonalizedDicts = prefs.getBoolean(Settings.PREF_KEY_USE_PERSONALIZED_DICTS, true); mUseDoubleSpacePeriod = prefs.getBoolean(Settings.PREF_KEY_USE_DOUBLE_SPACE_PERIOD, true); mBlockPotentiallyOffensive = Settings.readBlockPotentiallyOffensive(prefs, res); mAutoCorrectEnabled = Settings.readAutoCorrectEnabled(autoCorrectionThresholdRawValue, res); @@ -206,6 +206,7 @@ public final class SettingsValues { mIncludesOtherImesInLanguageSwitchList = false; mShowsLanguageSwitchKey = true; mUseContactsDict = true; + mUsePersonalizedDicts = true; mUseDoubleSpacePeriod = true; mBlockPotentiallyOffensive = true; mAutoCorrectEnabled = true; @@ -271,13 +272,6 @@ public final class SettingsValues { return mInputAttributes.mShouldInsertSpacesAutomatically; } - public boolean isVoiceKeyEnabled(final EditorInfo editorInfo) { - final boolean shortcutImeEnabled = SubtypeSwitcher.getInstance().isShortcutImeEnabled(); - final int inputType = (editorInfo != null) ? editorInfo.inputType : 0; - return shortcutImeEnabled && mShowsVoiceInputKey - && !InputTypeUtils.isPasswordInputType(inputType); - } - public boolean isLanguageSwitchKeyEnabled() { if (!mShowsLanguageSwitchKey) { return false; @@ -374,16 +368,20 @@ public final class SettingsValues { return autoCorrectionThreshold; } - private static boolean needsToShowVoiceInputKey(SharedPreferences prefs, Resources res) { - final String voiceModeMain = res.getString(R.string.voice_mode_main); - final String voiceMode = prefs.getString(Settings.PREF_VOICE_MODE_OBSOLETE, voiceModeMain); - final boolean showsVoiceInputKey = voiceMode == null || voiceMode.equals(voiceModeMain); - if (!showsVoiceInputKey) { - // Migrate settings from PREF_VOICE_MODE_OBSOLETE to PREF_VOICE_INPUT_KEY - // Set voiceModeMain as a value of obsolete voice mode settings. - prefs.edit().putString(Settings.PREF_VOICE_MODE_OBSOLETE, voiceModeMain).apply(); - // Disable voice input key. - prefs.edit().putBoolean(Settings.PREF_VOICE_INPUT_KEY, false).apply(); + private static boolean needsToShowVoiceInputKey(final SharedPreferences prefs, + final Resources res) { + if (!prefs.contains(Settings.PREF_VOICE_INPUT_KEY)) { + // Migrate preference from {@link Settings#PREF_VOICE_MODE_OBSOLETE} to + // {@link Settings#PREF_VOICE_INPUT_KEY}. + final String voiceModeMain = res.getString(R.string.voice_mode_main); + final String voiceMode = prefs.getString( + Settings.PREF_VOICE_MODE_OBSOLETE, voiceModeMain); + final boolean shouldShowVoiceInputKey = voiceModeMain.equals(voiceMode); + prefs.edit().putBoolean(Settings.PREF_VOICE_INPUT_KEY, shouldShowVoiceInputKey).apply(); + } + // Remove the obsolete preference if exists. + if (prefs.contains(Settings.PREF_VOICE_MODE_OBSOLETE)) { + prefs.edit().remove(Settings.PREF_VOICE_MODE_OBSOLETE).apply(); } return prefs.getBoolean(Settings.PREF_VOICE_INPUT_KEY, true); } diff --git a/java/src/com/android/inputmethod/latin/setup/SetupWizardActivity.java b/java/src/com/android/inputmethod/latin/setup/SetupWizardActivity.java index c4a813c24..5072fabd6 100644 --- a/java/src/com/android/inputmethod/latin/setup/SetupWizardActivity.java +++ b/java/src/com/android/inputmethod/latin/setup/SetupWizardActivity.java @@ -38,7 +38,7 @@ import com.android.inputmethod.compat.ViewCompatUtils; import com.android.inputmethod.latin.R; import com.android.inputmethod.latin.settings.SettingsActivity; import com.android.inputmethod.latin.utils.CollectionUtils; -import com.android.inputmethod.latin.utils.StaticInnerHandlerWrapper; +import com.android.inputmethod.latin.utils.LeakGuardHandlerWrapper; import java.util.ArrayList; @@ -74,21 +74,21 @@ public final class SetupWizardActivity extends Activity implements View.OnClickL private SettingsPoolingHandler mHandler; private static final class SettingsPoolingHandler - extends StaticInnerHandlerWrapper<SetupWizardActivity> { + extends LeakGuardHandlerWrapper<SetupWizardActivity> { private static final int MSG_POLLING_IME_SETTINGS = 0; private static final long IME_SETTINGS_POLLING_INTERVAL = 200; private final InputMethodManager mImmInHandler; - public SettingsPoolingHandler(final SetupWizardActivity outerInstance, + public SettingsPoolingHandler(final SetupWizardActivity ownerInstance, final InputMethodManager imm) { - super(outerInstance); + super(ownerInstance); mImmInHandler = imm; } @Override public void handleMessage(final Message msg) { - final SetupWizardActivity setupWizardActivity = getOuterInstance(); + final SetupWizardActivity setupWizardActivity = getOwnerInstance(); if (setupWizardActivity == null) { return; } diff --git a/java/src/com/android/inputmethod/latin/spellcheck/AndroidSpellCheckerService.java b/java/src/com/android/inputmethod/latin/spellcheck/AndroidSpellCheckerService.java index 503b18b1b..c108b20ed 100644 --- a/java/src/com/android/inputmethod/latin/spellcheck/AndroidSpellCheckerService.java +++ b/java/src/com/android/inputmethod/latin/spellcheck/AndroidSpellCheckerService.java @@ -428,7 +428,7 @@ public final class AndroidSpellCheckerService extends SpellCheckerService final String localeStr = locale.toString(); UserBinaryDictionary userDictionary = mUserDictionaries.get(localeStr); if (null == userDictionary) { - userDictionary = new SynchronouslyLoadedUserBinaryDictionary(this, localeStr, true); + userDictionary = new SynchronouslyLoadedUserBinaryDictionary(this, locale, true); mUserDictionaries.put(localeStr, userDictionary); } dictionaryCollection.addDictionary(userDictionary); diff --git a/java/src/com/android/inputmethod/latin/spellcheck/DictionaryPool.java b/java/src/com/android/inputmethod/latin/spellcheck/DictionaryPool.java index a0aed2829..b7a5a4026 100644 --- a/java/src/com/android/inputmethod/latin/spellcheck/DictionaryPool.java +++ b/java/src/com/android/inputmethod/latin/spellcheck/DictionaryPool.java @@ -49,6 +49,7 @@ public final class DictionaryPool extends LinkedBlockingQueue<DictAndKeyboard> { final static ArrayList<SuggestedWordInfo> noSuggestions = CollectionUtils.newArrayList(); private final static DictAndKeyboard dummyDict = new DictAndKeyboard( new Dictionary(Dictionary.TYPE_MAIN) { + // TODO: this dummy dictionary should be a singleton in the Dictionary class. @Override public ArrayList<SuggestedWordInfo> getSuggestions(final WordComposer composer, final String prevWord, final ProximityInfo proximityInfo, diff --git a/java/src/com/android/inputmethod/latin/suggestions/MoreSuggestions.java b/java/src/com/android/inputmethod/latin/suggestions/MoreSuggestions.java index acd47450b..52012c846 100644 --- a/java/src/com/android/inputmethod/latin/suggestions/MoreSuggestions.java +++ b/java/src/com/android/inputmethod/latin/suggestions/MoreSuggestions.java @@ -66,7 +66,8 @@ public final class MoreSuggestions extends Keyboard { clearKeys(); mDivider = res.getDrawable(R.drawable.more_suggestions_divider); mDividerWidth = mDivider.getIntrinsicWidth(); - final float padding = res.getDimension(R.dimen.more_suggestions_key_horizontal_padding); + final float padding = res.getDimension( + R.dimen.config_more_suggestions_key_horizontal_padding); int row = 0; int index = fromIndex; @@ -75,7 +76,7 @@ public final class MoreSuggestions extends Keyboard { while (index < size) { final String word = suggestedWords.getWord(index); // TODO: Should take care of text x-scaling. - mWidths[index] = (int)(TypefaceUtils.getLabelWidth(word, paint) + padding); + mWidths[index] = (int)(TypefaceUtils.getStringWidth(word, paint) + padding); final int numColumn = index - rowStartIndex + 1; final int columnWidth = (maxWidth - mDividerWidth * (numColumn - 1)) / numColumn; diff --git a/java/src/com/android/inputmethod/latin/suggestions/MoreSuggestionsView.java b/java/src/com/android/inputmethod/latin/suggestions/MoreSuggestionsView.java index 0ebe37782..549ff0d9d 100644 --- a/java/src/com/android/inputmethod/latin/suggestions/MoreSuggestionsView.java +++ b/java/src/com/android/inputmethod/latin/suggestions/MoreSuggestionsView.java @@ -54,7 +54,7 @@ public final class MoreSuggestionsView extends MoreKeysKeyboardView { public void adjustVerticalCorrectionForModalMode() { // Set vertical correction to zero (Reset more keys keyboard sliding allowance - // {@link R#dimen.more_keys_keyboard_slide_allowance}). + // {@link R#dimen.config_more_keys_keyboard_slide_allowance}). mKeyDetector.setKeyboard(getKeyboard(), -getPaddingLeft(), -getPaddingTop()); } diff --git a/java/src/com/android/inputmethod/latin/suggestions/SuggestionStripLayoutHelper.java b/java/src/com/android/inputmethod/latin/suggestions/SuggestionStripLayoutHelper.java index faa5560e4..72281e62c 100644 --- a/java/src/com/android/inputmethod/latin/suggestions/SuggestionStripLayoutHelper.java +++ b/java/src/com/android/inputmethod/latin/suggestions/SuggestionStripLayoutHelper.java @@ -119,7 +119,8 @@ final class SuggestionStripLayoutHelper { mDividerWidth = dividerView.getMeasuredWidth(); final Resources res = wordView.getResources(); - mSuggestionsStripHeight = res.getDimensionPixelSize(R.dimen.suggestions_strip_height); + mSuggestionsStripHeight = res.getDimensionPixelSize( + R.dimen.config_suggestions_strip_height); final TypedArray a = context.obtainStyledAttributes(attrs, R.styleable.SuggestionStripView, defStyle, R.style.SuggestionStripView); @@ -145,15 +146,17 @@ final class SuggestionStripLayoutHelper { a.recycle(); mMoreSuggestionsHint = getMoreSuggestionsHint(res, - res.getDimension(R.dimen.more_suggestions_hint_text_size), mColorAutoCorrect); + res.getDimension(R.dimen.config_more_suggestions_hint_text_size), + mColorAutoCorrect); mCenterPositionInStrip = mSuggestionsCountInStrip / 2; // Assuming there are at least three suggestions. Also, note that the suggestions are // laid out according to script direction, so this is left of the center for LTR scripts // and right of the center for RTL scripts. mTypedWordPositionWhenAutocorrect = mCenterPositionInStrip - 1; mMoreSuggestionsBottomGap = res.getDimensionPixelOffset( - R.dimen.more_suggestions_bottom_gap); - mMoreSuggestionsRowHeight = res.getDimensionPixelSize(R.dimen.more_suggestions_row_height); + R.dimen.config_more_suggestions_bottom_gap); + mMoreSuggestionsRowHeight = res.getDimensionPixelSize( + R.dimen.config_more_suggestions_row_height); final LayoutInflater inflater = LayoutInflater.from(context); mWordToSaveView = (TextView)inflater.inflate(R.layout.suggestion_word, null); diff --git a/java/src/com/android/inputmethod/latin/suggestions/SuggestionStripView.java b/java/src/com/android/inputmethod/latin/suggestions/SuggestionStripView.java index 75f17c559..073148a50 100644 --- a/java/src/com/android/inputmethod/latin/suggestions/SuggestionStripView.java +++ b/java/src/com/android/inputmethod/latin/suggestions/SuggestionStripView.java @@ -112,7 +112,7 @@ public final class SuggestionStripView extends RelativeLayout implements OnClick final Resources res = context.getResources(); mMoreSuggestionsModalTolerance = res.getDimensionPixelOffset( - R.dimen.more_suggestions_modal_tolerance); + R.dimen.config_more_suggestions_modal_tolerance); mMoreSuggestionsSlidingDetector = new GestureDetector( context, mMoreSuggestionsSlidingListener); } @@ -162,19 +162,19 @@ public final class SuggestionStripView extends RelativeLayout implements OnClick mSuggestionsStrip.removeAllViews(); removeAllViews(); addView(mSuggestionsStrip); - mMoreSuggestionsView.dismissMoreKeysPanel(); + dismissMoreSuggestionsPanel(); } private final MoreSuggestionsListener mMoreSuggestionsListener = new MoreSuggestionsListener() { @Override public void onSuggestionSelected(final int index, final SuggestedWordInfo wordInfo) { mListener.pickSuggestionManually(index, wordInfo); - mMoreSuggestionsView.dismissMoreKeysPanel(); + dismissMoreSuggestionsPanel(); } @Override public void onCancelInput() { - mMoreSuggestionsView.dismissMoreKeysPanel(); + dismissMoreSuggestionsPanel(); } }; @@ -192,10 +192,18 @@ public final class SuggestionStripView extends RelativeLayout implements OnClick @Override public void onCancelMoreKeysPanel(final MoreKeysPanel panel) { - mMoreSuggestionsView.dismissMoreKeysPanel(); + dismissMoreSuggestionsPanel(); } }; + public boolean isShowingMoreSuggestionPanel() { + return mMoreSuggestionsView.isShowingInParent(); + } + + public void dismissMoreSuggestionsPanel() { + mMoreSuggestionsView.dismissMoreKeysPanel(); + } + @Override public boolean onLongClick(final View view) { AudioAndHapticFeedbackManager.getInstance().performHapticAndAudioFeedback( @@ -322,6 +330,6 @@ public final class SuggestionStripView extends RelativeLayout implements OnClick @Override protected void onDetachedFromWindow() { super.onDetachedFromWindow(); - mMoreSuggestionsView.dismissMoreKeysPanel(); + dismissMoreSuggestionsPanel(); } } diff --git a/java/src/com/android/inputmethod/latin/utils/AdditionalSubtypeUtils.java b/java/src/com/android/inputmethod/latin/utils/AdditionalSubtypeUtils.java index d87f6f3c4..ef1d0f42c 100644 --- a/java/src/com/android/inputmethod/latin/utils/AdditionalSubtypeUtils.java +++ b/java/src/com/android/inputmethod/latin/utils/AdditionalSubtypeUtils.java @@ -17,21 +17,25 @@ package com.android.inputmethod.latin.utils; import static com.android.inputmethod.latin.Constants.Subtype.KEYBOARD_MODE; +import static com.android.inputmethod.latin.Constants.Subtype.ExtraValue.EMOJI_CAPABLE; import static com.android.inputmethod.latin.Constants.Subtype.ExtraValue.IS_ADDITIONAL_SUBTYPE; import static com.android.inputmethod.latin.Constants.Subtype.ExtraValue.KEYBOARD_LAYOUT_SET; import static com.android.inputmethod.latin.Constants.Subtype.ExtraValue.UNTRANSLATABLE_STRING_IN_SUBTYPE_NAME; import android.os.Build; import android.text.TextUtils; +import android.util.Log; import android.view.inputmethod.InputMethodSubtype; import com.android.inputmethod.compat.InputMethodSubtypeCompatUtils; -import com.android.inputmethod.latin.Constants; import com.android.inputmethod.latin.R; import java.util.ArrayList; +import java.util.Arrays; public final class AdditionalSubtypeUtils { + private static final String TAG = AdditionalSubtypeUtils.class.getSimpleName(); + private static final InputMethodSubtype[] EMPTY_SUBTYPE_ARRAY = new InputMethodSubtype[0]; private AdditionalSubtypeUtils() { @@ -43,6 +47,11 @@ public final class AdditionalSubtypeUtils { } private static final String LOCALE_AND_LAYOUT_SEPARATOR = ":"; + private static final int INDEX_OF_LOCALE = 0; + private static final int INDEX_OF_KEYBOARD_LAYOUT = 1; + private static final int INDEX_OF_EXTRA_VALUE = 2; + private static final int LENGTH_WITHOUT_EXTRA_VALUE = (INDEX_OF_KEYBOARD_LAYOUT + 1); + private static final int LENGTH_WITH_EXTRA_VALUE = (INDEX_OF_EXTRA_VALUE + 1); private static final String PREF_SUBTYPE_SEPARATOR = ";"; public static InputMethodSubtype createAdditionalSubtype(final String localeString, @@ -79,17 +88,6 @@ public final class AdditionalSubtypeUtils { : basePrefSubtype + LOCALE_AND_LAYOUT_SEPARATOR + extraValue; } - public static InputMethodSubtype createAdditionalSubtype(final String prefSubtype) { - final String elems[] = prefSubtype.split(LOCALE_AND_LAYOUT_SEPARATOR); - if (elems.length < 2 || elems.length > 3) { - throw new RuntimeException("Unknown additional subtype specified: " + prefSubtype); - } - final String localeString = elems[0]; - final String keyboardLayoutSetName = elems[1]; - final String extraValue = (elems.length == 3) ? elems[2] : null; - return createAdditionalSubtype(localeString, keyboardLayoutSetName, extraValue); - } - public static InputMethodSubtype[] createAdditionalSubtypesArray(final String prefSubtypes) { if (TextUtils.isEmpty(prefSubtypes)) { return EMPTY_SUBTYPE_ARRAY; @@ -98,7 +96,19 @@ public final class AdditionalSubtypeUtils { final ArrayList<InputMethodSubtype> subtypesList = CollectionUtils.newArrayList(prefSubtypeArray.length); for (final String prefSubtype : prefSubtypeArray) { - final InputMethodSubtype subtype = createAdditionalSubtype(prefSubtype); + final String elems[] = prefSubtype.split(LOCALE_AND_LAYOUT_SEPARATOR); + if (elems.length != LENGTH_WITHOUT_EXTRA_VALUE + && elems.length != LENGTH_WITH_EXTRA_VALUE) { + Log.w(TAG, "Unknown additional subtype specified: " + prefSubtype + " in " + + prefSubtypes); + continue; + } + final String localeString = elems[INDEX_OF_LOCALE]; + final String keyboardLayoutSetName = elems[INDEX_OF_KEYBOARD_LAYOUT]; + final String extraValue = (elems.length == LENGTH_WITH_EXTRA_VALUE) + ? elems[INDEX_OF_EXTRA_VALUE] : null; + final InputMethodSubtype subtype = createAdditionalSubtype( + localeString, keyboardLayoutSetName, extraValue); if (subtype.getNameResId() == SubtypeLocaleUtils.UNKNOWN_KEYBOARD_LAYOUT) { // Skip unknown keyboard layout subtype. This may happen when predefined keyboard // layout has been removed. @@ -137,31 +147,36 @@ public final class AdditionalSubtypeUtils { return sb.toString(); } - private static InputMethodSubtype buildInputMethodSubtype(int nameId, String localeString, - String layoutExtraValue, String additionalSubtypeExtraValue) { - // CAVEAT! If you want to change subtypeId after changing the extra values, - // you must change "getInputMethodSubtypeId". But it will remove the additional keyboard - // from the current users. So, you should be really careful to change it. - final int subtypeId = getInputMethodSubtypeId(nameId, localeString, layoutExtraValue, - additionalSubtypeExtraValue); + private static InputMethodSubtype buildInputMethodSubtype(final int nameId, + final String localeString, final String layoutExtraValue, + final String additionalSubtypeExtraValue) { + // To preserve additional subtype settings and user's selection across OS updates, subtype + // id shouldn't be changed. New attributes, such as emojiCapable, are carefully excluded + // from the calculation of subtype id. + final String compatibleExtraValue = StringUtils.joinCommaSplittableText( + layoutExtraValue, additionalSubtypeExtraValue); + final int compatibleSubtypeId = getInputMethodSubtypeId(localeString, compatibleExtraValue); final String extraValue; - if (Build.VERSION.SDK_INT >= Build.VERSION_CODES.JELLY_BEAN_MR1) { - extraValue = layoutExtraValue + "," + additionalSubtypeExtraValue - + "," + Constants.Subtype.ExtraValue.ASCII_CAPABLE - + "," + Constants.Subtype.ExtraValue.EMOJI_CAPABLE; + // Color Emoji is supported from KitKat. + if (Build.VERSION.SDK_INT >= Build.VERSION_CODES.KITKAT) { + extraValue = StringUtils.appendToCommaSplittableTextIfNotExists( + EMOJI_CAPABLE, compatibleExtraValue); } else { - extraValue = layoutExtraValue + "," + additionalSubtypeExtraValue; + extraValue = compatibleExtraValue; } return InputMethodSubtypeCompatUtils.newInputMethodSubtype(nameId, R.drawable.ic_ime_switcher_dark, localeString, KEYBOARD_MODE, extraValue, - false, false, subtypeId); + false, false, compatibleSubtypeId); } - private static int getInputMethodSubtypeId(int nameId, String localeString, - String layoutExtraValue, String additionalSubtypeExtraValue) { - // TODO: Use InputMethodSubtypeBuilder once we use SDK version 19. - return (new InputMethodSubtype(nameId, R.drawable.ic_ime_switcher_dark, - localeString, KEYBOARD_MODE, layoutExtraValue + "," + additionalSubtypeExtraValue, - false, false)).hashCode(); + private static int getInputMethodSubtypeId(final String localeString, final String extraValue) { + // From the compatibility point of view, the calculation of subtype id has been copied from + // {@link InputMethodSubtype} of JellyBean MR2. + return Arrays.hashCode(new Object[] { + localeString, + KEYBOARD_MODE, + extraValue, + false /* isAuxiliary */, + false /* overrideImplicitlyEnabledSubtype */ }); } } diff --git a/java/src/com/android/inputmethod/latin/utils/ApplicationUtils.java b/java/src/com/android/inputmethod/latin/utils/ApplicationUtils.java index 08a2a8c5a..e521ec807 100644 --- a/java/src/com/android/inputmethod/latin/utils/ApplicationUtils.java +++ b/java/src/com/android/inputmethod/latin/utils/ApplicationUtils.java @@ -62,4 +62,22 @@ public final class ApplicationUtils { } return ""; } + + /** + * A utility method to get the application's PackageInfo.versionCode + * @return the application's PackageInfo.versionCode + */ + public static int getVersionCode(final Context context) { + try { + if (context == null) { + return 0; + } + final String packageName = context.getPackageName(); + final PackageInfo info = context.getPackageManager().getPackageInfo(packageName, 0); + return info.versionCode; + } catch (final NameNotFoundException e) { + Log.e(TAG, "Could not find version info.", e); + } + return 0; + } } diff --git a/java/src/com/android/inputmethod/latin/utils/AsyncResultHolder.java b/java/src/com/android/inputmethod/latin/utils/AsyncResultHolder.java index c2e97a36f..d12aad639 100644 --- a/java/src/com/android/inputmethod/latin/utils/AsyncResultHolder.java +++ b/java/src/com/android/inputmethod/latin/utils/AsyncResultHolder.java @@ -20,7 +20,7 @@ import java.util.concurrent.CountDownLatch; import java.util.concurrent.TimeUnit; /** - * This class is a holder of a result of asynchronous computation. + * This class is a holder of the result of an asynchronous computation. * * @param <E> the type of the result. */ @@ -36,9 +36,9 @@ public class AsyncResultHolder<E> { } /** - * Sets the result value to this holder. + * Sets the result value of this holder. * - * @param result the value which is set. + * @param result the value to set. */ public void set(final E result) { synchronized(mLock) { @@ -54,12 +54,12 @@ public class AsyncResultHolder<E> { * Causes the current thread to wait unless the value is set or the specified time is elapsed. * * @param defaultValue the default value. - * @param timeOut the time to wait. - * @return if the result is set until the time limit then the result, otherwise defaultValue. + * @param timeOut the maximum time to wait. + * @return if the result is set before the time limit then the result, otherwise defaultValue. */ public E get(final E defaultValue, final long timeOut) { try { - if(mLatch.await(timeOut, TimeUnit.MILLISECONDS)) { + if (mLatch.await(timeOut, TimeUnit.MILLISECONDS)) { return mResult; } else { return defaultValue; diff --git a/java/src/com/android/inputmethod/latin/utils/FileUtils.java b/java/src/com/android/inputmethod/latin/utils/FileUtils.java new file mode 100644 index 000000000..83c1e7c4d --- /dev/null +++ b/java/src/com/android/inputmethod/latin/utils/FileUtils.java @@ -0,0 +1,33 @@ +/* + * Copyright (C) 2013 The Android Open Source Project + * + * Licensed under the Apache License, Version 2.0 (the "License"); you may not + * use this file except in compliance with the License. You may obtain a copy of + * the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, WITHOUT + * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the + * License for the specific language governing permissions and limitations under + * the License. + */ + +package com.android.inputmethod.latin.utils; + +import java.io.File; + +/** + * A simple class to help with removing directories recursively. + */ +public class FileUtils { + public static boolean deleteRecursively(final File path) { + if (path.isDirectory()) { + for (final File child : path.listFiles()) { + deleteRecursively(child); + } + } + return path.delete(); + } +} diff --git a/java/src/com/android/inputmethod/latin/utils/JsonUtils.java b/java/src/com/android/inputmethod/latin/utils/JsonUtils.java new file mode 100644 index 000000000..764ef72ce --- /dev/null +++ b/java/src/com/android/inputmethod/latin/utils/JsonUtils.java @@ -0,0 +1,103 @@ +/* + * Copyright (C) 2013 The Android Open Source Project + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package com.android.inputmethod.latin.utils; + +import android.util.JsonReader; +import android.util.JsonWriter; +import android.util.Log; + +import java.io.Closeable; +import java.io.IOException; +import java.io.StringReader; +import java.io.StringWriter; +import java.util.ArrayList; +import java.util.Collections; +import java.util.List; + +public final class JsonUtils { + private static final String TAG = JsonUtils.class.getSimpleName(); + + private static final String INTEGER_CLASS_NAME = Integer.class.getSimpleName(); + private static final String STRING_CLASS_NAME = String.class.getSimpleName(); + + private static final String EMPTY_STRING = ""; + + public static List<Object> jsonStrToList(final String s) { + final ArrayList<Object> list = CollectionUtils.newArrayList(); + final JsonReader reader = new JsonReader(new StringReader(s)); + try { + reader.beginArray(); + while (reader.hasNext()) { + reader.beginObject(); + while (reader.hasNext()) { + final String name = reader.nextName(); + if (name.equals(INTEGER_CLASS_NAME)) { + list.add(reader.nextInt()); + } else if (name.equals(STRING_CLASS_NAME)) { + list.add(reader.nextString()); + } else { + Log.w(TAG, "Invalid name: " + name); + reader.skipValue(); + } + } + reader.endObject(); + } + reader.endArray(); + return list; + } catch (final IOException e) { + } finally { + close(reader); + } + return Collections.<Object>emptyList(); + } + + public static String listToJsonStr(final List<Object> list) { + if (list == null || list.isEmpty()) { + return EMPTY_STRING; + } + final StringWriter sw = new StringWriter(); + final JsonWriter writer = new JsonWriter(sw); + try { + writer.beginArray(); + for (final Object o : list) { + writer.beginObject(); + if (o instanceof Integer) { + writer.name(INTEGER_CLASS_NAME).value((Integer)o); + } else if (o instanceof String) { + writer.name(STRING_CLASS_NAME).value((String)o); + } + writer.endObject(); + } + writer.endArray(); + return sw.toString(); + } catch (final IOException e) { + } finally { + close(writer); + } + return EMPTY_STRING; + } + + private static void close(final Closeable closeable) { + try { + if (closeable != null) { + closeable.close(); + } + } catch (final IOException e) { + // Ignore + } + } +} diff --git a/java/src/com/android/inputmethod/latin/utils/LatinImeLoggerUtils.java b/java/src/com/android/inputmethod/latin/utils/LatinImeLoggerUtils.java index e958a7e71..d14ba508b 100644 --- a/java/src/com/android/inputmethod/latin/utils/LatinImeLoggerUtils.java +++ b/java/src/com/android/inputmethod/latin/utils/LatinImeLoggerUtils.java @@ -35,7 +35,7 @@ public final class LatinImeLoggerUtils { public static void onSeparator(final int code, final int x, final int y) { // Helper method to log a single code point separator // TODO: cache this mapping of a code point to a string in a sparse array in StringUtils - onSeparator(new String(new int[]{code}, 0, 1), x, y); + onSeparator(StringUtils.newSingleCodePointString(code), x, y); } public static void onSeparator(final String separator, final int x, final int y) { diff --git a/java/src/com/android/inputmethod/latin/utils/StaticInnerHandlerWrapper.java b/java/src/com/android/inputmethod/latin/utils/LeakGuardHandlerWrapper.java index 44e5d17b4..8469c87b0 100644 --- a/java/src/com/android/inputmethod/latin/utils/StaticInnerHandlerWrapper.java +++ b/java/src/com/android/inputmethod/latin/utils/LeakGuardHandlerWrapper.java @@ -21,22 +21,22 @@ import android.os.Looper; import java.lang.ref.WeakReference; -public class StaticInnerHandlerWrapper<T> extends Handler { - private final WeakReference<T> mOuterInstanceRef; +public class LeakGuardHandlerWrapper<T> extends Handler { + private final WeakReference<T> mOwnerInstanceRef; - public StaticInnerHandlerWrapper(final T outerInstance) { - this(outerInstance, Looper.myLooper()); + public LeakGuardHandlerWrapper(final T ownerInstance) { + this(ownerInstance, Looper.myLooper()); } - public StaticInnerHandlerWrapper(final T outerInstance, final Looper looper) { + public LeakGuardHandlerWrapper(final T ownerInstance, final Looper looper) { super(looper); - if (outerInstance == null) { - throw new NullPointerException("outerInstance is null"); + if (ownerInstance == null) { + throw new NullPointerException("ownerInstance is null"); } - mOuterInstanceRef = new WeakReference<T>(outerInstance); + mOwnerInstanceRef = new WeakReference<T>(ownerInstance); } - public T getOuterInstance() { - return mOuterInstanceRef.get(); + public T getOwnerInstance() { + return mOwnerInstanceRef.get(); } } diff --git a/java/src/com/android/inputmethod/latin/utils/LocaleUtils.java b/java/src/com/android/inputmethod/latin/utils/LocaleUtils.java index 22045aa38..0c55484b4 100644 --- a/java/src/com/android/inputmethod/latin/utils/LocaleUtils.java +++ b/java/src/com/android/inputmethod/latin/utils/LocaleUtils.java @@ -30,9 +30,6 @@ import java.util.Locale; * dictionary pack. */ public final class LocaleUtils { - private static final HashMap<String, Long> EMPTY_LT_HASH_MAP = CollectionUtils.newHashMap(); - private static final String LOCALE_AND_TIME_STR_SEPARATER = ","; - private LocaleUtils() { // Intentional empty constructor for utility class. } @@ -168,12 +165,14 @@ public final class LocaleUtils { * Creates a locale from a string specification. */ public static Locale constructLocaleFromString(final String localeStr) { - if (localeStr == null) + if (localeStr == null) { return null; + } synchronized (sLocaleCache) { - if (sLocaleCache.containsKey(localeStr)) - return sLocaleCache.get(localeStr); - Locale retval = null; + Locale retval = sLocaleCache.get(localeStr); + if (retval != null) { + return retval; + } String[] localeParams = localeStr.split("_", 3); if (localeParams.length == 1) { retval = new Locale(localeParams[0]); @@ -188,38 +187,4 @@ public final class LocaleUtils { return retval; } } - - public static HashMap<String, Long> localeAndTimeStrToHashMap(String str) { - if (TextUtils.isEmpty(str)) { - return EMPTY_LT_HASH_MAP; - } - final String[] ss = str.split(LOCALE_AND_TIME_STR_SEPARATER); - final int N = ss.length; - if (N < 2 || N % 2 != 0) { - return EMPTY_LT_HASH_MAP; - } - final HashMap<String, Long> retval = CollectionUtils.newHashMap(); - for (int i = 0; i < N / 2; ++i) { - final String localeStr = ss[i * 2]; - final long time = Long.valueOf(ss[i * 2 + 1]); - retval.put(localeStr, time); - } - return retval; - } - - public static String localeAndTimeHashMapToStr(HashMap<String, Long> map) { - if (map == null || map.isEmpty()) { - return ""; - } - final StringBuilder builder = new StringBuilder(); - for (String localeStr : map.keySet()) { - if (builder.length() > 0) { - builder.append(LOCALE_AND_TIME_STR_SEPARATER); - } - final Long time = map.get(localeStr); - builder.append(localeStr).append(LOCALE_AND_TIME_STR_SEPARATER); - builder.append(String.valueOf(time)); - } - return builder.toString(); - } } diff --git a/java/src/com/android/inputmethod/latin/utils/PrioritizedSerialExecutor.java b/java/src/com/android/inputmethod/latin/utils/PrioritizedSerialExecutor.java index 201a70d42..b10d08af3 100644 --- a/java/src/com/android/inputmethod/latin/utils/PrioritizedSerialExecutor.java +++ b/java/src/com/android/inputmethod/latin/utils/PrioritizedSerialExecutor.java @@ -137,6 +137,7 @@ public class PrioritizedSerialExecutor { public void shutdown() { synchronized(mLock) { mIsShutdown = true; + mThreadPoolExecutor.shutdown(); } } diff --git a/java/src/com/android/inputmethod/latin/utils/ResourceUtils.java b/java/src/com/android/inputmethod/latin/utils/ResourceUtils.java index 22c92446a..deb28a08d 100644 --- a/java/src/com/android/inputmethod/latin/utils/ResourceUtils.java +++ b/java/src/com/android/inputmethod/latin/utils/ResourceUtils.java @@ -227,19 +227,19 @@ public final class ResourceUtils { final String keyboardHeightString = getDeviceOverrideValue(res, R.array.keyboard_heights); final float keyboardHeight; if (TextUtils.isEmpty(keyboardHeightString)) { - keyboardHeight = res.getDimension(R.dimen.keyboardHeight); + keyboardHeight = res.getDimension(R.dimen.config_default_keyboard_height); } else { keyboardHeight = Float.parseFloat(keyboardHeightString) * dm.density; } final float maxKeyboardHeight = res.getFraction( - R.fraction.maxKeyboardHeight, dm.heightPixels, dm.heightPixels); + R.fraction.config_max_keyboard_height, dm.heightPixels, dm.heightPixels); float minKeyboardHeight = res.getFraction( - R.fraction.minKeyboardHeight, dm.heightPixels, dm.heightPixels); + R.fraction.config_min_keyboard_height, dm.heightPixels, dm.heightPixels); if (minKeyboardHeight < 0.0f) { // Specified fraction was negative, so it should be calculated against display // width. minKeyboardHeight = -res.getFraction( - R.fraction.minKeyboardHeight, dm.widthPixels, dm.widthPixels); + R.fraction.config_min_keyboard_height, dm.widthPixels, dm.widthPixels); } // Keyboard height will not exceed maxKeyboardHeight and will not be less than // minKeyboardHeight. diff --git a/java/src/com/android/inputmethod/latin/utils/SpannableStringUtils.java b/java/src/com/android/inputmethod/latin/utils/SpannableStringUtils.java index b51fd9377..be0955456 100644 --- a/java/src/com/android/inputmethod/latin/utils/SpannableStringUtils.java +++ b/java/src/com/android/inputmethod/latin/utils/SpannableStringUtils.java @@ -40,12 +40,17 @@ public final class SpannableStringUtils { * are out of range in <code>dest</code>. */ public static void copyNonParagraphSuggestionSpansFrom(Spanned source, int start, int end, - Spannable dest, int destoff) { + Spannable dest, int destoff) { Object[] spans = source.getSpans(start, end, SuggestionSpan.class); for (int i = 0; i < spans.length; i++) { int fl = source.getSpanFlags(spans[i]); - if (0 != (fl & Spannable.SPAN_PARAGRAPH)) continue; + // We don't care about the PARAGRAPH flag in LatinIME code. However, if this flag + // is set, Spannable#setSpan will throw an exception unless the span is on the edge + // of a word. But the spans have been split into two by the getText{Before,After}Cursor + // methods, so after concatenation they may end in the middle of a word. + // Since we don't use them, we can just remove them and avoid crashing. + fl &= ~Spannable.SPAN_PARAGRAPH; int st = source.getSpanStart(spans[i]); int en = source.getSpanEnd(spans[i]); diff --git a/java/src/com/android/inputmethod/latin/utils/StringUtils.java b/java/src/com/android/inputmethod/latin/utils/StringUtils.java index a36548392..85f44541e 100644 --- a/java/src/com/android/inputmethod/latin/utils/StringUtils.java +++ b/java/src/com/android/inputmethod/latin/utils/StringUtils.java @@ -16,20 +16,15 @@ package com.android.inputmethod.latin.utils; +import android.text.TextUtils; +import android.util.Log; + import com.android.inputmethod.annotations.UsedForTesting; import com.android.inputmethod.latin.Constants; import com.android.inputmethod.latin.settings.SettingsValues; -import android.text.TextUtils; -import android.util.JsonReader; -import android.util.JsonWriter; -import android.util.Log; - import java.io.IOException; -import java.io.StringReader; -import java.io.StringWriter; import java.util.ArrayList; -import java.util.Collections; import java.util.List; import java.util.Locale; @@ -39,6 +34,8 @@ public final class StringUtils { public static final int CAPITALIZE_FIRST = 1; // First only public static final int CAPITALIZE_ALL = 2; // All caps + private static final String EMPTY_STRING = ""; + private StringUtils() { // This utility class is not publicly instantiable. } @@ -80,6 +77,20 @@ public final class StringUtils { return containsInArray(text, extraValues.split(SEPARATOR_FOR_COMMA_SPLITTABLE_TEXT)); } + public static String joinCommaSplittableText(final String head, final String tail) { + if (TextUtils.isEmpty(head) && TextUtils.isEmpty(tail)) { + return EMPTY_STRING; + } + // Here either head or tail is not null. + if (TextUtils.isEmpty(head)) { + return tail; + } + if (TextUtils.isEmpty(tail)) { + return head; + } + return head + SEPARATOR_FOR_COMMA_SPLITTABLE_TEXT + tail; + } + public static String appendToCommaSplittableTextIfNotExists(final String text, final String extraValues) { if (TextUtils.isEmpty(extraValues)) { @@ -94,7 +105,7 @@ public final class StringUtils { public static String removeFromCommaSplittableTextIfExists(final String text, final String extraValues) { if (TextUtils.isEmpty(extraValues)) { - return ""; + return EMPTY_STRING; } final String[] elements = extraValues.split(SEPARATOR_FOR_COMMA_SPLITTABLE_TEXT); if (!containsInArray(text, elements)) { @@ -239,6 +250,24 @@ public final class StringUtils { return true; } + /** + * Returns true if all code points in text are whitespace, false otherwise. Empty is true. + */ + // Interestingly enough, U+00A0 NO-BREAK SPACE and U+200B ZERO-WIDTH SPACE are not considered + // whitespace, while EN SPACE, EM SPACE and IDEOGRAPHIC SPACES are. + public static boolean containsOnlyWhitespace(final String text) { + final int length = text.length(); + int i = 0; + while (i < length) { + final int codePoint = text.codePointAt(i); + if (!Character.isWhitespace(codePoint)) { + return false; + } + i += Character.charCount(codePoint); + } + return true; + } + @UsedForTesting public static boolean looksValidForDictionaryInsertion(final CharSequence text, final SettingsValues settings) { @@ -367,7 +396,7 @@ public final class StringUtils { return false; } - public static boolean isEmptyStringOrWhiteSpaces(String s) { + public static boolean isEmptyStringOrWhiteSpaces(final String s) { final int N = codePointCount(s); for (int i = 0; i < N; ++i) { if (!Character.isWhitespace(s.codePointAt(i))) { @@ -378,9 +407,9 @@ public final class StringUtils { } @UsedForTesting - public static String byteArrayToHexString(byte[] bytes) { + public static String byteArrayToHexString(final byte[] bytes) { if (bytes == null || bytes.length == 0) { - return ""; + return EMPTY_STRING; } final StringBuilder sb = new StringBuilder(); for (byte b : bytes) { @@ -393,7 +422,7 @@ public final class StringUtils { * Convert hex string to byte array. The string length must be an even number. */ @UsedForTesting - public static byte[] hexStringToByteArray(String hexString) { + public static byte[] hexStringToByteArray(final String hexString) { if (TextUtils.isEmpty(hexString)) { return null; } @@ -409,67 +438,4 @@ public final class StringUtils { } return bytes; } - - public static List<Object> jsonStrToList(String s) { - final ArrayList<Object> retval = CollectionUtils.newArrayList(); - final JsonReader reader = new JsonReader(new StringReader(s)); - try { - reader.beginArray(); - while(reader.hasNext()) { - reader.beginObject(); - while (reader.hasNext()) { - final String name = reader.nextName(); - if (name.equals(Integer.class.getSimpleName())) { - retval.add(reader.nextInt()); - } else if (name.equals(String.class.getSimpleName())) { - retval.add(reader.nextString()); - } else { - Log.w(TAG, "Invalid name: " + name); - reader.skipValue(); - } - } - reader.endObject(); - } - reader.endArray(); - return retval; - } catch (IOException e) { - } finally { - try { - reader.close(); - } catch (IOException e) { - } - } - return Collections.<Object>emptyList(); - } - - public static String listToJsonStr(List<Object> list) { - if (list == null || list.isEmpty()) { - return ""; - } - final StringWriter sw = new StringWriter(); - final JsonWriter writer = new JsonWriter(sw); - try { - writer.beginArray(); - for (final Object o : list) { - writer.beginObject(); - if (o instanceof Integer) { - writer.name(Integer.class.getSimpleName()).value((Integer)o); - } else if (o instanceof String) { - writer.name(String.class.getSimpleName()).value((String)o); - } - writer.endObject(); - } - writer.endArray(); - return sw.toString(); - } catch (IOException e) { - } finally { - try { - if (writer != null) { - writer.close(); - } - } catch (IOException e) { - } - } - return ""; - } } diff --git a/java/src/com/android/inputmethod/latin/utils/SubtypeLocaleUtils.java b/java/src/com/android/inputmethod/latin/utils/SubtypeLocaleUtils.java index 102a41b4e..fdbe81ab6 100644 --- a/java/src/com/android/inputmethod/latin/utils/SubtypeLocaleUtils.java +++ b/java/src/com/android/inputmethod/latin/utils/SubtypeLocaleUtils.java @@ -197,7 +197,9 @@ public final class SubtypeLocaleUtils { // es_US spanish F Español (EE.UU.) exception // fr azerty F Français // fr_CA qwerty F Français (Canada) + // fr_CH swiss F Français (Suisse) // de qwertz F Deutsch + // de_CH swiss T Deutsch (Schweiz) // zz qwerty F No language (QWERTY) in system locale // fr qwertz T Français (QWERTZ) // de qwerty T Deutsch (QWERTY) @@ -298,7 +300,9 @@ public final class SubtypeLocaleUtils { // es_US spanish F Es Español Español (EE.UU.) exception // fr azerty F Fr Français Français // fr_CA qwerty F Fr Français Français (Canada) + // fr_CH swiss F Fr Français Français (Suisse) // de qwertz F De Deutsch Deutsch + // de_CH swiss T De Deutsch Deutsch (Schweiz) // zz qwerty F QWERTY QWERTY // fr qwertz T Fr Français Français // de qwerty T De Deutsch Deutsch diff --git a/java/src/com/android/inputmethod/latin/utils/TypefaceUtils.java b/java/src/com/android/inputmethod/latin/utils/TypefaceUtils.java index 47ea1ea75..087a7f255 100644 --- a/java/src/com/android/inputmethod/latin/utils/TypefaceUtils.java +++ b/java/src/com/android/inputmethod/latin/utils/TypefaceUtils.java @@ -22,6 +22,9 @@ import android.graphics.Typeface; import android.util.SparseArray; public final class TypefaceUtils { + private static final char[] KEY_LABEL_REFERENCE_CHAR = { 'M' }; + private static final char[] KEY_NUMERIC_HINT_LABEL_REFERENCE_CHAR = { '8' }; + private TypefaceUtils() { // This utility class is not publicly instantiable. } @@ -31,7 +34,7 @@ public final class TypefaceUtils { // Working variable for the following method. private static final Rect sTextHeightBounds = new Rect(); - public static float getCharHeight(final char[] referenceChar, final Paint paint) { + private static float getCharHeight(final char[] referenceChar, final Paint paint) { final int key = getCharGeometryCacheKey(referenceChar[0], paint); synchronized (sTextHeightCache) { final Float cachedValue = sTextHeightCache.get(key); @@ -51,7 +54,7 @@ public final class TypefaceUtils { // Working variable for the following method. private static final Rect sTextWidthBounds = new Rect(); - public static float getCharWidth(final char[] referenceChar, final Paint paint) { + private static float getCharWidth(final char[] referenceChar, final Paint paint) { final int key = getCharGeometryCacheKey(referenceChar[0], paint); synchronized (sTextWidthCache) { final Float cachedValue = sTextWidthCache.get(key); @@ -66,11 +69,6 @@ public final class TypefaceUtils { } } - public static float getStringWidth(final String string, final Paint paint) { - paint.getTextBounds(string, 0, string.length(), sTextWidthBounds); - return sTextWidthBounds.width(); - } - private static int getCharGeometryCacheKey(final char referenceChar, final Paint paint) { final int labelSize = (int)paint.getTextSize(); final Typeface face = paint.getTypeface(); @@ -86,9 +84,25 @@ public final class TypefaceUtils { } } - public static float getLabelWidth(final String label, final Paint paint) { - final Rect textBounds = new Rect(); - paint.getTextBounds(label, 0, label.length(), textBounds); - return textBounds.width(); + public static float getReferenceCharHeight(final Paint paint) { + return getCharHeight(KEY_LABEL_REFERENCE_CHAR, paint); + } + + public static float getReferenceCharWidth(final Paint paint) { + return getCharWidth(KEY_LABEL_REFERENCE_CHAR, paint); + } + + public static float getReferenceDigitWidth(final Paint paint) { + return getCharWidth(KEY_NUMERIC_HINT_LABEL_REFERENCE_CHAR, paint); + } + + // Working variable for the following method. + private static final Rect sStringWidthBounds = new Rect(); + + public static float getStringWidth(final String string, final Paint paint) { + synchronized (sStringWidthBounds) { + paint.getTextBounds(string, 0, string.length(), sStringWidthBounds); + return sStringWidthBounds.width(); + } } } diff --git a/java/src/com/android/inputmethod/latin/utils/UnigramProperty.java b/java/src/com/android/inputmethod/latin/utils/UnigramProperty.java new file mode 100644 index 000000000..4feee4393 --- /dev/null +++ b/java/src/com/android/inputmethod/latin/utils/UnigramProperty.java @@ -0,0 +1,82 @@ +/* + * Copyright (C) 2013 The Android Open Source Project + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + + +package com.android.inputmethod.latin.utils; + +import com.android.inputmethod.annotations.UsedForTesting; +import com.android.inputmethod.latin.BinaryDictionary; +import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; + +import java.util.ArrayList; + +// This has information that belong to a unigram. This class has some detailed attributes such as +// historical information but they have to be checked only for testing purpose. +@UsedForTesting +public class UnigramProperty { + public final String mCodePoints; + public final boolean mIsNotAWord; + public final boolean mIsBlacklisted; + public final boolean mHasBigrams; + public final boolean mHasShortcuts; + public final int mProbability; + // mTimestamp, mLevel and mCount are historical info. These values are depend on the + // implementation in native code; thus, we must not use them and have any assumptions about + // them except for tests. + public final int mTimestamp; + public final int mLevel; + public final int mCount; + public final ArrayList<WeightedString> mShortcutTargets = CollectionUtils.newArrayList(); + + private static int getCodePointCount(final int[] codePoints) { + for (int i = 0; i < codePoints.length; i++) { + if (codePoints[i] == 0) { + return i; + } + } + return codePoints.length; + } + + // This represents invalid unigram when the probability is BinaryDictionary.NOT_A_PROBABILITY. + public UnigramProperty(final int[] codePoints, final boolean isNotAWord, + final boolean isBlacklisted, final boolean hasBigram, + final boolean hasShortcuts, final int probability, final int timestamp, + final int level, final int count, final ArrayList<int[]> shortcutTargets, + final ArrayList<Integer> shortcutProbabilities) { + mCodePoints = new String(codePoints, 0 /* offset */, getCodePointCount(codePoints)); + mIsNotAWord = isNotAWord; + mIsBlacklisted = isBlacklisted; + mHasBigrams = hasBigram; + mHasShortcuts = hasShortcuts; + mProbability = probability; + mTimestamp = timestamp; + mLevel = level; + mCount = count; + final int shortcutTargetCount = shortcutTargets.size(); + for (int i = 0; i < shortcutTargetCount; i++) { + final int[] shortcutTargetCodePointArray = shortcutTargets.get(i); + final String shortcutTargetString = new String(shortcutTargetCodePointArray, + 0 /* offset */, getCodePointCount(shortcutTargetCodePointArray)); + mShortcutTargets.add( + new WeightedString(shortcutTargetString, shortcutProbabilities.get(i))); + } + } + + @UsedForTesting + public boolean isValid() { + return mProbability != BinaryDictionary.NOT_A_PROBABILITY; + } +}
\ No newline at end of file diff --git a/java/src/com/android/inputmethod/latin/utils/UserHistoryDictIOUtils.java b/java/src/com/android/inputmethod/latin/utils/UserHistoryDictIOUtils.java index 635afe7cc..db628fe18 100644 --- a/java/src/com/android/inputmethod/latin/utils/UserHistoryDictIOUtils.java +++ b/java/src/com/android/inputmethod/latin/utils/UserHistoryDictIOUtils.java @@ -70,10 +70,11 @@ public final class UserHistoryDictIOUtils { /** * Writes dictionary to file. */ + @UsedForTesting public static void writeDictionary(final DictEncoder dictEncoder, final BigramDictionaryInterface dict, final UserHistoryDictionaryBigramList bigrams, - final FormatOptions formatOptions) { - final FusionDictionary fusionDict = constructFusionDictionary(dict, bigrams); + final FormatOptions formatOptions, final HashMap<String, String> options) { + final FusionDictionary fusionDict = constructFusionDictionary(dict, bigrams, options); fusionDict.addOptionAttribute(USES_FORGETTING_CURVE_KEY, USES_FORGETTING_CURVE_VALUE); fusionDict.addOptionAttribute(LAST_UPDATED_TIME_KEY, String.valueOf(TimeUnit.MILLISECONDS.toSeconds(System.currentTimeMillis()))); @@ -91,11 +92,10 @@ public final class UserHistoryDictIOUtils { * Constructs a new FusionDictionary from BigramDictionaryInterface. */ @UsedForTesting - static FusionDictionary constructFusionDictionary( - final BigramDictionaryInterface dict, final UserHistoryDictionaryBigramList bigrams) { + static FusionDictionary constructFusionDictionary(final BigramDictionaryInterface dict, + final UserHistoryDictionaryBigramList bigrams, final HashMap<String, String> options) { final FusionDictionary fusionDict = new FusionDictionary(new PtNodeArray(), - new FusionDictionary.DictionaryOptions(new HashMap<String, String>(), false, - false)); + new FusionDictionary.DictionaryOptions(options)); int profTotal = 0; for (final String word1 : bigrams.keySet()) { final HashMap<String, Byte> word1Bigrams = bigrams.getBigrams(word1); diff --git a/java/src/com/android/inputmethod/latin/utils/UserHistoryForgettingCurveUtils.java b/java/src/com/android/inputmethod/latin/utils/UserHistoryForgettingCurveUtils.java deleted file mode 100644 index 1992b2f5d..000000000 --- a/java/src/com/android/inputmethod/latin/utils/UserHistoryForgettingCurveUtils.java +++ /dev/null @@ -1,233 +0,0 @@ -/* - * Copyright (C) 2012 The Android Open Source Project - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -package com.android.inputmethod.latin.utils; - -import android.util.Log; - -import java.util.concurrent.TimeUnit; - -public final class UserHistoryForgettingCurveUtils { - private static final String TAG = UserHistoryForgettingCurveUtils.class.getSimpleName(); - private static final boolean DEBUG = false; - private static final int DEFAULT_FC_FREQ = 127; - private static final int BOOSTED_FC_FREQ = 200; - private static int FC_FREQ_MAX = DEFAULT_FC_FREQ; - /* package */ static final int COUNT_MAX = 3; - private static final int FC_LEVEL_MAX = 3; - /* package */ static final int ELAPSED_TIME_MAX = 15; - private static final int ELAPSED_TIME_INTERVAL_HOURS = 6; - private static final long ELAPSED_TIME_INTERVAL_MILLIS = - TimeUnit.HOURS.toMillis(ELAPSED_TIME_INTERVAL_HOURS); - private static final int HALF_LIFE_HOURS = 48; - private static final int MAX_PUSH_ELAPSED = (FC_LEVEL_MAX + 1) * (ELAPSED_TIME_MAX + 1); - - public static void boostMaxFreqForDebug() { - FC_FREQ_MAX = BOOSTED_FC_FREQ; - } - - public static void resetMaxFreqForDebug() { - FC_FREQ_MAX = DEFAULT_FC_FREQ; - } - - private UserHistoryForgettingCurveUtils() { - // This utility class is not publicly instantiable. - } - - public static final class ForgettingCurveParams { - private byte mFc; - long mLastTouchedTime = 0; - private final boolean mIsValid; - - private void updateLastTouchedTime() { - mLastTouchedTime = System.currentTimeMillis(); - } - - public ForgettingCurveParams(boolean isValid) { - this(System.currentTimeMillis(), isValid); - } - - private ForgettingCurveParams(long now, boolean isValid) { - this(pushCount((byte)0, isValid), now, now, isValid); - } - - /** This constructor is called when the user history bigram dictionary is being restored. */ - public ForgettingCurveParams(int fc, long now, long last) { - // All words with level >= 1 had been saved. - // Invalid words with level == 0 had been saved. - // Valid words words with level == 0 had *not* been saved. - this(fc, now, last, fcToLevel((byte)fc) > 0); - } - - private ForgettingCurveParams(int fc, long now, long last, boolean isValid) { - mIsValid = isValid; - mFc = (byte)fc; - mLastTouchedTime = last; - updateElapsedTime(now); - } - - public boolean isValid() { - return mIsValid; - } - - public byte getFc() { - updateElapsedTime(System.currentTimeMillis()); - return mFc; - } - - public int getFrequency() { - updateElapsedTime(System.currentTimeMillis()); - return UserHistoryForgettingCurveUtils.fcToFreq(mFc); - } - - public int notifyTypedAgainAndGetFrequency() { - updateLastTouchedTime(); - // TODO: Check whether this word is valid or not - mFc = pushCount(mFc, false); - return UserHistoryForgettingCurveUtils.fcToFreq(mFc); - } - - private void updateElapsedTime(long now) { - final int elapsedTimeCount = - (int)((now - mLastTouchedTime) / ELAPSED_TIME_INTERVAL_MILLIS); - if (elapsedTimeCount <= 0) { - return; - } - if (elapsedTimeCount >= MAX_PUSH_ELAPSED) { - mLastTouchedTime = now; - mFc = 0; - return; - } - for (int i = 0; i < elapsedTimeCount; ++i) { - mLastTouchedTime += ELAPSED_TIME_INTERVAL_MILLIS; - mFc = pushElapsedTime(mFc); - } - } - } - - /* package */ static int fcToElapsedTime(byte fc) { - return fc & 0x0F; - } - - /* package */ static int fcToCount(byte fc) { - return (fc >> 4) & 0x03; - } - - /* package */ static int fcToLevel(byte fc) { - return (fc >> 6) & 0x03; - } - - private static int calcFreq(int elapsedTime, int count, int level) { - if (level <= 0) { - // Reserved words, just return -1 - return -1; - } - if (count == COUNT_MAX) { - // Temporary promote because it's frequently typed recently - ++level; - } - final int et = Math.min(FC_FREQ_MAX, Math.max(0, elapsedTime)); - final int l = Math.min(FC_LEVEL_MAX, Math.max(0, level)); - return MathUtils.SCORE_TABLE[l - 1][et]; - } - - /* pakcage */ static byte calcFc(int elapsedTime, int count, int level) { - final int et = Math.min(FC_FREQ_MAX, Math.max(0, elapsedTime)); - final int c = Math.min(COUNT_MAX, Math.max(0, count)); - final int l = Math.min(FC_LEVEL_MAX, Math.max(0, level)); - return (byte)(et | (c << 4) | (l << 6)); - } - - public static int fcToFreq(byte fc) { - final int elapsedTime = fcToElapsedTime(fc); - final int count = fcToCount(fc); - final int level = fcToLevel(fc); - return calcFreq(elapsedTime, count, level); - } - - public static byte pushElapsedTime(byte fc) { - int elapsedTime = fcToElapsedTime(fc); - int count = fcToCount(fc); - int level = fcToLevel(fc); - if (elapsedTime >= ELAPSED_TIME_MAX) { - // Downgrade level - elapsedTime = 0; - count = COUNT_MAX; - --level; - } else { - ++elapsedTime; - } - return calcFc(elapsedTime, count, level); - } - - public static byte pushCount(byte fc, boolean isValid) { - final int elapsedTime = fcToElapsedTime(fc); - int count = fcToCount(fc); - int level = fcToLevel(fc); - if ((elapsedTime == 0 && count >= COUNT_MAX) || (isValid && level == 0)) { - // Upgrade level - ++level; - count = 0; - if (DEBUG) { - Log.d(TAG, "Upgrade level."); - } - } else { - ++count; - } - return calcFc(0, count, level); - } - - // TODO: isValid should be false for a word whose frequency is 0, - // or that is not in the dictionary. - /** - * Check wheather we should save the bigram to the SQL DB or not - */ - public static boolean needsToSave(byte fc, boolean isValid, boolean addLevel0Bigram) { - int level = fcToLevel(fc); - if (level == 0) { - if (isValid || !addLevel0Bigram) { - return false; - } - } - final int elapsedTime = fcToElapsedTime(fc); - return (elapsedTime < ELAPSED_TIME_MAX - 1 || level > 0); - } - - private static final class MathUtils { - public static final int[][] SCORE_TABLE = new int[FC_LEVEL_MAX][ELAPSED_TIME_MAX + 1]; - static { - for (int i = 0; i < FC_LEVEL_MAX; ++i) { - final float initialFreq; - if (i >= 2) { - initialFreq = FC_FREQ_MAX; - } else if (i == 1) { - initialFreq = FC_FREQ_MAX / 2; - } else if (i == 0) { - initialFreq = FC_FREQ_MAX / 4; - } else { - continue; - } - for (int j = 0; j < ELAPSED_TIME_MAX; ++j) { - final float elapsedHours = j * ELAPSED_TIME_INTERVAL_HOURS; - final float freq = initialFreq - * (float)Math.pow(initialFreq, elapsedHours / HALF_LIFE_HOURS); - final int intFreq = Math.min(FC_FREQ_MAX, Math.max(0, (int)freq)); - SCORE_TABLE[i][j] = intFreq; - } - } - } - } -} |