diff options
Diffstat (limited to 'java/src')
16 files changed, 165 insertions, 141 deletions
diff --git a/java/src/com/android/inputmethod/latin/AbstractDictionaryWriter.java b/java/src/com/android/inputmethod/latin/AbstractDictionaryWriter.java index 1aee22baf..1c6a14efe 100644 --- a/java/src/com/android/inputmethod/latin/AbstractDictionaryWriter.java +++ b/java/src/com/android/inputmethod/latin/AbstractDictionaryWriter.java @@ -16,12 +16,12 @@ package com.android.inputmethod.latin; -import android.content.Context; import android.util.Log; import com.android.inputmethod.latin.makedict.DictEncoder; import com.android.inputmethod.latin.makedict.UnsupportedFormatException; -import com.android.inputmethod.latin.makedict.Ver2DictEncoder; +import com.android.inputmethod.latin.makedict.Ver4DictEncoder; +import com.android.inputmethod.latin.utils.FileUtils; import java.io.File; import java.io.IOException; @@ -31,10 +31,7 @@ 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) { - mContext = context; + public AbstractDictionaryWriter() { } abstract public void clear(); @@ -61,12 +58,11 @@ abstract public class AbstractDictionaryWriter { final Map<String, String> attributeMap) throws IOException, UnsupportedFormatException; 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 Ver2DictEncoder(tempFile); + FileUtils.deleteRecursively(file); + file.mkdir(); + final DictEncoder dictEncoder = new Ver4DictEncoder(file); writeDictionary(dictEncoder, attributeMap); - tempFile.renameTo(file); } catch (IOException e) { Log.e(TAG, "IO exception while writing file", e); } catch (UnsupportedFormatException e) { diff --git a/java/src/com/android/inputmethod/latin/BinaryDictionary.java b/java/src/com/android/inputmethod/latin/BinaryDictionary.java index 2c7998688..e66cfca49 100644 --- a/java/src/com/android/inputmethod/latin/BinaryDictionary.java +++ b/java/src/com/android/inputmethod/latin/BinaryDictionary.java @@ -165,6 +165,7 @@ public final class BinaryDictionary extends Dictionary { LanguageModelParam[] languageModelParams, int startIndex); private static native int calculateProbabilityNative(long dict, int unigramProbability, int bigramProbability); + private static native int setCurrentTimeForTestNative(int currentTime); private static native String getPropertyNative(long dict, String query); @UsedForTesting @@ -420,8 +421,22 @@ public final class BinaryDictionary extends Dictionary { return calculateProbabilityNative(mNativeDict, unigramProbability, bigramProbability); } + /** + * Control the current time to be used in the native code. If currentTime >= 0, this method sets + * the current time and gets into test mode. + * In test mode, set timestamp is used as the current time in the native code. + * If currentTime < 0, quit the test mode and returns to using time() to get the current time. + * + * @param currentTime seconds since the unix epoch + * @return current time got in the native code. + */ + @UsedForTesting + public static int setCurrentTimeForTest(final int currentTime) { + return setCurrentTimeForTestNative(currentTime); + } + @UsedForTesting - public String getPropertyForTests(String query) { + public String getPropertyForTest(final String query) { if (!isValidDictionary()) return ""; return getPropertyNative(mNativeDict, query); } diff --git a/java/src/com/android/inputmethod/latin/Constants.java b/java/src/com/android/inputmethod/latin/Constants.java index 00b54f593..77e99bfba 100644 --- a/java/src/com/android/inputmethod/latin/Constants.java +++ b/java/src/com/android/inputmethod/latin/Constants.java @@ -183,6 +183,7 @@ public final class Constants { public static final int CODE_TAB = '\t'; public static final int CODE_SPACE = ' '; public static final int CODE_PERIOD = '.'; + public static final int CODE_COMMA = ','; public static final int CODE_ARMENIAN_PERIOD = 0x0589; public static final int CODE_DASH = '-'; public static final int CODE_SINGLE_QUOTE = '\''; diff --git a/java/src/com/android/inputmethod/latin/DictionaryWriter.java b/java/src/com/android/inputmethod/latin/DictionaryWriter.java index 89ef96d7f..84898680e 100644 --- a/java/src/com/android/inputmethod/latin/DictionaryWriter.java +++ b/java/src/com/android/inputmethod/latin/DictionaryWriter.java @@ -16,8 +16,6 @@ package com.android.inputmethod.latin; -import android.content.Context; - import com.android.inputmethod.latin.makedict.DictEncoder; import com.android.inputmethod.latin.makedict.FormatSpec; import com.android.inputmethod.latin.makedict.FusionDictionary; @@ -35,14 +33,13 @@ 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 = 2; + private static final int BINARY_DICT_VERSION = FormatSpec.VERSION4; private static final FormatSpec.FormatOptions FORMAT_OPTIONS = - new FormatSpec.FormatOptions(BINARY_DICT_VERSION, false /* supportsDynamicUpdate */); + new FormatSpec.FormatOptions(BINARY_DICT_VERSION, false /* hasTimestamp */); private FusionDictionary mFusionDictionary; - public DictionaryWriter(final Context context) { - super(context); + public DictionaryWriter() { clear(); } diff --git a/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java b/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java index ecef20efc..8e1675b2a 100644 --- a/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java +++ b/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java @@ -136,12 +136,8 @@ 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; + private boolean matchesExpectedBinaryDictFormatVersionForThisType(final int formatVersion) { + return formatVersion == FormatSpec.VERSION4; } public boolean isValidDictionary() { @@ -194,12 +190,12 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { } } - private static AbstractDictionaryWriter getDictionaryWriter(final Context context, + private static AbstractDictionaryWriter getDictionaryWriter( final boolean isDynamicPersonalizationDictionary) { if (isDynamicPersonalizationDictionary) { return null; } else { - return new DictionaryWriter(context); + return new DictionaryWriter(); } } @@ -233,7 +229,7 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { mBinaryDictionary = null; mDictNameDictionaryUpdateController = getDictionaryUpdateController(dictName); // Currently, only dynamic personalization dictionary is updatable. - mDictionaryWriter = getDictionaryWriter(context, isUpdatable); + mDictionaryWriter = getDictionaryWriter(isUpdatable); } protected static String getDictNameWithLocale(final String name, final Locale locale) { diff --git a/java/src/com/android/inputmethod/latin/LatinIME.java b/java/src/com/android/inputmethod/latin/LatinIME.java index 3fca4fd19..d3e6a1bc2 100644 --- a/java/src/com/android/inputmethod/latin/LatinIME.java +++ b/java/src/com/android/inputmethod/latin/LatinIME.java @@ -1407,7 +1407,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // TODO[IL]: Move this to InputLogic public SuggestedWords maybeRetrieveOlderSuggestions(final String typedWord, - final SuggestedWords suggestedWords) { + final SuggestedWords suggestedWords, final SuggestedWords previousSuggestedWords) { // TODO: consolidate this into getSuggestedWords // We update the suggestion strip only when we have some suggestions to show, i.e. when // the suggestion count is > 1; else, we leave the old suggestions, with the typed word @@ -1420,28 +1420,22 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen || mSuggestionStripView.isShowingAddToDictionaryHint()) { return suggestedWords; } else { - return getOlderSuggestions(typedWord); - } - } - - private SuggestedWords getOlderSuggestions(final String typedWord) { - SuggestedWords previousSuggestedWords = mInputLogic.mSuggestedWords; - if (previousSuggestedWords - == mSettings.getCurrent().mSpacingAndPunctuations.mSuggestPuncList) { - previousSuggestedWords = SuggestedWords.EMPTY; - } - if (typedWord == null) { - return previousSuggestedWords; + final SuggestedWords punctuationList = + mSettings.getCurrent().mSpacingAndPunctuations.mSuggestPuncList; + final SuggestedWords oldSuggestedWords = previousSuggestedWords == punctuationList + ? SuggestedWords.EMPTY : previousSuggestedWords; + if (TextUtils.isEmpty(typedWord)) { + return oldSuggestedWords; + } + final ArrayList<SuggestedWords.SuggestedWordInfo> typedWordAndPreviousSuggestions = + SuggestedWords.getTypedWordAndPreviousSuggestions(typedWord, oldSuggestedWords); + return new SuggestedWords(typedWordAndPreviousSuggestions, + false /* typedWordValid */, + false /* hasAutoCorrectionCandidate */, + false /* isPunctuationSuggestions */, + true /* isObsoleteSuggestions */, + false /* isPrediction */); } - final ArrayList<SuggestedWords.SuggestedWordInfo> typedWordAndPreviousSuggestions = - SuggestedWords.getTypedWordAndPreviousSuggestions(typedWord, - previousSuggestedWords); - return new SuggestedWords(typedWordAndPreviousSuggestions, - false /* typedWordValid */, - false /* hasAutoCorrectionCandidate */, - false /* isPunctuationSuggestions */, - true /* isObsoleteSuggestions */, - false /* isPrediction */); } private void showSuggestionStripWithTypedWord(final SuggestedWords suggestedWords, diff --git a/java/src/com/android/inputmethod/latin/RichInputConnection.java b/java/src/com/android/inputmethod/latin/RichInputConnection.java index 79d66744b..325a0d981 100644 --- a/java/src/com/android/inputmethod/latin/RichInputConnection.java +++ b/java/src/com/android/inputmethod/latin/RichInputConnection.java @@ -27,7 +27,6 @@ import android.view.inputmethod.ExtractedTextRequest; import android.view.inputmethod.InputConnection; import com.android.inputmethod.latin.define.ProductionFlag; -import com.android.inputmethod.latin.settings.SettingsValues; import com.android.inputmethod.latin.settings.SpacingAndPunctuations; import com.android.inputmethod.latin.utils.CapsModeUtils; import com.android.inputmethod.latin.utils.DebugLogUtils; @@ -36,6 +35,7 @@ import com.android.inputmethod.latin.utils.StringUtils; import com.android.inputmethod.latin.utils.TextRange; import com.android.inputmethod.research.ResearchLogger; +import java.util.Arrays; import java.util.regex.Pattern; /** @@ -98,7 +98,7 @@ public final class RichInputConnection { final ExtractedText et = mIC.getExtractedText(r, 0); final CharSequence beforeCursor = getTextBeforeCursor(Constants.EDITOR_CONTENTS_CACHE_SIZE, 0); - final StringBuilder internal = new StringBuilder().append(mCommittedTextBeforeComposingText) + final StringBuilder internal = new StringBuilder(mCommittedTextBeforeComposingText) .append(mComposingText); if (null == et || null == beforeCursor) return; final int actualLength = Math.min(beforeCursor.length(), internal.length()); @@ -253,8 +253,7 @@ public final class RichInputConnection { } public CharSequence getSelectedText(final int flags) { - if (null == mIC) return null; - return mIC.getSelectedText(flags); + return (null == mIC) ? null : mIC.getSelectedText(flags); } public boolean canDeleteCharacters() { @@ -272,12 +271,12 @@ public final class RichInputConnection { * American English, it's just the most common set of rules for English). * * @param inputType a mask of the caps modes to test for. - * @param settingsValues the values of the settings to use for locale and separators. + * @param spacingAndPunctuations the values of the settings to use for locale and separators. * @param hasSpaceBefore if we should consider there should be a space after the string. * @return the caps modes that should be on as a set of bits */ - public int getCursorCapsMode(final int inputType, final SettingsValues settingsValues, - final boolean hasSpaceBefore) { + public int getCursorCapsMode(final int inputType, + final SpacingAndPunctuations spacingAndPunctuations, final boolean hasSpaceBefore) { mIC = mParent.getCurrentInputConnection(); if (null == mIC) return Constants.TextUtils.CAP_MODE_OFF; if (!TextUtils.isEmpty(mComposingText)) { @@ -304,13 +303,13 @@ public final class RichInputConnection { // This never calls InputConnection#getCapsMode - in fact, it's a static method that // never blocks or initiates IPC. return CapsModeUtils.getCapsMode(mCommittedTextBeforeComposingText, inputType, - settingsValues.mSpacingAndPunctuations, hasSpaceBefore); + spacingAndPunctuations, hasSpaceBefore); } public int getCodePointBeforeCursor() { - if (mCommittedTextBeforeComposingText.length() < 1) return Constants.NOT_A_CODE; - return Character.codePointBefore(mCommittedTextBeforeComposingText, - mCommittedTextBeforeComposingText.length()); + final int length = mCommittedTextBeforeComposingText.length(); + if (length < 1) return Constants.NOT_A_CODE; + return Character.codePointBefore(mCommittedTextBeforeComposingText, length); } public CharSequence getTextBeforeCursor(final int n, final int flags) { @@ -338,16 +337,12 @@ public final class RichInputConnection { return s; } mIC = mParent.getCurrentInputConnection(); - if (null != mIC) { - return mIC.getTextBeforeCursor(n, flags); - } - return null; + return (null == mIC) ? null : mIC.getTextBeforeCursor(n, flags); } public CharSequence getTextAfterCursor(final int n, final int flags) { mIC = mParent.getCurrentInputConnection(); - if (null != mIC) return mIC.getTextAfterCursor(n, flags); - return null; + return (null == mIC) ? null : mIC.getTextAfterCursor(n, flags); } public void deleteSurroundingText(final int beforeLength, final int afterLength) { @@ -563,8 +558,8 @@ public final class RichInputConnection { return getNthPreviousWord(prev, spacingAndPunctuations, n); } - private static boolean isSeparator(int code, String sep) { - return sep.indexOf(code) != -1; + private static boolean isSeparator(final int code, final int[] sortedSeparators) { + return Arrays.binarySearch(sortedSeparators, code) >= 0; } // Get the nth word before cursor. n = 1 retrieves the word immediately before the cursor, @@ -603,29 +598,29 @@ public final class RichInputConnection { } /** - * @param separators characters which may separate words + * @param sortedSeparators a sorted array of code points which may separate words * @return the word that surrounds the cursor, including up to one trailing * separator. For example, if the field contains "he|llo world", where | * represents the cursor, then "hello " will be returned. */ - public CharSequence getWordAtCursor(String separators) { + public CharSequence getWordAtCursor(final int[] sortedSeparators) { // getWordRangeAtCursor returns null if the connection is null - TextRange r = getWordRangeAtCursor(separators, 0); + final TextRange r = getWordRangeAtCursor(sortedSeparators, 0); return (r == null) ? null : r.mWord; } /** * Returns the text surrounding the cursor. * - * @param sep a string of characters that split words. + * @param sortedSeparators a sorted array of code points that split words. * @param additionalPrecedingWordsCount the number of words before the current word that should * be included in the returned range * @return a range containing the text surrounding the cursor */ - public TextRange getWordRangeAtCursor(final String sep, + public TextRange getWordRangeAtCursor(final int[] sortedSeparators, final int additionalPrecedingWordsCount) { mIC = mParent.getCurrentInputConnection(); - if (mIC == null || sep == null) { + if (mIC == null) { return null; } final CharSequence before = mIC.getTextBeforeCursor(Constants.EDITOR_CONTENTS_CACHE_SIZE, @@ -644,7 +639,7 @@ public final class RichInputConnection { while (true) { // see comments below for why this is guaranteed to halt while (startIndexInBefore > 0) { final int codePoint = Character.codePointBefore(before, startIndexInBefore); - if (isStoppingAtWhitespace == isSeparator(codePoint, sep)) { + if (isStoppingAtWhitespace == isSeparator(codePoint, sortedSeparators)) { break; // inner loop } --startIndexInBefore; @@ -665,7 +660,7 @@ public final class RichInputConnection { int endIndexInAfter = -1; while (++endIndexInAfter < after.length()) { final int codePoint = Character.codePointAt(after, endIndexInAfter); - if (isSeparator(codePoint, sep)) { + if (isSeparator(codePoint, sortedSeparators)) { break; } if (Character.isSupplementaryCodePoint(codePoint)) { @@ -681,23 +676,28 @@ public final class RichInputConnection { startIndexInBefore, before.length() + endIndexInAfter, before.length()); } - public boolean isCursorTouchingWord(final SettingsValues settingsValues) { + public boolean isCursorTouchingWord(final SpacingAndPunctuations spacingAndPunctuations) { final int codePointBeforeCursor = getCodePointBeforeCursor(); - if (Constants.NOT_A_CODE != codePointBeforeCursor - && !settingsValues.isWordSeparator(codePointBeforeCursor) - && !settingsValues.isWordConnector(codePointBeforeCursor)) { - return true; + if (Constants.NOT_A_CODE == codePointBeforeCursor + || spacingAndPunctuations.isWordSeparator(codePointBeforeCursor) + || spacingAndPunctuations.isWordConnector(codePointBeforeCursor)) { + return isCursorFollowedByWordCharacter(spacingAndPunctuations); } - return isCursorFollowedByWordCharacter(settingsValues); + return true; } - public boolean isCursorFollowedByWordCharacter(final SettingsValues settingsValues) { + public boolean isCursorFollowedByWordCharacter( + final SpacingAndPunctuations spacingAndPunctuations) { final CharSequence after = getTextAfterCursor(1, 0); - if (!TextUtils.isEmpty(after) && !settingsValues.isWordSeparator(after.charAt(0)) - && !settingsValues.isWordConnector(after.charAt(0))) { - return true; + if (TextUtils.isEmpty(after)) { + return false; + } + final int codePointAfterCursor = Character.codePointAt(after, 0); + if (spacingAndPunctuations.isWordSeparator(codePointAfterCursor) + || spacingAndPunctuations.isWordConnector(codePointAfterCursor)) { + return false; } - return false; + return true; } public void removeTrailingSpace() { diff --git a/java/src/com/android/inputmethod/latin/inputlogic/InputLogic.java b/java/src/com/android/inputmethod/latin/inputlogic/InputLogic.java index 0c33e6b72..c040ee87d 100644 --- a/java/src/com/android/inputmethod/latin/inputlogic/InputLogic.java +++ b/java/src/com/android/inputmethod/latin/inputlogic/InputLogic.java @@ -534,7 +534,7 @@ public final class InputLogic { && settingsValues.isSuggestionsRequested() && // In languages with spaces, we only start composing a word when we are not already // touching a word. In languages without spaces, the above conditions are sufficient. - (!mConnection.isCursorTouchingWord(settingsValues) + (!mConnection.isCursorTouchingWord(settingsValues.mSpacingAndPunctuations) || !settingsValues.mSpacingAndPunctuations.mCurrentLanguageHasSpaces)) { // Reset entirely the composing state anyway, then start composing a new word unless // the character is a single quote or a dash. The idea here is, single quote and dash @@ -816,7 +816,8 @@ public final class InputLogic { } if (settingsValues.isSuggestionStripVisible() && settingsValues.mSpacingAndPunctuations.mCurrentLanguageHasSpaces - && !mConnection.isCursorFollowedByWordCharacter(settingsValues)) { + && !mConnection.isCursorFollowedByWordCharacter( + settingsValues.mSpacingAndPunctuations)) { restartSuggestionsOnWordTouchedByCursor(settingsValues, true /* includeResumedWordInSuggestions */, keyboardSwitcher); } @@ -969,7 +970,8 @@ public final class InputLogic { if (TextUtils.isEmpty(selectedText)) return; // Race condition with the input connection mRecapitalizeStatus.initialize(mConnection.getExpectedSelectionStart(), mConnection.getExpectedSelectionEnd(), selectedText.toString(), - settingsValues.mLocale, settingsValues.mSpacingAndPunctuations.mWordSeparators); + settingsValues.mLocale, + settingsValues.mSpacingAndPunctuations.mSortedWordSeparators); // We trim leading and trailing whitespace. mRecapitalizeStatus.trim(); } @@ -1030,7 +1032,8 @@ public final class InputLogic { public void onGetSuggestedWords(final SuggestedWords suggestedWords) { final SuggestedWords suggestedWordsWithMaybeOlderSuggestions = mLatinIME.maybeRetrieveOlderSuggestions( - mWordComposer.getTypedWord(), suggestedWords); + mWordComposer.getTypedWord(), suggestedWords, + mSuggestedWords); holder.set(suggestedWordsWithMaybeOlderSuggestions); } } @@ -1069,9 +1072,9 @@ public final class InputLogic { // If we don't know the cursor location, return. if (mConnection.getExpectedSelectionStart() < 0) return; final int expectedCursorPosition = mConnection.getExpectedSelectionStart(); - if (!mConnection.isCursorTouchingWord(settingsValues)) return; + if (!mConnection.isCursorTouchingWord(settingsValues.mSpacingAndPunctuations)) return; final TextRange range = mConnection.getWordRangeAtCursor( - settingsValues.mSpacingAndPunctuations.mWordSeparators, + settingsValues.mSpacingAndPunctuations.mSortedWordSeparators, 0 /* additionalPrecedingWordsCount */); if (null == range) return; // Happens if we don't have an input connection at all if (range.length() <= 0) return; // Race condition. No text to resume on, so bail out. @@ -1290,7 +1293,7 @@ public final class InputLogic { final int inputType = ei.inputType; // Warning: this depends on mSpaceState, which may not be the most current value. If // mSpaceState gets updated later, whoever called this may need to be told about it. - return mConnection.getCursorCapsMode(inputType, settingsValues, + return mConnection.getCursorCapsMode(inputType, settingsValues.mSpacingAndPunctuations, SpaceState.PHANTOM == mSpaceState); } diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver2DictEncoder.java b/java/src/com/android/inputmethod/latin/makedict/Ver2DictEncoder.java index e5430423d..a3a6c2c34 100644 --- a/java/src/com/android/inputmethod/latin/makedict/Ver2DictEncoder.java +++ b/java/src/com/android/inputmethod/latin/makedict/Ver2DictEncoder.java @@ -16,6 +16,7 @@ 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.FormatOptions; import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode; @@ -33,6 +34,7 @@ import java.util.Iterator; /** * An implementation of DictEncoder for version 2 binary dictionary. */ +@UsedForTesting public class Ver2DictEncoder implements DictEncoder { private final File mDictFile; @@ -40,6 +42,7 @@ public class Ver2DictEncoder implements DictEncoder { private byte[] mBuffer; private int mPosition; + @UsedForTesting public Ver2DictEncoder(final File dictFile) { mDictFile = dictFile; mOutStream = null; @@ -49,6 +52,7 @@ public class Ver2DictEncoder 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. + @UsedForTesting public Ver2DictEncoder(final OutputStream outStream) { mDictFile = null; mOutStream = outStream; diff --git a/java/src/com/android/inputmethod/latin/personalization/DecayingExpandableBinaryDictionaryBase.java b/java/src/com/android/inputmethod/latin/personalization/DecayingExpandableBinaryDictionaryBase.java index cc57d13dc..4a610e6f9 100644 --- a/java/src/com/android/inputmethod/latin/personalization/DecayingExpandableBinaryDictionaryBase.java +++ b/java/src/com/android/inputmethod/latin/personalization/DecayingExpandableBinaryDictionaryBase.java @@ -116,13 +116,6 @@ 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) { diff --git a/java/src/com/android/inputmethod/latin/settings/Settings.java b/java/src/com/android/inputmethod/latin/settings/Settings.java index 7db10714a..9bf269b6e 100644 --- a/java/src/com/android/inputmethod/latin/settings/Settings.java +++ b/java/src/com/android/inputmethod/latin/settings/Settings.java @@ -181,10 +181,6 @@ public final class Settings implements SharedPreferences.OnSharedPreferenceChang return mSettingsValues.mIsInternal; } - public String getWordSeparators() { - return mSettingsValues.mSpacingAndPunctuations.mWordSeparators; - } - public boolean isWordSeparator(final int code) { return mSettingsValues.isWordSeparator(code); } diff --git a/java/src/com/android/inputmethod/latin/settings/SpacingAndPunctuations.java b/java/src/com/android/inputmethod/latin/settings/SpacingAndPunctuations.java index dbe30e260..8ba32ff76 100644 --- a/java/src/com/android/inputmethod/latin/settings/SpacingAndPunctuations.java +++ b/java/src/com/android/inputmethod/latin/settings/SpacingAndPunctuations.java @@ -32,38 +32,41 @@ import java.util.Arrays; import java.util.Locale; public final class SpacingAndPunctuations { - private final int[] mSymbolsPrecededBySpace; - private final int[] mSymbolsFollowedBySpace; - private final int[] mWordConnectors; + private final int[] mSortedSymbolsPrecededBySpace; + private final int[] mSortedSymbolsFollowedBySpace; + private final int[] mSortedWordConnectors; + public final int[] mSortedWordSeparators; public final SuggestedWords mSuggestPuncList; - public final String mWordSeparators; private final int mSentenceSeparator; public final String mSentenceSeparatorAndSpace; public final boolean mCurrentLanguageHasSpaces; public final boolean mUsesAmericanTypography; + public final boolean mUsesGermanRules; public SpacingAndPunctuations(final Resources res) { - mSymbolsPrecededBySpace = - StringUtils.toCodePointArray(res.getString(R.string.symbols_preceded_by_space)); - Arrays.sort(mSymbolsPrecededBySpace); - mSymbolsFollowedBySpace = - StringUtils.toCodePointArray(res.getString(R.string.symbols_followed_by_space)); - Arrays.sort(mSymbolsFollowedBySpace); - mWordConnectors = - StringUtils.toCodePointArray(res.getString(R.string.symbols_word_connectors)); - Arrays.sort(mWordConnectors); + // To be able to binary search the code point. See {@link #isUsuallyPrecededBySpace(int)}. + mSortedSymbolsPrecededBySpace = StringUtils.toSortedCodePointArray( + res.getString(R.string.symbols_preceded_by_space)); + // To be able to binary search the code point. See {@link #isUsuallyFollowedBySpace(int)}. + mSortedSymbolsFollowedBySpace = StringUtils.toSortedCodePointArray( + res.getString(R.string.symbols_followed_by_space)); + // To be able to binary search the code point. See {@link #isWordConnector(int)}. + mSortedWordConnectors = StringUtils.toSortedCodePointArray( + res.getString(R.string.symbols_word_connectors)); + mSortedWordSeparators = StringUtils.toSortedCodePointArray( + res.getString(R.string.symbols_word_separators)); final String[] suggestPuncsSpec = KeySpecParser.splitKeySpecs(res.getString( R.string.suggested_punctuations)); mSuggestPuncList = createSuggestPuncList(suggestPuncsSpec); - mWordSeparators = res.getString(R.string.symbols_word_separators); mSentenceSeparator = res.getInteger(R.integer.sentence_separator); mSentenceSeparatorAndSpace = new String(new int[] { mSentenceSeparator, Constants.CODE_SPACE }, 0, 2); mCurrentLanguageHasSpaces = res.getBoolean(R.bool.current_language_has_spaces); final Locale locale = res.getConfiguration().locale; // Heuristic: we use American Typography rules because it's the most common rules for all - // English variants. + // English variants. German rules (not "German typography") also have small gotchas. mUsesAmericanTypography = Locale.ENGLISH.getLanguage().equals(locale.getLanguage()); + mUsesGermanRules = Locale.GERMAN.getLanguage().equals(locale.getLanguage()); } // Helper functions to create member values. @@ -72,6 +75,7 @@ public final class SpacingAndPunctuations { if (puncs != null) { for (final String puncSpec : puncs) { // TODO: Stop using KeySpceParser.getLabel(). + // TODO: Punctuation suggestions should honor RTL languages. puncList.add(new SuggestedWordInfo(KeySpecParser.getLabel(puncSpec), SuggestedWordInfo.MAX_SCORE, SuggestedWordInfo.KIND_HARDCODED, Dictionary.DICTIONARY_HARDCODED, @@ -88,11 +92,11 @@ public final class SpacingAndPunctuations { } public boolean isWordSeparator(final int code) { - return mWordSeparators.contains(String.valueOf((char)code)); + return Arrays.binarySearch(mSortedWordSeparators, code) >= 0; } public boolean isWordConnector(final int code) { - return Arrays.binarySearch(mWordConnectors, code) >= 0; + return Arrays.binarySearch(mSortedWordConnectors, code) >= 0; } public boolean isWordCodePoint(final int code) { @@ -100,11 +104,11 @@ public final class SpacingAndPunctuations { } public boolean isUsuallyPrecededBySpace(final int code) { - return Arrays.binarySearch(mSymbolsPrecededBySpace, code) >= 0; + return Arrays.binarySearch(mSortedSymbolsPrecededBySpace, code) >= 0; } public boolean isUsuallyFollowedBySpace(final int code) { - return Arrays.binarySearch(mSymbolsFollowedBySpace, code) >= 0; + return Arrays.binarySearch(mSortedSymbolsFollowedBySpace, code) >= 0; } public boolean isSentenceSeparator(final int code) { diff --git a/java/src/com/android/inputmethod/latin/utils/CapsModeUtils.java b/java/src/com/android/inputmethod/latin/utils/CapsModeUtils.java index 057e332e9..702688f93 100644 --- a/java/src/com/android/inputmethod/latin/utils/CapsModeUtils.java +++ b/java/src/com/android/inputmethod/latin/utils/CapsModeUtils.java @@ -139,6 +139,20 @@ public final class CapsModeUtils { j--; } if (j <= 0 || Character.isWhitespace(prevChar)) { + if (spacingAndPunctuations.mUsesGermanRules) { + // In German typography rules, there is a specific case that the first character + // of a new line should not be capitalized if the previous line ends in a comma. + boolean hasNewLine = false; + while (--j >= 0 && Character.isWhitespace(prevChar)) { + if (Constants.CODE_ENTER == prevChar) { + hasNewLine = true; + } + prevChar = cs.charAt(j); + } + if (Constants.CODE_COMMA == prevChar && hasNewLine) { + return (TextUtils.CAP_MODE_CHARACTERS | TextUtils.CAP_MODE_WORDS) & reqModes; + } + } // There are only spacing chars between the start of the paragraph and the cursor, // defined as a isWhitespace() char that is neither a isSpaceChar() nor a tab. Both // MODE_WORDS and MODE_SENTENCES should be active. diff --git a/java/src/com/android/inputmethod/latin/utils/RecapitalizeStatus.java b/java/src/com/android/inputmethod/latin/utils/RecapitalizeStatus.java index 0f5cd80db..4521ec531 100644 --- a/java/src/com/android/inputmethod/latin/utils/RecapitalizeStatus.java +++ b/java/src/com/android/inputmethod/latin/utils/RecapitalizeStatus.java @@ -37,12 +37,12 @@ public class RecapitalizeStatus { CAPS_MODE_ALL_UPPER }; - private static final int getStringMode(final String string, final String separators) { + private static final int getStringMode(final String string, final int[] sortedSeparators) { if (StringUtils.isIdenticalAfterUpcase(string)) { return CAPS_MODE_ALL_UPPER; } else if (StringUtils.isIdenticalAfterDowncase(string)) { return CAPS_MODE_ALL_LOWER; - } else if (StringUtils.isIdenticalAfterCapitalizeEachWord(string, separators)) { + } else if (StringUtils.isIdenticalAfterCapitalizeEachWord(string, sortedSeparators)) { return CAPS_MODE_FIRST_WORD_UPPER; } else { return CAPS_MODE_ORIGINAL_MIXED_CASE; @@ -60,26 +60,28 @@ public class RecapitalizeStatus { private int mRotationStyleCurrentIndex; private boolean mSkipOriginalMixedCaseMode; private Locale mLocale; - private String mSeparators; + private int[] mSortedSeparators; private String mStringAfter; private boolean mIsActive; + private static final int[] EMPTY_STORTED_SEPARATORS = {}; + public RecapitalizeStatus() { // By default, initialize with dummy values that won't match any real recapitalize. - initialize(-1, -1, "", Locale.getDefault(), ""); + initialize(-1, -1, "", Locale.getDefault(), EMPTY_STORTED_SEPARATORS); deactivate(); } public void initialize(final int cursorStart, final int cursorEnd, final String string, - final Locale locale, final String separators) { + final Locale locale, final int[] sortedSeparators) { mCursorStartBefore = cursorStart; mStringBefore = string; mCursorStartAfter = cursorStart; mCursorEndAfter = cursorEnd; mStringAfter = string; - final int initialMode = getStringMode(mStringBefore, separators); + final int initialMode = getStringMode(mStringBefore, sortedSeparators); mLocale = locale; - mSeparators = separators; + mSortedSeparators = sortedSeparators; if (CAPS_MODE_ORIGINAL_MIXED_CASE == initialMode) { mRotationStyleCurrentIndex = 0; mSkipOriginalMixedCaseMode = false; @@ -131,7 +133,7 @@ public class RecapitalizeStatus { mStringAfter = mStringBefore.toLowerCase(mLocale); break; case CAPS_MODE_FIRST_WORD_UPPER: - mStringAfter = StringUtils.capitalizeEachWord(mStringBefore, mSeparators, + mStringAfter = StringUtils.capitalizeEachWord(mStringBefore, mSortedSeparators, mLocale); break; case CAPS_MODE_ALL_UPPER: diff --git a/java/src/com/android/inputmethod/latin/utils/StringUtils.java b/java/src/com/android/inputmethod/latin/utils/StringUtils.java index c5ed39310..5920c68f1 100644 --- a/java/src/com/android/inputmethod/latin/utils/StringUtils.java +++ b/java/src/com/android/inputmethod/latin/utils/StringUtils.java @@ -22,6 +22,7 @@ import com.android.inputmethod.annotations.UsedForTesting; import com.android.inputmethod.latin.Constants; import java.util.ArrayList; +import java.util.Arrays; import java.util.Locale; public final class StringUtils { @@ -183,6 +184,12 @@ public final class StringUtils { return codePoints; } + public static int[] toSortedCodePointArray(final String string) { + final int[] codePoints = toCodePointArray(string); + Arrays.sort(codePoints); + return codePoints; + } + // This method assumes the text is not null. For the empty string, it returns CAPITALIZE_NONE. public static int getCapitalizationType(final String text) { // If the first char is not uppercase, then the word is either all lower case or @@ -265,39 +272,39 @@ public final class StringUtils { } public static boolean isIdenticalAfterCapitalizeEachWord(final String text, - final String separators) { - boolean needCapsNext = true; + final int[] sortedSeparators) { + boolean needsCapsNext = true; final int len = text.length(); for (int i = 0; i < len; i = text.offsetByCodePoints(i, 1)) { final int codePoint = text.codePointAt(i); if (Character.isLetter(codePoint)) { - if ((needCapsNext && !Character.isUpperCase(codePoint)) - || (!needCapsNext && !Character.isLowerCase(codePoint))) { + if ((needsCapsNext && !Character.isUpperCase(codePoint)) + || (!needsCapsNext && !Character.isLowerCase(codePoint))) { return false; } } // We need a capital letter next if this is a separator. - needCapsNext = (-1 != separators.indexOf(codePoint)); + needsCapsNext = (Arrays.binarySearch(sortedSeparators, codePoint) >= 0); } return true; } // TODO: like capitalizeFirst*, this does not work perfectly for Dutch because of the IJ digraph // which should be capitalized together in *some* cases. - public static String capitalizeEachWord(final String text, final String separators, + public static String capitalizeEachWord(final String text, final int[] sortedSeparators, final Locale locale) { final StringBuilder builder = new StringBuilder(); - boolean needCapsNext = true; + boolean needsCapsNext = true; final int len = text.length(); for (int i = 0; i < len; i = text.offsetByCodePoints(i, 1)) { final String nextChar = text.substring(i, text.offsetByCodePoints(i, 1)); - if (needCapsNext) { + if (needsCapsNext) { builder.append(nextChar.toUpperCase(locale)); } else { builder.append(nextChar.toLowerCase(locale)); } // We need a capital letter next if this is a separator. - needCapsNext = (-1 != separators.indexOf(nextChar.codePointAt(0))); + needsCapsNext = (Arrays.binarySearch(sortedSeparators, nextChar.codePointAt(0)) >= 0); } return builder.toString(); } diff --git a/java/src/com/android/inputmethod/research/ResearchLogger.java b/java/src/com/android/inputmethod/research/ResearchLogger.java index e7f49a605..11fb3a156 100644 --- a/java/src/com/android/inputmethod/research/ResearchLogger.java +++ b/java/src/com/android/inputmethod/research/ResearchLogger.java @@ -59,6 +59,7 @@ import com.android.inputmethod.latin.RichInputConnection; import com.android.inputmethod.latin.SuggestedWords; import com.android.inputmethod.latin.define.ProductionFlag; import com.android.inputmethod.latin.utils.InputTypeUtils; +import com.android.inputmethod.latin.utils.StringUtils; import com.android.inputmethod.latin.utils.TextRange; import com.android.inputmethod.research.MotionEventReader.ReplayData; import com.android.inputmethod.research.ui.SplashScreen; @@ -131,7 +132,8 @@ public class ResearchLogger implements SharedPreferences.OnSharedPreferenceChang public static final String RESEARCH_KEY_OUTPUT_TEXT = ".research."; // constants related to specific log points - private static final String WHITESPACE_SEPARATORS = " \t\n\r"; + private static final int[] WHITESPACE_SEPARATORS = + StringUtils.toSortedCodePointArray(" \t\n\r"); private static final int MAX_INPUTVIEW_LENGTH_TO_CAPTURE = 8192; // must be >=1 private static final String PREF_RESEARCH_SAVED_CHANNEL = "pref_research_saved_channel"; |