diff options
Diffstat (limited to 'java/src/com/android/inputmethod/latin')
34 files changed, 2203 insertions, 896 deletions
diff --git a/java/src/com/android/inputmethod/latin/AdditionalSubtype.java b/java/src/com/android/inputmethod/latin/AdditionalSubtype.java index 4b47a261f..509fc1ba3 100644 --- a/java/src/com/android/inputmethod/latin/AdditionalSubtype.java +++ b/java/src/com/android/inputmethod/latin/AdditionalSubtype.java @@ -27,22 +27,22 @@ import android.view.inputmethod.InputMethodSubtype; import java.util.ArrayList; -public class AdditionalSubtype { +public final class AdditionalSubtype { private static final InputMethodSubtype[] EMPTY_SUBTYPE_ARRAY = new InputMethodSubtype[0]; private AdditionalSubtype() { // This utility class is not publicly instantiable. } - public static boolean isAdditionalSubtype(InputMethodSubtype subtype) { + public static boolean isAdditionalSubtype(final InputMethodSubtype subtype) { return subtype.containsExtraValueKey(IS_ADDITIONAL_SUBTYPE); } private static final String LOCALE_AND_LAYOUT_SEPARATOR = ":"; - public static final String PREF_SUBTYPE_SEPARATOR = ";"; + private static final String PREF_SUBTYPE_SEPARATOR = ";"; - public static InputMethodSubtype createAdditionalSubtype( - String localeString, String keyboardLayoutSetName, String extraValue) { + public static InputMethodSubtype createAdditionalSubtype(final String localeString, + final String keyboardLayoutSetName, final String extraValue) { final String layoutExtraValue = KEYBOARD_LAYOUT_SET + "=" + keyboardLayoutSetName; final String layoutDisplayNameExtraValue; if (Build.VERSION.SDK_INT >= /* JELLY_BEAN */ 15 @@ -62,7 +62,7 @@ public class AdditionalSubtype { layoutExtraValue + "," + additionalSubtypeExtraValue, false, false); } - public static String getPrefSubtype(InputMethodSubtype subtype) { + public static String getPrefSubtype(final InputMethodSubtype subtype) { final String localeString = subtype.getLocale(); final String keyboardLayoutSetName = SubtypeLocale.getKeyboardLayoutSetName(subtype); final String layoutExtraValue = KEYBOARD_LAYOUT_SET + "=" + keyboardLayoutSetName; @@ -74,7 +74,7 @@ public class AdditionalSubtype { : basePrefSubtype + LOCALE_AND_LAYOUT_SEPARATOR + extraValue; } - public static InputMethodSubtype createAdditionalSubtype(String prefSubtype) { + 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); @@ -85,7 +85,7 @@ public class AdditionalSubtype { return createAdditionalSubtype(localeString, keyboardLayoutSetName, extraValue); } - public static InputMethodSubtype[] createAdditionalSubtypesArray(String prefSubtypes) { + public static InputMethodSubtype[] createAdditionalSubtypesArray(final String prefSubtypes) { if (TextUtils.isEmpty(prefSubtypes)) { return EMPTY_SUBTYPE_ARRAY; } @@ -103,4 +103,32 @@ public class AdditionalSubtype { } return subtypesList.toArray(new InputMethodSubtype[subtypesList.size()]); } + + public static String createPrefSubtypes(final InputMethodSubtype[] subtypes) { + if (subtypes == null || subtypes.length == 0) { + return ""; + } + final StringBuilder sb = new StringBuilder(); + for (final InputMethodSubtype subtype : subtypes) { + if (sb.length() > 0) { + sb.append(PREF_SUBTYPE_SEPARATOR); + } + sb.append(getPrefSubtype(subtype)); + } + return sb.toString(); + } + + public static String createPrefSubtypes(final String[] prefSubtypes) { + if (prefSubtypes == null || prefSubtypes.length == 0) { + return ""; + } + final StringBuilder sb = new StringBuilder(); + for (final String prefSubtype : prefSubtypes) { + if (sb.length() > 0) { + sb.append(PREF_SUBTYPE_SEPARATOR); + } + sb.append(prefSubtype); + } + return sb.toString(); + } } diff --git a/java/src/com/android/inputmethod/latin/AdditionalSubtypeSettings.java b/java/src/com/android/inputmethod/latin/AdditionalSubtypeSettings.java index d01592a4d..ae51d2537 100644 --- a/java/src/com/android/inputmethod/latin/AdditionalSubtypeSettings.java +++ b/java/src/com/android/inputmethod/latin/AdditionalSubtypeSettings.java @@ -63,13 +63,13 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { private static final String KEY_IS_SUBTYPE_ENABLER_NOTIFICATION_DIALOG_OPEN = "is_subtype_enabler_notification_dialog_open"; private static final String KEY_SUBTYPE_FOR_SUBTYPE_ENABLER = "subtype_for_subtype_enabler"; - static class SubtypeLocaleItem extends Pair<String, String> + static final class SubtypeLocaleItem extends Pair<String, String> implements Comparable<SubtypeLocaleItem> { - public SubtypeLocaleItem(String localeString, String displayName) { + public SubtypeLocaleItem(final String localeString, final String displayName) { super(localeString, displayName); } - public SubtypeLocaleItem(String localeString) { + public SubtypeLocaleItem(final String localeString) { this(localeString, SubtypeLocale.getSubtypeLocaleDisplayName(localeString)); } @@ -79,13 +79,13 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } @Override - public int compareTo(SubtypeLocaleItem o) { + public int compareTo(final SubtypeLocaleItem o) { return first.compareTo(o.first); } } - static class SubtypeLocaleAdapter extends ArrayAdapter<SubtypeLocaleItem> { - public SubtypeLocaleAdapter(Context context) { + static final class SubtypeLocaleAdapter extends ArrayAdapter<SubtypeLocaleItem> { + public SubtypeLocaleAdapter(final Context context) { super(context, android.R.layout.simple_spinner_item); setDropDownViewResource(android.R.layout.simple_spinner_dropdown_item); @@ -102,7 +102,8 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { addAll(items); } - public static SubtypeLocaleItem createItem(Context context, String localeString) { + public static SubtypeLocaleItem createItem(final Context context, + final String localeString) { if (localeString.equals(SubtypeLocale.NO_LANGUAGE)) { final String displayName = context.getString(R.string.subtype_no_language); return new SubtypeLocaleItem(localeString, displayName); @@ -112,8 +113,8 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } } - static class KeyboardLayoutSetItem extends Pair<String, String> { - public KeyboardLayoutSetItem(InputMethodSubtype subtype) { + static final class KeyboardLayoutSetItem extends Pair<String, String> { + public KeyboardLayoutSetItem(final InputMethodSubtype subtype) { super(SubtypeLocale.getKeyboardLayoutSetName(subtype), SubtypeLocale.getKeyboardLayoutSetDisplayName(subtype)); } @@ -124,8 +125,8 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } } - static class KeyboardLayoutSetAdapter extends ArrayAdapter<KeyboardLayoutSetItem> { - public KeyboardLayoutSetAdapter(Context context) { + static final class KeyboardLayoutSetAdapter extends ArrayAdapter<KeyboardLayoutSetItem> { + public KeyboardLayoutSetAdapter(final Context context) { super(context, android.R.layout.simple_spinner_item); setDropDownViewResource(android.R.layout.simple_spinner_dropdown_item); @@ -147,7 +148,7 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { public KeyboardLayoutSetAdapter getKeyboardLayoutSetAdapter(); } - static class SubtypePreference extends DialogPreference + static final class SubtypePreference extends DialogPreference implements DialogInterface.OnCancelListener { private static final String KEY_PREFIX = "subtype_pref_"; private static final String KEY_NEW_SUBTYPE = KEY_PREFIX + "new"; @@ -159,13 +160,13 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { private Spinner mSubtypeLocaleSpinner; private Spinner mKeyboardLayoutSetSpinner; - public static SubtypePreference newIncompleteSubtypePreference( - Context context, SubtypeDialogProxy proxy) { + public static SubtypePreference newIncompleteSubtypePreference(final Context context, + final SubtypeDialogProxy proxy) { return new SubtypePreference(context, null, proxy); } - public SubtypePreference(Context context, InputMethodSubtype subtype, - SubtypeDialogProxy proxy) { + public SubtypePreference(final Context context, final InputMethodSubtype subtype, + final SubtypeDialogProxy proxy) { super(context, null); setDialogLayoutResource(R.layout.additional_subtype_dialog); setPersistent(false); @@ -185,7 +186,7 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { return mSubtype; } - public void setSubtype(InputMethodSubtype subtype) { + public void setSubtype(final InputMethodSubtype subtype) { mPreviousSubtype = mSubtype; mSubtype = subtype; if (isIncomplete()) { @@ -221,7 +222,7 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } @Override - protected void onPrepareDialogBuilder(AlertDialog.Builder builder) { + protected void onPrepareDialogBuilder(final AlertDialog.Builder builder) { final Context context = builder.getContext(); builder.setCancelable(true).setOnCancelListener(this); if (isIncomplete()) { @@ -239,7 +240,7 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } } - private static void setSpinnerPosition(Spinner spinner, Object itemToSelect) { + private static void setSpinnerPosition(final Spinner spinner, final Object itemToSelect) { final SpinnerAdapter adapter = spinner.getAdapter(); final int count = adapter.getCount(); for (int i = 0; i < count; i++) { @@ -252,14 +253,14 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } @Override - public void onCancel(DialogInterface dialog) { + public void onCancel(final DialogInterface dialog) { if (isIncomplete()) { mProxy.onRemovePressed(this); } } @Override - public void onClick(DialogInterface dialog, int which) { + public void onClick(final DialogInterface dialog, final int which) { super.onClick(dialog, which); switch (which) { case DialogInterface.BUTTON_POSITIVE: @@ -287,12 +288,12 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } } - private static int getSpinnerPosition(Spinner spinner) { + private static int getSpinnerPosition(final Spinner spinner) { if (spinner == null) return -1; return spinner.getSelectedItemPosition(); } - private static void setSpinnerPosition(Spinner spinner, int position) { + private static void setSpinnerPosition(final Spinner spinner, final int position) { if (spinner == null || position < 0) return; spinner.setSelection(position); } @@ -313,7 +314,7 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } @Override - protected void onRestoreInstanceState(Parcelable state) { + protected void onRestoreInstanceState(final Parcelable state) { if (!(state instanceof SavedState)) { super.onRestoreInstanceState(state); return; @@ -326,24 +327,24 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { setSubtype(myState.mSubtype); } - static class SavedState extends Preference.BaseSavedState { + static final class SavedState extends Preference.BaseSavedState { InputMethodSubtype mSubtype; int mSubtypeLocaleSelectedPos; int mKeyboardLayoutSetSelectedPos; - public SavedState(Parcelable superState) { + public SavedState(final Parcelable superState) { super(superState); } @Override - public void writeToParcel(Parcel dest, int flags) { + public void writeToParcel(final Parcel dest, final int flags) { super.writeToParcel(dest, flags); dest.writeInt(mSubtypeLocaleSelectedPos); dest.writeInt(mKeyboardLayoutSetSelectedPos); dest.writeParcelable(mSubtype, 0); } - public SavedState(Parcel source) { + public SavedState(final Parcel source) { super(source); mSubtypeLocaleSelectedPos = source.readInt(); mKeyboardLayoutSetSelectedPos = source.readInt(); @@ -354,12 +355,12 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { public static final Parcelable.Creator<SavedState> CREATOR = new Parcelable.Creator<SavedState>() { @Override - public SavedState createFromParcel(Parcel source) { + public SavedState createFromParcel(final Parcel source) { return new SavedState(source); } @Override - public SavedState[] newArray(int size) { + public SavedState[] newArray(final int size) { return new SavedState[size]; } }; @@ -371,7 +372,7 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } @Override - public void onCreate(Bundle savedInstanceState) { + public void onCreate(final Bundle savedInstanceState) { super.onCreate(savedInstanceState); addPreferencesFromResource(R.xml.additional_subtype_settings); @@ -381,7 +382,7 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } @Override - public void onActivityCreated(Bundle savedInstanceState) { + public void onActivityCreated(final Bundle savedInstanceState) { final Context context = getActivity(); mSubtypeLocaleAdapter = new SubtypeLocaleAdapter(context); mKeyboardLayoutSetAdapter = new KeyboardLayoutSetAdapter(context); @@ -411,7 +412,7 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } @Override - public void onSaveInstanceState(Bundle outState) { + public void onSaveInstanceState(final Bundle outState) { super.onSaveInstanceState(outState); if (mIsAddingNewSubtype) { outState.putBoolean(KEY_IS_ADDING_NEW_SUBTYPE, true); @@ -426,7 +427,7 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { private final SubtypeDialogProxy mSubtypeProxy = new SubtypeDialogProxy() { @Override - public void onRemovePressed(SubtypePreference subtypePref) { + public void onRemovePressed(final SubtypePreference subtypePref) { mIsAddingNewSubtype = false; final PreferenceGroup group = getPreferenceScreen(); group.removePreference(subtypePref); @@ -434,7 +435,7 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } @Override - public void onSavePressed(SubtypePreference subtypePref) { + public void onSavePressed(final SubtypePreference subtypePref) { final InputMethodSubtype subtype = subtypePref.getSubtype(); if (!subtypePref.hasBeenModified()) { return; @@ -453,7 +454,7 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } @Override - public void onAddPressed(SubtypePreference subtypePref) { + public void onAddPressed(final SubtypePreference subtypePref) { mIsAddingNewSubtype = false; final InputMethodSubtype subtype = subtypePref.getSubtype(); if (findDuplicatedSubtype(subtype) == null) { @@ -481,7 +482,7 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } }; - private void showSubtypeAlreadyExistsToast(InputMethodSubtype subtype) { + private void showSubtypeAlreadyExistsToast(final InputMethodSubtype subtype) { final Context context = getActivity(); final Resources res = context.getResources(); final String message = res.getString(R.string.custom_input_style_already_exists, @@ -489,14 +490,15 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { Toast.makeText(context, message, Toast.LENGTH_SHORT).show(); } - private InputMethodSubtype findDuplicatedSubtype(InputMethodSubtype subtype) { + private InputMethodSubtype findDuplicatedSubtype(final InputMethodSubtype subtype) { final String localeString = subtype.getLocale(); final String keyboardLayoutSetName = SubtypeLocale.getKeyboardLayoutSetName(subtype); return ImfUtils.findSubtypeByLocaleAndKeyboardLayoutSet( getActivity(), localeString, keyboardLayoutSetName); } - private AlertDialog createDialog(SubtypePreference subtypePref) { + private AlertDialog createDialog( + @SuppressWarnings("unused") final SubtypePreference subtypePref) { final AlertDialog.Builder builder = new AlertDialog.Builder(getActivity()); builder.setTitle(R.string.custom_input_styles_title) .setMessage(R.string.custom_input_style_note_message) @@ -519,7 +521,7 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { return builder.create(); } - private void setPrefSubtypes(String prefSubtypes, Context context) { + private void setPrefSubtypes(final String prefSubtypes, final Context context) { final PreferenceGroup group = getPreferenceScreen(); group.removeAll(); final InputMethodSubtype[] subtypesArray = @@ -547,23 +549,12 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { return subtypes.toArray(new InputMethodSubtype[subtypes.size()]); } - private String getPrefSubtypes(InputMethodSubtype[] subtypes) { - final StringBuilder sb = new StringBuilder(); - for (final InputMethodSubtype subtype : subtypes) { - if (sb.length() > 0) { - sb.append(AdditionalSubtype.PREF_SUBTYPE_SEPARATOR); - } - sb.append(AdditionalSubtype.getPrefSubtype(subtype)); - } - return sb.toString(); - } - @Override public void onPause() { super.onPause(); final String oldSubtypes = SettingsValues.getPrefAdditionalSubtypes(mPrefs, getResources()); final InputMethodSubtype[] subtypes = getSubtypes(); - final String prefSubtypes = getPrefSubtypes(subtypes); + final String prefSubtypes = AdditionalSubtype.createPrefSubtypes(subtypes); if (prefSubtypes.equals(oldSubtypes)) { return; } @@ -578,13 +569,13 @@ public class AdditionalSubtypeSettings extends PreferenceFragment { } @Override - public void onCreateOptionsMenu(Menu menu, MenuInflater inflater) { + public void onCreateOptionsMenu(final Menu menu, final MenuInflater inflater) { final MenuItem addSubtypeMenu = menu.add(0, MENU_ADD_SUBTYPE, 0, R.string.add_style); addSubtypeMenu.setShowAsAction(MenuItem.SHOW_AS_ACTION_IF_ROOM); } @Override - public boolean onOptionsItemSelected(MenuItem item) { + public boolean onOptionsItemSelected(final MenuItem item) { final int itemId = item.getItemId(); if (itemId == MENU_ADD_SUBTYPE) { final SubtypePreference newSubtype = diff --git a/java/src/com/android/inputmethod/latin/AutoCorrection.java b/java/src/com/android/inputmethod/latin/AutoCorrection.java index 01ba30077..f425e360a 100644 --- a/java/src/com/android/inputmethod/latin/AutoCorrection.java +++ b/java/src/com/android/inputmethod/latin/AutoCorrection.java @@ -73,11 +73,11 @@ public class AutoCorrection { return maxFreq; } - // Returns true if this isn't in any dictionary. - public static boolean isNotAWord( + // Returns true if this is in any of the dictionaries. + public static boolean isInTheDictionary( final ConcurrentHashMap<String, Dictionary> dictionaries, final CharSequence word, final boolean ignoreCase) { - return !isValidWord(dictionaries, word, ignoreCase); + return isValidWord(dictionaries, word, ignoreCase); } public static boolean suggestionExceedsAutoCorrectionThreshold(SuggestedWordInfo suggestion, diff --git a/java/src/com/android/inputmethod/latin/BinaryDictionary.java b/java/src/com/android/inputmethod/latin/BinaryDictionary.java index 8909526d8..c3ae81f3a 100644 --- a/java/src/com/android/inputmethod/latin/BinaryDictionary.java +++ b/java/src/com/android/inputmethod/latin/BinaryDictionary.java @@ -41,7 +41,7 @@ public class BinaryDictionary extends Dictionary { * It is necessary to keep it at this value because some languages e.g. German have * really long words. */ - public static final int MAX_WORD_LENGTH = 48; + public static final int MAX_WORD_LENGTH = Constants.Dictionary.MAX_WORD_LENGTH; public static final int MAX_WORDS = 18; public static final int MAX_SPACES = 16; diff --git a/java/src/com/android/inputmethod/latin/BinaryDictionaryGetter.java b/java/src/com/android/inputmethod/latin/BinaryDictionaryGetter.java index e1cb195bc..9a888ade4 100644 --- a/java/src/com/android/inputmethod/latin/BinaryDictionaryGetter.java +++ b/java/src/com/android/inputmethod/latin/BinaryDictionaryGetter.java @@ -17,6 +17,7 @@ package com.android.inputmethod.latin; import com.android.inputmethod.latin.makedict.BinaryDictInputOutput; +import com.android.inputmethod.latin.makedict.FormatSpec; import android.content.Context; import android.content.SharedPreferences; @@ -359,7 +360,7 @@ class BinaryDictionaryGetter { final ByteBuffer buffer = inStream.getChannel().map( FileChannel.MapMode.READ_ONLY, 0, f.length()); final int magic = buffer.getInt(); - if (magic != BinaryDictInputOutput.VERSION_2_MAGIC_NUMBER) { + if (magic != FormatSpec.VERSION_2_MAGIC_NUMBER) { return false; } final int formatVersion = buffer.getInt(); diff --git a/java/src/com/android/inputmethod/latin/CollectionUtils.java b/java/src/com/android/inputmethod/latin/CollectionUtils.java index baa2ee1cd..c75f2df5c 100644 --- a/java/src/com/android/inputmethod/latin/CollectionUtils.java +++ b/java/src/com/android/inputmethod/latin/CollectionUtils.java @@ -30,7 +30,7 @@ import java.util.TreeSet; import java.util.concurrent.ConcurrentHashMap; import java.util.concurrent.CopyOnWriteArrayList; -public class CollectionUtils { +public final class CollectionUtils { private CollectionUtils() { // This utility class is not publicly instantiable. } diff --git a/java/src/com/android/inputmethod/latin/Constants.java b/java/src/com/android/inputmethod/latin/Constants.java index d71c0f995..57e12a64f 100644 --- a/java/src/com/android/inputmethod/latin/Constants.java +++ b/java/src/com/android/inputmethod/latin/Constants.java @@ -16,8 +16,6 @@ package com.android.inputmethod.latin; -import android.view.inputmethod.EditorInfo; - public final class Constants { public static final class Color { /** @@ -54,7 +52,7 @@ public final class Constants { * The private IME option used to indicate that the given text field needs ASCII code points * input. * - * @deprecated Use {@link EditorInfo#IME_FLAG_FORCE_ASCII}. + * @deprecated Use EditorInfo#IME_FLAG_FORCE_ASCII. */ @SuppressWarnings("dep-ann") public static final String FORCE_ASCII = "forceAscii"; @@ -128,6 +126,14 @@ public final class Constants { } } + public static class Dictionary { + public static final int MAX_WORD_LENGTH = 48; + + private Dictionary() { + // This utility class is no publicly instantiable. + } + } + public static final int NOT_A_CODE = -1; // See {@link KeyboardActionListener.Adapter#isInvalidCoordinate(int)}. diff --git a/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java b/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java index cdf5247de..b93c17f11 100644 --- a/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java +++ b/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java @@ -21,6 +21,7 @@ import android.util.Log; import com.android.inputmethod.keyboard.ProximityInfo; import com.android.inputmethod.latin.SuggestedWords.SuggestedWordInfo; import com.android.inputmethod.latin.makedict.BinaryDictInputOutput; +import com.android.inputmethod.latin.makedict.FormatSpec; import com.android.inputmethod.latin.makedict.FusionDictionary; import com.android.inputmethod.latin.makedict.FusionDictionary.Node; import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; @@ -89,6 +90,10 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { /** Controls access to the local binary dictionary for this instance. */ private final DictionaryController mLocalDictionaryController = new DictionaryController(); + private static final int BINARY_DICT_VERSION = 1; + private static final FormatSpec.FormatOptions FORMAT_OPTIONS = + new FormatSpec.FormatOptions(BINARY_DICT_VERSION); + /** * Abstract method for loading the unigrams and bigrams of a given dictionary in a background * thread. @@ -172,12 +177,12 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { // considering performance regression. protected void addWord(final String word, final String shortcutTarget, final int frequency) { if (shortcutTarget == null) { - mFusionDictionary.add(word, frequency, null); + mFusionDictionary.add(word, frequency, null, false /* isNotAWord */); } else { // TODO: Do this in the subclass, with this class taking an arraylist. final ArrayList<WeightedString> shortcutTargets = CollectionUtils.newArrayList(); shortcutTargets.add(new WeightedString(shortcutTarget, frequency)); - mFusionDictionary.add(word, frequency, shortcutTargets); + mFusionDictionary.add(word, frequency, shortcutTargets, false /* isNotAWord */); } } @@ -310,7 +315,7 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { FileOutputStream out = null; try { out = new FileOutputStream(tempFile); - BinaryDictInputOutput.writeDictionaryBinary(out, mFusionDictionary, 1); + BinaryDictInputOutput.writeDictionaryBinary(out, mFusionDictionary, FORMAT_OPTIONS); out.flush(); out.close(); tempFile.renameTo(file); diff --git a/java/src/com/android/inputmethod/latin/ImfUtils.java b/java/src/com/android/inputmethod/latin/ImfUtils.java index 1461c0240..2674e4575 100644 --- a/java/src/com/android/inputmethod/latin/ImfUtils.java +++ b/java/src/com/android/inputmethod/latin/ImfUtils.java @@ -29,7 +29,7 @@ import java.util.List; /** * Utility class for Input Method Framework */ -public class ImfUtils { +public final class ImfUtils { private ImfUtils() { // This utility class is not publicly instantiable. } diff --git a/java/src/com/android/inputmethod/latin/InputTypeUtils.java b/java/src/com/android/inputmethod/latin/InputTypeUtils.java index 40c3b765e..500866a13 100644 --- a/java/src/com/android/inputmethod/latin/InputTypeUtils.java +++ b/java/src/com/android/inputmethod/latin/InputTypeUtils.java @@ -18,7 +18,7 @@ package com.android.inputmethod.latin; import android.text.InputType; -public class InputTypeUtils implements InputType { +public final class InputTypeUtils implements InputType { private static final int WEB_TEXT_PASSWORD_INPUT_TYPE = TYPE_CLASS_TEXT | TYPE_TEXT_VARIATION_WEB_PASSWORD; private static final int WEB_TEXT_EMAIL_ADDRESS_INPUT_TYPE = diff --git a/java/src/com/android/inputmethod/latin/JniUtils.java b/java/src/com/android/inputmethod/latin/JniUtils.java index 86a3826d8..f9305991a 100644 --- a/java/src/com/android/inputmethod/latin/JniUtils.java +++ b/java/src/com/android/inputmethod/latin/JniUtils.java @@ -20,7 +20,7 @@ import android.util.Log; import com.android.inputmethod.latin.define.JniLibName; -public class JniUtils { +public final class JniUtils { private static final String TAG = JniUtils.class.getSimpleName(); private JniUtils() { diff --git a/java/src/com/android/inputmethod/latin/LastComposedWord.java b/java/src/com/android/inputmethod/latin/LastComposedWord.java index bb39ce4f7..dd73a978c 100644 --- a/java/src/com/android/inputmethod/latin/LastComposedWord.java +++ b/java/src/com/android/inputmethod/latin/LastComposedWord.java @@ -38,12 +38,12 @@ public class LastComposedWord { // an auto-correction. public static final int COMMIT_TYPE_CANCEL_AUTO_CORRECT = 3; - public static final int NOT_A_SEPARATOR = -1; + public static final String NOT_A_SEPARATOR = ""; public final int[] mPrimaryKeyCodes; public final String mTypedWord; public final String mCommittedWord; - public final int mSeparatorCode; + public final String mSeparatorString; public final CharSequence mPrevWord; public final InputPointers mInputPointers = new InputPointers(BinaryDictionary.MAX_WORD_LENGTH); @@ -56,14 +56,14 @@ public class LastComposedWord { // immutable. Do not fiddle with their contents after you passed them to this constructor. public LastComposedWord(final int[] primaryKeyCodes, final InputPointers inputPointers, final String typedWord, final String committedWord, - final int separatorCode, final CharSequence prevWord) { + final String separatorString, final CharSequence prevWord) { mPrimaryKeyCodes = primaryKeyCodes; if (inputPointers != null) { mInputPointers.copy(inputPointers); } mTypedWord = typedWord; mCommittedWord = committedWord; - mSeparatorCode = separatorCode; + mSeparatorString = separatorString; mActive = true; mPrevWord = prevWord; } @@ -80,7 +80,7 @@ public class LastComposedWord { return TextUtils.equals(mTypedWord, mCommittedWord); } - public static int getSeparatorLength(final int separatorCode) { - return NOT_A_SEPARATOR == separatorCode ? 0 : 1; + public static int getSeparatorLength(final String separatorString) { + return StringUtils.codePointCount(separatorString); } } diff --git a/java/src/com/android/inputmethod/latin/LatinIME.java b/java/src/com/android/inputmethod/latin/LatinIME.java index 83a306818..db8f269eb 100644 --- a/java/src/com/android/inputmethod/latin/LatinIME.java +++ b/java/src/com/android/inputmethod/latin/LatinIME.java @@ -36,6 +36,8 @@ import android.inputmethodservice.InputMethodService; import android.media.AudioManager; import android.net.ConnectivityManager; import android.os.Debug; +import android.os.Handler; +import android.os.HandlerThread; import android.os.IBinder; import android.os.Message; import android.os.SystemClock; @@ -184,13 +186,16 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen 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; + private static final int MSG_SHOW_GESTURE_PREVIEW_AND_SUGGESTION_STRIP = 3; + + private static final int ARG1_DISMISS_GESTURE_FLOATING_PREVIEW_TEXT = 1; private int mDelayUpdateSuggestions; private int mDelayUpdateShiftState; private long mDoubleSpacesTurnIntoPeriodTimeout; private long mDoubleSpaceTimerStart; - public UIHandler(LatinIME outerInstance) { + public UIHandler(final LatinIME outerInstance) { super(outerInstance); } @@ -205,7 +210,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } @Override - public void handleMessage(Message msg) { + public void handleMessage(final Message msg) { final LatinIME latinIme = getOuterInstance(); final KeyboardSwitcher switcher = latinIme.mKeyboardSwitcher; switch (msg.what) { @@ -215,6 +220,10 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen case MSG_UPDATE_SHIFT_STATE: switcher.updateShiftState(); break; + case MSG_SHOW_GESTURE_PREVIEW_AND_SUGGESTION_STRIP: + latinIme.showGesturePreviewAndSuggestionStrip((SuggestedWords)msg.obj, + msg.arg1 == ARG1_DISMISS_GESTURE_FLOATING_PREVIEW_TEXT); + break; } } @@ -239,6 +248,15 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen removeMessages(MSG_UPDATE_SHIFT_STATE); } + public void showGesturePreviewAndSuggestionStrip(final SuggestedWords suggestedWords, + final boolean dismissGestureFloatingPreviewText) { + removeMessages(MSG_SHOW_GESTURE_PREVIEW_AND_SUGGESTION_STRIP); + final int arg1 = dismissGestureFloatingPreviewText + ? ARG1_DISMISS_GESTURE_FLOATING_PREVIEW_TEXT : 0; + obtainMessage(MSG_SHOW_GESTURE_PREVIEW_AND_SUGGESTION_STRIP, arg1, 0, suggestedWords) + .sendToTarget(); + } + public void startDoubleSpacesTimer() { mDoubleSpaceTimerStart = SystemClock.uptimeMillis(); } @@ -276,7 +294,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen mHasPendingStartInput = false; } - private void executePendingImsCallback(LatinIME latinIme, EditorInfo editorInfo, + private void executePendingImsCallback(final LatinIME latinIme, final EditorInfo editorInfo, boolean restarting) { if (mHasPendingFinishInputView) latinIme.onFinishInputViewInternal(mHasPendingFinishInput); @@ -287,7 +305,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen resetPendingImsCallback(); } - public void onStartInput(EditorInfo editorInfo, boolean restarting) { + public void onStartInput(final EditorInfo editorInfo, final boolean restarting) { if (hasMessages(MSG_PENDING_IMS_CALLBACK)) { // Typically this is the second onStartInput after orientation changed. mHasPendingStartInput = true; @@ -303,7 +321,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } } - public void onStartInputView(EditorInfo editorInfo, boolean restarting) { + public void onStartInputView(final EditorInfo editorInfo, final boolean restarting) { if (hasMessages(MSG_PENDING_IMS_CALLBACK) && KeyboardId.equivalentEditorInfoForKeyboard(editorInfo, mAppliedEditorInfo)) { // Typically this is the second onStartInputView after orientation changed. @@ -323,7 +341,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } } - public void onFinishInputView(boolean finishingInput) { + public void onFinishInputView(final boolean finishingInput) { if (hasMessages(MSG_PENDING_IMS_CALLBACK)) { // Typically this is the first onFinishInputView after orientation changed. mHasPendingFinishInputView = true; @@ -425,7 +443,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // Note that this method is called from a non-UI thread. @Override - public void onUpdateMainDictionaryAvailability(boolean isMainDictionaryAvailable) { + public void onUpdateMainDictionaryAvailability(final boolean isMainDictionaryAvailable) { mIsMainDictionaryAvailable = isMainDictionaryAvailable; final MainKeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); if (mainKeyboardView != null) { @@ -529,7 +547,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } @Override - public void onConfigurationChanged(Configuration conf) { + public void onConfigurationChanged(final Configuration conf) { // System locale has been changed. Needs to reload keyboard. if (mSubtypeSwitcher.onConfigurationChanged(conf, this)) { loadKeyboard(); @@ -555,7 +573,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } @Override - public void setInputView(View view) { + public void setInputView(final View view) { super.setInputView(view); mExtractArea = getWindow().getWindow().getDecorView() .findViewById(android.R.id.extractArea); @@ -570,23 +588,23 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } @Override - public void setCandidatesView(View view) { + public void setCandidatesView(final View view) { // To ensure that CandidatesView will never be set. return; } @Override - public void onStartInput(EditorInfo editorInfo, boolean restarting) { + public void onStartInput(final EditorInfo editorInfo, final boolean restarting) { mHandler.onStartInput(editorInfo, restarting); } @Override - public void onStartInputView(EditorInfo editorInfo, boolean restarting) { + public void onStartInputView(final EditorInfo editorInfo, final boolean restarting) { mHandler.onStartInputView(editorInfo, restarting); } @Override - public void onFinishInputView(boolean finishingInput) { + public void onFinishInputView(final boolean finishingInput) { mHandler.onFinishInputView(finishingInput); } @@ -596,19 +614,19 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } @Override - public void onCurrentInputMethodSubtypeChanged(InputMethodSubtype subtype) { + public void onCurrentInputMethodSubtypeChanged(final InputMethodSubtype subtype) { // Note that the calling sequence of onCreate() and onCurrentInputMethodSubtypeChanged() // is not guaranteed. It may even be called at the same time on a different thread. mSubtypeSwitcher.updateSubtype(subtype); loadKeyboard(); } - private void onStartInputInternal(EditorInfo editorInfo, boolean restarting) { + private void onStartInputInternal(final EditorInfo editorInfo, final boolean restarting) { super.onStartInput(editorInfo, restarting); } @SuppressWarnings("deprecation") - private void onStartInputViewInternal(EditorInfo editorInfo, boolean restarting) { + private void onStartInputViewInternal(final EditorInfo editorInfo, final boolean restarting) { super.onStartInputView(editorInfo, restarting); final KeyboardSwitcher switcher = mKeyboardSwitcher; final MainKeyboardView mainKeyboardView = switcher.getMainKeyboardView(); @@ -700,6 +718,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } } + mConnection.resetCachesUponCursorMove(mLastSelectionStart); + if (isDifferentTextField) { mainKeyboardView.closing(); loadSettings(); @@ -749,7 +769,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen getCurrentInputConnection()); } super.onWindowHidden(); - final KeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); + final MainKeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); if (mainKeyboardView != null) { mainKeyboardView.closing(); } @@ -763,16 +783,16 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen ResearchLogger.getInstance().latinIME_onFinishInputInternal(); } - final KeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); + final MainKeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); if (mainKeyboardView != null) { mainKeyboardView.closing(); } } - private void onFinishInputViewInternal(boolean finishingInput) { + private void onFinishInputViewInternal(final boolean finishingInput) { super.onFinishInputView(finishingInput); mKeyboardSwitcher.onFinishInputView(); - final KeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); + final MainKeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); if (mainKeyboardView != null) { mainKeyboardView.cancelAllMessages(); } @@ -781,9 +801,9 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } @Override - public void onUpdateSelection(int oldSelStart, int oldSelEnd, - int newSelStart, int newSelEnd, - int composingSpanStart, int composingSpanEnd) { + public void onUpdateSelection(final int oldSelStart, final int oldSelEnd, + final int newSelStart, final int newSelEnd, + final int composingSpanStart, final int composingSpanEnd) { super.onUpdateSelection(oldSelStart, oldSelEnd, newSelStart, newSelEnd, composingSpanStart, composingSpanEnd); if (DEBUG) { @@ -823,7 +843,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // we know for sure the cursor moved while we were composing and we should reset // the state. final boolean noComposingSpan = composingSpanStart == -1 && composingSpanEnd == -1; - if (!mExpectingUpdateSelection) { + if (!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 @@ -839,7 +860,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen mSpaceState = SPACE_STATE_NONE; if ((!mWordComposer.isComposingWord()) || selectionChanged || noComposingSpan) { - resetEntireInputState(); + resetEntireInputState(newSelStart); } mHandler.postUpdateShiftState(); @@ -880,7 +901,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen * cause the suggestions strip to disappear and re-appear. */ @Override - public void onExtractedCursorMovement(int dx, int dy) { + public void onExtractedCursorMovement(final int dx, final int dy) { if (mCurrentSettings.isSuggestionsRequested(mDisplayOrientation)) return; super.onExtractedCursorMovement(dx, dy); @@ -900,7 +921,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } @Override - public void onDisplayCompletions(CompletionInfo[] applicationSpecifiedCompletions) { + public void onDisplayCompletions(final CompletionInfo[] applicationSpecifiedCompletions) { if (DEBUG) { Log.i(TAG, "Received completions:"); if (applicationSpecifiedCompletions != null) { @@ -942,7 +963,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } } - private void setSuggestionStripShownInternal(boolean shown, boolean needsInputViewShown) { + private void setSuggestionStripShownInternal(final boolean shown, + final boolean needsInputViewShown) { // TODO: Modify this if we support suggestions with hard keyboard if (onEvaluateInputViewShown() && mSuggestionsContainer != null) { final MainKeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); @@ -960,7 +982,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } } - private void setSuggestionStripShown(boolean shown) { + private void setSuggestionStripShown(final boolean shown) { setSuggestionStripShownInternal(shown, /* needsInputViewShown */true); } @@ -970,7 +992,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen return currentHeight; } - final KeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); + final MainKeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); if (mainKeyboardView == null) { return 0; } @@ -990,9 +1012,9 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } @Override - public void onComputeInsets(InputMethodService.Insets outInsets) { + public void onComputeInsets(final InputMethodService.Insets outInsets) { super.onComputeInsets(outInsets); - final KeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); + final MainKeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); if (mainKeyboardView == null || mSuggestionsContainer == null) { return; } @@ -1043,10 +1065,14 @@ 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() { + private void resetEntireInputState(final int newCursorPosition) { resetComposingState(true /* alsoResetLastComposedWord */); - clearSuggestionStrip(); - mConnection.finishComposingText(); + if (mCurrentSettings.mBigramPredictionEnabled) { + clearSuggestionStrip(); + } else { + setSuggestionStrip(mCurrentSettings.mSuggestPuncList, false); + } + mConnection.resetCachesUponCursorMove(newCursorPosition); } private void resetComposingState(final boolean alsoResetLastComposedWord) { @@ -1055,7 +1081,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen mLastComposedWord = LastComposedWord.NOT_A_COMPOSED_WORD; } - private void commitTyped(final int separatorCode) { + private void commitTyped(final String separatorString) { if (!mWordComposer.isComposingWord()) return; final CharSequence typedWord = mWordComposer.getTypedWord(); if (typedWord.length() > 0) { @@ -1063,7 +1089,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen final CharSequence prevWord = addToUserHistoryDictionary(typedWord); mLastComposedWord = mWordComposer.commitWord( LastComposedWord.COMMIT_TYPE_USER_TYPED_WORD, typedWord.toString(), - separatorCode, prevWord); + separatorString, prevWord); } } @@ -1092,7 +1118,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // Note: getCursorCapsMode() returns the current capitalization mode that is any // combination of CAP_MODE_CHARACTERS, CAP_MODE_WORDS, and CAP_MODE_SENTENCES. 0 means none // of them. - return mConnection.getCursorCapsMode(inputType); + return mConnection.getCursorCapsMode(inputType, mSubtypeSwitcher.getCurrentSubtypeLocale()); } // Factor in auto-caps and manual caps and compute the current caps mode. @@ -1153,12 +1179,12 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // Callback for the {@link SuggestionStripView}, to call when the "add to dictionary" hint is // pressed. @Override - public boolean addWordToUserDictionary(String word) { + public boolean addWordToUserDictionary(final String word) { mUserDictionary.addWordToUserDictionary(word, 128); return true; } - private static boolean isAlphabet(int code) { + private static boolean isAlphabet(final int code) { return Character.isLetter(code); } @@ -1171,7 +1197,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen public static final int CODE_SHOW_INPUT_METHOD_PICKER = 1; @Override - public boolean onCustomRequest(int requestCode) { + public boolean onCustomRequest(final int requestCode) { if (isShowingOptionDialog()) return false; switch (requestCode) { case CODE_SHOW_INPUT_METHOD_PICKER: @@ -1189,11 +1215,11 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen return mOptionsDialog != null && mOptionsDialog.isShowing(); } - private static int getActionId(Keyboard keyboard) { + private static int getActionId(final Keyboard keyboard) { return keyboard != null ? keyboard.mId.imeActionId() : EditorInfo.IME_ACTION_NONE; } - private void performEditorAction(int actionId) { + private void performEditorAction(final int actionId) { mConnection.performEditorAction(actionId); } @@ -1216,7 +1242,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } } - private void sendUpDownEnterOrBackspace(final int code) { + private void sendDownUpKeyEventForBackwardCompatibility(final int code) { final long eventTime = SystemClock.uptimeMillis(); mConnection.sendKeyEvent(new KeyEvent(eventTime, eventTime, KeyEvent.ACTION_DOWN, code, 0, 0, KeyCharacterMap.VIRTUAL_KEYBOARD, 0, @@ -1226,11 +1252,11 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen KeyEvent.FLAG_SOFT_KEYBOARD | KeyEvent.FLAG_KEEP_TOUCH_MODE)); } - private void sendKeyCodePoint(int code) { + private void sendKeyCodePoint(final int code) { // TODO: Remove this special handling of digit letters. // For backward compatibility. See {@link InputMethodService#sendKeyChar(char)}. if (code >= '0' && code <= '9') { - super.sendKeyChar((char)code); + sendDownUpKeyEventForBackwardCompatibility(code - '0' + KeyEvent.KEYCODE_0); if (ProductionFlag.IS_EXPERIMENTAL) { ResearchLogger.latinIME_sendKeyCodePoint(code); } @@ -1245,7 +1271,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // a hardware keyboard event on pressing enter or delete. This is bad for many // reasons (there are race conditions with commits) but some applications are // relying on this behavior so we continue to support it for older apps. - sendUpDownEnterOrBackspace(KeyEvent.KEYCODE_ENTER); + sendDownUpKeyEventForBackwardCompatibility(KeyEvent.KEYCODE_ENTER); } else { final String text = new String(new int[] { code }, 0, 1); mConnection.commitText(text, text.length()); @@ -1254,7 +1280,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // Implementation of {@link KeyboardActionListener}. @Override - public void onCodeInput(int primaryCode, int x, int y) { + public void onCodeInput(final int primaryCode, final int x, final int y) { final long when = SystemClock.uptimeMillis(); if (primaryCode != Keyboard.CODE_DELETE || when > mLastKeyTime + QUICK_PRESS) { mDeleteCount = 0; @@ -1309,7 +1335,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen break; case Keyboard.CODE_RESEARCH: if (ProductionFlag.IS_EXPERIMENTAL) { - ResearchLogger.getInstance().presentResearchDialog(this); + ResearchLogger.getInstance().onResearchKeySelected(this); } break; default: @@ -1340,7 +1366,9 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen if (!didAutoCorrect && primaryCode != Keyboard.CODE_SHIFT && primaryCode != Keyboard.CODE_SWITCH_ALPHA_SYMBOL) mLastComposedWord.deactivate(); - mEnteredText = null; + if (Keyboard.CODE_DELETE != primaryCode) { + mEnteredText = null; + } mConnection.endBatchEdit(); if (ProductionFlag.IS_EXPERIMENTAL) { ResearchLogger.latinIME_onCodeInput(primaryCode, x, y); @@ -1349,10 +1377,12 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // Called from PointerTracker through the KeyboardActionListener interface @Override - public void onTextInput(CharSequence rawText) { + public void onTextInput(final CharSequence rawText) { mConnection.beginBatchEdit(); if (mWordComposer.isComposingWord()) { - commitCurrentAutoCorrection(LastComposedWord.NOT_A_SEPARATOR); + commitCurrentAutoCorrection(rawText.toString()); + } else { + resetComposingState(true /* alsoResetLastComposedWord */); } mHandler.postUpdateSuggestionStrip(); final CharSequence text = specificTldProcessingOnTextInput(rawText); @@ -1365,14 +1395,13 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen mKeyboardSwitcher.onCodeInput(Keyboard.CODE_OUTPUT_TEXT); mSpaceState = SPACE_STATE_NONE; mEnteredText = text; - resetComposingState(true /* alsoResetLastComposedWord */); } @Override public void onStartBatchInput() { mConnection.beginBatchEdit(); if (mWordComposer.isComposingWord()) { - commitCurrentAutoCorrection(LastComposedWord.NOT_A_SEPARATOR); + commitTyped(LastComposedWord.NOT_A_SEPARATOR); mExpectingUpdateSelection = true; // TODO: Can we remove this? mSpaceState = SPACE_STATE_PHANTOM; @@ -1382,40 +1411,102 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen mWordComposer.setCapitalizedModeAtStartComposingTime(getActualCapsMode()); } - @Override - public void onUpdateBatchInput(InputPointers batchPointers) { - mWordComposer.setBatchInputPointers(batchPointers); - final SuggestedWords suggestedWords = getSuggestedWords(); - showSuggestionStrip(suggestedWords, null); - final String gestureFloatingPreviewText = (suggestedWords.size() > 0) + private static final class BatchInputUpdater implements Handler.Callback { + private final Handler mHandler; + private LatinIME mLatinIme; + + private BatchInputUpdater() { + final HandlerThread handlerThread = new HandlerThread( + BatchInputUpdater.class.getSimpleName()); + handlerThread.start(); + mHandler = new Handler(handlerThread.getLooper(), this); + } + + // Initialization-on-demand holder + private static final class OnDemandInitializationHolder { + public static final BatchInputUpdater sInstance = new BatchInputUpdater(); + } + + public static BatchInputUpdater getInstance() { + return OnDemandInitializationHolder.sInstance; + } + + private static final int MSG_UPDATE_GESTURE_PREVIEW_AND_SUGGESTION_STRIP = 1; + + @Override + public boolean handleMessage(final Message msg) { + switch (msg.what) { + case MSG_UPDATE_GESTURE_PREVIEW_AND_SUGGESTION_STRIP: + final SuggestedWords suggestedWords = getSuggestedWordsGesture( + (InputPointers)msg.obj, mLatinIme); + showGesturePreviewAndSuggestionStrip( + suggestedWords, false /* dismissGestureFloatingPreviewText */, mLatinIme); + break; + } + return true; + } + + public void updateGesturePreviewAndSuggestionStrip(final InputPointers batchPointers, + final LatinIME latinIme) { + mLatinIme = latinIme; + if (mHandler.hasMessages(MSG_UPDATE_GESTURE_PREVIEW_AND_SUGGESTION_STRIP)) { + return; + } + mHandler.obtainMessage( + MSG_UPDATE_GESTURE_PREVIEW_AND_SUGGESTION_STRIP, batchPointers) + .sendToTarget(); + } + + public void showGesturePreviewAndSuggestionStrip(final SuggestedWords suggestedWords, + final boolean dismissGestureFloatingPreviewText, final LatinIME latinIme) { + latinIme.mHandler.showGesturePreviewAndSuggestionStrip( + suggestedWords, dismissGestureFloatingPreviewText); + } + + // {@link LatinIME#getSuggestedWords(int)} method calls with same session id have to + // be synchronized. + public synchronized SuggestedWords getSuggestedWordsGesture( + final InputPointers batchPointers, final LatinIME latinIme) { + latinIme.mWordComposer.setBatchInputPointers(batchPointers); + return latinIme.getSuggestedWords(Suggest.SESSION_GESTURE); + } + } + + private void showGesturePreviewAndSuggestionStrip(final SuggestedWords suggestedWords, + final boolean dismissGestureFloatingPreviewText) { + final String batchInputText = (suggestedWords.size() > 0) ? suggestedWords.getWord(0) : null; - mKeyboardSwitcher.getMainKeyboardView() - .showGestureFloatingPreviewText(gestureFloatingPreviewText); + final KeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); + mainKeyboardView.showGestureFloatingPreviewText(batchInputText); + showSuggestionStrip(suggestedWords, null); + if (dismissGestureFloatingPreviewText) { + mainKeyboardView.dismissGestureFloatingPreviewText(); + } } @Override - public void onEndBatchInput(InputPointers batchPointers) { - mWordComposer.setBatchInputPointers(batchPointers); - final SuggestedWords suggestedWords = getSuggestedWords(); - showSuggestionStrip(suggestedWords, null); - final String gestureFloatingPreviewText = (suggestedWords.size() > 0) + public void onUpdateBatchInput(final InputPointers batchPointers) { + BatchInputUpdater.getInstance().updateGesturePreviewAndSuggestionStrip(batchPointers, this); + } + + @Override + public void onEndBatchInput(final InputPointers batchPointers) { + final BatchInputUpdater batchInputUpdater = BatchInputUpdater.getInstance(); + final SuggestedWords suggestedWords = batchInputUpdater.getSuggestedWordsGesture( + batchPointers, this); + batchInputUpdater.showGesturePreviewAndSuggestionStrip( + suggestedWords, true /* dismissGestureFloatingPreviewText */, this); + final String batchInputText = (suggestedWords.size() > 0) ? suggestedWords.getWord(0) : null; - final MainKeyboardView mainKeyboardView = mKeyboardSwitcher.getMainKeyboardView(); - mainKeyboardView.showGestureFloatingPreviewText(gestureFloatingPreviewText); - mainKeyboardView.dismissGestureFloatingPreviewText(); - if (suggestedWords == null || suggestedWords.size() == 0) { + if (TextUtils.isEmpty(batchInputText)) { return; } - final CharSequence text = suggestedWords.getWord(0); - if (TextUtils.isEmpty(text)) { - return; - } - mWordComposer.setBatchInputWord(text); + mWordComposer.setBatchInputWord(batchInputText); mConnection.beginBatchEdit(); if (SPACE_STATE_PHANTOM == mSpaceState) { sendKeyCodePoint(Keyboard.CODE_SPACE); } - mConnection.setComposingText(text, 1); + mConnection.setComposingText(batchInputText, 1); mExpectingUpdateSelection = true; mConnection.endBatchEdit(); mKeyboardSwitcher.updateShiftState(); @@ -1451,18 +1542,6 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // In many cases, we may have to put the keyboard in auto-shift state again. mHandler.postUpdateShiftState(); - if (mEnteredText != null && mConnection.sameAsTextBeforeCursor(mEnteredText)) { - // Cancel multi-character input: remove the text we just entered. - // This is triggered on backspace after a key that inputs multiple characters, - // like the smiley key or the .com key. - final int length = mEnteredText.length(); - mConnection.deleteSurroundingText(length, 0); - // If we have mEnteredText, then we know that mHasUncommittedTypedChars == false. - // In addition we know that spaceState is false, and that we should not be - // reverting any autocorrect at this point. So we can safely return. - return; - } - if (mWordComposer.isComposingWord()) { final int length = mWordComposer.size(); if (length > 0) { @@ -1483,6 +1562,18 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen revertCommit(); return; } + if (mEnteredText != null && mConnection.sameAsTextBeforeCursor(mEnteredText)) { + // Cancel multi-character input: remove the text we just entered. + // This is triggered on backspace after a key that inputs multiple characters, + // like the smiley key or the .com key. + final int length = mEnteredText.length(); + mConnection.deleteSurroundingText(length, 0); + mEnteredText = null; + // If we have mEnteredText, then we know that mHasUncommittedTypedChars == false. + // In addition we know that spaceState is false, and that we should not be + // reverting any autocorrect at this point. So we can safely return. + return; + } if (SPACE_STATE_DOUBLE == spaceState) { mHandler.cancelDoubleSpacesTimer(); if (mConnection.revertDoubleSpace()) { @@ -1518,7 +1609,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // a hardware keyboard event on pressing enter or delete. This is bad for many // reasons (there are race conditions with commits) but some applications are // relying on this behavior so we continue to support it for older apps. - sendUpDownEnterOrBackspace(KeyEvent.KEYCODE_DEL); + sendDownUpKeyEventForBackwardCompatibility(KeyEvent.KEYCODE_DEL); } else { mConnection.deleteSurroundingText(1, 0); } @@ -1626,10 +1717,11 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // Handle separator if (mWordComposer.isComposingWord()) { if (mCurrentSettings.mCorrectionEnabled) { - commitCurrentAutoCorrection(primaryCode); + // TODO: maybe cache Strings in an <String> sparse array or something + commitCurrentAutoCorrection(new String(new int[]{primaryCode}, 0, 1)); didAutoCorrect = true; } else { - commitTyped(primaryCode); + commitTyped(new String(new int[]{primaryCode}, 0, 1)); } } @@ -1660,7 +1752,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen swapSwapperAndSpace(); mSpaceState = SPACE_STATE_SWAP_PUNCTUATION; } else if (SPACE_STATE_PHANTOM == spaceState - && !mCurrentSettings.isWeakSpaceStripper(primaryCode)) { + && !mCurrentSettings.isWeakSpaceStripper(primaryCode) + && !mCurrentSettings.isPhantomSpacePromotingSymbol(primaryCode)) { // If we are in phantom space state, and the user presses a separator, we want to // stay in phantom space state so that the next keypress has a chance to add the // space. For example, if I type "Good dat", pick "day" from the suggestion strip @@ -1761,12 +1854,12 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen return; } - final SuggestedWords suggestedWords = getSuggestedWords(); + final SuggestedWords suggestedWords = getSuggestedWords(Suggest.SESSION_TYPING); final String typedWord = mWordComposer.getTypedWord(); showSuggestionStrip(suggestedWords, typedWord); } - private SuggestedWords getSuggestedWords() { + private SuggestedWords getSuggestedWords(final int sessionId) { final String typedWord = mWordComposer.getTypedWord(); // Get the word on which we should search the bigrams. If we are composing a word, it's // whatever is *before* the half-committed word in the buffer, hence 2; if we aren't, we @@ -1777,7 +1870,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen mWordComposer.isComposingWord() ? 2 : 1); final SuggestedWords suggestedWords = mSuggest.getSuggestedWords(mWordComposer, prevWord, mKeyboardSwitcher.getKeyboard().getProximityInfo(), - mCurrentSettings.mCorrectionEnabled); + mCurrentSettings.mCorrectionEnabled, sessionId); return maybeRetrieveOlderSuggestions(typedWord, suggestedWords); } @@ -1834,7 +1927,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen setSuggestionStripShown(isSuggestionsStripVisible()); } - private void commitCurrentAutoCorrection(final int separatorCodePoint) { + private void commitCurrentAutoCorrection(final String separatorString) { // Complete any pending suggestions query first if (mHandler.hasPendingUpdateSuggestions()) { updateSuggestionStrip(); @@ -1848,13 +1941,17 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen throw new RuntimeException("We have an auto-correction but the typed word " + "is empty? Impossible! I must commit suicide."); } - Utils.Stats.onAutoCorrection(typedWord, autoCorrection.toString(), separatorCodePoint); + Utils.Stats.onAutoCorrection(typedWord, autoCorrection.toString(), separatorString); mExpectingUpdateSelection = true; commitChosenWord(autoCorrection, LastComposedWord.COMMIT_TYPE_DECIDED_WORD, - separatorCodePoint); + separatorString); if (!typedWord.equals(autoCorrection)) { // This will make the correction flash for a short while as a visual clue - // to the user that auto-correction happened. + // to the user that auto-correction happened. It has no other effect; in particular + // note that this won't affect the text inside the text field AT ALL: it only makes + // the segment of text starting at the supplied index and running for the length + // of the auto-correction flash. At this moment, the "typedWord" argument is + // ignored by TextView. mConnection.commitCorrection( new CorrectionInfo(mLastSelectionEnd - typedWord.length(), typedWord, autoCorrection)); @@ -1949,7 +2046,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen * Commits the chosen word to the text field and saves it for later retrieval. */ private void commitChosenWord(final CharSequence chosenWord, final int commitType, - final int separatorCode) { + final String separatorString) { final SuggestedWords suggestedWords = mSuggestionStripView.getSuggestions(); mConnection.commitText(SuggestionSpanUtils.getTextWithSuggestionSpan( this, chosenWord, suggestedWords, mIsMainDictionaryAvailable), 1); @@ -1960,7 +2057,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // LastComposedWord#didCommitTypedWord by string equality of the remembered // strings. mLastComposedWord = mWordComposer.commitWord(commitType, chosenWord.toString(), - separatorCode, prevWord); + separatorString, prevWord); } private void setPunctuationSuggestions() { @@ -2030,7 +2127,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen final CharSequence committedWord = mLastComposedWord.mCommittedWord; final int cancelLength = committedWord.length(); final int separatorLength = LastComposedWord.getSeparatorLength( - mLastComposedWord.mSeparatorCode); + mLastComposedWord.mSeparatorString); // TODO: should we check our saved separator against the actual contents of the text view? final int deleteLength = cancelLength + separatorLength; if (DEBUG) { @@ -2051,10 +2148,8 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen mUserHistoryDictionary.cancelAddingUserHistory( previousWord.toString(), committedWord.toString()); } - mConnection.commitText(originallyTypedWord, 1); - // Re-insert the separator - sendKeyCodePoint(mLastComposedWord.mSeparatorCode); - Utils.Stats.onSeparator(mLastComposedWord.mSeparatorCode, + mConnection.commitText(originallyTypedWord + mLastComposedWord.mSeparatorString, 1); + Utils.Stats.onSeparator(mLastComposedWord.mSeparatorString, Constants.NOT_A_COORDINATE, Constants.NOT_A_COORDINATE); if (ProductionFlag.IS_EXPERIMENTAL) { ResearchLogger.latinIME_revertCommit(originallyTypedWord); @@ -2067,7 +2162,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen } // Used by the RingCharBuffer - public boolean isWordSeparator(int code) { + public boolean isWordSeparator(final int code) { return mCurrentSettings.isWordSeparator(code); } @@ -2099,14 +2194,14 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // Callback called by PointerTracker through the KeyboardActionListener. This is called when a // key is depressed; release matching call is onReleaseKey below. @Override - public void onPressKey(int primaryCode) { + public void onPressKey(final int primaryCode) { mKeyboardSwitcher.onPressKey(primaryCode); } // Callback by PointerTracker through the KeyboardActionListener. This is called when a key // is released; press matching call is onPressKey above. @Override - public void onReleaseKey(int primaryCode, boolean withSliding) { + public void onReleaseKey(final int primaryCode, final boolean withSliding) { mKeyboardSwitcher.onReleaseKey(primaryCode, withSliding); // If accessibility is on, ensure the user receives keyboard state updates. @@ -2135,7 +2230,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen // receive ringer mode change and network state change. private BroadcastReceiver mReceiver = new BroadcastReceiver() { @Override - public void onReceive(Context context, Intent intent) { + public void onReceive(final Context context, final Intent intent) { final String action = intent.getAction(); if (action.equals(ConnectivityManager.CONNECTIVITY_ACTION)) { mSubtypeSwitcher.onNetworkStateChanged(intent); @@ -2156,14 +2251,14 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen launchSubActivity(DebugSettingsActivity.class); } - public void launchKeyboardedDialogActivity(Class<? extends Activity> activityClass) { + public void launchKeyboardedDialogActivity(final Class<? extends Activity> activityClass) { // Put the text in the attached EditText into a safe, saved state before switching to a // new activity that will also use the soft keyboard. commitTyped(LastComposedWord.NOT_A_SEPARATOR); launchSubActivity(activityClass); } - private void launchSubActivity(Class<? extends Activity> activityClass) { + private void launchSubActivity(final Class<? extends Activity> activityClass) { Intent intent = new Intent(); intent.setClass(LatinIME.this, activityClass); intent.setFlags(Intent.FLAG_ACTIVITY_NEW_TASK); @@ -2203,7 +2298,7 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen showOptionDialog(builder.create()); } - public void showOptionDialog(AlertDialog dialog) { + public void showOptionDialog(final AlertDialog dialog) { final IBinder windowToken = mKeyboardSwitcher.getMainKeyboardView().getWindowToken(); if (windowToken == null) { return; @@ -2223,8 +2318,19 @@ public class LatinIME extends InputMethodService implements KeyboardActionListen dialog.show(); } + public void debugDumpStateAndCrashWithException(final String context) { + final StringBuilder s = new StringBuilder(); + s.append("Target application : ").append(mTargetApplicationInfo.name) + .append("\nPackage : ").append(mTargetApplicationInfo.packageName) + .append("\nTarget app sdk version : ") + .append(mTargetApplicationInfo.targetSdkVersion) + .append("\nAttributes : ").append(mCurrentSettings.getInputAttributesDebugString()) + .append("\nContext : ").append(context); + throw new RuntimeException(s.toString()); + } + @Override - protected void dump(FileDescriptor fd, PrintWriter fout, String[] args) { + protected void dump(final FileDescriptor fd, final PrintWriter fout, final String[] args) { super.dump(fd, fout, args); final Printer p = new PrintWriterPrinter(fout); diff --git a/java/src/com/android/inputmethod/latin/LocaleUtils.java b/java/src/com/android/inputmethod/latin/LocaleUtils.java index 3b08cab01..feb1b2d0e 100644 --- a/java/src/com/android/inputmethod/latin/LocaleUtils.java +++ b/java/src/com/android/inputmethod/latin/LocaleUtils.java @@ -31,7 +31,10 @@ import java.util.Locale; * update/bugfix to this file, consider also updating/fixing the version in the * dictionary pack. */ -public class LocaleUtils { +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. } @@ -219,4 +222,38 @@ public 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/ResourceUtils.java b/java/src/com/android/inputmethod/latin/ResourceUtils.java new file mode 100644 index 000000000..5021ad384 --- /dev/null +++ b/java/src/com/android/inputmethod/latin/ResourceUtils.java @@ -0,0 +1,128 @@ +/* + * 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; + +import android.content.res.Resources; +import android.content.res.TypedArray; +import android.os.Build; +import android.util.TypedValue; + +import java.util.HashMap; + +public final class ResourceUtils { + public static final float UNDEFINED_RATIO = -1.0f; + public static final int UNDEFINED_DIMENSION = -1; + + private ResourceUtils() { + // This utility class is not publicly instantiable. + } + + private static final String HARDWARE_PREFIX = Build.HARDWARE + ","; + private static final HashMap<String, String> sDeviceOverrideValueMap = + CollectionUtils.newHashMap(); + + public static String getDeviceOverrideValue(Resources res, int overrideResId, String defValue) { + final int orientation = res.getConfiguration().orientation; + final String key = overrideResId + "-" + orientation; + if (!sDeviceOverrideValueMap.containsKey(key)) { + String overrideValue = defValue; + for (final String element : res.getStringArray(overrideResId)) { + if (element.startsWith(HARDWARE_PREFIX)) { + overrideValue = element.substring(HARDWARE_PREFIX.length()); + break; + } + } + sDeviceOverrideValueMap.put(key, overrideValue); + } + return sDeviceOverrideValueMap.get(key); + } + + public static boolean isValidFraction(final float fraction) { + return fraction >= 0.0f; + } + + // {@link Resources#getDimensionPixelSize(int)} returns at least one pixel size. + public static boolean isValidDimensionPixelSize(final int dimension) { + return dimension > 0; + } + + // {@link Resources#getDimensionPixelOffset(int)} may return zero pixel offset. + public static boolean isValidDimensionPixelOffset(final int dimension) { + return dimension >= 0; + } + + public static float getFraction(final TypedArray a, final int index, final float defValue) { + final TypedValue value = a.peekValue(index); + if (value == null || !isFractionValue(value)) { + return defValue; + } + return a.getFraction(index, 1, 1, defValue); + } + + public static float getFraction(final TypedArray a, final int index) { + return getFraction(a, index, UNDEFINED_RATIO); + } + + public static int getDimensionPixelSize(final TypedArray a, final int index) { + final TypedValue value = a.peekValue(index); + if (value == null || !isDimensionValue(value)) { + return ResourceUtils.UNDEFINED_DIMENSION; + } + return a.getDimensionPixelSize(index, ResourceUtils.UNDEFINED_DIMENSION); + } + + public static float getDimensionOrFraction(TypedArray a, int index, int base, + float defValue) { + final TypedValue value = a.peekValue(index); + if (value == null) { + return defValue; + } + if (isFractionValue(value)) { + return a.getFraction(index, base, base, defValue); + } else if (isDimensionValue(value)) { + return a.getDimension(index, defValue); + } + return defValue; + } + + public static int getEnumValue(TypedArray a, int index, int defValue) { + final TypedValue value = a.peekValue(index); + if (value == null) { + return defValue; + } + if (isIntegerValue(value)) { + return a.getInt(index, defValue); + } + return defValue; + } + + public static boolean isFractionValue(TypedValue v) { + return v.type == TypedValue.TYPE_FRACTION; + } + + public static boolean isDimensionValue(TypedValue v) { + return v.type == TypedValue.TYPE_DIMENSION; + } + + public static boolean isIntegerValue(TypedValue v) { + return v.type >= TypedValue.TYPE_FIRST_INT && v.type <= TypedValue.TYPE_LAST_INT; + } + + public static boolean isStringValue(TypedValue v) { + return v.type == TypedValue.TYPE_STRING; + } +} diff --git a/java/src/com/android/inputmethod/latin/RichInputConnection.java b/java/src/com/android/inputmethod/latin/RichInputConnection.java index 41e59e92d..b85f9dcd7 100644 --- a/java/src/com/android/inputmethod/latin/RichInputConnection.java +++ b/java/src/com/android/inputmethod/latin/RichInputConnection.java @@ -30,19 +30,51 @@ import com.android.inputmethod.keyboard.Keyboard; import com.android.inputmethod.latin.define.ProductionFlag; import com.android.inputmethod.research.ResearchLogger; +import java.util.Locale; import java.util.regex.Pattern; /** - * Wrapper for InputConnection to simplify interaction + * Enrichment class for InputConnection to simplify interaction and add functionality. + * + * This class serves as a wrapper to be able to simply add hooks to any calls to the underlying + * InputConnection. It also keeps track of a number of things to avoid having to call upon IPC + * all the time to find out what text is in the buffer, when we need it to determine caps mode + * for example. */ public class RichInputConnection { private static final String TAG = RichInputConnection.class.getSimpleName(); private static final boolean DBG = false; + private static final boolean DEBUG_PREVIOUS_TEXT = false; // Provision for a long word pair and a separator private static final int LOOKBACK_CHARACTER_NUM = BinaryDictionary.MAX_WORD_LENGTH * 2 + 1; private static final Pattern spaceRegex = Pattern.compile("\\s+"); private static final int INVALID_CURSOR_POSITION = -1; + /** + * This variable contains the value LatinIME thinks the cursor position should be at now. + * This is a few steps in advance of what the TextView thinks it is, because TextView will + * only know after the IPC calls gets through. + */ + private int mCurrentCursorPosition = 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. + */ + private StringBuilder mCommittedTextBeforeComposingText = new StringBuilder(); + /** + * This contains the currently composing text, as LatinIME thinks the TextView is seeing it. + */ + private StringBuilder mComposingText = new StringBuilder(); + /** + * This is a one-character string containing the character after the cursor. Since LatinIME + * never touches it directly, it's never modified by any means other than re-reading from the + * TextView when the cursor position is changed by the user. + */ + private CharSequence mCharAfterTheCursor = ""; + // A hint on how many characters to cache from the TextView. A good value of this is given by + // how many characters we need to be able to almost always find the caps mode. + private static final int DEFAULT_TEXT_CACHE_SIZE = 100; + private final InputMethodService mParent; InputConnection mIC; int mNestLevel; @@ -52,6 +84,37 @@ public class RichInputConnection { mNestLevel = 0; } + private void checkConsistencyForDebug() { + final ExtractedTextRequest r = new ExtractedTextRequest(); + r.hintMaxChars = 0; + r.hintMaxLines = 0; + r.token = 1; + r.flags = 0; + final ExtractedText et = mIC.getExtractedText(r, 0); + final CharSequence beforeCursor = getTextBeforeCursor(DEFAULT_TEXT_CACHE_SIZE, 0); + final StringBuilder internal = new StringBuilder().append(mCommittedTextBeforeComposingText) + .append(mComposingText); + if (null == et || null == beforeCursor) return; + final int actualLength = Math.min(beforeCursor.length(), internal.length()); + if (internal.length() > actualLength) { + internal.delete(0, internal.length() - actualLength); + } + final String reference = (beforeCursor.length() <= actualLength) ? beforeCursor.toString() + : beforeCursor.subSequence(beforeCursor.length() - actualLength, + beforeCursor.length()).toString(); + if (et.selectionStart != mCurrentCursorPosition + || !(reference.equals(internal.toString()))) { + final String context = "Expected cursor position = " + mCurrentCursorPosition + + "\nActual cursor position = " + et.selectionStart + + "\nExpected text = " + internal.length() + " " + internal + + "\nActual text = " + reference.length() + " " + reference; + ((LatinIME)mParent).debugDumpStateAndCrashWithException(context); + } else { + Log.e(TAG, Utils.getStackTrace(2)); + Log.e(TAG, "Exp <> Actual : " + mCurrentCursorPosition + " <> " + et.selectionStart); + } + } + public void beginBatchEdit() { if (++mNestLevel == 1) { mIC = mParent.getCurrentInputConnection(); @@ -65,12 +128,30 @@ public class RichInputConnection { Log.e(TAG, "Nest level too deep : " + mNestLevel); } } + checkBatchEdit(); + if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); } + public void endBatchEdit() { if (mNestLevel <= 0) Log.e(TAG, "Batch edit not in progress!"); // TODO: exception instead if (--mNestLevel == 0 && null != mIC) { mIC.endBatchEdit(); } + if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); + } + + public void resetCachesUponCursorMove(final int newCursorPosition) { + mCurrentCursorPosition = newCursorPosition; + mComposingText.setLength(0); + mCommittedTextBeforeComposingText.setLength(0); + mCommittedTextBeforeComposingText.append(getTextBeforeCursor(DEFAULT_TEXT_CACHE_SIZE, 0)); + mCharAfterTheCursor = getTextAfterCursor(1, 0); + if (null != mIC) { + mIC.finishComposingText(); + if (ProductionFlag.IS_EXPERIMENTAL) { + ResearchLogger.richInputConnection_finishComposingText(); + } + } } private void checkBatchEdit() { @@ -83,6 +164,10 @@ public class RichInputConnection { public void finishComposingText() { checkBatchEdit(); + if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); + mCommittedTextBeforeComposingText.append(mComposingText); + mCurrentCursorPosition += mComposingText.length(); + mComposingText.setLength(0); if (null != mIC) { mIC.finishComposingText(); if (ProductionFlag.IS_EXPERIMENTAL) { @@ -93,6 +178,10 @@ public class RichInputConnection { public void commitText(final CharSequence text, final int i) { checkBatchEdit(); + if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); + mCommittedTextBeforeComposingText.append(text); + mCurrentCursorPosition += text.length() - mComposingText.length(); + mComposingText.setLength(0); if (null != mIC) { mIC.commitText(text, i); if (ProductionFlag.IS_EXPERIMENTAL) { @@ -101,10 +190,22 @@ public class RichInputConnection { } } - public int getCursorCapsMode(final int inputType) { + public int getCursorCapsMode(final int inputType, final Locale locale) { mIC = mParent.getCurrentInputConnection(); if (null == mIC) return Constants.TextUtils.CAP_MODE_OFF; - return mIC.getCursorCapsMode(inputType); + if (!TextUtils.isEmpty(mComposingText)) return Constants.TextUtils.CAP_MODE_OFF; + // TODO: this will generally work, but there may be cases where the buffer contains SOME + // information but not enough to determine the caps mode accurately. This may happen after + // 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 != mCurrentCursorPosition) { + mCommittedTextBeforeComposingText.append( + getTextBeforeCursor(DEFAULT_TEXT_CACHE_SIZE, 0)); + } + // This never calls InputConnection#getCapsMode - in fact, it's a static method that + // never blocks or initiates IPC. + return StringUtils.getCapsMode(mCommittedTextBeforeComposingText, inputType, locale); } public CharSequence getTextBeforeCursor(final int i, final int j) { @@ -121,12 +222,28 @@ public class RichInputConnection { public void deleteSurroundingText(final int i, final int j) { checkBatchEdit(); + final int remainingChars = mComposingText.length() - i; + if (remainingChars >= 0) { + mComposingText.setLength(remainingChars); + } else { + mComposingText.setLength(0); + // Never cut under 0 + final int len = Math.max(mCommittedTextBeforeComposingText.length() + + remainingChars, 0); + mCommittedTextBeforeComposingText.setLength(len); + } + if (mCurrentCursorPosition > i) { + mCurrentCursorPosition -= i; + } else { + mCurrentCursorPosition = 0; + } if (null != mIC) { mIC.deleteSurroundingText(i, j); if (ProductionFlag.IS_EXPERIMENTAL) { ResearchLogger.richInputConnection_deleteSurroundingText(i, j); } } + if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); } public void performEditorAction(final int actionId) { @@ -141,6 +258,44 @@ public class RichInputConnection { public void sendKeyEvent(final KeyEvent keyEvent) { checkBatchEdit(); + if (keyEvent.getAction() == KeyEvent.ACTION_DOWN) { + if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); + // This method is only called for enter or backspace when speaking to old + // applications (target SDK <= 15), or for digits. + // When talking to new applications we never use this method because it's inherently + // racy and has unpredictable results, but for backward compatibility we continue + // sending the key events for only Enter and Backspace because some applications + // mistakenly catch them to do some stuff. + switch (keyEvent.getKeyCode()) { + case KeyEvent.KEYCODE_ENTER: + mCommittedTextBeforeComposingText.append("\n"); + mCurrentCursorPosition += 1; + break; + case KeyEvent.KEYCODE_DEL: + if (0 == mComposingText.length()) { + if (mCommittedTextBeforeComposingText.length() > 0) { + mCommittedTextBeforeComposingText.delete( + mCommittedTextBeforeComposingText.length() - 1, + mCommittedTextBeforeComposingText.length()); + } + } else { + mComposingText.delete(mComposingText.length() - 1, mComposingText.length()); + } + if (mCurrentCursorPosition > 0) mCurrentCursorPosition -= 1; + break; + case KeyEvent.KEYCODE_UNKNOWN: + if (null != keyEvent.getCharacters()) { + mCommittedTextBeforeComposingText.append(keyEvent.getCharacters()); + mCurrentCursorPosition += keyEvent.getCharacters().length(); + } + break; + default: + final String text = new String(new int[] { keyEvent.getUnicodeChar() }, 0, 1); + mCommittedTextBeforeComposingText.append(text); + mCurrentCursorPosition += text.length(); + break; + } + } if (null != mIC) { mIC.sendKeyEvent(keyEvent); if (ProductionFlag.IS_EXPERIMENTAL) { @@ -151,48 +306,83 @@ public class RichInputConnection { public void setComposingText(final CharSequence text, final int i) { checkBatchEdit(); + if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); + mCurrentCursorPosition += text.length() - mComposingText.length(); + mComposingText.setLength(0); + mComposingText.append(text); + // TODO: support values of i != 1. At this time, this is never called with i != 1. if (null != mIC) { mIC.setComposingText(text, i); if (ProductionFlag.IS_EXPERIMENTAL) { ResearchLogger.richInputConnection_setComposingText(text, i); } } + if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); } public void setSelection(final int from, final int to) { checkBatchEdit(); + if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); if (null != mIC) { mIC.setSelection(from, to); if (ProductionFlag.IS_EXPERIMENTAL) { ResearchLogger.richInputConnection_setSelection(from, to); } } + mCurrentCursorPosition = from; + mCommittedTextBeforeComposingText.setLength(0); + mCommittedTextBeforeComposingText.append(getTextBeforeCursor(DEFAULT_TEXT_CACHE_SIZE, 0)); } public void commitCorrection(final CorrectionInfo correctionInfo) { checkBatchEdit(); + if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); + // This has no effect on the text field and does not change its content. It only makes + // TextView flash the text for a second based on indices contained in the argument. if (null != mIC) { mIC.commitCorrection(correctionInfo); if (ProductionFlag.IS_EXPERIMENTAL) { ResearchLogger.richInputConnection_commitCorrection(correctionInfo); } } + if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); } public void commitCompletion(final CompletionInfo completionInfo) { checkBatchEdit(); + if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); + final CharSequence text = completionInfo.getText(); + mCommittedTextBeforeComposingText.append(text); + mCurrentCursorPosition += text.length() - mComposingText.length(); + mComposingText.setLength(0); if (null != mIC) { mIC.commitCompletion(completionInfo); if (ProductionFlag.IS_EXPERIMENTAL) { ResearchLogger.richInputConnection_commitCompletion(completionInfo); } } + if (DEBUG_PREVIOUS_TEXT) checkConsistencyForDebug(); } public CharSequence getNthPreviousWord(final String sentenceSeperators, final int n) { mIC = mParent.getCurrentInputConnection(); if (null == mIC) return null; final CharSequence prev = mIC.getTextBeforeCursor(LOOKBACK_CHARACTER_NUM, 0); + if (DEBUG_PREVIOUS_TEXT && null != prev) { + final int checkLength = LOOKBACK_CHARACTER_NUM - 1; + final String reference = prev.length() <= checkLength ? prev.toString() + : prev.subSequence(prev.length() - checkLength, prev.length()).toString(); + final StringBuilder internal = new StringBuilder() + .append(mCommittedTextBeforeComposingText).append(mComposingText); + if (internal.length() > checkLength) { + internal.delete(0, internal.length() - checkLength); + if (!(reference.equals(internal.toString()))) { + final String context = + "Expected text = " + internal + "\nActual text = " + reference; + ((LatinIME)mParent).debugDumpStateAndCrashWithException(context); + } + } + } return getNthPreviousWord(prev, sentenceSeperators, n); } @@ -452,4 +642,34 @@ public class RichInputConnection { commitText(" " + textBeforeCursor.subSequence(0, 1), 1); return true; } + + /** + * Heuristic to determine if this is an expected update of the cursor. + * + * Sometimes updates to the cursor position are late because of their asynchronous nature. + * This method tries to determine if this update is one, based on the values of the cursor + * position in the update, and the currently expected position of the cursor according to + * LatinIME's internal accounting. If this is not a belated expected update, then it should + * mean that the user moved the cursor explicitly. + * This is quite robust, but of course it's not perfect. In particular, it will fail in the + * case we get an update A, the user types in N characters so as to move the cursor to A+N but + * we don't get those, and then the user places the cursor between A and A+N, and we get only + * 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. + * @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 == mCurrentCursorPosition) return true; + // If this is an update that moves the cursor from our expected position, it must be + // an explicit move. + if (oldSelStart == mCurrentCursorPosition) 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) * (mCurrentCursorPosition - newSelStart) >= 0; + } } diff --git a/java/src/com/android/inputmethod/latin/SettingsValues.java b/java/src/com/android/inputmethod/latin/SettingsValues.java index dcd2532c1..5e9c870d4 100644 --- a/java/src/com/android/inputmethod/latin/SettingsValues.java +++ b/java/src/com/android/inputmethod/latin/SettingsValues.java @@ -35,7 +35,7 @@ import java.util.HashMap; * When you call the constructor of this class, you may want to change the current system locale by * using {@link LocaleUtils.RunInLocale}. */ -public class SettingsValues { +public final class SettingsValues { private static final String TAG = SettingsValues.class.getSimpleName(); private static final int SUGGESTION_VISIBILITY_SHOW_VALUE @@ -246,64 +246,65 @@ public class SettingsValues { && orientation == Configuration.ORIENTATION_PORTRAIT); } - public boolean isWordSeparator(int code) { + public boolean isWordSeparator(final int code) { return mWordSeparators.contains(String.valueOf((char)code)); } - public boolean isSymbolExcludedFromWordSeparators(int code) { + public boolean isSymbolExcludedFromWordSeparators(final int code) { return mSymbolsExcludedFromWordSeparators.contains(String.valueOf((char)code)); } - public boolean isWeakSpaceStripper(int code) { + public boolean isWeakSpaceStripper(final int code) { // TODO: this does not work if the code does not fit in a char return mWeakSpaceStrippers.contains(String.valueOf((char)code)); } - public boolean isWeakSpaceSwapper(int code) { + public boolean isWeakSpaceSwapper(final int code) { // TODO: this does not work if the code does not fit in a char return mWeakSpaceSwappers.contains(String.valueOf((char)code)); } - public boolean isPhantomSpacePromotingSymbol(int code) { + public boolean isPhantomSpacePromotingSymbol(final int code) { // TODO: this does not work if the code does not fit in a char return mPhantomSpacePromotingSymbols.contains(String.valueOf((char)code)); } - private static boolean isAutoCorrectEnabled(final Resources resources, + private static boolean isAutoCorrectEnabled(final Resources res, final String currentAutoCorrectionSetting) { - final String autoCorrectionOff = resources.getString( + final String autoCorrectionOff = res.getString( R.string.auto_correction_threshold_mode_index_off); return !currentAutoCorrectionSetting.equals(autoCorrectionOff); } // Public to access from KeyboardSwitcher. Should it have access to some // process-global instance instead? - public static boolean isKeyPreviewPopupEnabled(SharedPreferences sp, Resources resources) { - final boolean showPopupOption = resources.getBoolean( + public static boolean isKeyPreviewPopupEnabled(final SharedPreferences prefs, + final Resources res) { + final boolean showPopupOption = res.getBoolean( R.bool.config_enable_show_popup_on_keypress_option); - if (!showPopupOption) return resources.getBoolean(R.bool.config_default_popup_preview); - return sp.getBoolean(Settings.PREF_POPUP_ON, - resources.getBoolean(R.bool.config_default_popup_preview)); + if (!showPopupOption) return res.getBoolean(R.bool.config_default_popup_preview); + return prefs.getBoolean(Settings.PREF_POPUP_ON, + res.getBoolean(R.bool.config_default_popup_preview)); } // Likewise - public static int getKeyPreviewPopupDismissDelay(SharedPreferences sp, - Resources resources) { + public static int getKeyPreviewPopupDismissDelay(final SharedPreferences prefs, + final Resources res) { // TODO: use mKeyPreviewPopupDismissDelayRawValue instead of reading it again here. - return Integer.parseInt(sp.getString(Settings.PREF_KEY_PREVIEW_POPUP_DISMISS_DELAY, - Integer.toString(resources.getInteger( + return Integer.parseInt(prefs.getString(Settings.PREF_KEY_PREVIEW_POPUP_DISMISS_DELAY, + Integer.toString(res.getInteger( R.integer.config_key_preview_linger_timeout)))); } - private static boolean isBigramPredictionEnabled(final SharedPreferences sp, - final Resources resources) { - return sp.getBoolean(Settings.PREF_BIGRAM_PREDICTIONS, resources.getBoolean( + private static boolean isBigramPredictionEnabled(final SharedPreferences prefs, + final Resources res) { + return prefs.getBoolean(Settings.PREF_BIGRAM_PREDICTIONS, res.getBoolean( R.bool.config_default_next_word_prediction)); } - private static float getAutoCorrectionThreshold(final Resources resources, + private static float getAutoCorrectionThreshold(final Resources res, final String currentAutoCorrectionSetting) { - final String[] autoCorrectionThresholdValues = resources.getStringArray( + final String[] autoCorrectionThresholdValues = res.getStringArray( R.array.auto_correction_threshold_values); // When autoCorrectionThreshold is greater than 1.0, it's like auto correction is off. float autoCorrectionThreshold = Float.MAX_VALUE; @@ -335,11 +336,11 @@ public class SettingsValues { return mVoiceKeyOnMain; } - public static boolean isLanguageSwitchKeySupressed(SharedPreferences sp) { - return sp.getBoolean(Settings.PREF_SUPPRESS_LANGUAGE_SWITCH_KEY, false); + public static boolean isLanguageSwitchKeySupressed(final SharedPreferences prefs) { + return prefs.getBoolean(Settings.PREF_SUPPRESS_LANGUAGE_SWITCH_KEY, false); } - public boolean isLanguageSwitchKeyEnabled(Context context) { + public boolean isLanguageSwitchKeyEnabled(final Context context) { if (mIsLanguageSwitchKeySuppressed) { return false; } @@ -352,7 +353,7 @@ public class SettingsValues { } } - public boolean isFullscreenModeAllowed(Resources res) { + public boolean isFullscreenModeAllowed(final Resources res) { return res.getBoolean(R.bool.config_use_fullscreen_mode); } @@ -362,34 +363,35 @@ public class SettingsValues { public static String getPrefAdditionalSubtypes(final SharedPreferences prefs, final Resources res) { - final String prefSubtypes = res.getString(R.string.predefined_subtypes, ""); - return prefs.getString(Settings.PREF_CUSTOM_INPUT_STYLES, prefSubtypes); + final String predefinedPrefSubtypes = AdditionalSubtype.createPrefSubtypes( + res.getStringArray(R.array.predefined_subtypes)); + return prefs.getString(Settings.PREF_CUSTOM_INPUT_STYLES, predefinedPrefSubtypes); } // Accessed from the settings interface, hence public - public static float getCurrentKeypressSoundVolume(final SharedPreferences sp, - final Resources res) { + public static float getCurrentKeypressSoundVolume(final SharedPreferences prefs, + final Resources res) { // TODO: use mVibrationDurationSettingsRawValue instead of reading it again here - final float volume = sp.getFloat(Settings.PREF_KEYPRESS_SOUND_VOLUME, -1.0f); + final float volume = prefs.getFloat(Settings.PREF_KEYPRESS_SOUND_VOLUME, -1.0f); if (volume >= 0) { return volume; } - return Float.parseFloat( - Utils.getDeviceOverrideValue(res, R.array.keypress_volumes, "-1.0f")); + return Float.parseFloat(ResourceUtils.getDeviceOverrideValue( + res, R.array.keypress_volumes, "-1.0f")); } // Likewise - public static int getCurrentVibrationDuration(final SharedPreferences sp, - final Resources res) { + public static int getCurrentVibrationDuration(final SharedPreferences prefs, + final Resources res) { // TODO: use mKeypressVibrationDuration instead of reading it again here - final int ms = sp.getInt(Settings.PREF_VIBRATION_DURATION_SETTINGS, -1); + final int ms = prefs.getInt(Settings.PREF_VIBRATION_DURATION_SETTINGS, -1); if (ms >= 0) { return ms; } - return Integer.parseInt( - Utils.getDeviceOverrideValue(res, R.array.keypress_vibration_durations, "-1")); + return Integer.parseInt(ResourceUtils.getDeviceOverrideValue( + res, R.array.keypress_vibration_durations, "-1")); } // Likewise @@ -398,22 +400,22 @@ public class SettingsValues { return prefs.getBoolean(Settings.PREF_USABILITY_STUDY_MODE, true); } - public static long getLastUserHistoryWriteTime( - final SharedPreferences prefs, final String locale) { + public static long getLastUserHistoryWriteTime(final SharedPreferences prefs, + final String locale) { final String str = prefs.getString(Settings.PREF_LAST_USER_DICTIONARY_WRITE_TIME, ""); - final HashMap<String, Long> map = Utils.localeAndTimeStrToHashMap(str); + final HashMap<String, Long> map = LocaleUtils.localeAndTimeStrToHashMap(str); if (map.containsKey(locale)) { return map.get(locale); } return 0; } - public static void setLastUserHistoryWriteTime( - final SharedPreferences prefs, final String locale) { + public static void setLastUserHistoryWriteTime(final SharedPreferences prefs, + final String locale) { final String oldStr = prefs.getString(Settings.PREF_LAST_USER_DICTIONARY_WRITE_TIME, ""); - final HashMap<String, Long> map = Utils.localeAndTimeStrToHashMap(oldStr); + final HashMap<String, Long> map = LocaleUtils.localeAndTimeStrToHashMap(oldStr); map.put(locale, System.currentTimeMillis()); - final String newStr = Utils.localeAndTimeHashMapToStr(map); + final String newStr = LocaleUtils.localeAndTimeHashMapToStr(map); prefs.edit().putString(Settings.PREF_LAST_USER_DICTIONARY_WRITE_TIME, newStr).apply(); } diff --git a/java/src/com/android/inputmethod/latin/StringUtils.java b/java/src/com/android/inputmethod/latin/StringUtils.java index 39c59b44c..6dc1ea807 100644 --- a/java/src/com/android/inputmethod/latin/StringUtils.java +++ b/java/src/com/android/inputmethod/latin/StringUtils.java @@ -18,10 +18,12 @@ package com.android.inputmethod.latin; import android.text.TextUtils; +import com.android.inputmethod.keyboard.Keyboard; // For character constants + import java.util.ArrayList; import java.util.Locale; -public class StringUtils { +public final class StringUtils { private StringUtils() { // This utility class is not publicly instantiable. } @@ -123,23 +125,6 @@ public class StringUtils { } /** - * Returns true if cs contains any upper case characters. - * - * @param cs the CharSequence to check - * @return {@code true} if cs contains any upper case characters, {@code false} otherwise. - */ - public static boolean hasUpperCase(final CharSequence cs) { - final int length = cs.length(); - for (int i = 0, cp = 0; i < length; i += Character.charCount(cp)) { - cp = Character.codePointAt(cs, i); - if (Character.isUpperCase(cp)) { - return true; - } - } - return false; - } - - /** * Remove duplicates from an array of strings. * * This method will always keep the first occurrence of all strings at their position @@ -197,4 +182,200 @@ public class StringUtils { codePoints[dsti] = codePoint; return codePoints; } + + /** + * Determine what caps mode should be in effect at the current offset in + * the text. Only the mode bits set in <var>reqModes</var> will be + * checked. Note that the caps mode flags here are explicitly defined + * to match those in {@link InputType}. + * + * This code is a straight copy of TextUtils.getCapsMode (modulo namespace and formatting + * issues). This will change in the future as we simplify the code for our use and fix bugs. + * + * @param cs The text that should be checked for caps modes. + * @param reqModes The modes to be checked: may be any combination of + * {@link TextUtils#CAP_MODE_CHARACTERS}, {@link TextUtils#CAP_MODE_WORDS}, and + * {@link TextUtils#CAP_MODE_SENTENCES}. + * @param locale The locale to consider for capitalization rules + * + * @return Returns the actual capitalization modes that can be in effect + * at the current position, which is any combination of + * {@link TextUtils#CAP_MODE_CHARACTERS}, {@link TextUtils#CAP_MODE_WORDS}, and + * {@link TextUtils#CAP_MODE_SENTENCES}. + */ + public static int getCapsMode(final CharSequence cs, final int reqModes, final Locale locale) { + // Quick description of what we want to do: + // CAP_MODE_CHARACTERS is always on. + // CAP_MODE_WORDS is on if there is some whitespace before the cursor. + // CAP_MODE_SENTENCES is on if there is some whitespace before the cursor, and the end + // of a sentence just before that. + // We ignore opening parentheses and the like just before the cursor for purposes of + // finding whitespace for WORDS and SENTENCES modes. + // The end of a sentence ends with a period, question mark or exclamation mark. If it's + // a period, it also needs not to be an abbreviation, which means it also needs to either + // be immediately preceded by punctuation, or by a string of only letters with single + // periods interleaved. + + // Step 1 : check for cap MODE_CHARACTERS. If it's looked for, it's always on. + if ((reqModes & (TextUtils.CAP_MODE_WORDS | TextUtils.CAP_MODE_SENTENCES)) == 0) { + // Here we are not looking for MODE_WORDS or MODE_SENTENCES, so since we already + // evaluated MODE_CHARACTERS, we can return. + return TextUtils.CAP_MODE_CHARACTERS & reqModes; + } + + // Step 2 : Skip (ignore at the end of input) any opening punctuation. This includes + // opening parentheses, brackets, opening quotes, everything that *opens* a span of + // text in the linguistic sense. In RTL languages, this is still an opening sign, although + // it may look like a right parenthesis for example. We also include double quote and + // single quote since they aren't start punctuation in the unicode sense, but should still + // be skipped for English. TODO: does this depend on the language? + int i; + for (i = cs.length(); i > 0; i--) { + final char c = cs.charAt(i - 1); + if (c != Keyboard.CODE_DOUBLE_QUOTE && c != Keyboard.CODE_SINGLE_QUOTE + && Character.getType(c) != Character.START_PUNCTUATION) { + break; + } + } + + // We are now on the character that precedes any starting punctuation, so in the most + // frequent case this will be whitespace or a letter, although it may occasionally be a + // start of line, or some symbol. + + // Step 3 : Search for the start of a paragraph. From the starting point computed in step 2, + // we go back over any space or tab char sitting there. We find the start of a paragraph + // if the first char that's not a space or tab is a start of line (as in, either \n or + // start of text). + int j = i; + while (j > 0 && Character.isWhitespace(cs.charAt(j - 1))) { + j--; + } + if (j == 0) { + // There is only whitespace between the start of the text and the cursor. Both + // MODE_WORDS and MODE_SENTENCES should be active. + return (TextUtils.CAP_MODE_CHARACTERS | TextUtils.CAP_MODE_WORDS + | TextUtils.CAP_MODE_SENTENCES) & reqModes; + } + if (i == j) { + // If we don't have whitespace before index i, it means neither MODE_WORDS + // nor mode sentences should be on so we can return right away. + return TextUtils.CAP_MODE_CHARACTERS & reqModes; + } + if ((reqModes & TextUtils.CAP_MODE_SENTENCES) == 0) { + // Here we know we have whitespace before the cursor (if not, we returned in the above + // if i == j clause), so we need MODE_WORDS to be on. And we don't need to evaluate + // MODE_SENTENCES so we can return right away. + return (TextUtils.CAP_MODE_CHARACTERS | TextUtils.CAP_MODE_WORDS) & reqModes; + } + // Please note that because of the reqModes & CAP_MODE_SENTENCES test a few lines above, + // we know that MODE_SENTENCES is being requested. + + // Step 4 : Search for MODE_SENTENCES. + // English is a special case in that "American typography" rules, which are the most common + // in English, state that a sentence terminator immediately following a quotation mark + // should be swapped with it and de-duplicated (included in the quotation mark), + // e.g. <<Did he say, "let's go home?">> + // No other language has such a rule as far as I know, instead putting inside the quotation + // mark as the exact thing quoted and handling the surrounding punctuation independently, + // e.g. <<Did he say, "let's go home"?>> + // Hence, specifically for English, we treat this special case here. + if (Locale.ENGLISH.getLanguage().equals(locale.getLanguage())) { + for (; j > 0; j--) { + // Here we look to go over any closing punctuation. This is because in dominant + // variants of English, the final period is placed within double quotes and maybe + // other closing punctuation signs. This is generally not true in other languages. + final char c = cs.charAt(j - 1); + if (c != Keyboard.CODE_DOUBLE_QUOTE && c != Keyboard.CODE_SINGLE_QUOTE + && Character.getType(c) != Character.END_PUNCTUATION) { + break; + } + } + } + + if (j <= 0) return TextUtils.CAP_MODE_CHARACTERS & reqModes; + char c = cs.charAt(--j); + + // We found the next interesting chunk of text ; next we need to determine if it's the + // end of a sentence. If we have a question mark or an exclamation mark, it's the end of + // a sentence. If it's neither, the only remaining case is the period so we get the opposite + // case out of the way. + if (c == Keyboard.CODE_QUESTION_MARK || c == Keyboard.CODE_EXCLAMATION_MARK) { + return (TextUtils.CAP_MODE_CHARACTERS | TextUtils.CAP_MODE_SENTENCES) & reqModes; + } + if (c != Keyboard.CODE_PERIOD || j <= 0) { + return (TextUtils.CAP_MODE_CHARACTERS | TextUtils.CAP_MODE_WORDS) & reqModes; + } + + // We found out that we have a period. We need to determine if this is a full stop or + // otherwise sentence-ending period, or an abbreviation like "e.g.". An abbreviation + // looks like (\w\.){2,} + // To find out, we will have a simple state machine with the following states : + // START, WORD, PERIOD, ABBREVIATION + // On START : (just before the first period) + // letter => WORD + // whitespace => end with no caps (it was a stand-alone period) + // otherwise => end with caps (several periods/symbols in a row) + // On WORD : (within the word just before the first period) + // letter => WORD + // period => PERIOD + // otherwise => end with caps (it was a word with a full stop at the end) + // On PERIOD : (period within a potential abbreviation) + // letter => LETTER + // otherwise => end with caps (it was not an abbreviation) + // On LETTER : (letter within a potential abbreviation) + // letter => LETTER + // period => PERIOD + // otherwise => end with no caps (it was an abbreviation) + // "Not an abbreviation" in the above chart essentially covers cases like "...yes.". This + // should capitalize. + + final int START = 0; + final int WORD = 1; + final int PERIOD = 2; + final int LETTER = 3; + final int caps = (TextUtils.CAP_MODE_CHARACTERS | TextUtils.CAP_MODE_WORDS + | TextUtils.CAP_MODE_SENTENCES) & reqModes; + final int noCaps = (TextUtils.CAP_MODE_CHARACTERS | TextUtils.CAP_MODE_WORDS) & reqModes; + int state = START; + while (j > 0) { + c = cs.charAt(--j); + switch (state) { + case START: + if (Character.isLetter(c)) { + state = WORD; + } else if (Character.isWhitespace(c)) { + return noCaps; + } else { + return caps; + } + break; + case WORD: + if (Character.isLetter(c)) { + state = WORD; + } else if (c == Keyboard.CODE_PERIOD) { + state = PERIOD; + } else { + return caps; + } + break; + case PERIOD: + if (Character.isLetter(c)) { + state = LETTER; + } else { + return caps; + } + break; + case LETTER: + if (Character.isLetter(c)) { + state = LETTER; + } else if (c == Keyboard.CODE_PERIOD) { + state = PERIOD; + } else { + return noCaps; + } + } + } + // Here we arrived at the start of the line. This should behave exactly like whitespace. + return (START == state || LETTER == state) ? noCaps : caps; + } } diff --git a/java/src/com/android/inputmethod/latin/Suggest.java b/java/src/com/android/inputmethod/latin/Suggest.java index 51ed09604..0418d3166 100644 --- a/java/src/com/android/inputmethod/latin/Suggest.java +++ b/java/src/com/android/inputmethod/latin/Suggest.java @@ -37,6 +37,11 @@ import java.util.concurrent.ConcurrentHashMap; public class Suggest { public static final String TAG = Suggest.class.getSimpleName(); + // Session id for + // {@link #getSuggestedWords(WordComposer,CharSequence,ProximityInfo,boolean,int)}. + public static final int SESSION_TYPING = 0; + public static final int SESSION_GESTURE = 1; + // TODO: rename this to CORRECTION_OFF public static final int CORRECTION_NONE = 0; // TODO: rename this to CORRECTION_ON @@ -157,13 +162,6 @@ public class Suggest { public SuggestedWords getSuggestedWords( final WordComposer wordComposer, CharSequence prevWordForBigram, - final ProximityInfo proximityInfo, final boolean isCorrectionEnabled) { - return getSuggestedWordsWithSessionId( - wordComposer, prevWordForBigram, proximityInfo, isCorrectionEnabled, 0); - } - - public SuggestedWords getSuggestedWordsWithSessionId( - final WordComposer wordComposer, CharSequence prevWordForBigram, final ProximityInfo proximityInfo, final boolean isCorrectionEnabled, int sessionId) { LatinImeLogger.onStartSuggestion(prevWordForBigram); if (wordComposer.isBatchMode()) { @@ -214,10 +212,12 @@ public class Suggest { whitelistedWord = suggestionsSet.first().mWord; } + // The word can be auto-corrected if it has a whitelist entry that is not itself, + // or if it's a 2+ characters non-word (i.e. it's not in the dictionary). final boolean allowsToBeAutoCorrected = (null != whitelistedWord && !whitelistedWord.equals(consideredWord)) - || AutoCorrection.isNotAWord(mDictionaries, consideredWord, - wordComposer.isFirstCharCapitalized()); + || (consideredWord.length() > 1 && !AutoCorrection.isInTheDictionary(mDictionaries, + consideredWord, wordComposer.isFirstCharCapitalized())); final boolean hasAutoCorrection; // TODO: using isCorrectionEnabled here is not very good. It's probably useless, because diff --git a/java/src/com/android/inputmethod/latin/SuggestedWords.java b/java/src/com/android/inputmethod/latin/SuggestedWords.java index 68ecfa0d7..d9f48c4a4 100644 --- a/java/src/com/android/inputmethod/latin/SuggestedWords.java +++ b/java/src/com/android/inputmethod/latin/SuggestedWords.java @@ -177,7 +177,7 @@ public class SuggestedWords { return; } int i = 1; - while(i < candidates.size()) { + while (i < candidates.size()) { final SuggestedWordInfo cur = candidates.get(i); for (int j = 0; j < i; ++j) { final SuggestedWordInfo previous = candidates.get(j); diff --git a/java/src/com/android/inputmethod/latin/UserHistoryDictIOUtils.java b/java/src/com/android/inputmethod/latin/UserHistoryDictIOUtils.java new file mode 100644 index 000000000..550e4e58b --- /dev/null +++ b/java/src/com/android/inputmethod/latin/UserHistoryDictIOUtils.java @@ -0,0 +1,199 @@ +/* + * 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; + +import android.util.Log; + +import com.android.inputmethod.latin.makedict.BinaryDictInputOutput; +import com.android.inputmethod.latin.makedict.BinaryDictInputOutput.FusionDictionaryBufferInterface; +import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions; +import com.android.inputmethod.latin.makedict.FusionDictionary; +import com.android.inputmethod.latin.makedict.FusionDictionary.Node; +import com.android.inputmethod.latin.makedict.PendingAttribute; +import com.android.inputmethod.latin.makedict.UnsupportedFormatException; + +import java.io.IOException; +import java.io.OutputStream; +import java.util.ArrayList; +import java.util.HashMap; +import java.util.Map; + +/** + * Reads and writes Binary files for a UserHistoryDictionary. + * + * All the methods in this class are static. + */ +public class UserHistoryDictIOUtils { + private static final String TAG = UserHistoryDictIOUtils.class.getSimpleName(); + private static final boolean DEBUG = false; + + public interface OnAddWordListener { + public void setUnigram(final String word, final String shortcutTarget, final int frequency); + public void setBigram(final String word1, final String word2, final int frequency); + } + + public interface BigramDictionaryInterface { + public int getFrequency(final String word1, final String word2); + } + + public static final class ByteArrayWrapper implements FusionDictionaryBufferInterface { + private byte[] mBuffer; + private int mPosition; + + public ByteArrayWrapper(final byte[] buffer) { + mBuffer = buffer; + mPosition = 0; + } + + @Override + public int readUnsignedByte() { + return ((int)mBuffer[mPosition++]) & 0xFF; + } + + @Override + public int readUnsignedShort() { + final int retval = readUnsignedByte(); + return (retval << 8) + readUnsignedByte(); + } + + @Override + public int readUnsignedInt24() { + final int retval = readUnsignedShort(); + return (retval << 8) + readUnsignedByte(); + } + + @Override + public int readInt() { + final int retval = readUnsignedShort(); + return (retval << 16) + readUnsignedShort(); + } + + @Override + public int position() { + return mPosition; + } + + @Override + public void position(int position) { + mPosition = position; + } + + @Override + public void put(final byte b) { + mBuffer[mPosition++] = b; + } + } + + /** + * Writes dictionary to file. + */ + public static void writeDictionaryBinary(final OutputStream destination, + final BigramDictionaryInterface dict, final UserHistoryDictionaryBigramList bigrams, + final FormatOptions formatOptions) { + + final FusionDictionary fusionDict = constructFusionDictionary(dict, bigrams); + + try { + BinaryDictInputOutput.writeDictionaryBinary(destination, fusionDict, formatOptions); + } catch (IOException e) { + Log.e(TAG, "IO exception while writing file: " + e); + } catch (UnsupportedFormatException e) { + Log.e(TAG, "Unsupported fomat: " + e); + } + } + + /** + * Constructs a new FusionDictionary from BigramDictionaryInterface. + */ + /* packages for test */ static FusionDictionary constructFusionDictionary( + final BigramDictionaryInterface dict, final UserHistoryDictionaryBigramList bigrams) { + + final FusionDictionary fusionDict = new FusionDictionary(new Node(), + new FusionDictionary.DictionaryOptions( + new HashMap<String,String>(), false, false)); + + for (final String word1 : bigrams.keySet()) { + final HashMap<String, Byte> word1Bigrams = bigrams.getBigrams(word1); + for (final String word2 : word1Bigrams.keySet()) { + final int freq = dict.getFrequency(word1, word2); + + if (DEBUG) { + if (word1 == null) { + Log.d(TAG, "add unigram: " + word2 + "," + Integer.toString(freq)); + } else { + Log.d(TAG, "add bigram: " + word1 + + "," + word2 + "," + Integer.toString(freq)); + } + } + + if (word1 == null) { // unigram + fusionDict.add(word2, freq, null, false /* isNotAWord */); + } else { // bigram + fusionDict.setBigram(word1, word2, freq); + } + bigrams.updateBigram(word1, word2, (byte)freq); + } + } + + return fusionDict; + } + + /** + * Reads dictionary from file. + */ + public static void readDictionaryBinary(final FusionDictionaryBufferInterface buffer, + final OnAddWordListener dict) { + final Map<Integer, String> unigrams = CollectionUtils.newTreeMap(); + final Map<Integer, Integer> frequencies = CollectionUtils.newTreeMap(); + final Map<Integer, ArrayList<PendingAttribute>> bigrams = CollectionUtils.newTreeMap(); + + try { + BinaryDictInputOutput.readUnigramsAndBigramsBinary(buffer, unigrams, frequencies, + bigrams); + addWordsFromWordMap(unigrams, frequencies, bigrams, dict); + } catch (IOException e) { + Log.e(TAG, "IO exception while reading file: " + e); + } catch (UnsupportedFormatException e) { + Log.e(TAG, "Unsupported format: " + e); + } + } + + /** + * Adds all unigrams and bigrams in maps to OnAddWordListener. + */ + /* package for test */ static void addWordsFromWordMap(final Map<Integer, String> unigrams, + final Map<Integer, Integer> frequencies, + final Map<Integer, ArrayList<PendingAttribute>> bigrams, final OnAddWordListener to) { + + for (Map.Entry<Integer, String> entry : unigrams.entrySet()) { + final String word1 = entry.getValue(); + final int unigramFrequency = frequencies.get(entry.getKey()); + to.setUnigram(word1, null, unigramFrequency); + + final ArrayList<PendingAttribute> attrList = bigrams.get(entry.getKey()); + + if (attrList != null) { + for (final PendingAttribute attr : attrList) { + to.setBigram(word1, unigrams.get(attr.mAddress), + BinaryDictInputOutput.reconstructBigramFrequency(unigramFrequency, + attr.mFrequency)); + } + } + } + + } +}
\ No newline at end of file diff --git a/java/src/com/android/inputmethod/latin/UserHistoryDictionary.java b/java/src/com/android/inputmethod/latin/UserHistoryDictionary.java index 6c9d1c250..683ee4f5c 100644 --- a/java/src/com/android/inputmethod/latin/UserHistoryDictionary.java +++ b/java/src/com/android/inputmethod/latin/UserHistoryDictionary.java @@ -182,6 +182,10 @@ public class UserHistoryDictionary extends ExpandableDictionary { * The second word may not be null (a NullPointerException would be thrown). */ public int addToUserHistory(final String word1, String word2, boolean isValid) { + if (word2.length() >= BinaryDictionary.MAX_WORD_LENGTH || + (word1 != null && word1.length() >= BinaryDictionary.MAX_WORD_LENGTH)) { + return -1; + } if (mBigramListLock.tryLock()) { try { super.addWord( diff --git a/java/src/com/android/inputmethod/latin/UserHistoryForgettingCurveUtils.java b/java/src/com/android/inputmethod/latin/UserHistoryForgettingCurveUtils.java index 5a2fdf48e..3d3bd980c 100644 --- a/java/src/com/android/inputmethod/latin/UserHistoryForgettingCurveUtils.java +++ b/java/src/com/android/inputmethod/latin/UserHistoryForgettingCurveUtils.java @@ -19,7 +19,7 @@ package com.android.inputmethod.latin; import android.text.format.DateUtils; import android.util.Log; -public class UserHistoryForgettingCurveUtils { +public final class UserHistoryForgettingCurveUtils { private static final String TAG = UserHistoryForgettingCurveUtils.class.getSimpleName(); private static final boolean DEBUG = false; private static final int FC_FREQ_MAX = 127; diff --git a/java/src/com/android/inputmethod/latin/Utils.java b/java/src/com/android/inputmethod/latin/Utils.java index fc7a42100..1c98b92cd 100644 --- a/java/src/com/android/inputmethod/latin/Utils.java +++ b/java/src/com/android/inputmethod/latin/Utils.java @@ -16,20 +16,16 @@ package com.android.inputmethod.latin; -import android.content.Context; import android.content.Intent; import android.content.pm.PackageManager; -import android.content.res.Resources; import android.inputmethodservice.InputMethodService; import android.net.Uri; import android.os.AsyncTask; -import android.os.Build; import android.os.Environment; import android.os.Handler; import android.os.HandlerThread; import android.os.Process; import android.text.TextUtils; -import android.text.format.DateUtils; import android.util.Log; import com.android.inputmethod.latin.SuggestedWords.SuggestedWordInfo; @@ -45,9 +41,8 @@ import java.io.PrintWriter; import java.nio.channels.FileChannel; import java.text.SimpleDateFormat; import java.util.Date; -import java.util.HashMap; -public class Utils { +public final class Utils { private Utils() { // This utility class is not publicly instantiable. } @@ -184,7 +179,7 @@ public class Utils { return getStackTrace(Integer.MAX_VALUE - 1); } - public static class UsabilityStudyLogUtils { + public static final class UsabilityStudyLogUtils { // TODO: remove code duplication with ResearchLog class private static final String USABILITY_TAG = UsabilityStudyLogUtils.class.getSimpleName(); private static final String FILENAME = "log.txt"; @@ -393,34 +388,38 @@ public class Utils { } } - public static float getDipScale(Context context) { - final float scale = context.getResources().getDisplayMetrics().density; - return scale; - } - - /** Convert pixel to DIP */ - public static int dipToPixel(float scale, int dip) { - return (int) (dip * scale + 0.5); - } - - public static class Stats { + public static final class Stats { public static void onNonSeparator(final char code, final int x, final int y) { RingCharBuffer.getInstance().push(code, x, y); LatinImeLogger.logOnInputChar(); } - public static void onSeparator(final int code, final int x, - final int y) { - // TODO: accept code points - RingCharBuffer.getInstance().push((char)code, x, y); + 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); + } + + public static void onSeparator(final String separator, final int x, final int y) { + final int length = separator.length(); + for (int i = 0; i < length; i = Character.offsetByCodePoints(separator, i, 1)) { + int codePoint = Character.codePointAt(separator, i); + // TODO: accept code points + RingCharBuffer.getInstance().push((char)codePoint, x, y); + } LatinImeLogger.logOnInputSeparator(); } public static void onAutoCorrection(final String typedWord, final String correctedWord, - final int separatorCode) { + final String separatorString) { if (TextUtils.isEmpty(typedWord)) return; - LatinImeLogger.logOnAutoCorrection(typedWord, correctedWord, separatorCode); + // TODO: this fails when the separator is more than 1 code point long, but + // the backend can't handle it yet. The only case when this happens is with + // smileys and other multi-character keys. + final int codePoint = TextUtils.isEmpty(separatorString) ? Constants.NOT_A_CODE + : separatorString.codePointAt(0); + LatinImeLogger.logOnAutoCorrection(typedWord, correctedWord, codePoint); } public static void onAutoCorrectionCancellation() { @@ -436,60 +435,4 @@ public class Utils { if (TextUtils.isEmpty(info)) return null; return info; } - - private static final String HARDWARE_PREFIX = Build.HARDWARE + ","; - private static final HashMap<String, String> sDeviceOverrideValueMap = - CollectionUtils.newHashMap(); - - public static String getDeviceOverrideValue(Resources res, int overrideResId, String defValue) { - final int orientation = res.getConfiguration().orientation; - final String key = overrideResId + "-" + orientation; - if (!sDeviceOverrideValueMap.containsKey(key)) { - String overrideValue = defValue; - for (final String element : res.getStringArray(overrideResId)) { - if (element.startsWith(HARDWARE_PREFIX)) { - overrideValue = element.substring(HARDWARE_PREFIX.length()); - break; - } - } - sDeviceOverrideValueMap.put(key, overrideValue); - } - return sDeviceOverrideValueMap.get(key); - } - - private static final HashMap<String, Long> EMPTY_LT_HASH_MAP = CollectionUtils.newHashMap(); - private static final String LOCALE_AND_TIME_STR_SEPARATER = ","; - 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/VibratorUtils.java b/java/src/com/android/inputmethod/latin/VibratorUtils.java index 33ffdd9c9..b6696cec0 100644 --- a/java/src/com/android/inputmethod/latin/VibratorUtils.java +++ b/java/src/com/android/inputmethod/latin/VibratorUtils.java @@ -19,7 +19,7 @@ package com.android.inputmethod.latin; import android.content.Context; import android.os.Vibrator; -public class VibratorUtils { +public final class VibratorUtils { private static final VibratorUtils sInstance = new VibratorUtils(); private Vibrator mVibrator; diff --git a/java/src/com/android/inputmethod/latin/WordComposer.java b/java/src/com/android/inputmethod/latin/WordComposer.java index ecec60f89..4b7adf26b 100644 --- a/java/src/com/android/inputmethod/latin/WordComposer.java +++ b/java/src/com/android/inputmethod/latin/WordComposer.java @@ -336,14 +336,14 @@ public class WordComposer { // `type' should be one of the LastComposedWord.COMMIT_TYPE_* constants above. public LastComposedWord commitWord(final int type, final String committedWord, - final int separatorCode, final CharSequence prevWord) { + final String separatorString, final CharSequence prevWord) { // Note: currently, we come here whenever we commit a word. If it's a MANUAL_PICK // or a DECIDED_WORD we may cancel the commit later; otherwise, we should deactivate // the last composed word to ensure this does not happen. final int[] primaryKeyCodes = mPrimaryKeyCodes; mPrimaryKeyCodes = new int[N]; final LastComposedWord lastComposedWord = new LastComposedWord(primaryKeyCodes, - mInputPointers, mTypedWord.toString(), committedWord, separatorCode, + mInputPointers, mTypedWord.toString(), committedWord, separatorString, prevWord); mInputPointers.reset(); if (type != LastComposedWord.COMMIT_TYPE_DECIDED_WORD diff --git a/java/src/com/android/inputmethod/latin/XmlParseUtils.java b/java/src/com/android/inputmethod/latin/XmlParseUtils.java index 481cdfa47..b5cbaf19e 100644 --- a/java/src/com/android/inputmethod/latin/XmlParseUtils.java +++ b/java/src/com/android/inputmethod/latin/XmlParseUtils.java @@ -23,7 +23,7 @@ import org.xmlpull.v1.XmlPullParserException; import java.io.IOException; -public class XmlParseUtils { +public final class XmlParseUtils { private XmlParseUtils() { // This utility class is not publicly instantiable. } diff --git a/java/src/com/android/inputmethod/latin/makedict/BinaryDictInputOutput.java b/java/src/com/android/inputmethod/latin/makedict/BinaryDictInputOutput.java index 161b94ca0..6f508695e 100644 --- a/java/src/com/android/inputmethod/latin/makedict/BinaryDictInputOutput.java +++ b/java/src/com/android/inputmethod/latin/makedict/BinaryDictInputOutput.java @@ -16,6 +16,8 @@ package com.android.inputmethod.latin.makedict; +import com.android.inputmethod.latin.makedict.FormatSpec.FileHeader; +import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions; import com.android.inputmethod.latin.makedict.FusionDictionary.CharGroup; import com.android.inputmethod.latin.makedict.FusionDictionary.DictionaryOptions; import com.android.inputmethod.latin.makedict.FusionDictionary.Node; @@ -34,6 +36,7 @@ import java.util.Arrays; import java.util.HashMap; import java.util.Iterator; import java.util.Map; +import java.util.Stack; import java.util.TreeMap; /** @@ -43,143 +46,7 @@ import java.util.TreeMap; */ public class BinaryDictInputOutput { - final static boolean DBG = MakedictLog.DBG; - - /* Node layout is as follows: - * | addressType xx : mask with MASK_GROUP_ADDRESS_TYPE - * 2 bits, 00 = no children : FLAG_GROUP_ADDRESS_TYPE_NOADDRESS - * f | 01 = 1 byte : FLAG_GROUP_ADDRESS_TYPE_ONEBYTE - * l | 10 = 2 bytes : FLAG_GROUP_ADDRESS_TYPE_TWOBYTES - * a | 11 = 3 bytes : FLAG_GROUP_ADDRESS_TYPE_THREEBYTES - * g | has several chars ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_MULTIPLE_CHARS - * s | 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 - * | has bigrams ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_BIGRAMS - * - * c | IF FLAG_HAS_MULTIPLE_CHARS - * h | char, char, char, char n * (1 or 3 bytes) : use CharGroupInfo for i/o helpers - * a | end 1 byte, = 0 - * r | ELSE - * s | char 1 or 3 bytes - * | END - * - * f | - * r | IF FLAG_IS_TERMINAL - * e | frequency 1 byte - * q | - * - * c | IF 00 = FLAG_GROUP_ADDRESS_TYPE_NOADDRESS = addressType - * h | // nothing - * i | ELSIF 01 = FLAG_GROUP_ADDRESS_TYPE_ONEBYTE == addressType - * l | children address, 1 byte - * d | ELSIF 10 = FLAG_GROUP_ADDRESS_TYPE_TWOBYTES == addressType - * r | children address, 2 bytes - * e | ELSE // 11 = FLAG_GROUP_ADDRESS_TYPE_THREEBYTES = addressType - * n | children address, 3 bytes - * A | END - * d - * dress - * - * | IF FLAG_IS_TERMINAL && FLAG_HAS_SHORTCUT_TARGETS - * | shortcut string list - * | IF FLAG_IS_TERMINAL && FLAG_HAS_BIGRAMS - * | bigrams address list - * - * Char format is: - * 1 byte = bbbbbbbb match - * case 000xxxxx: xxxxx << 16 + next byte << 8 + next byte - * else: if 00011111 (= 0x1F) : this is the terminator. This is a relevant choice because - * unicode code points range from 0 to 0x10FFFF, so any 3-byte value starting with - * 00011111 would be outside unicode. - * else: iso-latin-1 code - * This allows for the whole unicode range to be encoded, including chars outside of - * the BMP. Also everything in the iso-latin-1 charset is only 1 byte, except control - * characters which should never happen anyway (and still work, but take 3 bytes). - * - * bigram address list is: - * <flags> = | hasNext = 1 bit, 1 = yes, 0 = no : FLAG_ATTRIBUTE_HAS_NEXT - * | addressSign = 1 bit, : FLAG_ATTRIBUTE_OFFSET_NEGATIVE - * | 1 = must take -address, 0 = must take +address - * | xx : mask with MASK_ATTRIBUTE_ADDRESS_TYPE - * | addressFormat = 2 bits, 00 = unused : FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE - * | 01 = 1 byte : FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE - * | 10 = 2 bytes : FLAG_ATTRIBUTE_ADDRESS_TYPE_TWOBYTES - * | 11 = 3 bytes : FLAG_ATTRIBUTE_ADDRESS_TYPE_THREEBYTES - * | 4 bits : frequency : mask with FLAG_ATTRIBUTE_FREQUENCY - * <address> | IF (01 == FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE == addressFormat) - * | read 1 byte, add top 4 bits - * | ELSIF (10 == FLAG_ATTRIBUTE_ADDRESS_TYPE_TWOBYTES == addressFormat) - * | read 2 bytes, add top 4 bits - * | ELSE // 11 == FLAG_ATTRIBUTE_ADDRESS_TYPE_THREEBYTES == addressFormat - * | read 3 bytes, add top 4 bits - * | END - * | if (FLAG_ATTRIBUTE_OFFSET_NEGATIVE) then address = -address - * if (FLAG_ATTRIBUTE_HAS_NEXT) goto bigram_and_shortcut_address_list_is - * - * shortcut string list is: - * <byte size> = GROUP_SHORTCUT_LIST_SIZE_SIZE bytes, big-endian: size of the list, in bytes. - * <flags> = | hasNext = 1 bit, 1 = yes, 0 = no : FLAG_ATTRIBUTE_HAS_NEXT - * | reserved = 3 bits, must be 0 - * | 4 bits : frequency : mask with FLAG_ATTRIBUTE_FREQUENCY - * <shortcut> = | string of characters at the char format described above, with the terminator - * | used to signal the end of the string. - * if (FLAG_ATTRIBUTE_HAS_NEXT goto flags - */ - - private static final int VERSION_1_MAGIC_NUMBER = 0x78B1; - public static final int VERSION_2_MAGIC_NUMBER = 0x9BC13AFE; - private static final int MINIMUM_SUPPORTED_VERSION = 1; - private static final int MAXIMUM_SUPPORTED_VERSION = 2; - private static final int NOT_A_VERSION_NUMBER = -1; - private static final int FIRST_VERSION_WITH_HEADER_SIZE = 2; - - // These options need to be the same numeric values as the one in the native reading code. - private static final int GERMAN_UMLAUT_PROCESSING_FLAG = 0x1; - private static final int FRENCH_LIGATURE_PROCESSING_FLAG = 0x4; - private static final int CONTAINS_BIGRAMS_FLAG = 0x8; - - // TODO: Make this value adaptative to content data, store it in the header, and - // use it in the reading code. - private static final int MAX_WORD_LENGTH = 48; - - private static final int MASK_GROUP_ADDRESS_TYPE = 0xC0; - private static final int FLAG_GROUP_ADDRESS_TYPE_NOADDRESS = 0x00; - private static final int FLAG_GROUP_ADDRESS_TYPE_ONEBYTE = 0x40; - private static final int FLAG_GROUP_ADDRESS_TYPE_TWOBYTES = 0x80; - private static final int FLAG_GROUP_ADDRESS_TYPE_THREEBYTES = 0xC0; - - private static final int FLAG_HAS_MULTIPLE_CHARS = 0x20; - - private static final int FLAG_IS_TERMINAL = 0x10; - private static final int FLAG_HAS_SHORTCUT_TARGETS = 0x08; - private static final int FLAG_HAS_BIGRAMS = 0x04; - - private static final int FLAG_ATTRIBUTE_HAS_NEXT = 0x80; - private static final int FLAG_ATTRIBUTE_OFFSET_NEGATIVE = 0x40; - private static final int MASK_ATTRIBUTE_ADDRESS_TYPE = 0x30; - private static final int FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE = 0x10; - private static final int FLAG_ATTRIBUTE_ADDRESS_TYPE_TWOBYTES = 0x20; - private static final int FLAG_ATTRIBUTE_ADDRESS_TYPE_THREEBYTES = 0x30; - private static final int FLAG_ATTRIBUTE_FREQUENCY = 0x0F; - - private static final int GROUP_CHARACTERS_TERMINATOR = 0x1F; - - private static final int GROUP_TERMINATOR_SIZE = 1; - private static final int GROUP_FLAGS_SIZE = 1; - private static final int GROUP_FREQUENCY_SIZE = 1; - private static final int GROUP_MAX_ADDRESS_SIZE = 3; - private static final int GROUP_ATTRIBUTE_FLAGS_SIZE = 1; - private static final int GROUP_ATTRIBUTE_MAX_ADDRESS_SIZE = 3; - private static final int GROUP_SHORTCUT_LIST_SIZE_SIZE = 2; - - private static final int NO_CHILDREN_ADDRESS = Integer.MIN_VALUE; - private static final int INVALID_CHARACTER = -1; - - private static final int MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT = 0x7F; // 127 - private static final int MAX_CHARGROUPS_IN_A_NODE = 0x7FFF; // 32767 - - private static final int MAX_TERMINAL_FREQUENCY = 255; - private static final int MAX_BIGRAM_FREQUENCY = 15; + private static final boolean DBG = MakedictLog.DBG; // Arbitrary limit to how much passes we consider address size compression should // terminate in. At the time of this writing, our largest dictionary completes @@ -188,6 +55,60 @@ public class BinaryDictInputOutput { // suspicion that a bug might be causing an infinite loop. private static final int MAX_PASSES = 24; + public interface FusionDictionaryBufferInterface { + public int readUnsignedByte(); + public int readUnsignedShort(); + public int readUnsignedInt24(); + public int readInt(); + public int position(); + public void position(int newPosition); + public void put(final byte b); + } + + public static final class ByteBufferWrapper implements FusionDictionaryBufferInterface { + private ByteBuffer mBuffer; + + public ByteBufferWrapper(final ByteBuffer buffer) { + mBuffer = buffer; + } + + @Override + public int readUnsignedByte() { + return ((int)mBuffer.get()) & 0xFF; + } + + @Override + public int readUnsignedShort() { + return ((int)mBuffer.getShort()) & 0xFFFF; + } + + @Override + public int readUnsignedInt24() { + final int retval = readUnsignedByte(); + return (retval << 16) + readUnsignedShort(); + } + + @Override + public int readInt() { + return mBuffer.getInt(); + } + + @Override + public int position() { + return mBuffer.position(); + } + + @Override + public void position(int newPos) { + mBuffer.position(newPos); + } + + @Override + public void put(final byte b) { + mBuffer.put(b); + } + } + /** * A class grouping utility function for our specific character encoding. */ @@ -199,7 +120,7 @@ public class BinaryDictInputOutput { /** * Helper method to find out whether this code fits on one byte */ - private static boolean fitsOnOneByte(int character) { + private static boolean fitsOnOneByte(final int character) { return character >= MINIMAL_ONE_BYTE_CHARACTER_VALUE && character <= MAXIMAL_ONE_BYTE_CHARACTER_VALUE; } @@ -221,10 +142,10 @@ public class BinaryDictInputOutput { * @param character the character code. * @return the size in binary encoded-form, either 1 or 3 bytes. */ - private static int getCharSize(int character) { + private static int getCharSize(final int character) { // See char encoding in FusionDictionary.java if (fitsOnOneByte(character)) return 1; - if (INVALID_CHARACTER == character) return 1; + if (FormatSpec.INVALID_CHARACTER == character) return 1; return 3; } @@ -282,7 +203,7 @@ public class BinaryDictInputOutput { buffer[index++] = (byte)(0xFF & codePoint); } } - buffer[index++] = GROUP_CHARACTERS_TERMINATOR; + buffer[index++] = FormatSpec.GROUP_CHARACTERS_TERMINATOR; return index - origin; } @@ -294,7 +215,7 @@ public class BinaryDictInputOutput { * @param buffer the ByteArrayOutputStream to write to. * @param word the string to write. */ - private static void writeString(ByteArrayOutputStream buffer, final String word) { + private static void writeString(final ByteArrayOutputStream buffer, final String word) { final int length = word.length(); for (int i = 0; i < length; i = word.offsetByCodePoints(i, 1)) { final int codePoint = word.codePointAt(i); @@ -306,16 +227,16 @@ public class BinaryDictInputOutput { buffer.write((byte) (0xFF & codePoint)); } } - buffer.write(GROUP_CHARACTERS_TERMINATOR); + buffer.write(FormatSpec.GROUP_CHARACTERS_TERMINATOR); } /** - * Reads a string from a ByteBuffer. This is the converse of the above method. + * Reads a string from a buffer. This is the converse of the above method. */ - private static String readString(final ByteBuffer buffer) { + private static String readString(final FusionDictionaryBufferInterface buffer) { final StringBuilder s = new StringBuilder(); int character = readChar(buffer); - while (character != INVALID_CHARACTER) { + while (character != FormatSpec.INVALID_CHARACTER) { s.appendCodePoint(character); character = readChar(buffer); } @@ -323,19 +244,21 @@ public class BinaryDictInputOutput { } /** - * Reads a character from the ByteBuffer. + * Reads a character from the buffer. * * This follows the character format documented earlier in this source file. * * @param buffer the buffer, positioned over an encoded character. * @return the character code. */ - private static int readChar(final ByteBuffer buffer) { - int character = readUnsignedByte(buffer); + private static int readChar(final FusionDictionaryBufferInterface buffer) { + int character = buffer.readUnsignedByte(); if (!fitsOnOneByte(character)) { - if (GROUP_CHARACTERS_TERMINATOR == character) return INVALID_CHARACTER; + if (FormatSpec.GROUP_CHARACTERS_TERMINATOR == character) { + return FormatSpec.INVALID_CHARACTER; + } character <<= 16; - character += readUnsignedShort(buffer); + character += buffer.readUnsignedShort(); } return character; } @@ -350,9 +273,9 @@ public class BinaryDictInputOutput { * @param group the group * @return the size of the char array, including the terminator if any */ - private static int getGroupCharactersSize(CharGroup group) { + private static int getGroupCharactersSize(final CharGroup group) { int size = CharEncoding.getCharArraySize(group.mChars); - if (group.hasSeveralChars()) size += GROUP_TERMINATOR_SIZE; + if (group.hasSeveralChars()) size += FormatSpec.GROUP_TERMINATOR_SIZE; return size; } @@ -362,13 +285,14 @@ public class BinaryDictInputOutput { * @return the size of the group count, either 1 or 2 bytes. */ private static int getGroupCountSize(final int count) { - if (MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT >= count) { + if (FormatSpec.MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT >= count) { return 1; - } else if (MAX_CHARGROUPS_IN_A_NODE >= count) { + } else if (FormatSpec.MAX_CHARGROUPS_IN_A_NODE >= count) { return 2; } else { - throw new RuntimeException("Can't have more than " + MAX_CHARGROUPS_IN_A_NODE - + " groups in a node (found " + count +")"); + throw new RuntimeException("Can't have more than " + + FormatSpec.MAX_CHARGROUPS_IN_A_NODE + " groups in a node (found " + count + + ")"); } } @@ -385,14 +309,14 @@ public class BinaryDictInputOutput { * Compute the size of a shortcut in bytes. */ private static int getShortcutSize(final WeightedString shortcut) { - int size = GROUP_ATTRIBUTE_FLAGS_SIZE; + int size = FormatSpec.GROUP_ATTRIBUTE_FLAGS_SIZE; final String word = shortcut.mWord; final int length = word.length(); for (int i = 0; i < length; i = word.offsetByCodePoints(i, 1)) { final int codePoint = word.codePointAt(i); size += CharEncoding.getCharSize(codePoint); } - size += GROUP_TERMINATOR_SIZE; + size += FormatSpec.GROUP_TERMINATOR_SIZE; return size; } @@ -404,7 +328,7 @@ public class BinaryDictInputOutput { */ private static int getShortcutListSize(final ArrayList<WeightedString> shortcutList) { if (null == shortcutList) return 0; - int size = GROUP_SHORTCUT_LIST_SIZE_SIZE; + int size = FormatSpec.GROUP_SHORTCUT_LIST_SIZE_SIZE; for (final WeightedString shortcut : shortcutList) { size += getShortcutSize(shortcut); } @@ -415,16 +339,18 @@ public class BinaryDictInputOutput { * Compute the maximum size of a CharGroup, assuming 3-byte addresses for everything. * * @param group the CharGroup to compute the size of. + * @param options file format options. * @return the maximum size of the group. */ - private static int getCharGroupMaximumSize(CharGroup group) { - int size = getGroupCharactersSize(group) + GROUP_FLAGS_SIZE; + private static int getCharGroupMaximumSize(final CharGroup group, final FormatOptions options) { + int size = getGroupHeaderSize(group, options); // If terminal, one byte for the frequency - if (group.isTerminal()) size += GROUP_FREQUENCY_SIZE; - size += GROUP_MAX_ADDRESS_SIZE; // For children address + if (group.isTerminal()) size += FormatSpec.GROUP_FREQUENCY_SIZE; + size += FormatSpec.GROUP_MAX_ADDRESS_SIZE; // For children address size += getShortcutListSize(group.mShortcutTargets); if (null != group.mBigrams) { - size += (GROUP_ATTRIBUTE_FLAGS_SIZE + GROUP_ATTRIBUTE_MAX_ADDRESS_SIZE) + size += (FormatSpec.GROUP_ATTRIBUTE_FLAGS_SIZE + + FormatSpec.GROUP_ATTRIBUTE_MAX_ADDRESS_SIZE) * group.mBigrams.size(); } return size; @@ -435,11 +361,12 @@ public class BinaryDictInputOutput { * it in the 'actualSize' member of the node. * * @param node the node to compute the maximum size of. + * @param options file format options. */ - private static void setNodeMaximumSize(Node node) { + private static void setNodeMaximumSize(final Node node, final FormatOptions options) { int size = getGroupCountSize(node); for (CharGroup g : node.mData) { - final int groupSize = getCharGroupMaximumSize(g); + final int groupSize = getCharGroupMaximumSize(g, options); g.mCachedSize = groupSize; size += groupSize; } @@ -449,8 +376,31 @@ public class BinaryDictInputOutput { /** * Helper method to hide the actual value of the no children address. */ - private static boolean hasChildrenAddress(int address) { - return NO_CHILDREN_ADDRESS != address; + private static boolean hasChildrenAddress(final int address) { + return FormatSpec.NO_CHILDREN_ADDRESS != address; + } + + /** + * Helper method to check whether the CharGroup has a parent address. + */ + private static boolean hasParentAddress(final FormatOptions options) { + return options.mVersion >= FormatSpec.FIRST_VERSION_WITH_PARENT_ADDRESS + && options.mHasParentAddress; + } + + /** + * Compute the size of the header (flag + [parent address] + characters size) of a CharGroup. + * + * @param group the group of which to compute the size of the header + * @param options file format options. + */ + private static int getGroupHeaderSize(final CharGroup group, final FormatOptions options) { + if (hasParentAddress(options)) { + return FormatSpec.GROUP_FLAGS_SIZE + FormatSpec.PARENT_ADDRESS_SIZE + + getGroupCharactersSize(group); + } else { + return FormatSpec.GROUP_FLAGS_SIZE + getGroupCharactersSize(group); + } } /** @@ -463,7 +413,7 @@ public class BinaryDictInputOutput { * @param address the address * @return the byte size. */ - private static int getByteSize(int address) { + private static int getByteSize(final int address) { assert(address < 0x1000000); if (!hasChildrenAddress(address)) { return 0; @@ -479,14 +429,14 @@ public class BinaryDictInputOutput { // This method is responsible for finding a nice ordering of the nodes that favors run-time // cache performance and dictionary size. - /* package for tests */ static ArrayList<Node> flattenTree(Node root) { + /* package for tests */ static ArrayList<Node> flattenTree(final Node root) { final int treeSize = FusionDictionary.countCharGroups(root); MakedictLog.i("Counted nodes : " + treeSize); final ArrayList<Node> flatTree = new ArrayList<Node>(treeSize); return flattenTreeInner(flatTree, root); } - private static ArrayList<Node> flattenTreeInner(ArrayList<Node> list, Node node) { + private static ArrayList<Node> flattenTreeInner(final ArrayList<Node> list, final Node node) { // Removing the node is necessary if the tails are merged, because we would then // add the same node several times when we only want it once. A number of places in // the code also depends on any node being only once in the list. @@ -536,9 +486,11 @@ public class BinaryDictInputOutput { * * @param node the node to compute the size of. * @param dict the dictionary in which the word/attributes are to be found. + * @param formatOptions file format options. * @return false if none of the cached addresses inside the node changed, true otherwise. */ - private static boolean computeActualNodeSize(Node node, FusionDictionary dict) { + private static boolean computeActualNodeSize(final Node node, final FusionDictionary dict, + final FormatOptions formatOptions) { boolean changed = false; int size = getGroupCountSize(node); for (CharGroup group : node.mData) { @@ -546,21 +498,24 @@ public class BinaryDictInputOutput { changed = true; group.mCachedAddress = node.mCachedAddress + size; } - int groupSize = GROUP_FLAGS_SIZE + getGroupCharactersSize(group); - if (group.isTerminal()) groupSize += GROUP_FREQUENCY_SIZE; + int groupSize = getGroupHeaderSize(group, formatOptions); + if (group.isTerminal()) groupSize += FormatSpec.GROUP_FREQUENCY_SIZE; if (null != group.mChildren) { - final int offsetBasePoint= groupSize + node.mCachedAddress + size; + final int offsetBasePoint = groupSize + node.mCachedAddress + size; final int offset = group.mChildren.mCachedAddress - offsetBasePoint; + // assign my address to children's parent address + group.mChildren.mCachedParentAddress = group.mCachedAddress + - group.mChildren.mCachedAddress; groupSize += getByteSize(offset); } groupSize += getShortcutListSize(group.mShortcutTargets); if (null != group.mBigrams) { for (WeightedString bigram : group.mBigrams) { final int offsetBasePoint = groupSize + node.mCachedAddress + size - + GROUP_FLAGS_SIZE; + + FormatSpec.GROUP_FLAGS_SIZE; final int addressOfBigram = findAddressOfWord(dict, bigram.mWord); final int offset = addressOfBigram - offsetBasePoint; - groupSize += getByteSize(offset) + GROUP_FLAGS_SIZE; + groupSize += getByteSize(offset) + FormatSpec.GROUP_FLAGS_SIZE; } } group.mCachedSize = groupSize; @@ -579,7 +534,7 @@ public class BinaryDictInputOutput { * @param flatNodes the array of nodes. * @return the byte size of the entire stack. */ - private static int stackNodes(ArrayList<Node> flatNodes) { + private static int stackNodes(final ArrayList<Node> flatNodes) { int nodeOffset = 0; for (Node n : flatNodes) { n.mCachedAddress = nodeOffset; @@ -609,12 +564,13 @@ public class BinaryDictInputOutput { * * @param dict the dictionary * @param flatNodes the ordered array of nodes + * @param formatOptions file format options. * @return the same array it was passed. The nodes have been updated for address and size. */ - private static ArrayList<Node> computeAddresses(FusionDictionary dict, - ArrayList<Node> flatNodes) { + private static ArrayList<Node> computeAddresses(final FusionDictionary dict, + final ArrayList<Node> flatNodes, final FormatOptions formatOptions) { // First get the worst sizes and offsets - for (Node n : flatNodes) setNodeMaximumSize(n); + for (Node n : flatNodes) setNodeMaximumSize(n, formatOptions); final int offset = stackNodes(flatNodes); MakedictLog.i("Compressing the array addresses. Original size : " + offset); @@ -626,7 +582,7 @@ public class BinaryDictInputOutput { changesDone = false; for (Node n : flatNodes) { final int oldNodeSize = n.mCachedSize; - final boolean changed = computeActualNodeSize(n, dict); + final boolean changed = computeActualNodeSize(n, dict, formatOptions); final int newNodeSize = n.mCachedSize; if (oldNodeSize < newNodeSize) throw new RuntimeException("Increased size ?!"); changesDone |= changed; @@ -654,7 +610,7 @@ public class BinaryDictInputOutput { * * @param array the array node to check */ - private static void checkFlatNodeArray(ArrayList<Node> array) { + private static void checkFlatNodeArray(final ArrayList<Node> array) { int offset = 0; int index = 0; for (Node n : array) { @@ -699,20 +655,20 @@ public class BinaryDictInputOutput { private static byte makeCharGroupFlags(final CharGroup group, final int groupAddress, final int childrenOffset) { byte flags = 0; - if (group.mChars.length > 1) flags |= FLAG_HAS_MULTIPLE_CHARS; + if (group.mChars.length > 1) flags |= FormatSpec.FLAG_HAS_MULTIPLE_CHARS; if (group.mFrequency >= 0) { - flags |= FLAG_IS_TERMINAL; + flags |= FormatSpec.FLAG_IS_TERMINAL; } if (null != group.mChildren) { switch (getByteSize(childrenOffset)) { case 1: - flags |= FLAG_GROUP_ADDRESS_TYPE_ONEBYTE; + flags |= FormatSpec.FLAG_GROUP_ADDRESS_TYPE_ONEBYTE; break; case 2: - flags |= FLAG_GROUP_ADDRESS_TYPE_TWOBYTES; + flags |= FormatSpec.FLAG_GROUP_ADDRESS_TYPE_TWOBYTES; break; case 3: - flags |= FLAG_GROUP_ADDRESS_TYPE_THREEBYTES; + flags |= FormatSpec.FLAG_GROUP_ADDRESS_TYPE_THREEBYTES; break; default: throw new RuntimeException("Node with a strange address"); @@ -722,13 +678,19 @@ public class BinaryDictInputOutput { if (DBG && 0 == group.mShortcutTargets.size()) { throw new RuntimeException("0-sized shortcut list must be null"); } - flags |= FLAG_HAS_SHORTCUT_TARGETS; + flags |= FormatSpec.FLAG_HAS_SHORTCUT_TARGETS; } if (null != group.mBigrams) { if (DBG && 0 == group.mBigrams.size()) { throw new RuntimeException("0-sized bigram list must be null"); } - flags |= FLAG_HAS_BIGRAMS; + flags |= FormatSpec.FLAG_HAS_BIGRAMS; + } + if (group.mIsNotAWord) { + flags |= FormatSpec.FLAG_IS_NOT_A_WORD; + } + if (group.mIsBlacklistEntry) { + flags |= FormatSpec.FLAG_IS_BLACKLISTED; } return flags; } @@ -745,17 +707,17 @@ public class BinaryDictInputOutput { */ private static final int makeBigramFlags(final boolean more, final int offset, int bigramFrequency, final int unigramFrequency, final String word) { - int bigramFlags = (more ? FLAG_ATTRIBUTE_HAS_NEXT : 0) - + (offset < 0 ? FLAG_ATTRIBUTE_OFFSET_NEGATIVE : 0); + int bigramFlags = (more ? FormatSpec.FLAG_ATTRIBUTE_HAS_NEXT : 0) + + (offset < 0 ? FormatSpec.FLAG_ATTRIBUTE_OFFSET_NEGATIVE : 0); switch (getByteSize(offset)) { case 1: - bigramFlags |= FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE; + bigramFlags |= FormatSpec.FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE; break; case 2: - bigramFlags |= FLAG_ATTRIBUTE_ADDRESS_TYPE_TWOBYTES; + bigramFlags |= FormatSpec.FLAG_ATTRIBUTE_ADDRESS_TYPE_TWOBYTES; break; case 3: - bigramFlags |= FLAG_ATTRIBUTE_ADDRESS_TYPE_THREEBYTES; + bigramFlags |= FormatSpec.FLAG_ATTRIBUTE_ADDRESS_TYPE_THREEBYTES; break; default: throw new RuntimeException("Strange offset size"); @@ -790,7 +752,8 @@ public class BinaryDictInputOutput { // approximation. (0.5 to get the first step start, and 0.5 to get the middle of the // step pointed by the discretized frequency. final float stepSize = - (MAX_TERMINAL_FREQUENCY - unigramFrequency) / (1.5f + MAX_BIGRAM_FREQUENCY); + (FormatSpec.MAX_TERMINAL_FREQUENCY - unigramFrequency) + / (1.5f + FormatSpec.MAX_BIGRAM_FREQUENCY); final float firstStepStart = 1 + unigramFrequency + (stepSize / 2.0f); final int discretizedFrequency = (int)((bigramFrequency - firstStepStart) / stepSize); // If the bigram freq is less than half-a-step higher than the unigram freq, we get -1 @@ -799,19 +762,21 @@ public class BinaryDictInputOutput { // 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 & FLAG_ATTRIBUTE_FREQUENCY; + bigramFlags += finalBigramFrequency & FormatSpec.FLAG_ATTRIBUTE_FREQUENCY; return bigramFlags; } /** * Makes the 2-byte value for options flags. */ - private static final int makeOptionsValue(final FusionDictionary dictionary) { + private static final int makeOptionsValue(final FusionDictionary dictionary, + final FormatOptions formatOptions) { final DictionaryOptions options = dictionary.mOptions; final boolean hasBigrams = dictionary.hasBigrams(); - return (options.mFrenchLigatureProcessing ? FRENCH_LIGATURE_PROCESSING_FLAG : 0) - + (options.mGermanUmlautProcessing ? GERMAN_UMLAUT_PROCESSING_FLAG : 0) - + (hasBigrams ? CONTAINS_BIGRAMS_FLAG : 0); + return (options.mFrenchLigatureProcessing ? FormatSpec.FRENCH_LIGATURE_PROCESSING_FLAG : 0) + + (options.mGermanUmlautProcessing ? FormatSpec.GERMAN_UMLAUT_PROCESSING_FLAG : 0) + + (hasBigrams ? FormatSpec.CONTAINS_BIGRAMS_FLAG : 0) + + (formatOptions.mHasParentAddress ? FormatSpec.HAS_PARENT_ADDRESS : 0); } /** @@ -822,7 +787,8 @@ public class BinaryDictInputOutput { * @return the flags */ private static final int makeShortcutFlags(final boolean more, final int frequency) { - return (more ? FLAG_ATTRIBUTE_HAS_NEXT : 0) + (frequency & FLAG_ATTRIBUTE_FREQUENCY); + return (more ? FormatSpec.FLAG_ATTRIBUTE_HAS_NEXT : 0) + + (frequency & FormatSpec.FLAG_ATTRIBUTE_FREQUENCY); } /** @@ -834,13 +800,16 @@ public class BinaryDictInputOutput { * @param dict the dictionary the node is a part of (for relative offsets). * @param buffer the memory buffer to write to. * @param node the node to write. + * @param formatOptions file format options. * @return the address of the END of the node. */ - private static int writePlacedNode(FusionDictionary dict, byte[] buffer, Node node) { + private static int writePlacedNode(final FusionDictionary dict, byte[] buffer, + final Node node, final FormatOptions formatOptions) { int index = node.mCachedAddress; final int groupCount = node.mData.size(); final int countSize = getGroupCountSize(node); + final int parentAddress = node.mCachedParentAddress; if (1 == countSize) { buffer[index++] = (byte)groupCount; } else if (2 == countSize) { @@ -857,20 +826,38 @@ public class BinaryDictInputOutput { if (index != group.mCachedAddress) throw new RuntimeException("Bug: write index is not " + "the same as the cached address of the group : " + index + " <> " + group.mCachedAddress); - groupAddress += GROUP_FLAGS_SIZE + getGroupCharactersSize(group); + groupAddress += getGroupHeaderSize(group, formatOptions); // Sanity checks. - if (DBG && group.mFrequency > MAX_TERMINAL_FREQUENCY) { - throw new RuntimeException("A node has a frequency > " + MAX_TERMINAL_FREQUENCY + if (DBG && group.mFrequency > FormatSpec.MAX_TERMINAL_FREQUENCY) { + throw new RuntimeException("A node has a frequency > " + + FormatSpec.MAX_TERMINAL_FREQUENCY + " : " + group.mFrequency); } - if (group.mFrequency >= 0) groupAddress += GROUP_FREQUENCY_SIZE; + if (group.mFrequency >= 0) groupAddress += FormatSpec.GROUP_FREQUENCY_SIZE; final int childrenOffset = null == group.mChildren - ? NO_CHILDREN_ADDRESS : group.mChildren.mCachedAddress - groupAddress; + ? FormatSpec.NO_CHILDREN_ADDRESS + : group.mChildren.mCachedAddress - groupAddress; byte flags = makeCharGroupFlags(group, groupAddress, childrenOffset); buffer[index++] = flags; + + if (hasParentAddress(formatOptions)) { + if (parentAddress == FormatSpec.NO_PARENT_ADDRESS) { + // this node is the root node. + buffer[index] = buffer[index + 1] = buffer[index + 2] = 0; + } else { + // write parent address. (version 3) + final int actualParentAddress = Math.abs(parentAddress + + (node.mCachedAddress - group.mCachedAddress)); + buffer[index] = (byte)((actualParentAddress >> 16) & 0xFF); + buffer[index + 1] = (byte)((actualParentAddress >> 8) & 0xFF); + buffer[index + 2] = (byte)(actualParentAddress & 0xFF); + } + index += 3; + } + index = CharEncoding.writeCharArray(group.mChars, buffer, index); if (group.hasSeveralChars()) { - buffer[index++] = GROUP_CHARACTERS_TERMINATOR; + buffer[index++] = FormatSpec.GROUP_CHARACTERS_TERMINATOR; } if (group.mFrequency >= 0) { buffer[index++] = (byte) group.mFrequency; @@ -882,8 +869,8 @@ public class BinaryDictInputOutput { // Write shortcuts if (null != group.mShortcutTargets) { final int indexOfShortcutByteSize = index; - index += GROUP_SHORTCUT_LIST_SIZE_SIZE; - groupAddress += GROUP_SHORTCUT_LIST_SIZE_SIZE; + index += FormatSpec.GROUP_SHORTCUT_LIST_SIZE_SIZE; + groupAddress += FormatSpec.GROUP_SHORTCUT_LIST_SIZE_SIZE; final Iterator<WeightedString> shortcutIterator = group.mShortcutTargets.iterator(); while (shortcutIterator.hasNext()) { final WeightedString target = shortcutIterator.next(); @@ -992,10 +979,10 @@ public class BinaryDictInputOutput { * * @param destination the stream to write the binary data to. * @param dict the dictionary to write. - * @param version the version of the format to write, currently either 1 or 2. + * @param formatOptions file format options. */ public static void writeDictionaryBinary(final OutputStream destination, - final FusionDictionary dict, final int version) + final FusionDictionary dict, final FormatOptions formatOptions) throws IOException, UnsupportedFormatException { // Addresses are limited to 3 bytes, but since addresses can be relative to each node, the @@ -1004,36 +991,39 @@ public class BinaryDictInputOutput { // does not have a size limit, each node must still be within 16MB of all its children and // parents. As long as this is ensured, the dictionary file may grow to any size. - if (version < MINIMUM_SUPPORTED_VERSION || version > MAXIMUM_SUPPORTED_VERSION) { + final int version = formatOptions.mVersion; + if (version < FormatSpec.MINIMUM_SUPPORTED_VERSION + || version > FormatSpec.MAXIMUM_SUPPORTED_VERSION) { throw new UnsupportedFormatException("Requested file format version " + version + ", but this implementation only supports versions " - + MINIMUM_SUPPORTED_VERSION + " through " + MAXIMUM_SUPPORTED_VERSION); + + FormatSpec.MINIMUM_SUPPORTED_VERSION + " through " + + FormatSpec.MAXIMUM_SUPPORTED_VERSION); } ByteArrayOutputStream headerBuffer = new ByteArrayOutputStream(256); // The magic number in big-endian order. - if (version >= FIRST_VERSION_WITH_HEADER_SIZE) { + if (version >= FormatSpec.FIRST_VERSION_WITH_HEADER_SIZE) { // Magic number for version 2+. - headerBuffer.write((byte) (0xFF & (VERSION_2_MAGIC_NUMBER >> 24))); - headerBuffer.write((byte) (0xFF & (VERSION_2_MAGIC_NUMBER >> 16))); - headerBuffer.write((byte) (0xFF & (VERSION_2_MAGIC_NUMBER >> 8))); - headerBuffer.write((byte) (0xFF & VERSION_2_MAGIC_NUMBER)); + headerBuffer.write((byte) (0xFF & (FormatSpec.VERSION_2_MAGIC_NUMBER >> 24))); + headerBuffer.write((byte) (0xFF & (FormatSpec.VERSION_2_MAGIC_NUMBER >> 16))); + headerBuffer.write((byte) (0xFF & (FormatSpec.VERSION_2_MAGIC_NUMBER >> 8))); + headerBuffer.write((byte) (0xFF & FormatSpec.VERSION_2_MAGIC_NUMBER)); // Dictionary version. headerBuffer.write((byte) (0xFF & (version >> 8))); headerBuffer.write((byte) (0xFF & version)); } else { // Magic number for version 1. - headerBuffer.write((byte) (0xFF & (VERSION_1_MAGIC_NUMBER >> 8))); - headerBuffer.write((byte) (0xFF & VERSION_1_MAGIC_NUMBER)); + headerBuffer.write((byte) (0xFF & (FormatSpec.VERSION_1_MAGIC_NUMBER >> 8))); + headerBuffer.write((byte) (0xFF & FormatSpec.VERSION_1_MAGIC_NUMBER)); // Dictionary version. headerBuffer.write((byte) (0xFF & version)); } // Options flags - final int options = makeOptionsValue(dict); + final int options = makeOptionsValue(dict, formatOptions); headerBuffer.write((byte) (0xFF & (options >> 8))); headerBuffer.write((byte) (0xFF & options)); - if (version >= FIRST_VERSION_WITH_HEADER_SIZE) { + if (version >= FormatSpec.FIRST_VERSION_WITH_HEADER_SIZE) { final int headerSizeOffset = headerBuffer.size(); // Placeholder to be written later with header size. for (int i = 0; i < 4; ++i) { @@ -1064,20 +1054,20 @@ public class BinaryDictInputOutput { ArrayList<Node> flatNodes = flattenTree(dict.mRoot); MakedictLog.i("Computing addresses..."); - computeAddresses(dict, flatNodes); + computeAddresses(dict, flatNodes, formatOptions); MakedictLog.i("Checking array..."); if (DBG) checkFlatNodeArray(flatNodes); // Create a buffer that matches the final dictionary size. final Node lastNode = flatNodes.get(flatNodes.size() - 1); - final int bufferSize =(lastNode.mCachedAddress + lastNode.mCachedSize); + final int bufferSize = lastNode.mCachedAddress + lastNode.mCachedSize; final byte[] buffer = new byte[bufferSize]; int index = 0; MakedictLog.i("Writing file..."); int dataEndOffset = 0; for (Node n : flatNodes) { - dataEndOffset = writePlacedNode(dict, buffer, n); + dataEndOffset = writePlacedNode(dict, buffer, n, formatOptions); } if (DBG) showStatistics(flatNodes); @@ -1092,113 +1082,127 @@ public class BinaryDictInputOutput { // Input methods: Read a binary dictionary to memory. // readDictionaryBinary is the public entry point for them. - static final int[] characterBuffer = new int[MAX_WORD_LENGTH]; - private static CharGroupInfo readCharGroup(final ByteBuffer buffer, - final int originalGroupAddress) { + private static final int[] CHARACTER_BUFFER = new int[FormatSpec.MAX_WORD_LENGTH]; + private static CharGroupInfo readCharGroup(final FusionDictionaryBufferInterface buffer, + final int originalGroupAddress, final FormatOptions options) { int addressPointer = originalGroupAddress; - final int flags = readUnsignedByte(buffer); + final int flags = buffer.readUnsignedByte(); ++addressPointer; + + final int parentAddress; + if (hasParentAddress(options)) { + // read the parent address. (version 3) + parentAddress = -buffer.readUnsignedInt24(); + addressPointer += 3; + } else { + parentAddress = FormatSpec.NO_PARENT_ADDRESS; + } + final int characters[]; - if (0 != (flags & FLAG_HAS_MULTIPLE_CHARS)) { + if (0 != (flags & FormatSpec.FLAG_HAS_MULTIPLE_CHARS)) { int index = 0; int character = CharEncoding.readChar(buffer); addressPointer += CharEncoding.getCharSize(character); while (-1 != character) { - characterBuffer[index++] = character; + // FusionDictionary is making sure that the length of the word is smaller than + // MAX_WORD_LENGTH. + // So we'll never write past the end of CHARACTER_BUFFER. + CHARACTER_BUFFER[index++] = character; character = CharEncoding.readChar(buffer); addressPointer += CharEncoding.getCharSize(character); } - characters = Arrays.copyOfRange(characterBuffer, 0, index); + characters = Arrays.copyOfRange(CHARACTER_BUFFER, 0, index); } else { final int character = CharEncoding.readChar(buffer); addressPointer += CharEncoding.getCharSize(character); characters = new int[] { character }; } final int frequency; - if (0 != (FLAG_IS_TERMINAL & flags)) { + if (0 != (FormatSpec.FLAG_IS_TERMINAL & flags)) { ++addressPointer; - frequency = readUnsignedByte(buffer); + frequency = buffer.readUnsignedByte(); } else { frequency = CharGroup.NOT_A_TERMINAL; } int childrenAddress = addressPointer; - switch (flags & MASK_GROUP_ADDRESS_TYPE) { - case FLAG_GROUP_ADDRESS_TYPE_ONEBYTE: - childrenAddress += readUnsignedByte(buffer); + switch (flags & FormatSpec.MASK_GROUP_ADDRESS_TYPE) { + case FormatSpec.FLAG_GROUP_ADDRESS_TYPE_ONEBYTE: + childrenAddress += buffer.readUnsignedByte(); addressPointer += 1; break; - case FLAG_GROUP_ADDRESS_TYPE_TWOBYTES: - childrenAddress += readUnsignedShort(buffer); + case FormatSpec.FLAG_GROUP_ADDRESS_TYPE_TWOBYTES: + childrenAddress += buffer.readUnsignedShort(); addressPointer += 2; break; - case FLAG_GROUP_ADDRESS_TYPE_THREEBYTES: - childrenAddress += readUnsignedInt24(buffer); + case FormatSpec.FLAG_GROUP_ADDRESS_TYPE_THREEBYTES: + childrenAddress += buffer.readUnsignedInt24(); addressPointer += 3; break; - case FLAG_GROUP_ADDRESS_TYPE_NOADDRESS: + case FormatSpec.FLAG_GROUP_ADDRESS_TYPE_NOADDRESS: default: - childrenAddress = NO_CHILDREN_ADDRESS; + childrenAddress = FormatSpec.NO_CHILDREN_ADDRESS; break; } ArrayList<WeightedString> shortcutTargets = null; - if (0 != (flags & FLAG_HAS_SHORTCUT_TARGETS)) { + if (0 != (flags & FormatSpec.FLAG_HAS_SHORTCUT_TARGETS)) { final int pointerBefore = buffer.position(); shortcutTargets = new ArrayList<WeightedString>(); - buffer.getShort(); // Skip the size + buffer.readUnsignedShort(); // Skip the size while (true) { - final int targetFlags = readUnsignedByte(buffer); + final int targetFlags = buffer.readUnsignedByte(); final String word = CharEncoding.readString(buffer); shortcutTargets.add(new WeightedString(word, - targetFlags & FLAG_ATTRIBUTE_FREQUENCY)); - if (0 == (targetFlags & FLAG_ATTRIBUTE_HAS_NEXT)) break; + targetFlags & FormatSpec.FLAG_ATTRIBUTE_FREQUENCY)); + if (0 == (targetFlags & FormatSpec.FLAG_ATTRIBUTE_HAS_NEXT)) break; } addressPointer += buffer.position() - pointerBefore; } ArrayList<PendingAttribute> bigrams = null; - if (0 != (flags & FLAG_HAS_BIGRAMS)) { + if (0 != (flags & FormatSpec.FLAG_HAS_BIGRAMS)) { bigrams = new ArrayList<PendingAttribute>(); while (true) { - final int bigramFlags = readUnsignedByte(buffer); + final int bigramFlags = buffer.readUnsignedByte(); ++addressPointer; - final int sign = 0 == (bigramFlags & FLAG_ATTRIBUTE_OFFSET_NEGATIVE) ? 1 : -1; + final int sign = 0 == (bigramFlags & FormatSpec.FLAG_ATTRIBUTE_OFFSET_NEGATIVE) + ? 1 : -1; int bigramAddress = addressPointer; - switch (bigramFlags & MASK_ATTRIBUTE_ADDRESS_TYPE) { - case FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE: - bigramAddress += sign * readUnsignedByte(buffer); + switch (bigramFlags & FormatSpec.MASK_ATTRIBUTE_ADDRESS_TYPE) { + case FormatSpec.FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE: + bigramAddress += sign * buffer.readUnsignedByte(); addressPointer += 1; break; - case FLAG_ATTRIBUTE_ADDRESS_TYPE_TWOBYTES: - bigramAddress += sign * readUnsignedShort(buffer); + case FormatSpec.FLAG_ATTRIBUTE_ADDRESS_TYPE_TWOBYTES: + bigramAddress += sign * buffer.readUnsignedShort(); addressPointer += 2; break; - case FLAG_ATTRIBUTE_ADDRESS_TYPE_THREEBYTES: - final int offset = (readUnsignedByte(buffer) << 16) - + readUnsignedShort(buffer); + case FormatSpec.FLAG_ATTRIBUTE_ADDRESS_TYPE_THREEBYTES: + final int offset = (buffer.readUnsignedByte() << 16) + + buffer.readUnsignedShort(); bigramAddress += sign * offset; addressPointer += 3; break; default: throw new RuntimeException("Has bigrams with no address"); } - bigrams.add(new PendingAttribute(bigramFlags & FLAG_ATTRIBUTE_FREQUENCY, + bigrams.add(new PendingAttribute(bigramFlags & FormatSpec.FLAG_ATTRIBUTE_FREQUENCY, bigramAddress)); - if (0 == (bigramFlags & FLAG_ATTRIBUTE_HAS_NEXT)) break; + if (0 == (bigramFlags & FormatSpec.FLAG_ATTRIBUTE_HAS_NEXT)) break; } } return new CharGroupInfo(originalGroupAddress, addressPointer, flags, characters, frequency, - childrenAddress, shortcutTargets, bigrams); + parentAddress, childrenAddress, shortcutTargets, bigrams); } /** * Reads and returns the char group count out of a buffer and forwards the pointer. */ - private static int readCharGroupCount(final ByteBuffer buffer) { - final int msb = readUnsignedByte(buffer); - if (MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT >= msb) { + private static int readCharGroupCount(final FusionDictionaryBufferInterface buffer) { + final int msb = buffer.readUnsignedByte(); + if (FormatSpec.MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT >= msb) { return msb; } else { - return ((MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT & msb) << 8) - + readUnsignedByte(buffer); + return ((FormatSpec.MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT & msb) << 8) + + buffer.readUnsignedByte(); } } @@ -1213,13 +1217,56 @@ public class BinaryDictInputOutput { * @param buffer the buffer to read from. * @param headerSize the size of the header. * @param address the address to seek. + * @param formatOptions file format options. * @return the word, as a string. */ - private static String getWordAtAddress(final ByteBuffer buffer, final int headerSize, - final int address) { + private static String getWordAtAddress(final FusionDictionaryBufferInterface buffer, + final int headerSize, final int address, final FormatOptions formatOptions) { final String cachedString = wordCache.get(address); if (null != cachedString) return cachedString; + + final String result; final int originalPointer = buffer.position(); + + if (hasParentAddress(formatOptions)) { + result = getWordAtAddressWithParentAddress(buffer, headerSize, address, formatOptions); + } else { + result = getWordAtAddressWithoutParentAddress(buffer, headerSize, address, + formatOptions); + } + + wordCache.put(address, result); + buffer.position(originalPointer); + return result; + } + + private static int[] sGetWordBuffer = new int[FormatSpec.MAX_WORD_LENGTH]; + private static String getWordAtAddressWithParentAddress( + final FusionDictionaryBufferInterface buffer, final int headerSize, final int address, + final FormatOptions options) { + final StringBuilder builder = new StringBuilder(); + + int currentAddress = address; + int index = FormatSpec.MAX_WORD_LENGTH - 1; + // the length of the path from the root to the leaf is limited by MAX_WORD_LENGTH + for (int count = 0; count < FormatSpec.MAX_WORD_LENGTH; ++count) { + buffer.position(currentAddress + headerSize); + final CharGroupInfo currentInfo = readCharGroup(buffer, currentAddress, options); + for (int i = 0; i < currentInfo.mCharacters.length; ++i) { + sGetWordBuffer[index--] = + currentInfo.mCharacters[currentInfo.mCharacters.length - i - 1]; + } + + if (currentInfo.mParentAddress == FormatSpec.NO_PARENT_ADDRESS) break; + currentAddress = currentInfo.mParentAddress + currentInfo.mOriginalAddress; + } + + return new String(sGetWordBuffer, index + 1, FormatSpec.MAX_WORD_LENGTH - index - 1); + } + + private static String getWordAtAddressWithoutParentAddress( + final FusionDictionaryBufferInterface buffer, final int headerSize, final int address, + final FormatOptions options) { buffer.position(headerSize); final int count = readCharGroupCount(buffer); int groupOffset = getGroupCountSize(count); @@ -1228,7 +1275,7 @@ public class BinaryDictInputOutput { CharGroupInfo last = null; for (int i = count - 1; i >= 0; --i) { - CharGroupInfo info = readCharGroup(buffer, groupOffset); + CharGroupInfo info = readCharGroup(buffer, groupOffset, options); groupOffset = info.mEndAddress; if (info.mOriginalAddress == address) { builder.append(new String(info.mCharacters, 0, info.mCharacters.length)); @@ -1241,7 +1288,7 @@ public class BinaryDictInputOutput { builder.append(new String(last.mCharacters, 0, last.mCharacters.length)); buffer.position(last.mChildrenAddress + headerSize); groupOffset = last.mChildrenAddress + 1; - i = readUnsignedByte(buffer); + i = buffer.readUnsignedByte(); last = null; continue; } @@ -1251,21 +1298,19 @@ public class BinaryDictInputOutput { builder.append(new String(last.mCharacters, 0, last.mCharacters.length)); buffer.position(last.mChildrenAddress + headerSize); groupOffset = last.mChildrenAddress + 1; - i = readUnsignedByte(buffer); + i = buffer.readUnsignedByte(); last = null; continue; } } - buffer.position(originalPointer); - wordCache.put(address, result); return result; } /** - * Reads a single node from a binary file. + * Reads a single node from a buffer. * - * This methods reads the file at the current position of its file pointer. A node is - * fully expected to start at the current position. + * This methods reads the file at the current position. A node is fully expected to start at + * the current position. * This will recursively read other nodes into the structure, populating the reverse * maps on the fly and using them to keep track of already read nodes. * @@ -1273,24 +1318,26 @@ public class BinaryDictInputOutput { * @param headerSize the size, in bytes, of the file header. * @param reverseNodeMap a mapping from addresses to already read nodes. * @param reverseGroupMap a mapping from addresses to already read character groups. + * @param options file format options. * @return the read node with all his children already read. */ - private static Node readNode(final ByteBuffer buffer, final int headerSize, - final Map<Integer, Node> reverseNodeMap, final Map<Integer, CharGroup> reverseGroupMap) + private static Node readNode(final FusionDictionaryBufferInterface buffer, final int headerSize, + final Map<Integer, Node> reverseNodeMap, final Map<Integer, CharGroup> reverseGroupMap, + final FormatOptions options) throws IOException { final int nodeOrigin = buffer.position() - headerSize; final int count = readCharGroupCount(buffer); final ArrayList<CharGroup> nodeContents = new ArrayList<CharGroup>(); int groupOffset = nodeOrigin + getGroupCountSize(count); for (int i = count; i > 0; --i) { - CharGroupInfo info =readCharGroup(buffer, groupOffset); + CharGroupInfo info = readCharGroup(buffer, groupOffset, options); ArrayList<WeightedString> shortcutTargets = info.mShortcutTargets; ArrayList<WeightedString> bigrams = null; if (null != info.mBigrams) { bigrams = new ArrayList<WeightedString>(); for (PendingAttribute bigram : info.mBigrams) { final String word = getWordAtAddress( - buffer, headerSize, bigram.mAddress); + buffer, headerSize, bigram.mAddress, options); bigrams.add(new WeightedString(word, bigram.mFrequency)); } } @@ -1300,16 +1347,18 @@ public class BinaryDictInputOutput { final int currentPosition = buffer.position(); buffer.position(info.mChildrenAddress + headerSize); children = readNode( - buffer, headerSize, reverseNodeMap, reverseGroupMap); + buffer, headerSize, reverseNodeMap, reverseGroupMap, options); buffer.position(currentPosition); } nodeContents.add( - new CharGroup(info.mCharacters, shortcutTargets, - bigrams, info.mFrequency, children)); + new CharGroup(info.mCharacters, shortcutTargets, bigrams, info.mFrequency, + 0 != (info.mFlags & FormatSpec.FLAG_IS_NOT_A_WORD), + 0 != (info.mFlags & FormatSpec.FLAG_IS_BLACKLISTED), children)); } else { nodeContents.add( - new CharGroup(info.mCharacters, shortcutTargets, - bigrams, info.mFrequency)); + new CharGroup(info.mCharacters, shortcutTargets, bigrams, info.mFrequency, + 0 != (info.mFlags & FormatSpec.FLAG_IS_NOT_A_WORD), + 0 != (info.mFlags & FormatSpec.FLAG_IS_BLACKLISTED))); } groupOffset = info.mEndAddress; } @@ -1319,86 +1368,220 @@ public class BinaryDictInputOutput { return node; } + // TODO: move these methods (readUnigramsAndBigramsBinary(|Inner)) and an inner class (Position) + // out of this class. + private static class Position { + public static final int NOT_READ_GROUPCOUNT = -1; + + public int mAddress; + public int mNumOfCharGroup; + public int mPosition; + public int mLength; + + public Position(int address, int length) { + mAddress = address; + mLength = length; + mNumOfCharGroup = NOT_READ_GROUPCOUNT; + } + } + + /** + * Tours all node without recursive call. + */ + private static void readUnigramsAndBigramsBinaryInner( + final FusionDictionaryBufferInterface buffer, final int headerSize, + final Map<Integer, String> words, final Map<Integer, Integer> frequencies, + final Map<Integer, ArrayList<PendingAttribute>> bigrams, + final FormatOptions formatOptions) { + int[] pushedChars = new int[FormatSpec.MAX_WORD_LENGTH + 1]; + + Stack<Position> stack = new Stack<Position>(); + int index = 0; + + Position initPos = new Position(headerSize, 0); + stack.push(initPos); + + while (!stack.empty()) { + Position p = stack.peek(); + + if (DBG) { + MakedictLog.d("read: address=" + p.mAddress + ", numOfCharGroup=" + + p.mNumOfCharGroup + ", position=" + p.mPosition + ", length=" + p.mLength); + } + + if (buffer.position() != p.mAddress) buffer.position(p.mAddress); + if (index != p.mLength) index = p.mLength; + + if (p.mNumOfCharGroup == Position.NOT_READ_GROUPCOUNT) { + p.mNumOfCharGroup = readCharGroupCount(buffer); + p.mAddress += getGroupCountSize(p.mNumOfCharGroup); + p.mPosition = 0; + } + + CharGroupInfo info = readCharGroup(buffer, p.mAddress - headerSize, formatOptions); + for (int i = 0; i < info.mCharacters.length; ++i) { + pushedChars[index++] = info.mCharacters[i]; + } + p.mPosition++; + + if (info.mFrequency != FusionDictionary.CharGroup.NOT_A_TERMINAL) { // found word + words.put(info.mOriginalAddress, new String(pushedChars, 0, index)); + frequencies.put(info.mOriginalAddress, info.mFrequency); + if (info.mBigrams != null) bigrams.put(info.mOriginalAddress, info.mBigrams); + } + + if (p.mPosition == p.mNumOfCharGroup) { + stack.pop(); + } else { + // the node has more groups. + p.mAddress = buffer.position(); + } + + if (hasChildrenAddress(info.mChildrenAddress)) { + Position childrenPos = new Position(info.mChildrenAddress + headerSize, index); + stack.push(childrenPos); + } + } + } + + /** + * Reads unigrams and bigrams from the binary file. + * Doesn't make the memory representation of the dictionary. + * + * @param buffer the buffer to read. + * @param words the map to store the address as a key and the word as a value. + * @param frequencies the map to store the address as a key and the frequency as a value. + * @param bigrams the map to store the address as a key and the list of address as a value. + * @throws IOException + * @throws UnsupportedFormatException + */ + public static void readUnigramsAndBigramsBinary(final FusionDictionaryBufferInterface buffer, + final Map<Integer, String> words, final Map<Integer, Integer> frequencies, + final Map<Integer, ArrayList<PendingAttribute>> bigrams) throws IOException, + UnsupportedFormatException { + // Read header + final FileHeader header = readHeader(buffer); + readUnigramsAndBigramsBinaryInner(buffer, header.mHeaderSize, words, frequencies, bigrams, + header.mFormatOptions); + } + /** * Helper function to get the binary format version from the header. * @throws IOException */ - private static int getFormatVersion(final ByteBuffer buffer) throws IOException { - final int magic_v1 = readUnsignedShort(buffer); - if (VERSION_1_MAGIC_NUMBER == magic_v1) return readUnsignedByte(buffer); - final int magic_v2 = (magic_v1 << 16) + readUnsignedShort(buffer); - if (VERSION_2_MAGIC_NUMBER == magic_v2) return readUnsignedShort(buffer); - return NOT_A_VERSION_NUMBER; + private static int getFormatVersion(final FusionDictionaryBufferInterface buffer) + throws IOException { + final int magic_v1 = buffer.readUnsignedShort(); + if (FormatSpec.VERSION_1_MAGIC_NUMBER == magic_v1) return buffer.readUnsignedByte(); + final int magic_v2 = (magic_v1 << 16) + buffer.readUnsignedShort(); + if (FormatSpec.VERSION_2_MAGIC_NUMBER == magic_v2) return buffer.readUnsignedShort(); + return FormatSpec.NOT_A_VERSION_NUMBER; + } + + /** + * Helper function to get and validate the binary format version. + * @throws UnsupportedFormatException + * @throws IOException + */ + private static int checkFormatVersion(final FusionDictionaryBufferInterface buffer) + throws IOException, UnsupportedFormatException { + final int version = getFormatVersion(buffer); + if (version < FormatSpec.MINIMUM_SUPPORTED_VERSION + || version > FormatSpec.MAXIMUM_SUPPORTED_VERSION) { + throw new UnsupportedFormatException("This file has version " + version + + ", but this implementation does not support versions above " + + FormatSpec.MAXIMUM_SUPPORTED_VERSION); + } + return version; } /** - * Reads options from a file and populate a map with their contents. + * Reads a header from a buffer. + * @param buffer the buffer to read. + * @throws IOException + * @throws UnsupportedFormatException + */ + private static FileHeader readHeader(final FusionDictionaryBufferInterface buffer) + throws IOException, UnsupportedFormatException { + final int version = checkFormatVersion(buffer); + final int optionsFlags = buffer.readUnsignedShort(); + + final HashMap<String, String> attributes = new HashMap<String, String>(); + final int headerSize; + if (version < FormatSpec.FIRST_VERSION_WITH_HEADER_SIZE) { + headerSize = buffer.position(); + } else { + headerSize = buffer.readInt(); + populateOptions(buffer, headerSize, attributes); + buffer.position(headerSize); + } + + if (headerSize < 0) { + throw new UnsupportedFormatException("header size can't be negative."); + } + + 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.HAS_PARENT_ADDRESS))); + return header; + } + + /** + * Reads options from a buffer and populate a map with their contents. * - * The file is read at the current file pointer, so the caller must take care the pointer + * The buffer is read at the current position, so the caller must take care the pointer * is in the right place before calling this. */ - public static void populateOptions(final ByteBuffer buffer, final int headerSize, - final HashMap<String, String> options) { + public static void populateOptions(final FusionDictionaryBufferInterface buffer, + final int headerSize, final HashMap<String, String> options) { while (buffer.position() < headerSize) { final String key = CharEncoding.readString(buffer); final String value = CharEncoding.readString(buffer); options.put(key, value); } } + // TODO: remove this method. + public static void populateOptions(final ByteBuffer buffer, final int headerSize, + final HashMap<String, String> options) { + populateOptions(new ByteBufferWrapper(buffer), headerSize, options); + } /** - * Reads a byte buffer and returns the memory representation of the dictionary. + * Reads a buffer and returns the memory representation of the dictionary. * - * This high-level method takes a binary file and reads its contents, populating a + * This high-level method takes a buffer and reads its contents, populating a * FusionDictionary structure. The optional dict argument is an existing dictionary to - * which words from the file should be added. If it is null, a new dictionary is created. + * which words from the buffer should be added. If it is null, a new dictionary is created. * * @param buffer the buffer to read. * @param dict an optional dictionary to add words to, or null. * @return the created (or merged) dictionary. */ - public static FusionDictionary readDictionaryBinary(final ByteBuffer buffer, - final FusionDictionary dict) throws IOException, UnsupportedFormatException { - // Check file version - final int version = getFormatVersion(buffer); - if (version < MINIMUM_SUPPORTED_VERSION || version > MAXIMUM_SUPPORTED_VERSION) { - throw new UnsupportedFormatException("This file has version " + version - + ", but this implementation does not support versions above " - + MAXIMUM_SUPPORTED_VERSION); - } - + public static FusionDictionary readDictionaryBinary( + final FusionDictionaryBufferInterface buffer, final FusionDictionary dict) + throws IOException, UnsupportedFormatException { // clear cache wordCache.clear(); - // Read options - final int optionsFlags = readUnsignedShort(buffer); - - final int headerSize; - final HashMap<String, String> options = new HashMap<String, String>(); - if (version < FIRST_VERSION_WITH_HEADER_SIZE) { - headerSize = buffer.position(); - } else { - headerSize = buffer.getInt(); - populateOptions(buffer, headerSize, options); - buffer.position(headerSize); - } - - if (headerSize < 0) { - throw new UnsupportedFormatException("header size can't be negative."); - } + // Read header + final FileHeader header = readHeader(buffer); Map<Integer, Node> reverseNodeMapping = new TreeMap<Integer, Node>(); Map<Integer, CharGroup> reverseGroupMapping = new TreeMap<Integer, CharGroup>(); - final Node root = readNode( - buffer, headerSize, reverseNodeMapping, reverseGroupMapping); + final Node root = readNode(buffer, header.mHeaderSize, reverseNodeMapping, + reverseGroupMapping, header.mFormatOptions); - FusionDictionary newDict = new FusionDictionary(root, - new FusionDictionary.DictionaryOptions(options, - 0 != (optionsFlags & GERMAN_UMLAUT_PROCESSING_FLAG), - 0 != (optionsFlags & FRENCH_LIGATURE_PROCESSING_FLAG))); + FusionDictionary newDict = new FusionDictionary(root, header.mDictionaryOptions); if (null != dict) { for (final Word w : dict) { - newDict.add(w.mWord, w.mFrequency, w.mShortcutTargets); + if (w.mIsBlacklistEntry) { + newDict.addBlacklistEntry(w.mWord, w.mShortcutTargets, w.mIsNotAWord); + } else { + newDict.add(w.mWord, w.mFrequency, w.mShortcutTargets, w.mIsNotAWord); + } } for (final Word w : dict) { // By construction a binary dictionary may not have bigrams pointing to @@ -1414,28 +1597,6 @@ public class BinaryDictInputOutput { } /** - * Helper function to read one byte from ByteBuffer. - */ - private static int readUnsignedByte(final ByteBuffer buffer) { - return ((int)buffer.get()) & 0xFF; - } - - /** - * Helper function to read two byte from ByteBuffer. - */ - private static int readUnsignedShort(final ByteBuffer buffer) { - return ((int)buffer.getShort()) & 0xFFFF; - } - - /** - * Helper function to read three byte from ByteBuffer. - */ - private static int readUnsignedInt24(final ByteBuffer buffer) { - final int value = readUnsignedByte(buffer) << 16; - return value + readUnsignedShort(buffer); - } - - /** * Basic test to find out whether the file is a binary dictionary or not. * * Concretely this only tests the magic number. @@ -1450,8 +1611,9 @@ public class BinaryDictInputOutput { inStream = new FileInputStream(file); final ByteBuffer buffer = inStream.getChannel().map( FileChannel.MapMode.READ_ONLY, 0, file.length()); - final int version = getFormatVersion(buffer); - return (version >= MINIMUM_SUPPORTED_VERSION && version <= MAXIMUM_SUPPORTED_VERSION); + final int version = getFormatVersion(new ByteBufferWrapper(buffer)); + return (version >= FormatSpec.MINIMUM_SUPPORTED_VERSION + && version <= FormatSpec.MAXIMUM_SUPPORTED_VERSION); } catch (FileNotFoundException e) { return false; } catch (IOException e) { @@ -1478,8 +1640,8 @@ public class BinaryDictInputOutput { */ public static int reconstructBigramFrequency(final int unigramFrequency, final int bigramFrequency) { - final float stepSize = (MAX_TERMINAL_FREQUENCY - unigramFrequency) - / (1.5f + MAX_BIGRAM_FREQUENCY); + final float stepSize = (FormatSpec.MAX_TERMINAL_FREQUENCY - unigramFrequency) + / (1.5f + FormatSpec.MAX_BIGRAM_FREQUENCY); final float resultFreqFloat = (float)unigramFrequency + stepSize * (bigramFrequency + 1.0f); return (int)resultFreqFloat; diff --git a/java/src/com/android/inputmethod/latin/makedict/CharGroupInfo.java b/java/src/com/android/inputmethod/latin/makedict/CharGroupInfo.java index ef7dbb251..ed9388409 100644 --- a/java/src/com/android/inputmethod/latin/makedict/CharGroupInfo.java +++ b/java/src/com/android/inputmethod/latin/makedict/CharGroupInfo.java @@ -31,18 +31,20 @@ public class CharGroupInfo { public final int[] mCharacters; public final int mFrequency; public final int mChildrenAddress; + public final int mParentAddress; public final ArrayList<WeightedString> mShortcutTargets; public final ArrayList<PendingAttribute> mBigrams; public CharGroupInfo(final int originalAddress, final int endAddress, final int flags, - final int[] characters, final int frequency, final int childrenAddress, - final ArrayList<WeightedString> shortcutTargets, + final int[] characters, final int frequency, final int parentAddress, + final int childrenAddress, final ArrayList<WeightedString> shortcutTargets, final ArrayList<PendingAttribute> bigrams) { mOriginalAddress = originalAddress; mEndAddress = endAddress; mFlags = flags; mCharacters = characters; mFrequency = frequency; + mParentAddress = parentAddress; mChildrenAddress = childrenAddress; mShortcutTargets = shortcutTargets; mBigrams = bigrams; diff --git a/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java b/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java new file mode 100644 index 000000000..1707ccc39 --- /dev/null +++ b/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java @@ -0,0 +1,235 @@ +/* + * 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.makedict; + +import com.android.inputmethod.latin.Constants; +import com.android.inputmethod.latin.makedict.FusionDictionary.DictionaryOptions; + +/** + * Dictionary File Format Specification. + */ +public final class FormatSpec { + + /* + * Array of Node(FusionDictionary.Node) layout is as follows: + * + * g | + * r | the number of groups, 1 or 2 bytes. + * o | 1 byte = bbbbbbbb match + * u | case 1xxxxxxx => xxxxxxx << 8 + next byte + * p | otherwise => bbbbbbbb + * c | + * ount + * + * g | + * r | sequence of groups, + * o | the layout of each group is described below. + * u | + * ps + * + */ + + /* Node(CharGroup) layout is as follows: + * | addressType xx : mask with MASK_GROUP_ADDRESS_TYPE + * 2 bits, 00 = no children : FLAG_GROUP_ADDRESS_TYPE_NOADDRESS + * f | 01 = 1 byte : FLAG_GROUP_ADDRESS_TYPE_ONEBYTE + * l | 10 = 2 bytes : FLAG_GROUP_ADDRESS_TYPE_TWOBYTES + * a | 11 = 3 bytes : FLAG_GROUP_ADDRESS_TYPE_THREEBYTES + * g | has several chars ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_MULTIPLE_CHARS + * s | 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 + * | 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 HAS_PARENT_ADDRESS (defined in the file header) + * r | parent address, 3byte + * e | the address must be negative, so the absolute value of the address is stored. + * n | + * taddress + * + * c | IF FLAG_HAS_MULTIPLE_CHARS + * h | char, char, char, char n * (1 or 3 bytes) : use CharGroupInfo for i/o helpers + * a | end 1 byte, = 0 + * r | ELSE + * s | char 1 or 3 bytes + * | END + * + * f | + * r | IF FLAG_IS_TERMINAL + * e | frequency 1 byte + * q | + * + * c | IF 00 = FLAG_GROUP_ADDRESS_TYPE_NOADDRESS = addressType + * h | // nothing + * i | ELSIF 01 = FLAG_GROUP_ADDRESS_TYPE_ONEBYTE == addressType + * l | children address, 1 byte + * d | ELSIF 10 = FLAG_GROUP_ADDRESS_TYPE_TWOBYTES == addressType + * r | children address, 2 bytes + * e | ELSE // 11 = FLAG_GROUP_ADDRESS_TYPE_THREEBYTES = addressType + * n | children address, 3 bytes + * A | END + * d + * dress + * + * | IF FLAG_IS_TERMINAL && FLAG_HAS_SHORTCUT_TARGETS + * | shortcut string list + * | IF FLAG_IS_TERMINAL && FLAG_HAS_BIGRAMS + * | bigrams address list + * + * Char format is: + * 1 byte = bbbbbbbb match + * case 000xxxxx: xxxxx << 16 + next byte << 8 + next byte + * else: if 00011111 (= 0x1F) : this is the terminator. This is a relevant choice because + * unicode code points range from 0 to 0x10FFFF, so any 3-byte value starting with + * 00011111 would be outside unicode. + * else: iso-latin-1 code + * This allows for the whole unicode range to be encoded, including chars outside of + * the BMP. Also everything in the iso-latin-1 charset is only 1 byte, except control + * characters which should never happen anyway (and still work, but take 3 bytes). + * + * bigram address list is: + * <flags> = | hasNext = 1 bit, 1 = yes, 0 = no : FLAG_ATTRIBUTE_HAS_NEXT + * | addressSign = 1 bit, : FLAG_ATTRIBUTE_OFFSET_NEGATIVE + * | 1 = must take -address, 0 = must take +address + * | xx : mask with MASK_ATTRIBUTE_ADDRESS_TYPE + * | addressFormat = 2 bits, 00 = unused : FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE + * | 01 = 1 byte : FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE + * | 10 = 2 bytes : FLAG_ATTRIBUTE_ADDRESS_TYPE_TWOBYTES + * | 11 = 3 bytes : FLAG_ATTRIBUTE_ADDRESS_TYPE_THREEBYTES + * | 4 bits : frequency : mask with FLAG_ATTRIBUTE_FREQUENCY + * <address> | IF (01 == FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE == addressFormat) + * | read 1 byte, add top 4 bits + * | ELSIF (10 == FLAG_ATTRIBUTE_ADDRESS_TYPE_TWOBYTES == addressFormat) + * | read 2 bytes, add top 4 bits + * | ELSE // 11 == FLAG_ATTRIBUTE_ADDRESS_TYPE_THREEBYTES == addressFormat + * | read 3 bytes, add top 4 bits + * | END + * | if (FLAG_ATTRIBUTE_OFFSET_NEGATIVE) then address = -address + * if (FLAG_ATTRIBUTE_HAS_NEXT) goto bigram_and_shortcut_address_list_is + * + * shortcut string list is: + * <byte size> = GROUP_SHORTCUT_LIST_SIZE_SIZE bytes, big-endian: size of the list, in bytes. + * <flags> = | hasNext = 1 bit, 1 = yes, 0 = no : FLAG_ATTRIBUTE_HAS_NEXT + * | reserved = 3 bits, must be 0 + * | 4 bits : frequency : mask with FLAG_ATTRIBUTE_FREQUENCY + * <shortcut> = | string of characters at the char format described above, with the terminator + * | used to signal the end of the string. + * if (FLAG_ATTRIBUTE_HAS_NEXT goto flags + */ + + static final int VERSION_1_MAGIC_NUMBER = 0x78B1; + public static final int VERSION_2_MAGIC_NUMBER = 0x9BC13AFE; + static final int MINIMUM_SUPPORTED_VERSION = 1; + static final int MAXIMUM_SUPPORTED_VERSION = 3; + static final int NOT_A_VERSION_NUMBER = -1; + static final int FIRST_VERSION_WITH_HEADER_SIZE = 2; + static final int FIRST_VERSION_WITH_PARENT_ADDRESS = 3; + + // 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; + static final int HAS_PARENT_ADDRESS = 0x2; + static final int FRENCH_LIGATURE_PROCESSING_FLAG = 0x4; + static final int CONTAINS_BIGRAMS_FLAG = 0x8; + + // TODO: Make this value adaptative to content data, store it in the header, and + // use it in the reading code. + static final int MAX_WORD_LENGTH = Constants.Dictionary.MAX_WORD_LENGTH; + + static final int PARENT_ADDRESS_SIZE = 3; + + static final int MASK_GROUP_ADDRESS_TYPE = 0xC0; + static final int FLAG_GROUP_ADDRESS_TYPE_NOADDRESS = 0x00; + static final int FLAG_GROUP_ADDRESS_TYPE_ONEBYTE = 0x40; + static final int FLAG_GROUP_ADDRESS_TYPE_TWOBYTES = 0x80; + static final int FLAG_GROUP_ADDRESS_TYPE_THREEBYTES = 0xC0; + + static final int FLAG_HAS_MULTIPLE_CHARS = 0x20; + + static final int FLAG_IS_TERMINAL = 0x10; + static final int FLAG_HAS_SHORTCUT_TARGETS = 0x08; + static final int FLAG_HAS_BIGRAMS = 0x04; + static final int FLAG_IS_NOT_A_WORD = 0x02; + static final int FLAG_IS_BLACKLISTED = 0x01; + + static final int FLAG_ATTRIBUTE_HAS_NEXT = 0x80; + static final int FLAG_ATTRIBUTE_OFFSET_NEGATIVE = 0x40; + static final int MASK_ATTRIBUTE_ADDRESS_TYPE = 0x30; + static final int FLAG_ATTRIBUTE_ADDRESS_TYPE_ONEBYTE = 0x10; + static final int FLAG_ATTRIBUTE_ADDRESS_TYPE_TWOBYTES = 0x20; + static final int FLAG_ATTRIBUTE_ADDRESS_TYPE_THREEBYTES = 0x30; + static final int FLAG_ATTRIBUTE_FREQUENCY = 0x0F; + + static final int GROUP_CHARACTERS_TERMINATOR = 0x1F; + + static final int GROUP_TERMINATOR_SIZE = 1; + static final int GROUP_FLAGS_SIZE = 1; + static final int GROUP_FREQUENCY_SIZE = 1; + static final int GROUP_MAX_ADDRESS_SIZE = 3; + static final int GROUP_ATTRIBUTE_FLAGS_SIZE = 1; + static final int GROUP_ATTRIBUTE_MAX_ADDRESS_SIZE = 3; + static final int GROUP_SHORTCUT_LIST_SIZE_SIZE = 2; + + static final int NO_CHILDREN_ADDRESS = Integer.MIN_VALUE; + static final int NO_PARENT_ADDRESS = 0; + static final int INVALID_CHARACTER = -1; + + static final int MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT = 0x7F; // 127 + static final int MAX_CHARGROUPS_IN_A_NODE = 0x7FFF; // 32767 + + static final int MAX_TERMINAL_FREQUENCY = 255; + static final int MAX_BIGRAM_FREQUENCY = 15; + + /** + * Options about file format. + */ + public static class FormatOptions { + public final int mVersion; + public final boolean mHasParentAddress; + public FormatOptions(final int version) { + this(version, false); + } + public FormatOptions(final int version, final boolean hasParentAddress) { + mVersion = version; + if (version < FormatSpec.FIRST_VERSION_WITH_PARENT_ADDRESS && hasParentAddress) { + throw new RuntimeException("Parent addresses are only supported with versions " + + FormatSpec.FIRST_VERSION_WITH_PARENT_ADDRESS + " and ulterior."); + } + mHasParentAddress = hasParentAddress; + } + } + + /** + * Class representing file header. + */ + static final class FileHeader { + public final int mHeaderSize; + public final DictionaryOptions mDictionaryOptions; + public final FormatOptions mFormatOptions; + public FileHeader(final int headerSize, final DictionaryOptions dictionaryOptions, + final FormatOptions formatOptions) { + mHeaderSize = headerSize; + mDictionaryOptions = dictionaryOptions; + mFormatOptions = formatOptions; + } + } + + private FormatSpec() { + // This utility class is not publicly instantiable. + } +} diff --git a/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java b/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java index 7c15ba54d..6775de8a8 100644 --- a/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java +++ b/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java @@ -16,6 +16,8 @@ package com.android.inputmethod.latin.makedict; +import com.android.inputmethod.latin.Constants; + import java.util.ArrayList; import java.util.Arrays; import java.util.Collections; @@ -41,17 +43,15 @@ public class FusionDictionary implements Iterable<Word> { public static class Node { ArrayList<CharGroup> mData; // To help with binary generation - int mCachedSize; - int mCachedAddress; + int mCachedSize = Integer.MIN_VALUE; + int mCachedAddress = Integer.MIN_VALUE; + int mCachedParentAddress = 0; + public Node() { mData = new ArrayList<CharGroup>(); - mCachedSize = Integer.MIN_VALUE; - mCachedAddress = Integer.MIN_VALUE; } public Node(ArrayList<CharGroup> data) { mData = data; - mCachedSize = Integer.MIN_VALUE; - mCachedAddress = Integer.MIN_VALUE; } } @@ -101,26 +101,34 @@ public class FusionDictionary implements Iterable<Word> { ArrayList<WeightedString> mBigrams; int mFrequency; // NOT_A_TERMINAL == mFrequency indicates this is not a terminal. Node mChildren; + boolean mIsNotAWord; // Only a shortcut + boolean mIsBlacklistEntry; // The two following members to help with binary generation int mCachedSize; int mCachedAddress; public CharGroup(final int[] chars, final ArrayList<WeightedString> shortcutTargets, - final ArrayList<WeightedString> bigrams, final int frequency) { + final ArrayList<WeightedString> bigrams, final int frequency, + final boolean isNotAWord, final boolean isBlacklistEntry) { mChars = chars; mFrequency = frequency; mShortcutTargets = shortcutTargets; mBigrams = bigrams; mChildren = null; + mIsNotAWord = isNotAWord; + mIsBlacklistEntry = isBlacklistEntry; } public CharGroup(final int[] chars, final ArrayList<WeightedString> shortcutTargets, - final ArrayList<WeightedString> bigrams, final int frequency, final Node children) { + final ArrayList<WeightedString> bigrams, final int frequency, + final boolean isNotAWord, final boolean isBlacklistEntry, final Node children) { mChars = chars; mFrequency = frequency; mShortcutTargets = shortcutTargets; mBigrams = bigrams; mChildren = children; + mIsNotAWord = isNotAWord; + mIsBlacklistEntry = isBlacklistEntry; } public void addChild(CharGroup n) { @@ -197,8 +205,9 @@ public class FusionDictionary implements Iterable<Word> { * the existing ones if any. Note: unigram, bigram, and shortcut frequencies are only * updated if they are higher than the existing ones. */ - public void update(int frequency, ArrayList<WeightedString> shortcutTargets, - ArrayList<WeightedString> bigrams) { + public void update(final int frequency, final ArrayList<WeightedString> shortcutTargets, + final ArrayList<WeightedString> bigrams, + final boolean isNotAWord, final boolean isBlacklistEntry) { if (frequency > mFrequency) { mFrequency = frequency; } @@ -234,6 +243,8 @@ public class FusionDictionary implements Iterable<Word> { } } } + mIsNotAWord = isNotAWord; + mIsBlacklistEntry = isBlacklistEntry; } } @@ -296,10 +307,24 @@ public class FusionDictionary implements Iterable<Word> { * @param word the word to add. * @param frequency the frequency of the word, in the range [0..255]. * @param shortcutTargets a list of shortcut targets for this word, or null. + * @param isNotAWord true if this should not be considered a word (e.g. shortcut only) */ public void add(final String word, final int frequency, - final ArrayList<WeightedString> shortcutTargets) { - add(getCodePoints(word), frequency, shortcutTargets); + final ArrayList<WeightedString> shortcutTargets, final boolean isNotAWord) { + add(getCodePoints(word), frequency, shortcutTargets, isNotAWord, + false /* isBlacklistEntry */); + } + + /** + * Helper method to add a blacklist entry as a string. + * + * @param word the word to add as a blacklist entry. + * @param shortcutTargets a list of shortcut targets for this word, or null. + * @param isNotAWord true if this is not a word for spellcheking purposes (shortcut only or so) + */ + public void addBlacklistEntry(final String word, + final ArrayList<WeightedString> shortcutTargets, final boolean isNotAWord) { + add(getCodePoints(word), 0, shortcutTargets, isNotAWord, true /* isBlacklistEntry */); } /** @@ -332,7 +357,8 @@ public class FusionDictionary implements Iterable<Word> { if (charGroup != null) { final CharGroup charGroup2 = findWordInTree(mRoot, word2); if (charGroup2 == null) { - add(getCodePoints(word2), 0, null); + add(getCodePoints(word2), 0, null, false /* isNotAWord */, + false /* isBlacklistEntry */); } charGroup.addBigram(word2, frequency); } else { @@ -349,10 +375,18 @@ public class FusionDictionary implements Iterable<Word> { * @param word the word, as an int array. * @param frequency the frequency of the word, in the range [0..255]. * @param shortcutTargets an optional list of shortcut targets for this word (null if none). + * @param isNotAWord true if this is not a word for spellcheking purposes (shortcut only or so) + * @param isBlacklistEntry true if this is a blacklisted word, false otherwise */ private void add(final int[] word, final int frequency, - final ArrayList<WeightedString> shortcutTargets) { + final ArrayList<WeightedString> shortcutTargets, + final boolean isNotAWord, final boolean isBlacklistEntry) { assert(frequency >= 0 && frequency <= 255); + if (word.length >= Constants.Dictionary.MAX_WORD_LENGTH) { + MakedictLog.w("Ignoring a word that is too long: word.length = " + word.length); + return; + } + Node currentNode = mRoot; int charIndex = 0; @@ -376,7 +410,7 @@ public class FusionDictionary implements Iterable<Word> { final int insertionIndex = findInsertionIndex(currentNode, word[charIndex]); final CharGroup newGroup = new CharGroup( Arrays.copyOfRange(word, charIndex, word.length), - shortcutTargets, null /* bigrams */, frequency); + shortcutTargets, null /* bigrams */, frequency, isNotAWord, isBlacklistEntry); currentNode.mData.add(insertionIndex, newGroup); if (DBG) checkStack(currentNode); } else { @@ -386,13 +420,15 @@ public class FusionDictionary implements Iterable<Word> { // The new word is a prefix of an existing word, but the node on which it // should end already exists as is. Since the old CharNode was not a terminal, // make it one by filling in its frequency and other attributes - currentGroup.update(frequency, shortcutTargets, null); + currentGroup.update(frequency, shortcutTargets, null, isNotAWord, + isBlacklistEntry); } else { // The new word matches the full old word and extends past it. // We only have to create a new node and add it to the end of this. final CharGroup newNode = new CharGroup( Arrays.copyOfRange(word, charIndex + differentCharIndex, word.length), - shortcutTargets, null /* bigrams */, frequency); + shortcutTargets, null /* bigrams */, frequency, isNotAWord, + isBlacklistEntry); currentGroup.mChildren = new Node(); currentGroup.mChildren.mData.add(newNode); } @@ -400,7 +436,9 @@ public class FusionDictionary implements Iterable<Word> { if (0 == differentCharIndex) { // Exact same word. Update the frequency if higher. This will also add the // new shortcuts to the existing shortcut list if it already exists. - currentGroup.update(frequency, shortcutTargets, null); + currentGroup.update(frequency, shortcutTargets, null, + currentGroup.mIsNotAWord && isNotAWord, + currentGroup.mIsBlacklistEntry || isBlacklistEntry); } else { // Partial prefix match only. We have to replace the current node with a node // containing the current prefix and create two new ones for the tails. @@ -408,21 +446,26 @@ public class FusionDictionary implements Iterable<Word> { final CharGroup newOldWord = new CharGroup( Arrays.copyOfRange(currentGroup.mChars, differentCharIndex, currentGroup.mChars.length), currentGroup.mShortcutTargets, - currentGroup.mBigrams, currentGroup.mFrequency, currentGroup.mChildren); + currentGroup.mBigrams, currentGroup.mFrequency, + currentGroup.mIsNotAWord, currentGroup.mIsBlacklistEntry, + currentGroup.mChildren); newChildren.mData.add(newOldWord); final CharGroup newParent; if (charIndex + differentCharIndex >= word.length) { newParent = new CharGroup( Arrays.copyOfRange(currentGroup.mChars, 0, differentCharIndex), - shortcutTargets, null /* bigrams */, frequency, newChildren); + shortcutTargets, null /* bigrams */, frequency, + isNotAWord, isBlacklistEntry, newChildren); } else { newParent = new CharGroup( Arrays.copyOfRange(currentGroup.mChars, 0, differentCharIndex), - null /* shortcutTargets */, null /* bigrams */, -1, newChildren); + null /* shortcutTargets */, null /* bigrams */, -1, + false /* isNotAWord */, false /* isBlacklistEntry */, newChildren); final CharGroup newWord = new CharGroup(Arrays.copyOfRange(word, charIndex + differentCharIndex, word.length), - shortcutTargets, null /* bigrams */, frequency); + shortcutTargets, null /* bigrams */, frequency, + isNotAWord, isBlacklistEntry); final int addIndex = word[charIndex + differentCharIndex] > currentGroup.mChars[differentCharIndex] ? 1 : 0; newChildren.mData.add(addIndex, newWord); @@ -483,7 +526,8 @@ public class FusionDictionary implements Iterable<Word> { private static int findInsertionIndex(final Node node, int character) { final ArrayList<CharGroup> data = node.mData; final CharGroup reference = new CharGroup(new int[] { character }, - null /* shortcutTargets */, null /* bigrams */, 0); + null /* shortcutTargets */, null /* bigrams */, 0, false /* isNotAWord */, + false /* isBlacklistEntry */); int result = Collections.binarySearch(data, reference, CHARGROUP_COMPARATOR); return result >= 0 ? result : -result - 1; } @@ -689,7 +733,7 @@ public class FusionDictionary implements Iterable<Word> { // StringBuilder s = new StringBuilder(); // for (CharGroup g : node.data) { // s.append(g.frequency); -// for (int ch : g.chars){ +// for (int ch : g.chars) { // s.append(Character.toChars(ch)); // } // } @@ -748,13 +792,14 @@ public class FusionDictionary implements Iterable<Word> { } if (currentGroup.mFrequency >= 0) return new Word(mCurrentString.toString(), currentGroup.mFrequency, - currentGroup.mShortcutTargets, currentGroup.mBigrams); + currentGroup.mShortcutTargets, currentGroup.mBigrams, + currentGroup.mIsNotAWord, currentGroup.mIsBlacklistEntry); } else { mPositions.removeLast(); currentPos = mPositions.getLast(); mCurrentString.setLength(mCurrentString.length() - mPositions.getLast().length); } - } while(true); + } while (true); } @Override diff --git a/java/src/com/android/inputmethod/latin/makedict/Word.java b/java/src/com/android/inputmethod/latin/makedict/Word.java index 65fc72c40..4683ef154 100644 --- a/java/src/com/android/inputmethod/latin/makedict/Word.java +++ b/java/src/com/android/inputmethod/latin/makedict/Word.java @@ -31,16 +31,21 @@ public class Word implements Comparable<Word> { public final int mFrequency; public final ArrayList<WeightedString> mShortcutTargets; public final ArrayList<WeightedString> mBigrams; + public final boolean mIsNotAWord; + public final boolean mIsBlacklistEntry; private int mHashCode = 0; public Word(final String word, final int frequency, final ArrayList<WeightedString> shortcutTargets, - final ArrayList<WeightedString> bigrams) { + final ArrayList<WeightedString> bigrams, + final boolean isNotAWord, final boolean isBlacklistEntry) { mWord = word; mFrequency = frequency; mShortcutTargets = shortcutTargets; mBigrams = bigrams; + mIsNotAWord = isNotAWord; + mIsBlacklistEntry = isBlacklistEntry; } private static int computeHashCode(Word word) { @@ -48,7 +53,9 @@ public class Word implements Comparable<Word> { word.mWord, word.mFrequency, word.mShortcutTargets.hashCode(), - word.mBigrams.hashCode() + word.mBigrams.hashCode(), + word.mIsNotAWord, + word.mIsBlacklistEntry }); } @@ -78,7 +85,9 @@ public class Word implements Comparable<Word> { Word w = (Word)o; return mFrequency == w.mFrequency && mWord.equals(w.mWord) && mShortcutTargets.equals(w.mShortcutTargets) - && mBigrams.equals(w.mBigrams); + && mBigrams.equals(w.mBigrams) + && mIsNotAWord == w.mIsNotAWord + && mIsBlacklistEntry == w.mIsBlacklistEntry; } @Override diff --git a/java/src/com/android/inputmethod/latin/suggestions/MoreSuggestions.java b/java/src/com/android/inputmethod/latin/suggestions/MoreSuggestions.java index 58b01aa55..1f883aa60 100644 --- a/java/src/com/android/inputmethod/latin/suggestions/MoreSuggestions.java +++ b/java/src/com/android/inputmethod/latin/suggestions/MoreSuggestions.java @@ -23,7 +23,9 @@ import android.graphics.drawable.Drawable; import com.android.inputmethod.keyboard.Key; import com.android.inputmethod.keyboard.Keyboard; import com.android.inputmethod.keyboard.KeyboardSwitcher; +import com.android.inputmethod.keyboard.internal.KeyboardBuilder; import com.android.inputmethod.keyboard.internal.KeyboardIconsSet; +import com.android.inputmethod.keyboard.internal.KeyboardParams; import com.android.inputmethod.latin.R; import com.android.inputmethod.latin.SuggestedWords; import com.android.inputmethod.latin.Utils; @@ -31,145 +33,149 @@ import com.android.inputmethod.latin.Utils; public class MoreSuggestions extends Keyboard { public static final int SUGGESTION_CODE_BASE = 1024; - MoreSuggestions(Builder.MoreSuggestionsParam params) { + MoreSuggestions(final MoreSuggestionsParam params) { super(params); } - public static class Builder extends Keyboard.Builder<Builder.MoreSuggestionsParam> { - private final MoreSuggestionsView mPaneView; - private SuggestedWords mSuggestions; - private int mFromPos; - private int mToPos; + private static class MoreSuggestionsParam extends KeyboardParams { + private final int[] mWidths = new int[SuggestionStripView.MAX_SUGGESTIONS]; + private final int[] mRowNumbers = new int[SuggestionStripView.MAX_SUGGESTIONS]; + private final int[] mColumnOrders = new int[SuggestionStripView.MAX_SUGGESTIONS]; + private final int[] mNumColumnsInRow = new int[SuggestionStripView.MAX_SUGGESTIONS]; + private static final int MAX_COLUMNS_IN_ROW = 3; + private int mNumRows; + public Drawable mDivider; + public int mDividerWidth; + + public MoreSuggestionsParam() { + super(); + } - public static class MoreSuggestionsParam extends Keyboard.Params { - private final int[] mWidths = new int[SuggestionStripView.MAX_SUGGESTIONS]; - private final int[] mRowNumbers = new int[SuggestionStripView.MAX_SUGGESTIONS]; - private final int[] mColumnOrders = new int[SuggestionStripView.MAX_SUGGESTIONS]; - private final int[] mNumColumnsInRow = new int[SuggestionStripView.MAX_SUGGESTIONS]; - private static final int MAX_COLUMNS_IN_ROW = 3; - private int mNumRows; - public Drawable mDivider; - public int mDividerWidth; - - public int layout(SuggestedWords suggestions, int fromPos, int maxWidth, int minWidth, - int maxRow, MoreSuggestionsView view) { - clearKeys(); - final Resources res = view.getContext().getResources(); - mDivider = res.getDrawable(R.drawable.more_suggestions_divider); - mDividerWidth = mDivider.getIntrinsicWidth(); - final int padding = (int) res.getDimension( - R.dimen.more_suggestions_key_horizontal_padding); - final Paint paint = view.newDefaultLabelPaint(); - - int row = 0; - int pos = fromPos, rowStartPos = fromPos; - final int size = Math.min(suggestions.size(), SuggestionStripView.MAX_SUGGESTIONS); - while (pos < size) { - final String word = suggestions.getWord(pos).toString(); - // TODO: Should take care of text x-scaling. - mWidths[pos] = (int)view.getLabelWidth(word, paint) + padding; - final int numColumn = pos - rowStartPos + 1; - final int columnWidth = - (maxWidth - mDividerWidth * (numColumn - 1)) / numColumn; - if (numColumn > MAX_COLUMNS_IN_ROW - || !fitInWidth(rowStartPos, pos + 1, columnWidth)) { - if ((row + 1) >= maxRow) { - break; - } - mNumColumnsInRow[row] = pos - rowStartPos; - rowStartPos = pos; - row++; + public int layout(final SuggestedWords suggestions, final int fromPos, final int maxWidth, + final int minWidth, final int maxRow, final MoreSuggestionsView view) { + clearKeys(); + final Resources res = view.getContext().getResources(); + mDivider = res.getDrawable(R.drawable.more_suggestions_divider); + mDividerWidth = mDivider.getIntrinsicWidth(); + final int padding = (int) res.getDimension( + R.dimen.more_suggestions_key_horizontal_padding); + final Paint paint = view.newDefaultLabelPaint(); + + int row = 0; + int pos = fromPos, rowStartPos = fromPos; + final int size = Math.min(suggestions.size(), SuggestionStripView.MAX_SUGGESTIONS); + while (pos < size) { + final String word = suggestions.getWord(pos).toString(); + // TODO: Should take care of text x-scaling. + mWidths[pos] = (int)view.getLabelWidth(word, paint) + padding; + final int numColumn = pos - rowStartPos + 1; + final int columnWidth = + (maxWidth - mDividerWidth * (numColumn - 1)) / numColumn; + if (numColumn > MAX_COLUMNS_IN_ROW + || !fitInWidth(rowStartPos, pos + 1, columnWidth)) { + if ((row + 1) >= maxRow) { + break; } - mColumnOrders[pos] = pos - rowStartPos; - mRowNumbers[pos] = row; - pos++; + mNumColumnsInRow[row] = pos - rowStartPos; + rowStartPos = pos; + row++; } - mNumColumnsInRow[row] = pos - rowStartPos; - mNumRows = row + 1; - mBaseWidth = mOccupiedWidth = Math.max( - minWidth, calcurateMaxRowWidth(fromPos, pos)); - mBaseHeight = mOccupiedHeight = mNumRows * mDefaultRowHeight + mVerticalGap; - return pos - fromPos; + mColumnOrders[pos] = pos - rowStartPos; + mRowNumbers[pos] = row; + pos++; } + mNumColumnsInRow[row] = pos - rowStartPos; + mNumRows = row + 1; + mBaseWidth = mOccupiedWidth = Math.max( + minWidth, calcurateMaxRowWidth(fromPos, pos)); + mBaseHeight = mOccupiedHeight = mNumRows * mDefaultRowHeight + mVerticalGap; + return pos - fromPos; + } - private boolean fitInWidth(int startPos, int endPos, int width) { - for (int pos = startPos; pos < endPos; pos++) { - if (mWidths[pos] > width) - return false; - } - return true; + private boolean fitInWidth(final int startPos, final int endPos, final int width) { + for (int pos = startPos; pos < endPos; pos++) { + if (mWidths[pos] > width) + return false; } + return true; + } - private int calcurateMaxRowWidth(int startPos, int endPos) { - int maxRowWidth = 0; - int pos = startPos; - for (int row = 0; row < mNumRows; row++) { - final int numColumnInRow = mNumColumnsInRow[row]; - int maxKeyWidth = 0; - while (pos < endPos && mRowNumbers[pos] == row) { - maxKeyWidth = Math.max(maxKeyWidth, mWidths[pos]); - pos++; - } - maxRowWidth = Math.max(maxRowWidth, - maxKeyWidth * numColumnInRow + mDividerWidth * (numColumnInRow - 1)); + private int calcurateMaxRowWidth(final int startPos, final int endPos) { + int maxRowWidth = 0; + int pos = startPos; + for (int row = 0; row < mNumRows; row++) { + final int numColumnInRow = mNumColumnsInRow[row]; + int maxKeyWidth = 0; + while (pos < endPos && mRowNumbers[pos] == row) { + maxKeyWidth = Math.max(maxKeyWidth, mWidths[pos]); + pos++; } - return maxRowWidth; + maxRowWidth = Math.max(maxRowWidth, + maxKeyWidth * numColumnInRow + mDividerWidth * (numColumnInRow - 1)); } + return maxRowWidth; + } - private static final int[][] COLUMN_ORDER_TO_NUMBER = { - { 0, }, - { 1, 0, }, - { 2, 0, 1}, - }; - - public int getNumColumnInRow(int pos) { - return mNumColumnsInRow[mRowNumbers[pos]]; - } + private static final int[][] COLUMN_ORDER_TO_NUMBER = { + { 0, }, + { 1, 0, }, + { 2, 0, 1}, + }; - public int getColumnNumber(int pos) { - final int columnOrder = mColumnOrders[pos]; - final int numColumn = getNumColumnInRow(pos); - return COLUMN_ORDER_TO_NUMBER[numColumn - 1][columnOrder]; - } + public int getNumColumnInRow(final int pos) { + return mNumColumnsInRow[mRowNumbers[pos]]; + } - public int getX(int pos) { - final int columnNumber = getColumnNumber(pos); - return columnNumber * (getWidth(pos) + mDividerWidth); - } + public int getColumnNumber(final int pos) { + final int columnOrder = mColumnOrders[pos]; + final int numColumn = getNumColumnInRow(pos); + return COLUMN_ORDER_TO_NUMBER[numColumn - 1][columnOrder]; + } - public int getY(int pos) { - final int row = mRowNumbers[pos]; - return (mNumRows -1 - row) * mDefaultRowHeight + mTopPadding; - } + public int getX(final int pos) { + final int columnNumber = getColumnNumber(pos); + return columnNumber * (getWidth(pos) + mDividerWidth); + } - public int getWidth(int pos) { - final int numColumnInRow = getNumColumnInRow(pos); - return (mOccupiedWidth - mDividerWidth * (numColumnInRow - 1)) / numColumnInRow; - } + public int getY(final int pos) { + final int row = mRowNumbers[pos]; + return (mNumRows -1 - row) * mDefaultRowHeight + mTopPadding; + } - public void markAsEdgeKey(Key key, int pos) { - final int row = mRowNumbers[pos]; - if (row == 0) - key.markAsBottomEdge(this); - if (row == mNumRows - 1) - key.markAsTopEdge(this); + public int getWidth(final int pos) { + final int numColumnInRow = getNumColumnInRow(pos); + return (mOccupiedWidth - mDividerWidth * (numColumnInRow - 1)) / numColumnInRow; + } - final int numColumnInRow = mNumColumnsInRow[row]; - final int column = getColumnNumber(pos); - if (column == 0) - key.markAsLeftEdge(this); - if (column == numColumnInRow - 1) - key.markAsRightEdge(this); - } + public void markAsEdgeKey(final Key key, final int pos) { + final int row = mRowNumbers[pos]; + if (row == 0) + key.markAsBottomEdge(this); + if (row == mNumRows - 1) + key.markAsTopEdge(this); + + final int numColumnInRow = mNumColumnsInRow[row]; + final int column = getColumnNumber(pos); + if (column == 0) + key.markAsLeftEdge(this); + if (column == numColumnInRow - 1) + key.markAsRightEdge(this); } + } - public Builder(MoreSuggestionsView paneView) { + public static class Builder extends KeyboardBuilder<MoreSuggestionsParam> { + private final MoreSuggestionsView mPaneView; + private SuggestedWords mSuggestions; + private int mFromPos; + private int mToPos; + + public Builder(final MoreSuggestionsView paneView) { super(paneView.getContext(), new MoreSuggestionsParam()); mPaneView = paneView; } - public Builder layout(SuggestedWords suggestions, int fromPos, int maxWidth, - int minWidth, int maxRow) { + public Builder layout(final SuggestedWords suggestions, final int fromPos, + final int maxWidth, final int minWidth, final int maxRow) { final Keyboard keyboard = KeyboardSwitcher.getInstance().getKeyboard(); final int xmlId = R.xml.kbd_suggestions_pane_template; load(xmlId, keyboard.mId); @@ -183,25 +189,6 @@ public class MoreSuggestions extends Keyboard { return this; } - private static class Divider extends Key.Spacer { - private final Drawable mIcon; - - public Divider(Keyboard.Params params, Drawable icon, int x, int y, int width, - int height) { - super(params, x, y, width, height); - mIcon = icon; - } - - @Override - public Drawable getIcon(KeyboardIconsSet iconSet, int alpha) { - // KeyboardIconsSet and alpha are unused. Use the icon that has been passed to the - // constructor. - // TODO: Drawable itself should have an alpha value. - mIcon.setAlpha(128); - return mIcon; - } - } - @Override public MoreSuggestions build() { final MoreSuggestionsParam params = mParams; @@ -228,4 +215,23 @@ public class MoreSuggestions extends Keyboard { return new MoreSuggestions(params); } } + + private static class Divider extends Key.Spacer { + private final Drawable mIcon; + + public Divider(final KeyboardParams params, final Drawable icon, final int x, + final int y, final int width, final int height) { + super(params, x, y, width, height); + mIcon = icon; + } + + @Override + public Drawable getIcon(final KeyboardIconsSet iconSet, final int alpha) { + // KeyboardIconsSet and alpha are unused. Use the icon that has been passed to the + // constructor. + // TODO: Drawable itself should have an alpha value. + mIcon.setAlpha(128); + return mIcon; + } + } } diff --git a/java/src/com/android/inputmethod/latin/suggestions/SuggestionStripView.java b/java/src/com/android/inputmethod/latin/suggestions/SuggestionStripView.java index 03263d274..9e8ab81b0 100644 --- a/java/src/com/android/inputmethod/latin/suggestions/SuggestionStripView.java +++ b/java/src/com/android/inputmethod/latin/suggestions/SuggestionStripView.java @@ -61,6 +61,7 @@ import com.android.inputmethod.latin.AutoCorrection; import com.android.inputmethod.latin.CollectionUtils; import com.android.inputmethod.latin.LatinImeLogger; import com.android.inputmethod.latin.R; +import com.android.inputmethod.latin.ResourceUtils; import com.android.inputmethod.latin.StaticInnerHandlerWrapper; import com.android.inputmethod.latin.SuggestedWords; import com.android.inputmethod.latin.Utils; @@ -196,15 +197,15 @@ public class SuggestionStripView extends RelativeLayout implements OnClickListen R.styleable.SuggestionStripView, defStyle, R.style.SuggestionStripViewStyle); mSuggestionStripOption = a.getInt( R.styleable.SuggestionStripView_suggestionStripOption, 0); - final float alphaValidTypedWord = getFraction(a, + final float alphaValidTypedWord = ResourceUtils.getFraction(a, R.styleable.SuggestionStripView_alphaValidTypedWord, 1.0f); - final float alphaTypedWord = getFraction(a, + final float alphaTypedWord = ResourceUtils.getFraction(a, R.styleable.SuggestionStripView_alphaTypedWord, 1.0f); - final float alphaAutoCorrect = getFraction(a, + final float alphaAutoCorrect = ResourceUtils.getFraction(a, R.styleable.SuggestionStripView_alphaAutoCorrect, 1.0f); - final float alphaSuggested = getFraction(a, + final float alphaSuggested = ResourceUtils.getFraction(a, R.styleable.SuggestionStripView_alphaSuggested, 1.0f); - mAlphaObsoleted = getFraction(a, + mAlphaObsoleted = ResourceUtils.getFraction(a, R.styleable.SuggestionStripView_alphaSuggested, 1.0f); mColorValidTypedWord = applyAlpha(a.getColor( R.styleable.SuggestionStripView_colorValidTypedWord, 0), alphaValidTypedWord); @@ -217,13 +218,13 @@ public class SuggestionStripView extends RelativeLayout implements OnClickListen mSuggestionsCountInStrip = a.getInt( R.styleable.SuggestionStripView_suggestionsCountInStrip, DEFAULT_SUGGESTIONS_COUNT_IN_STRIP); - mCenterSuggestionWeight = getFraction(a, + mCenterSuggestionWeight = ResourceUtils.getFraction(a, R.styleable.SuggestionStripView_centerSuggestionPercentile, DEFAULT_CENTER_SUGGESTION_PERCENTILE); mMaxMoreSuggestionsRow = a.getInt( R.styleable.SuggestionStripView_maxMoreSuggestionsRow, DEFAULT_MAX_MORE_SUGGESTIONS_ROW); - mMinMoreSuggestionsWidth = getFraction(a, + mMinMoreSuggestionsWidth = ResourceUtils.getFraction(a, R.styleable.SuggestionStripView_minMoreSuggestionsWidth, 1.0f); a.recycle(); @@ -278,10 +279,6 @@ public class SuggestionStripView extends RelativeLayout implements OnClickListen return new BitmapDrawable(res, buffer); } - static float getFraction(final TypedArray a, final int index, final float defValue) { - return a.getFraction(index, 1, 1, defValue); - } - private CharSequence getStyledSuggestionWord(SuggestedWords suggestedWords, int pos) { final CharSequence word = suggestedWords.getWord(pos); final boolean isAutoCorrect = pos == 1 && suggestedWords.willAutoCorrect(); |