diff options
Diffstat (limited to 'java/src/com/android/inputmethod/latin')
8 files changed, 331 insertions, 58 deletions
diff --git a/java/src/com/android/inputmethod/latin/BinaryDictionary.java b/java/src/com/android/inputmethod/latin/BinaryDictionary.java index 448d25c73..878633ee0 100644 --- a/java/src/com/android/inputmethod/latin/BinaryDictionary.java +++ b/java/src/com/android/inputmethod/latin/BinaryDictionary.java @@ -42,17 +42,13 @@ public final class BinaryDictionary extends Dictionary { * really long words. */ private static final int MAX_WORD_LENGTH = Constants.Dictionary.MAX_WORD_LENGTH; - private static final int MAX_WORDS = 18; - private static final int MAX_SPACES = 16; - - private static final int MAX_PREDICTIONS = 60; - private static final int MAX_RESULTS = Math.max(MAX_PREDICTIONS, MAX_WORDS); + private static final int MAX_RESULTS = 18; /* Must be identical to MAX_RESULTS in defines.h */ private long mNativeDict; private final Locale mLocale; private final int[] mInputCodePoints = new int[MAX_WORD_LENGTH]; private final int[] mOutputCodePoints = new int[MAX_WORD_LENGTH * MAX_RESULTS]; - private final int[] mSpaceIndices = new int[MAX_SPACES]; + private final int[] mSpaceIndices = new int[MAX_RESULTS]; private final int[] mOutputScores = new int[MAX_RESULTS]; private final int[] mOutputTypes = new int[MAX_RESULTS]; @@ -80,16 +76,14 @@ public final class BinaryDictionary extends Dictionary { /** * Constructor for the binary dictionary. This is supposed to be called from the * dictionary factory. - * @param context the context to access the environment from. * @param filename the name of the file to read through native code. * @param offset the offset of the dictionary data within the file. * @param length the length of the binary data. * @param useFullEditDistance whether to use the full edit distance in suggestions * @param dictType the dictionary type, as a human-readable string */ - public BinaryDictionary(final Context context, final String filename, final long offset, - final long length, final boolean useFullEditDistance, final Locale locale, - final String dictType) { + public BinaryDictionary(final String filename, final long offset, final long length, + final boolean useFullEditDistance, final Locale locale, final String dictType) { super(dictType); mLocale = locale; mUseFullEditDistance = useFullEditDistance; @@ -101,7 +95,7 @@ public final class BinaryDictionary extends Dictionary { } private native long openNative(String sourceDir, long dictOffset, long dictSize, - int maxWordLength, int maxWords, int maxPredictions); + int maxWordLength); private native void closeNative(long dict); private native int getFrequencyNative(long dict, int[] word); private native boolean isValidBigramNative(long dict, int[] word1, int[] word2); @@ -116,8 +110,7 @@ public final class BinaryDictionary extends Dictionary { // TODO: Move native dict into session private final void loadDictionary(final String path, final long startOffset, final long length) { - mNativeDict = openNative(path, startOffset, length, MAX_WORD_LENGTH, MAX_WORDS, - MAX_PREDICTIONS); + mNativeDict = openNative(path, startOffset, length, MAX_WORD_LENGTH); } @Override @@ -148,14 +141,12 @@ public final class BinaryDictionary extends Dictionary { final InputPointers ips = composer.getInputPointers(); final int codesSize = isGesture ? ips.getPointerSize() : composerSize; // proximityInfo and/or prevWordForBigrams may not be null. - final int tmpCount = getSuggestionsNative(mNativeDict, - proximityInfo.getNativeProximityInfo(), getTraverseSession(sessionId).getSession(), - ips.getXCoordinates(), ips.getYCoordinates(), ips.getTimes(), ips.getPointerIds(), - mInputCodePoints, codesSize, 0 /* commitPoint */, isGesture, prevWordCodePointArray, + final int count = getSuggestionsNative(mNativeDict, proximityInfo.getNativeProximityInfo(), + getTraverseSession(sessionId).getSession(), ips.getXCoordinates(), + ips.getYCoordinates(), ips.getTimes(), ips.getPointerIds(), mInputCodePoints, + codesSize, 0 /* commitPoint */, isGesture, prevWordCodePointArray, mUseFullEditDistance, mOutputCodePoints, mOutputScores, mSpaceIndices, mOutputTypes); - final int count = Math.min(tmpCount, MAX_PREDICTIONS); - final ArrayList<SuggestedWordInfo> suggestions = CollectionUtils.newArrayList(); for (int j = 0; j < count; ++j) { if (composerSize > 0 && mOutputScores[j] < 1) break; diff --git a/java/src/com/android/inputmethod/latin/DictionaryFactory.java b/java/src/com/android/inputmethod/latin/DictionaryFactory.java index f381973ae..22cf5b38d 100644 --- a/java/src/com/android/inputmethod/latin/DictionaryFactory.java +++ b/java/src/com/android/inputmethod/latin/DictionaryFactory.java @@ -58,9 +58,8 @@ public final class DictionaryFactory { BinaryDictionaryGetter.getDictionaryFiles(locale, context); if (null != assetFileList) { for (final AssetFileAddress f : assetFileList) { - final BinaryDictionary binaryDictionary = - new BinaryDictionary(context, f.mFilename, f.mOffset, f.mLength, - useFullEditDistance, locale, Dictionary.TYPE_MAIN); + final BinaryDictionary binaryDictionary = new BinaryDictionary(f.mFilename, + f.mOffset, f.mLength, useFullEditDistance, locale, Dictionary.TYPE_MAIN); if (binaryDictionary.isValidDictionary()) { dictList.add(binaryDictionary); } @@ -112,7 +111,7 @@ public final class DictionaryFactory { Log.e(TAG, "sourceDir is not a file: " + sourceDir); return null; } - return new BinaryDictionary(context, sourceDir, afd.getStartOffset(), afd.getLength(), + return new BinaryDictionary(sourceDir, afd.getStartOffset(), afd.getLength(), false /* useFullEditDistance */, locale, Dictionary.TYPE_MAIN); } catch (android.content.res.Resources.NotFoundException e) { Log.e(TAG, "Could not find the resource"); @@ -130,17 +129,16 @@ public final class DictionaryFactory { /** * Create a dictionary from passed data. This is intended for unit tests only. - * @param context the test context to create this data from. * @param dictionary the file to read * @param startOffset the offset in the file where the data starts * @param length the length of the data * @param useFullEditDistance whether to use the full edit distance in suggestions * @return the created dictionary, or null. */ - public static Dictionary createDictionaryForTest(Context context, File dictionary, - long startOffset, long length, final boolean useFullEditDistance, Locale locale) { + public static Dictionary createDictionaryForTest(File dictionary, long startOffset, long length, + final boolean useFullEditDistance, Locale locale) { if (dictionary.isFile()) { - return new BinaryDictionary(context, dictionary.getAbsolutePath(), startOffset, length, + return new BinaryDictionary(dictionary.getAbsolutePath(), startOffset, length, useFullEditDistance, locale, Dictionary.TYPE_MAIN); } else { Log.e(TAG, "Could not find the file. path=" + dictionary.getAbsolutePath()); diff --git a/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java b/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java index 47adaa8ed..2c7fdcc93 100644 --- a/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java +++ b/java/src/com/android/inputmethod/latin/ExpandableBinaryDictionary.java @@ -279,9 +279,8 @@ abstract public class ExpandableBinaryDictionary extends Dictionary { final long length = file.length(); // Build the new binary dictionary - final BinaryDictionary newBinaryDictionary = - new BinaryDictionary(mContext, filename, 0, length, true /* useFullEditDistance */, - null, mDictType); + final BinaryDictionary newBinaryDictionary = new BinaryDictionary(filename, 0, length, + true /* useFullEditDistance */, null, mDictType); if (mBinaryDictionary != null) { // Ensure all threads accessing the current dictionary have finished before swapping in diff --git a/java/src/com/android/inputmethod/latin/LatinIME.java b/java/src/com/android/inputmethod/latin/LatinIME.java index df733c55a..4a3511472 100644 --- a/java/src/com/android/inputmethod/latin/LatinIME.java +++ b/java/src/com/android/inputmethod/latin/LatinIME.java @@ -1301,13 +1301,13 @@ public final class LatinIME extends InputMethodService implements KeyboardAction } private void sendKeyCodePoint(final int code) { + if (ProductionFlag.IS_EXPERIMENTAL) { + ResearchLogger.latinIME_sendKeyCodePoint(code); + } // TODO: Remove this special handling of digit letters. // For backward compatibility. See {@link InputMethodService#sendKeyChar(char)}. if (code >= '0' && code <= '9') { sendDownUpKeyEventForBackwardCompatibility(code - '0' + KeyEvent.KEYCODE_0); - if (ProductionFlag.IS_EXPERIMENTAL) { - ResearchLogger.latinIME_sendKeyCodePoint(code); - } return; } @@ -1327,6 +1327,9 @@ public final class LatinIME extends InputMethodService implements KeyboardAction // Implementation of {@link KeyboardActionListener}. @Override public void onCodeInput(final int primaryCode, final int x, final int y) { + if (ProductionFlag.IS_EXPERIMENTAL) { + ResearchLogger.latinIME_onCodeInput(primaryCode, x, y); + } final long when = SystemClock.uptimeMillis(); if (primaryCode != Constants.CODE_DELETE || when > mLastKeyTime + QUICK_PRESS) { mDeleteCount = 0; @@ -1420,9 +1423,6 @@ public final class LatinIME extends InputMethodService implements KeyboardAction mEnteredText = null; } mConnection.endBatchEdit(); - if (ProductionFlag.IS_EXPERIMENTAL) { - ResearchLogger.latinIME_onCodeInput(primaryCode, x, y); - } } // Called from PointerTracker through the KeyboardActionListener interface @@ -1667,7 +1667,7 @@ public final class LatinIME extends InputMethodService implements KeyboardAction if (mWordComposer.isBatchMode()) { if (ProductionFlag.IS_EXPERIMENTAL) { final String word = mWordComposer.getTypedWord(); - ResearchLogger.latinIME_handleBackspace_batch(word); + ResearchLogger.latinIME_handleBackspace_batch(word, 1); ResearchLogger.getInstance().uncommitCurrentLogUnit( word, false /* dumpCurrentLogUnit */); } @@ -1718,14 +1718,17 @@ public final class LatinIME extends InputMethodService implements KeyboardAction // We should backspace one char and restart suggestion if at the end of a word. if (mLastSelectionStart != mLastSelectionEnd) { // If there is a selection, remove it. - final int lengthToDelete = mLastSelectionEnd - mLastSelectionStart; + final int numCharsDeleted = mLastSelectionEnd - mLastSelectionStart; mConnection.setSelection(mLastSelectionEnd, mLastSelectionEnd); // Reset mLastSelectionEnd to mLastSelectionStart. This is what is supposed to // happen, and if it's wrong, the next call to onUpdateSelection will correct it, // but we want to set it right away to avoid it being used with the wrong values // later (typically, in a subsequent press on backspace). mLastSelectionEnd = mLastSelectionStart; - mConnection.deleteSurroundingText(lengthToDelete, 0); + mConnection.deleteSurroundingText(numCharsDeleted, 0); + if (ProductionFlag.IS_EXPERIMENTAL) { + ResearchLogger.latinIME_handleBackspace(numCharsDeleted); + } } else { // There is no selection, just delete one character. if (NOT_A_CURSOR_POSITION == mLastSelectionEnd) { @@ -1742,8 +1745,14 @@ public final class LatinIME extends InputMethodService implements KeyboardAction } else { mConnection.deleteSurroundingText(1, 0); } + if (ProductionFlag.IS_EXPERIMENTAL) { + ResearchLogger.latinIME_handleBackspace(1); + } if (mDeleteCount > DELETE_ACCELERATE_AT) { mConnection.deleteSurroundingText(1, 0); + if (ProductionFlag.IS_EXPERIMENTAL) { + ResearchLogger.latinIME_handleBackspace(1); + } } } if (mSettings.getCurrent().isSuggestionsRequested(mDisplayOrientation)) { @@ -1843,6 +1852,9 @@ public final class LatinIME extends InputMethodService implements KeyboardAction // Returns true if we did an autocorrection, false otherwise. private boolean handleSeparator(final int primaryCode, final int x, final int y, final int spaceState) { + if (ProductionFlag.IS_EXPERIMENTAL) { + ResearchLogger.latinIME_handleSeparator(); + } boolean didAutoCorrect = false; // Handle separator if (mWordComposer.isComposingWord()) { @@ -2122,7 +2134,7 @@ public final class LatinIME extends InputMethodService implements KeyboardAction Constants.SUGGESTION_STRIP_COORDINATE, Constants.SUGGESTION_STRIP_COORDINATE); if (ProductionFlag.IS_EXPERIMENTAL) { ResearchLogger.latinIME_punctuationSuggestion(index, suggestion, - false /* isBatchMode */); + false /* isBatchMode */, suggestedWords.mIsPrediction); } return; } @@ -2309,7 +2321,9 @@ public final class LatinIME extends InputMethodService implements KeyboardAction } if (ProductionFlag.IS_EXPERIMENTAL) { ResearchLogger.latinIME_revertCommit(committedWord, originallyTypedWord, - mWordComposer.isBatchMode()); + mWordComposer.isBatchMode(), mLastComposedWord.mSeparatorString); + ResearchLogger.getInstance().uncommitCurrentLogUnit(committedWord, + true /* dumpCurrentLogUnit */); } // Don't restart suggestion yet. We'll restart if the user deletes the // separator. @@ -2322,6 +2336,9 @@ public final class LatinIME extends InputMethodService implements KeyboardAction public void promotePhantomSpace() { if (mSettings.getCurrent().shouldInsertSpacesAutomatically()) { sendKeyCodePoint(Constants.CODE_SPACE); + if (ProductionFlag.IS_EXPERIMENTAL) { + ResearchLogger.latinIME_promotePhantomSpace(); + } } } diff --git a/java/src/com/android/inputmethod/latin/Suggest.java b/java/src/com/android/inputmethod/latin/Suggest.java index 3dc2ba95b..2abf75da4 100644 --- a/java/src/com/android/inputmethod/latin/Suggest.java +++ b/java/src/com/android/inputmethod/latin/Suggest.java @@ -72,9 +72,8 @@ public final class Suggest { } @UsedForTesting - Suggest(final Context context, final File dictionary, - final long startOffset, final long length, final Locale locale) { - final Dictionary mainDict = DictionaryFactory.createDictionaryForTest(context, dictionary, + Suggest(final File dictionary, final long startOffset, final long length, final Locale locale) { + final Dictionary mainDict = DictionaryFactory.createDictionaryForTest(dictionary, startOffset, length /* useFullEditDistance */, false, locale); mLocale = locale; mMainDictionary = mainDict; diff --git a/java/src/com/android/inputmethod/latin/spellcheck/AndroidSpellCheckerService.java b/java/src/com/android/inputmethod/latin/spellcheck/AndroidSpellCheckerService.java index 89d6c9010..907c0cdca 100644 --- a/java/src/com/android/inputmethod/latin/spellcheck/AndroidSpellCheckerService.java +++ b/java/src/com/android/inputmethod/latin/spellcheck/AndroidSpellCheckerService.java @@ -81,6 +81,7 @@ public final class AndroidSpellCheckerService extends SpellCheckerService public static final int SCRIPT_LATIN = 0; public static final int SCRIPT_CYRILLIC = 1; + public static final int SCRIPT_GREEK = 2; public static final String SINGLE_QUOTE = "\u0027"; public static final String APOSTROPHE = "\u2019"; private static final TreeMap<String, Integer> mLanguageToScript; @@ -94,18 +95,23 @@ public final class AndroidSpellCheckerService extends SpellCheckerService // IMPORTANT: this only contains languages - do not write countries in there. // Only the language is searched from the map. mLanguageToScript = CollectionUtils.newTreeMap(); - mLanguageToScript.put("en", SCRIPT_LATIN); - mLanguageToScript.put("fr", SCRIPT_LATIN); - mLanguageToScript.put("de", SCRIPT_LATIN); - mLanguageToScript.put("nl", SCRIPT_LATIN); mLanguageToScript.put("cs", SCRIPT_LATIN); + mLanguageToScript.put("da", SCRIPT_LATIN); + mLanguageToScript.put("de", SCRIPT_LATIN); + mLanguageToScript.put("el", SCRIPT_GREEK); + mLanguageToScript.put("en", SCRIPT_LATIN); mLanguageToScript.put("es", SCRIPT_LATIN); - mLanguageToScript.put("it", SCRIPT_LATIN); + mLanguageToScript.put("fi", SCRIPT_LATIN); + mLanguageToScript.put("fr", SCRIPT_LATIN); mLanguageToScript.put("hr", SCRIPT_LATIN); + mLanguageToScript.put("it", SCRIPT_LATIN); + mLanguageToScript.put("lt", SCRIPT_LATIN); + mLanguageToScript.put("lv", SCRIPT_LATIN); + mLanguageToScript.put("nb", SCRIPT_LATIN); + mLanguageToScript.put("nl", SCRIPT_LATIN); mLanguageToScript.put("pt", SCRIPT_LATIN); + mLanguageToScript.put("sl", SCRIPT_LATIN); mLanguageToScript.put("ru", SCRIPT_CYRILLIC); - // TODO: Make a persian proximity, and activate the Farsi subtype. - // mLanguageToScript.put("fa", SCRIPT_PERSIAN); } @Override public void onCreate() { diff --git a/java/src/com/android/inputmethod/latin/spellcheck/AndroidWordLevelSpellCheckerSession.java b/java/src/com/android/inputmethod/latin/spellcheck/AndroidWordLevelSpellCheckerSession.java index 470943be1..6581978c9 100644 --- a/java/src/com/android/inputmethod/latin/spellcheck/AndroidWordLevelSpellCheckerSession.java +++ b/java/src/com/android/inputmethod/latin/spellcheck/AndroidWordLevelSpellCheckerSession.java @@ -143,8 +143,17 @@ public abstract class AndroidWordLevelSpellCheckerSession extends Session { case AndroidSpellCheckerService.SCRIPT_CYRILLIC: // All Cyrillic characters are in the 400~52F block. There are some in the upper // Unicode range, but they are archaic characters that are not used in modern - // russian and are not used by our dictionary. + // Russian and are not used by our dictionary. return codePoint >= 0x400 && codePoint <= 0x52F && Character.isLetter(codePoint); + case AndroidSpellCheckerService.SCRIPT_GREEK: + // Greek letters are either in the 370~3FF range (Greek & Coptic), or in the + // 1F00~1FFF range (Greek extended). Our dictionary contains both sort of characters. + // Our dictionary also contains a few words with 0xF2; it would be best to check + // if that's correct, but a Google search does return results for these words so + // they are probably okay. + return (codePoint >= 0x370 && codePoint <= 0x3FF) + || (codePoint >= 0x1F00 && codePoint <= 0x1FFF) + || codePoint == 0xF2; default: // Should never come here throw new RuntimeException("Impossible value of script: " + script); diff --git a/java/src/com/android/inputmethod/latin/spellcheck/SpellCheckerProximityInfo.java b/java/src/com/android/inputmethod/latin/spellcheck/SpellCheckerProximityInfo.java index 6c0d79c2b..572a826a5 100644 --- a/java/src/com/android/inputmethod/latin/spellcheck/SpellCheckerProximityInfo.java +++ b/java/src/com/android/inputmethod/latin/spellcheck/SpellCheckerProximityInfo.java @@ -73,6 +73,12 @@ public final class SpellCheckerProximityInfo { // to spell check has been entered with one of the keyboards above. Also, specifically // to English, many spelling errors consist of the last vowel of the word being wrong // because in English vowels tend to merge with each other in pronunciation. + /* + The Qwerty layout this represents looks like the following: + q w e r t y u i o p + a s d f g h j k l + z x c v b n m + */ final static int[] PROXIMITY = { // Proximity for row 1. This must have exactly ROW_SIZE entries for each letter, // and exactly PROXIMITY_GRID_WIDTH letters for a row. Pad with NUL's. @@ -127,10 +133,13 @@ public final class SpellCheckerProximityInfo { final private static TreeMap<Integer, Integer> INDICES = CollectionUtils.newTreeMap(); // TODO: The following table is solely based on the keyboard layout. Consult with Russian // speakers on commonly misspelled words/letters. - final static int[] PROXIMITY = { - // Proximity for row 1. This must have exactly ROW_SIZE entries for each letter, - // and exactly PROXIMITY_GRID_WIDTH letters for a row. Pad with NUL's. - // The number of rows must be exactly PROXIMITY_GRID_HEIGHT. + /* + The Russian layout this represents looks like the following: + й ц у к е н г ш щ з х + ф ы в а п р о л д ж э + я ч с м и т ь б ю + + This gives us the following table: 'й', 'ц', 'ф', 'ы', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, 'ц', 'й', 'ф', 'ы', 'в', 'у', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, 'у', 'ц', 'ы', 'в', 'а', 'к', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, @@ -143,7 +152,6 @@ public final class SpellCheckerProximityInfo { 'з', 'щ', 'д', 'ж', 'э', 'х', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, 'х', 'з', 'ж', 'э', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, - // Proximity for row 2. See comment above about size. 'ф', 'й', 'ц', 'ы', 'я', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, 'ы', 'й', 'ц', 'у', 'ф', 'в', 'я', 'ч', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, 'в', 'ц', 'у', 'к', 'ы', 'а', 'я', 'ч', 'с', NUL, NUL, NUL, NUL, NUL, NUL, NUL, @@ -156,7 +164,6 @@ public final class SpellCheckerProximityInfo { 'ж', 'щ', 'з', 'х', 'д', 'э', 'б', 'ю', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, 'э', 'з', 'х', 'ю', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, - // Proximity for row 3. See comment above about size. 'я', 'ф', 'ы', 'в', 'ч', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, 'ч', 'ы', 'в', 'а', 'я', 'с', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, 'с', 'в', 'а', 'п', 'ч', 'м', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, @@ -166,6 +173,249 @@ public final class SpellCheckerProximityInfo { 'ь', 'о', 'л', 'д', 'т', 'б', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, 'б', 'л', 'д', 'ж', 'ь', 'ю', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, 'ю', 'д', 'ж', 'э', 'б', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + + Using the following characters: + */ + private static final int CY_SHORT_I = '\u0439'; // й + private static final int CY_TSE = '\u0446'; // ц + private static final int CY_U = '\u0443'; // у + private static final int CY_KA = '\u043A'; // к + private static final int CY_IE = '\u0435'; // е + private static final int CY_EN = '\u043D'; // н + private static final int CY_GHE = '\u0433'; // г + private static final int CY_SHA = '\u0448'; // ш + private static final int CY_SHCHA = '\u0449'; // щ + private static final int CY_ZE = '\u0437'; // з + private static final int CY_HA = '\u0445'; // х + private static final int CY_EF = '\u0444'; // ф + private static final int CY_YERU = '\u044B'; // ы + private static final int CY_VE = '\u0432'; // в + private static final int CY_A = '\u0430'; // а + private static final int CY_PE = '\u043F'; // п + private static final int CY_ER = '\u0440'; // р + private static final int CY_O = '\u043E'; // о + private static final int CY_EL = '\u043B'; // л + private static final int CY_DE = '\u0434'; // д + private static final int CY_ZHE = '\u0436'; // ж + private static final int CY_E = '\u044D'; // э + private static final int CY_YA = '\u044F'; // я + private static final int CY_CHE = '\u0447'; // ч + private static final int CY_ES = '\u0441'; // с + private static final int CY_EM = '\u043C'; // м + private static final int CY_I = '\u0438'; // и + private static final int CY_TE = '\u0442'; // т + private static final int CY_SOFT_SIGN = '\u044C'; // ь + private static final int CY_BE = '\u0431'; // б + private static final int CY_YU = '\u044E'; // ю + final static int[] PROXIMITY = { + // Proximity for row 1. This must have exactly ROW_SIZE entries for each letter, + // and exactly PROXIMITY_GRID_WIDTH letters for a row. Pad with NUL's. + // The number of rows must be exactly PROXIMITY_GRID_HEIGHT. + CY_SHORT_I, CY_TSE, CY_EF, CY_YERU, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_TSE, CY_SHORT_I, CY_EF, CY_YERU, CY_VE, CY_U, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_U, CY_TSE, CY_YERU, CY_VE, CY_A, CY_KA, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_KA, CY_U, CY_VE, CY_A, CY_PE, CY_IE, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_IE, CY_KA, CY_A, CY_PE, CY_ER, CY_EN, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_EN, CY_IE, CY_PE, CY_ER, CY_O, CY_GHE, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_GHE, CY_EN, CY_ER, CY_O, CY_EL, CY_SHA, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_SHA, CY_GHE, CY_O, CY_EL, CY_DE, CY_SHCHA, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_SHCHA, CY_SHA, CY_EL, CY_DE, CY_ZHE, CY_ZE, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_ZE, CY_SHCHA, CY_DE, CY_ZHE, CY_E, CY_HA, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_HA, CY_ZE, CY_ZHE, CY_E, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + + // Proximity for row 2. See comment above about size. + CY_EF, CY_SHORT_I, CY_TSE, CY_YERU, CY_YA, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_YERU, CY_SHORT_I, CY_TSE, CY_U, CY_EF, CY_VE, CY_YA, CY_CHE, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_VE, CY_TSE, CY_U, CY_KA, CY_YERU, CY_A, CY_YA, CY_CHE, + CY_ES, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_A, CY_U, CY_KA, CY_IE, CY_VE, CY_PE, CY_CHE, CY_ES, + CY_EM, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_PE, CY_KA, CY_IE, CY_EN, CY_A, CY_ER, CY_ES, CY_EM, + CY_I, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_ER, CY_IE, CY_EN, CY_GHE, CY_PE, CY_O, CY_EM, CY_I, + CY_TE, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_O, CY_EN, CY_GHE, CY_SHA, CY_ER, CY_EL, CY_I, CY_TE, + CY_SOFT_SIGN, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_EL, CY_GHE, CY_SHA, CY_SHCHA, CY_O, CY_DE, CY_TE, CY_SOFT_SIGN, + CY_BE, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_DE, CY_SHA, CY_SHCHA, CY_ZE, CY_EL, CY_ZHE, CY_SOFT_SIGN, CY_BE, + CY_YU, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_ZHE, CY_SHCHA, CY_ZE, CY_HA, CY_DE, CY_E, CY_BE, CY_YU, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_E, CY_ZE, CY_HA, CY_YU, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + + // Proximity for row 3. See comment above about size. + CY_YA, CY_EF, CY_YERU, CY_VE, CY_CHE, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_CHE, CY_YERU, CY_VE, CY_A, CY_YA, CY_ES, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_ES, CY_VE, CY_A, CY_PE, CY_CHE, CY_EM, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_EM, CY_A, CY_PE, CY_ER, CY_ES, CY_I, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_I, CY_PE, CY_ER, CY_O, CY_EM, CY_TE, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_TE, CY_ER, CY_O, CY_EL, CY_I, CY_SOFT_SIGN, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_SOFT_SIGN, CY_O, CY_EL, CY_DE, CY_TE, CY_BE, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_BE, CY_EL, CY_DE, CY_ZHE, CY_SOFT_SIGN, CY_YU, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + CY_YU, CY_DE, CY_ZHE, CY_E, CY_BE, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + }; + static { + buildProximityIndices(PROXIMITY, INDICES); + } + static int getIndexOf(int characterCode) { + return computeIndex(characterCode, INDICES); + } + } + + private static final class Greek { + final private static TreeMap<Integer, Integer> INDICES = CollectionUtils.newTreeMap(); + // TODO: The following table is solely based on the keyboard layout. Consult with Greek + // speakers on commonly misspelled words/letters. + /* + The Greek layout this represents looks like the following: + ; ς ε ρ τ υ θ ι ο π + α σ δ φ γ η ξ κ λ + ζ χ ψ ω β ν μ + + This gives us the following table: + 'ς', 'ε', 'α', 'σ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'ε', 'ς', 'ρ', 'σ', 'δ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'ρ', 'ε', 'τ', 'δ', 'φ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'τ', 'ρ', 'υ', 'φ', 'γ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'υ', 'τ', 'θ', 'γ', 'η', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'θ', 'υ', 'ι', 'η', 'ξ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'ι', 'θ', 'ο', 'ξ', 'κ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'ο', 'ι', 'π', 'κ', 'λ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'π', 'ο', 'λ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + + 'α', 'ς', 'σ', 'ζ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'σ', 'ς', 'ε', 'α', 'δ', 'ζ', 'χ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'δ', 'ε', 'ρ', 'σ', 'φ', 'ζ', 'χ', 'ψ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'φ', 'ρ', 'τ', 'δ', 'γ', 'χ', 'ψ', 'ω', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'γ', 'τ', 'υ', 'φ', 'η', 'ψ', 'ω', 'β', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'η', 'υ', 'θ', 'γ', 'ξ', 'ω', 'β', 'ν', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'ξ', 'θ', 'ι', 'η', 'κ', 'β', 'ν', 'μ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'κ', 'ι', 'ο', 'ξ', 'λ', 'ν', 'μ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'λ', 'ο', 'π', 'κ', 'μ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + + 'ζ', 'α', 'σ', 'δ', 'χ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'χ', 'σ', 'δ', 'φ', 'ζ', 'ψ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'ψ', 'δ', 'φ', 'γ', 'χ', 'ω', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'ω', 'φ', 'γ', 'η', 'ψ', 'β', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'β', 'γ', 'η', 'ξ', 'ω', 'ν', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'ν', 'η', 'ξ', 'κ', 'β', 'μ', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + 'μ', 'ξ', 'κ', 'λ', 'ν', NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + + Using the following characters: + */ + private static final int GR_FINAL_SIGMA = '\u03C2'; // ς + private static final int GR_EPSILON = '\u03B5'; // ε + private static final int GR_RHO = '\u03C1'; // ρ + private static final int GR_TAU = '\u03C4'; // τ + private static final int GR_UPSILON = '\u03C5'; // υ + private static final int GR_THETA = '\u03B8'; // θ + private static final int GR_IOTA = '\u03B9'; // ι + private static final int GR_OMICRON = '\u03BF'; // ο + private static final int GR_PI = '\u03C0'; // π + private static final int GR_ALPHA = '\u03B1'; // α + private static final int GR_SIGMA = '\u03C3'; // σ + private static final int GR_DELTA = '\u03B4'; // δ + private static final int GR_PHI = '\u03C6'; // φ + private static final int GR_GAMMA = '\u03B3'; // γ + private static final int GR_ETA = '\u03B7'; // η + private static final int GR_XI = '\u03BE'; // ξ + private static final int GR_KAPPA = '\u03BA'; // κ + private static final int GR_LAMDA = '\u03BB'; // λ + private static final int GR_ZETA = '\u03B6'; // ζ + private static final int GR_CHI = '\u03C7'; // χ + private static final int GR_PSI = '\u03C8'; // ψ + private static final int GR_OMEGA = '\u03C9'; // ω + private static final int GR_BETA = '\u03B2'; // β + private static final int GR_NU = '\u03BD'; // ν + private static final int GR_MU = '\u03BC'; // μ + final static int[] PROXIMITY = { + // Proximity for row 1. This must have exactly ROW_SIZE entries for each letter, + // and exactly PROXIMITY_GRID_WIDTH letters for a row. Pad with NUL's. + // The number of rows must be exactly PROXIMITY_GRID_HEIGHT. + GR_FINAL_SIGMA, GR_EPSILON, GR_ALPHA, GR_SIGMA, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_EPSILON, GR_FINAL_SIGMA, GR_RHO, GR_SIGMA, GR_DELTA, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_RHO, GR_EPSILON, GR_TAU, GR_DELTA, GR_PHI, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_TAU, GR_RHO, GR_UPSILON, GR_PHI, GR_GAMMA, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_UPSILON, GR_TAU, GR_THETA, GR_GAMMA, GR_ETA, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_THETA, GR_UPSILON, GR_IOTA, GR_ETA, GR_XI, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_IOTA, GR_THETA, GR_OMICRON, GR_XI, GR_KAPPA, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_OMICRON, GR_IOTA, GR_PI, GR_KAPPA, GR_LAMDA, NUL, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_PI, GR_OMICRON, GR_LAMDA, NUL, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + + GR_ALPHA, GR_FINAL_SIGMA, GR_SIGMA, GR_ZETA, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_SIGMA, GR_FINAL_SIGMA, GR_EPSILON, GR_ALPHA, GR_DELTA, GR_ZETA, GR_CHI, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_DELTA, GR_EPSILON, GR_RHO, GR_SIGMA, GR_PHI, GR_ZETA, GR_CHI, GR_PSI, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_PHI, GR_RHO, GR_TAU, GR_DELTA, GR_GAMMA, GR_CHI, GR_PSI, GR_OMEGA, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_GAMMA, GR_TAU, GR_UPSILON, GR_PHI, GR_ETA, GR_PSI, GR_OMEGA, GR_BETA, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_ETA, GR_UPSILON, GR_THETA, GR_GAMMA, GR_XI, GR_OMEGA, GR_BETA, GR_NU, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_XI, GR_THETA, GR_IOTA, GR_ETA, GR_KAPPA, GR_BETA, GR_NU, GR_MU, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_KAPPA, GR_IOTA, GR_OMICRON, GR_XI, GR_LAMDA, GR_NU, GR_MU, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_LAMDA, GR_OMICRON, GR_PI, GR_KAPPA, GR_MU, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + + GR_ZETA, GR_ALPHA, GR_SIGMA, GR_DELTA, GR_CHI, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_CHI, GR_SIGMA, GR_DELTA, GR_PHI, GR_ZETA, GR_PSI, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_PSI, GR_DELTA, GR_PHI, GR_GAMMA, GR_CHI, GR_OMEGA, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_OMEGA, GR_PHI, GR_GAMMA, GR_ETA, GR_PSI, GR_BETA, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_BETA, GR_GAMMA, GR_ETA, GR_XI, GR_OMEGA, GR_NU, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_NU, GR_ETA, GR_XI, GR_KAPPA, GR_BETA, GR_MU, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + GR_MU, GR_XI, GR_KAPPA, GR_LAMDA, GR_NU, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, + NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, NUL, }; @@ -183,6 +433,8 @@ public final class SpellCheckerProximityInfo { return Latin.PROXIMITY; case AndroidSpellCheckerService.SCRIPT_CYRILLIC: return Cyrillic.PROXIMITY; + case AndroidSpellCheckerService.SCRIPT_GREEK: + return Greek.PROXIMITY; default: throw new RuntimeException("Wrong script supplied: " + script); } @@ -194,6 +446,8 @@ public final class SpellCheckerProximityInfo { return Latin.getIndexOf(codePoint); case AndroidSpellCheckerService.SCRIPT_CYRILLIC: return Cyrillic.getIndexOf(codePoint); + case AndroidSpellCheckerService.SCRIPT_GREEK: + return Greek.getIndexOf(codePoint); default: throw new RuntimeException("Wrong script supplied: " + script); } |