diff options
Diffstat (limited to 'java/src/com/android/inputmethod/latin/spellcheck')
3 files changed, 63 insertions, 107 deletions
diff --git a/java/src/com/android/inputmethod/latin/spellcheck/AndroidSpellCheckerService.java b/java/src/com/android/inputmethod/latin/spellcheck/AndroidSpellCheckerService.java index 38a26486d..2d0a89bb3 100644 --- a/java/src/com/android/inputmethod/latin/spellcheck/AndroidSpellCheckerService.java +++ b/java/src/com/android/inputmethod/latin/spellcheck/AndroidSpellCheckerService.java @@ -58,10 +58,6 @@ public final class AndroidSpellCheckerService extends SpellCheckerService public static final String PREF_USE_CONTACTS_KEY = "pref_spellcheck_use_contacts"; - public static final int CAPITALIZE_NONE = 0; // No caps, or mixed case - public static final int CAPITALIZE_FIRST = 1; // First only - public static final int CAPITALIZE_ALL = 2; // All caps - private final static String[] EMPTY_STRING_ARRAY = new String[0]; private Map<String, DictionaryPool> mDictionaryPools = CollectionUtils.newSynchronizedTreeMap(); private Map<String, UserBinaryDictionary> mUserDictionaries = @@ -325,16 +321,16 @@ public final class AndroidSpellCheckerService extends SpellCheckerService } Collections.reverse(mSuggestions); StringUtils.removeDupes(mSuggestions); - if (CAPITALIZE_ALL == capitalizeType) { + if (StringUtils.CAPITALIZE_ALL == capitalizeType) { for (int i = 0; i < mSuggestions.size(); ++i) { // get(i) returns a CharSequence which is actually a String so .toString() // should return the same object. mSuggestions.set(i, mSuggestions.get(i).toString().toUpperCase(locale)); } - } else if (CAPITALIZE_FIRST == capitalizeType) { + } else if (StringUtils.CAPITALIZE_FIRST == capitalizeType) { for (int i = 0; i < mSuggestions.size(); ++i) { // Likewise - mSuggestions.set(i, StringUtils.toTitleCase( + mSuggestions.set(i, StringUtils.capitalizeFirstCodePoint( mSuggestions.get(i).toString(), locale)); } } @@ -407,11 +403,7 @@ public final class AndroidSpellCheckerService extends SpellCheckerService public DictAndProximity createDictAndProximity(final Locale locale) { final int script = getScriptFromLocale(locale); - final ProximityInfo proximityInfo = ProximityInfo.createSpellCheckerProximityInfo( - SpellCheckerProximityInfo.getProximityForScript(script), - SpellCheckerProximityInfo.ROW_SIZE, - SpellCheckerProximityInfo.PROXIMITY_GRID_WIDTH, - SpellCheckerProximityInfo.PROXIMITY_GRID_HEIGHT); + final ProximityInfo proximityInfo = new SpellCheckerProximityInfo(script); final DictionaryCollection dictionaryCollection = DictionaryFactory.createMainDictionaryFromManager(this, locale, true /* useFullEditDistance */); @@ -438,31 +430,4 @@ public final class AndroidSpellCheckerService extends SpellCheckerService } return new DictAndProximity(dictionaryCollection, proximityInfo); } - - // This method assumes the text is not empty or null. - public static int getCapitalizationType(String text) { - // If the first char is not uppercase, then the word is either all lower case, - // and in either case we return CAPITALIZE_NONE. - if (!Character.isUpperCase(text.codePointAt(0))) return CAPITALIZE_NONE; - final int len = text.length(); - int capsCount = 1; - int letterCount = 1; - for (int i = 1; i < len; i = text.offsetByCodePoints(i, 1)) { - if (1 != capsCount && letterCount != capsCount) break; - final int codePoint = text.codePointAt(i); - if (Character.isUpperCase(codePoint)) { - ++capsCount; - ++letterCount; - } else if (Character.isLetter(codePoint)) { - // We need to discount non-letters since they may not be upper-case, but may - // still be part of a word (e.g. single quote or dash, as in "IT'S" or "FULL-TIME") - ++letterCount; - } - } - // We know the first char is upper case. So we want to test if either every letter other - // than the first is lower case, or if they are all upper case. If the string is exactly - // one char long, then we will arrive here with letterCount 1, and this is correct, too. - if (1 == capsCount) return CAPITALIZE_FIRST; - return (letterCount == capsCount ? CAPITALIZE_ALL : CAPITALIZE_NONE); - } } diff --git a/java/src/com/android/inputmethod/latin/spellcheck/AndroidWordLevelSpellCheckerSession.java b/java/src/com/android/inputmethod/latin/spellcheck/AndroidWordLevelSpellCheckerSession.java index 4f86a3175..96b2c818d 100644 --- a/java/src/com/android/inputmethod/latin/spellcheck/AndroidWordLevelSpellCheckerSession.java +++ b/java/src/com/android/inputmethod/latin/spellcheck/AndroidWordLevelSpellCheckerSession.java @@ -150,7 +150,7 @@ public abstract class AndroidWordLevelSpellCheckerSession extends Session { // 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 + // if that's correct, but a web search does return results for these words so // they are probably okay. return (codePoint >= 0x370 && codePoint <= 0x3FF) || (codePoint >= 0x1F00 && codePoint <= 0x1FFF) @@ -214,19 +214,19 @@ public abstract class AndroidWordLevelSpellCheckerSession extends Session { // If the word is in there as is, then it's in the dictionary. If not, we'll test lower // case versions, but only if the word is not already all-lower case or mixed case. if (dict.isValidWord(text)) return true; - if (AndroidSpellCheckerService.CAPITALIZE_NONE == capitalizeType) return false; + if (StringUtils.CAPITALIZE_NONE == capitalizeType) return false; // If we come here, we have a capitalized word (either First- or All-). // Downcase the word and look it up again. If the word is only capitalized, we // tested all possibilities, so if it's still negative we can return false. final String lowerCaseText = text.toLowerCase(mLocale); if (dict.isValidWord(lowerCaseText)) return true; - if (AndroidSpellCheckerService.CAPITALIZE_FIRST == capitalizeType) return false; + if (StringUtils.CAPITALIZE_FIRST == capitalizeType) return false; // If the lower case version is not in the dictionary, it's still possible // that we have an all-caps version of a word that needs to be capitalized // according to the dictionary. E.g. "GERMANS" only exists in the dictionary as "Germans". - return dict.isValidWord(StringUtils.toTitleCase(lowerCaseText, mLocale)); + return dict.isValidWord(StringUtils.capitalizeFirstAndDowncaseRest(lowerCaseText, mLocale)); } // Note : this must be reentrant @@ -296,7 +296,7 @@ public abstract class AndroidWordLevelSpellCheckerSession extends Session { } } - final int capitalizeType = AndroidSpellCheckerService.getCapitalizationType(text); + final int capitalizeType = StringUtils.getCapitalizationType(text); boolean isInDict = true; DictAndProximity dictInfo = null; try { diff --git a/java/src/com/android/inputmethod/latin/spellcheck/SpellCheckerProximityInfo.java b/java/src/com/android/inputmethod/latin/spellcheck/SpellCheckerProximityInfo.java index 49dca21e6..0c480eaba 100644 --- a/java/src/com/android/inputmethod/latin/spellcheck/SpellCheckerProximityInfo.java +++ b/java/src/com/android/inputmethod/latin/spellcheck/SpellCheckerProximityInfo.java @@ -16,53 +16,40 @@ package com.android.inputmethod.latin.spellcheck; -import com.android.inputmethod.annotations.UsedForTesting; +import android.util.SparseIntArray; + import com.android.inputmethod.keyboard.ProximityInfo; -import com.android.inputmethod.latin.CollectionUtils; import com.android.inputmethod.latin.Constants; -import java.util.TreeMap; +public final class SpellCheckerProximityInfo extends ProximityInfo { + public SpellCheckerProximityInfo(final int script) { + super(getProximityForScript(script), PROXIMITY_GRID_WIDTH, PROXIMITY_GRID_HEIGHT); + } -public final class SpellCheckerProximityInfo { - @UsedForTesting - final public static int NUL = Constants.NOT_A_CODE; + private static final int NUL = Constants.NOT_A_CODE; // This must be the same as MAX_PROXIMITY_CHARS_SIZE else it will not work inside // native code - this value is passed at creation of the binary object and reused // as the size of the passed array afterwards so they can't be different. - final public static int ROW_SIZE = ProximityInfo.MAX_PROXIMITY_CHARS_SIZE; + private static final int ROW_SIZE = ProximityInfo.MAX_PROXIMITY_CHARS_SIZE; // The number of keys in a row of the grid used by the spell checker. - final public static int PROXIMITY_GRID_WIDTH = 11; + private static final int PROXIMITY_GRID_WIDTH = 11; // The number of rows in the grid used by the spell checker. - final public static int PROXIMITY_GRID_HEIGHT = 3; + private static final int PROXIMITY_GRID_HEIGHT = 3; - final private static int NOT_AN_INDEX = -1; - final public static int NOT_A_COORDINATE_PAIR = -1; + private static final int NOT_AN_INDEX = -1; + public static final int NOT_A_COORDINATE_PAIR = -1; // Helper methods - final protected static void buildProximityIndices(final int[] proximity, - final TreeMap<Integer, Integer> indices) { - for (int i = 0; i < proximity.length; i += ROW_SIZE) { - if (NUL != proximity[i]) indices.put(proximity[i], i / ROW_SIZE); + static void buildProximityIndices(final int[] proximity, final int rowSize, + final SparseIntArray indices) { + for (int i = 0; i < proximity.length; i += rowSize) { + if (NUL != proximity[i]) indices.put(proximity[i], i / rowSize); } } - final protected static int computeIndex(final int characterCode, - final TreeMap<Integer, Integer> indices) { - final Integer result = indices.get(characterCode); - if (null == result) return NOT_AN_INDEX; - return result; - } private static final class Latin { - // This is a map from the code point to the index in the PROXIMITY array. - // At the time the native code to read the binary dictionary needs the proximity info be - // passed as a flat array spaced by MAX_PROXIMITY_CHARS_SIZE columns, one for each input - // character. - // Since we need to build such an array, we want to be able to search in our big proximity - // data quickly by character, and a map is probably the best way to do this. - final private static TreeMap<Integer, Integer> INDICES = CollectionUtils.newTreeMap(); - // The proximity here is the union of // - the proximity for a QWERTY keyboard. // - the proximity for an AZERTY keyboard. @@ -79,7 +66,7 @@ public final class SpellCheckerProximityInfo { a s d f g h j k l z x c v b n m */ - final static int[] PROXIMITY = { + static final 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. @@ -121,16 +108,21 @@ public final class SpellCheckerProximityInfo { 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, }; + + // This is a mapping array from the code point to the index in the PROXIMITY array. + // When we check the spelling of a word, we need to pass (x,y) coordinates to the native + // code for each letter of the word. These are most easily computed from the index in the + // PROXIMITY array. Since we'll need to do that very often, the index lookup from the code + // point needs to be as fast as possible, and a map is probably the best way to do this. + // To avoid unnecessary boxing conversion to Integer, here we use SparseIntArray. + static final SparseIntArray INDICES = new SparseIntArray(PROXIMITY.length / ROW_SIZE); + static { - buildProximityIndices(PROXIMITY, INDICES); - } - static int getIndexOf(int characterCode) { - return computeIndex(characterCode, INDICES); + buildProximityIndices(PROXIMITY, ROW_SIZE, INDICES); } } private static final class Cyrillic { - 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. /* @@ -207,7 +199,7 @@ public final class SpellCheckerProximityInfo { 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 = { + static final 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. @@ -280,16 +272,15 @@ public final class SpellCheckerProximityInfo { 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 final SparseIntArray INDICES = new SparseIntArray(PROXIMITY.length / ROW_SIZE); + static { - buildProximityIndices(PROXIMITY, INDICES); - } - static int getIndexOf(int characterCode) { - return computeIndex(characterCode, INDICES); + buildProximityIndices(PROXIMITY, ROW_SIZE, 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. /* @@ -354,7 +345,7 @@ public final class SpellCheckerProximityInfo { 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 = { + static final 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. @@ -419,37 +410,37 @@ public final class SpellCheckerProximityInfo { 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 final SparseIntArray INDICES = new SparseIntArray(PROXIMITY.length / ROW_SIZE); + static { - buildProximityIndices(PROXIMITY, INDICES); - } - static int getIndexOf(int characterCode) { - return computeIndex(characterCode, INDICES); + buildProximityIndices(PROXIMITY, ROW_SIZE, INDICES); } } - public static int[] getProximityForScript(final int script) { + private static int[] getProximityForScript(final int script) { switch (script) { - case AndroidSpellCheckerService.SCRIPT_LATIN: - 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); + case AndroidSpellCheckerService.SCRIPT_LATIN: + 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); } } private static int getIndexOfCodeForScript(final int codePoint, final int script) { switch (script) { - case AndroidSpellCheckerService.SCRIPT_LATIN: - 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); + case AndroidSpellCheckerService.SCRIPT_LATIN: + return Latin.INDICES.get(codePoint, NOT_AN_INDEX); + case AndroidSpellCheckerService.SCRIPT_CYRILLIC: + return Cyrillic.INDICES.get(codePoint, NOT_AN_INDEX); + case AndroidSpellCheckerService.SCRIPT_GREEK: + return Greek.INDICES.get(codePoint, NOT_AN_INDEX); + default: + throw new RuntimeException("Wrong script supplied: " + script); } } |