diff options
Diffstat (limited to 'java/src/com/android/inputmethod/latin/Suggest.java')
| -rw-r--r-- | java/src/com/android/inputmethod/latin/Suggest.java | 204 |
1 files changed, 71 insertions, 133 deletions
diff --git a/java/src/com/android/inputmethod/latin/Suggest.java b/java/src/com/android/inputmethod/latin/Suggest.java index 892245402..03ed9d98c 100644 --- a/java/src/com/android/inputmethod/latin/Suggest.java +++ b/java/src/com/android/inputmethod/latin/Suggest.java @@ -26,6 +26,7 @@ import com.android.inputmethod.latin.SuggestedWords.SuggestedWordInfo; import java.io.File; import java.util.ArrayList; +import java.util.HashMap; import java.util.HashSet; import java.util.Locale; import java.util.concurrent.ConcurrentHashMap; @@ -34,7 +35,7 @@ import java.util.concurrent.ConcurrentHashMap; * This class loads a dictionary and provides a list of suggestions for a given sequence of * characters. This includes corrections and completions. */ -public class Suggest implements Dictionary.WordCallback { +public class Suggest { public static final String TAG = Suggest.class.getSimpleName(); public static final int APPROX_MAX_WORD_LENGTH = 32; @@ -58,30 +59,34 @@ public class Suggest implements Dictionary.WordCallback { public static final String DICT_KEY_CONTACTS = "contacts"; // User dictionary, the system-managed one. public static final String DICT_KEY_USER = "user"; - // User history dictionary for the unigram map, internal to LatinIME - public static final String DICT_KEY_USER_HISTORY_UNIGRAM = "history_unigram"; - // User history dictionary for the bigram map, internal to LatinIME - public static final String DICT_KEY_USER_HISTORY_BIGRAM = "history_bigram"; + // User history dictionary internal to LatinIME + public static final String DICT_KEY_USER_HISTORY = "history"; public static final String DICT_KEY_WHITELIST ="whitelist"; + // TODO: remove this map. This only serves as backward compatibility with a feature + // that has never been used and has been broken for a while. + private static final HashMap<String, Integer> sDictKeyToDictIndex + = new HashMap<String, Integer>(); + static { + sDictKeyToDictIndex.put(DICT_KEY_MAIN, DIC_MAIN); + sDictKeyToDictIndex.put(DICT_KEY_USER, DIC_USER); + sDictKeyToDictIndex.put(DICT_KEY_USER_HISTORY, DIC_USER_HISTORY); + sDictKeyToDictIndex.put(DICT_KEY_CONTACTS, DIC_CONTACTS); + sDictKeyToDictIndex.put(DICT_KEY_WHITELIST, DIC_WHITELIST); + } private static final boolean DBG = LatinImeLogger.sDBG; private Dictionary mMainDictionary; private ContactsBinaryDictionary mContactsDict; private WhitelistDictionary mWhiteListDictionary; - private final ConcurrentHashMap<String, Dictionary> mUnigramDictionaries = - new ConcurrentHashMap<String, Dictionary>(); - private final ConcurrentHashMap<String, Dictionary> mBigramDictionaries = + private final ConcurrentHashMap<String, Dictionary> mDictionaries = new ConcurrentHashMap<String, Dictionary>(); public static final int MAX_SUGGESTIONS = 18; - private static final int PREF_MAX_BIGRAMS = 60; - private float mAutoCorrectionThreshold; private ArrayList<SuggestedWordInfo> mSuggestions = new ArrayList<SuggestedWordInfo>(); - private ArrayList<SuggestedWordInfo> mBigramSuggestions = new ArrayList<SuggestedWordInfo>(); private CharSequence mConsideredWord; // TODO: Remove these member variables by passing more context to addWord() callback method @@ -100,14 +105,13 @@ public class Suggest implements Dictionary.WordCallback { final Dictionary mainDict = DictionaryFactory.createDictionaryForTest(context, dictionary, startOffset, length /* useFullEditDistance */, false, locale); mMainDictionary = mainDict; - addOrReplaceDictionary(mUnigramDictionaries, DICT_KEY_MAIN, mainDict); - addOrReplaceDictionary(mBigramDictionaries, DICT_KEY_MAIN, mainDict); + addOrReplaceDictionary(mDictionaries, DICT_KEY_MAIN, mainDict); initWhitelistAndAutocorrectAndPool(context, locale); } private void initWhitelistAndAutocorrectAndPool(final Context context, final Locale locale) { mWhiteListDictionary = new WhitelistDictionary(context, locale); - addOrReplaceDictionary(mUnigramDictionaries, DICT_KEY_WHITELIST, mWhiteListDictionary); + addOrReplaceDictionary(mDictionaries, DICT_KEY_WHITELIST, mWhiteListDictionary); } private void initAsynchronously(final Context context, final Locale locale) { @@ -136,8 +140,7 @@ public class Suggest implements Dictionary.WordCallback { public void run() { final DictionaryCollection newMainDict = DictionaryFactory.createMainDictionaryFromManager(context, locale); - addOrReplaceDictionary(mUnigramDictionaries, DICT_KEY_MAIN, newMainDict); - addOrReplaceDictionary(mBigramDictionaries, DICT_KEY_MAIN, newMainDict); + addOrReplaceDictionary(mDictionaries, DICT_KEY_MAIN, newMainDict); mMainDictionary = newMainDict; } }.start(); @@ -158,7 +161,7 @@ public class Suggest implements Dictionary.WordCallback { } public ConcurrentHashMap<String, Dictionary> getUnigramDictionaries() { - return mUnigramDictionaries; + return mDictionaries; } public static int getApproxMaxWordLength() { @@ -170,7 +173,7 @@ public class Suggest implements Dictionary.WordCallback { * before the main dictionary, if set. This refers to the system-managed user dictionary. */ public void setUserDictionary(UserBinaryDictionary userDictionary) { - addOrReplaceDictionary(mUnigramDictionaries, DICT_KEY_USER, userDictionary); + addOrReplaceDictionary(mDictionaries, DICT_KEY_USER, userDictionary); } /** @@ -180,15 +183,11 @@ public class Suggest implements Dictionary.WordCallback { */ public void setContactsDictionary(ContactsBinaryDictionary contactsDictionary) { mContactsDict = contactsDictionary; - addOrReplaceDictionary(mUnigramDictionaries, DICT_KEY_CONTACTS, contactsDictionary); - addOrReplaceDictionary(mBigramDictionaries, DICT_KEY_CONTACTS, contactsDictionary); + addOrReplaceDictionary(mDictionaries, DICT_KEY_CONTACTS, contactsDictionary); } public void setUserHistoryDictionary(UserHistoryDictionary userHistoryDictionary) { - addOrReplaceDictionary(mUnigramDictionaries, DICT_KEY_USER_HISTORY_UNIGRAM, - userHistoryDictionary); - addOrReplaceDictionary(mBigramDictionaries, DICT_KEY_USER_HISTORY_BIGRAM, - userHistoryDictionary); + addOrReplaceDictionary(mDictionaries, DICT_KEY_USER_HISTORY, userHistoryDictionary); } public void setAutoCorrectionThreshold(float threshold) { @@ -212,50 +211,14 @@ public class Suggest implements Dictionary.WordCallback { return sb; } - protected void addBigramToSuggestions(SuggestedWordInfo bigram) { - mSuggestions.add(bigram); - } - - private static final WordComposer sEmptyWordComposer = new WordComposer(); - public SuggestedWords getBigramPredictions(CharSequence prevWordForBigram) { - LatinImeLogger.onStartSuggestion(prevWordForBigram); - mIsFirstCharCapitalized = false; - mIsAllUpperCase = false; - mTrailingSingleQuotesCount = 0; - mSuggestions = new ArrayList<SuggestedWordInfo>(MAX_SUGGESTIONS); - - // Treating USER_TYPED as UNIGRAM suggestion for logging now. - LatinImeLogger.onAddSuggestedWord("", Suggest.DIC_USER_TYPED, Dictionary.UNIGRAM); - mConsideredWord = ""; - - mBigramSuggestions = new ArrayList<SuggestedWordInfo>(PREF_MAX_BIGRAMS); - - getAllBigrams(prevWordForBigram, sEmptyWordComposer); - - // Nothing entered: return all bigrams for the previous word - int insertCount = Math.min(mBigramSuggestions.size(), MAX_SUGGESTIONS); - for (int i = 0; i < insertCount; ++i) { - addBigramToSuggestions(mBigramSuggestions.get(i)); - } - - SuggestedWordInfo.removeDups(mSuggestions); - - return new SuggestedWords(mSuggestions, - false /* typedWordValid */, - false /* hasAutoCorrectionCandidate */, - false /* allowsToBeAutoCorrected */, - false /* isPunctuationSuggestions */, - false /* isObsoleteSuggestions */, - true /* isPrediction */); - } - // TODO: cleanup dictionaries looking up and suggestions building with SuggestedWords.Builder public SuggestedWords getSuggestedWords( final WordComposer wordComposer, CharSequence prevWordForBigram, - final ProximityInfo proximityInfo, final boolean isCorrectionEnabled) { + final ProximityInfo proximityInfo, final boolean isCorrectionEnabled, + final boolean isPrediction) { LatinImeLogger.onStartSuggestion(prevWordForBigram); - mIsFirstCharCapitalized = wordComposer.isFirstCharCapitalized(); - mIsAllUpperCase = wordComposer.isAllUpperCase(); + mIsFirstCharCapitalized = !isPrediction && wordComposer.isFirstCharCapitalized(); + mIsAllUpperCase = !isPrediction && wordComposer.isAllUpperCase(); mTrailingSingleQuotesCount = wordComposer.trailingSingleQuotesCount(); mSuggestions = new ArrayList<SuggestedWordInfo>(MAX_SUGGESTIONS); @@ -269,37 +232,29 @@ public class Suggest implements Dictionary.WordCallback { if (wordComposer.size() <= 1 && isCorrectionEnabled) { // At first character typed, search only the bigrams - mBigramSuggestions = new ArrayList<SuggestedWordInfo>(PREF_MAX_BIGRAMS); - if (!TextUtils.isEmpty(prevWordForBigram)) { - getAllBigrams(prevWordForBigram, wordComposer); - if (TextUtils.isEmpty(consideredWord)) { - // Nothing entered: return all bigrams for the previous word - int insertCount = Math.min(mBigramSuggestions.size(), MAX_SUGGESTIONS); - for (int i = 0; i < insertCount; ++i) { - addBigramToSuggestions(mBigramSuggestions.get(i)); - } - } else { - // Word entered: return only bigrams that match the first char of the typed word - final char currentChar = consideredWord.charAt(0); + final CharSequence lowerPrevWord; + if (StringUtils.hasUpperCase(prevWordForBigram)) { // TODO: Must pay attention to locale when changing case. - // TODO: Use codepoint instead of char - final char currentCharUpper = Character.toUpperCase(currentChar); - int count = 0; - final int bigramSuggestionSize = mBigramSuggestions.size(); - for (int i = 0; i < bigramSuggestionSize; i++) { - final SuggestedWordInfo bigramSuggestion = mBigramSuggestions.get(i); - final char bigramSuggestionFirstChar = - (char)bigramSuggestion.codePointAt(0); - if (bigramSuggestionFirstChar == currentChar - || bigramSuggestionFirstChar == currentCharUpper) { - addBigramToSuggestions(bigramSuggestion); - if (++count > MAX_SUGGESTIONS) break; - } + lowerPrevWord = prevWordForBigram.toString().toLowerCase(); + } else { + lowerPrevWord = null; + } + for (final String key : mDictionaries.keySet()) { + final int dicTypeId = sDictKeyToDictIndex.get(key); + final Dictionary dictionary = mDictionaries.get(key); + final ArrayList<SuggestedWordInfo> suggestions = + dictionary.getBigrams(wordComposer, prevWordForBigram); + if (null != lowerPrevWord) { + suggestions.addAll(dictionary.getBigrams(wordComposer, lowerPrevWord)); + } + for (final SuggestedWordInfo suggestion : suggestions) { + final String suggestionStr = suggestion.mWord.toString(); + addWord(suggestionStr.toCharArray(), null, 0, suggestionStr.length(), + suggestion.mScore, dicTypeId, Dictionary.BIGRAM, mSuggestions); } } } - } else if (wordComposer.size() > 1) { final WordComposer wordComposerForLookup; if (mTrailingSingleQuotesCount > 0) { @@ -311,12 +266,19 @@ public class Suggest implements Dictionary.WordCallback { wordComposerForLookup = wordComposer; } // At second character typed, search the unigrams (scores being affected by bigrams) - for (final String key : mUnigramDictionaries.keySet()) { + for (final String key : mDictionaries.keySet()) { // Skip UserUnigramDictionary and WhitelistDictionary to lookup - if (key.equals(DICT_KEY_USER_HISTORY_UNIGRAM) || key.equals(DICT_KEY_WHITELIST)) + if (key.equals(DICT_KEY_USER_HISTORY) || key.equals(DICT_KEY_WHITELIST)) continue; - final Dictionary dictionary = mUnigramDictionaries.get(key); - dictionary.getWords(wordComposerForLookup, prevWordForBigram, this, proximityInfo); + final int dicTypeId = sDictKeyToDictIndex.get(key); + final Dictionary dictionary = mDictionaries.get(key); + final ArrayList<SuggestedWordInfo> suggestions = dictionary.getWords( + wordComposerForLookup, prevWordForBigram, proximityInfo); + for (final SuggestedWordInfo suggestion : suggestions) { + final String suggestionStr = suggestion.mWord.toString(); + addWord(suggestionStr.toCharArray(), null, 0, suggestionStr.length(), + suggestion.mScore, dicTypeId, Dictionary.UNIGRAM, mSuggestions); + } } } @@ -326,7 +288,7 @@ public class Suggest implements Dictionary.WordCallback { final boolean hasAutoCorrection; if (isCorrectionEnabled) { final CharSequence autoCorrection = - AutoCorrection.computeAutoCorrectionWord(mUnigramDictionaries, wordComposer, + AutoCorrection.computeAutoCorrectionWord(mDictionaries, wordComposer, mSuggestions, consideredWord, mAutoCorrectionThreshold, whitelistedWord); hasAutoCorrection = (null != autoCorrection); @@ -348,12 +310,14 @@ public class Suggest implements Dictionary.WordCallback { } } - mSuggestions.add(0, new SuggestedWordInfo(typedWord, SuggestedWordInfo.MAX_SCORE, - SuggestedWordInfo.KIND_TYPED)); + if (!isPrediction) { + mSuggestions.add(0, new SuggestedWordInfo(typedWord, SuggestedWordInfo.MAX_SCORE, + SuggestedWordInfo.KIND_TYPED)); + } SuggestedWordInfo.removeDups(mSuggestions); final ArrayList<SuggestedWordInfo> suggestionsList; - if (DBG) { + if (DBG && !mSuggestions.isEmpty()) { suggestionsList = getSuggestionsInfoListWithDebugInfo(typedWord, mSuggestions); } else { suggestionsList = mSuggestions; @@ -386,29 +350,12 @@ public class Suggest implements Dictionary.WordCallback { autoCorrectionAvailable = false; } return new SuggestedWords(suggestionsList, - !allowsToBeAutoCorrected /* typedWordValid */, - autoCorrectionAvailable /* hasAutoCorrectionCandidate */, - allowsToBeAutoCorrected /* allowsToBeAutoCorrected */, + !isPrediction && !allowsToBeAutoCorrected /* typedWordValid */, + !isPrediction && autoCorrectionAvailable /* hasAutoCorrectionCandidate */, + !isPrediction && allowsToBeAutoCorrected /* allowsToBeAutoCorrected */, false /* isPunctuationSuggestions */, false /* isObsoleteSuggestions */, - false /* isPrediction */); - } - - /** - * Adds all bigram predictions for prevWord. Also checks the lower case version of prevWord if - * it contains any upper case characters. - */ - private void getAllBigrams(final CharSequence prevWord, final WordComposer wordComposer) { - if (StringUtils.hasUpperCase(prevWord)) { - // TODO: Must pay attention to locale when changing case. - final CharSequence lowerPrevWord = prevWord.toString().toLowerCase(); - for (final Dictionary dictionary : mBigramDictionaries.values()) { - dictionary.getBigrams(wordComposer, lowerPrevWord, this); - } - } - for (final Dictionary dictionary : mBigramDictionaries.values()) { - dictionary.getBigrams(wordComposer, prevWord, this); - } + isPrediction); } private static ArrayList<SuggestedWordInfo> getSuggestionsInfoListWithDebugInfo( @@ -438,19 +385,11 @@ public class Suggest implements Dictionary.WordCallback { } // TODO: Use codepoint instead of char - @Override - public boolean addWord(final char[] word, final int offset, final int length, int score, - final int dicTypeId, final int dataType) { + public boolean addWord(final char[] word, int[] indices, final int offset, final int length, + int score, final int dicTypeId, final int dataType, + final ArrayList<SuggestedWordInfo> suggestions) { int dataTypeForLog = dataType; - final ArrayList<SuggestedWordInfo> suggestions; - final int prefMaxSuggestions; - if (dataType == Dictionary.BIGRAM) { - suggestions = mBigramSuggestions; - prefMaxSuggestions = PREF_MAX_BIGRAMS; - } else { - suggestions = mSuggestions; - prefMaxSuggestions = MAX_SUGGESTIONS; - } + final int prefMaxSuggestions = MAX_SUGGESTIONS; int pos = 0; @@ -513,8 +452,7 @@ public class Suggest implements Dictionary.WordCallback { public void close() { final HashSet<Dictionary> dictionaries = new HashSet<Dictionary>(); - dictionaries.addAll(mUnigramDictionaries.values()); - dictionaries.addAll(mBigramDictionaries.values()); + dictionaries.addAll(mDictionaries.values()); for (final Dictionary dictionary : dictionaries) { dictionary.close(); } |