1 /* 2 * Copyright (C) 2011 The Android Open Source Project 3 * 4 * Licensed under the Apache License, Version 2.0 (the "License"); 5 * you may not use this file except in compliance with the License. 6 * You may obtain a copy of the License at 7 * 8 * http://www.apache.org/licenses/LICENSE-2.0 9 * 10 * Unless required by applicable law or agreed to in writing, software 11 * distributed under the License is distributed on an "AS IS" BASIS, 12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 * See the License for the specific language governing permissions and 14 * limitations under the License. 15 */ 16 17 package com.android.inputmethod.latin; 18 19 import com.android.inputmethod.latin.SuggestedWords.SuggestedWordInfo; 20 21 import android.text.TextUtils; 22 import android.util.Log; 23 24 import java.util.ArrayList; 25 import java.util.concurrent.ConcurrentHashMap; 26 27 public class AutoCorrection { 28 private static final boolean DBG = LatinImeLogger.sDBG; 29 private static final String TAG = AutoCorrection.class.getSimpleName(); 30 31 private AutoCorrection() { 32 // Purely static class: can't instantiate. 33 } 34 35 public static CharSequence computeAutoCorrectionWord( 36 final ConcurrentHashMap<String, Dictionary> dictionaries, 37 final WordComposer wordComposer, final ArrayList<SuggestedWordInfo> suggestions, 38 final CharSequence consideredWord, final float autoCorrectionThreshold, 39 final CharSequence whitelistedWord) { 40 if (hasAutoCorrectionForWhitelistedWord(whitelistedWord)) { 41 return whitelistedWord; 42 } else if (hasAutoCorrectionForConsideredWord( 43 dictionaries, wordComposer, suggestions, consideredWord)) { 44 return consideredWord; 45 } else if (hasAutoCorrectionForBinaryDictionary(wordComposer, suggestions, 46 consideredWord, autoCorrectionThreshold)) { 47 return suggestions.get(0).mWord; 48 } 49 return null; 50 } 51 52 public static boolean isValidWord(final ConcurrentHashMap<String, Dictionary> dictionaries, 53 CharSequence word, boolean ignoreCase) { 54 if (TextUtils.isEmpty(word)) { 55 return false; 56 } 57 final CharSequence lowerCasedWord = word.toString().toLowerCase(); 58 for (final String key : dictionaries.keySet()) { 59 if (key.equals(Suggest.DICT_KEY_WHITELIST)) continue; 60 final Dictionary dictionary = dictionaries.get(key); 61 // It's unclear how realistically 'dictionary' can be null, but the monkey is somehow 62 // managing to get null in here. Presumably the language is changing to a language with 63 // no main dictionary and the monkey manages to type a whole word before the thread 64 // that reads the dictionary is started or something? 65 // Ideally the passed map would come out of a {@link java.util.concurrent.Future} and 66 // would be immutable once it's finished initializing, but concretely a null test is 67 // probably good enough for the time being. 68 if (null == dictionary) continue; 69 if (dictionary.isValidWord(word) 70 || (ignoreCase && dictionary.isValidWord(lowerCasedWord))) { 71 return true; 72 } 73 } 74 return false; 75 } 76 77 public static int getMaxFrequency(final ConcurrentHashMap<String, Dictionary> dictionaries, 78 CharSequence word) { 79 if (TextUtils.isEmpty(word)) { 80 return Dictionary.NOT_A_PROBABILITY; 81 } 82 int maxFreq = -1; 83 for (final String key : dictionaries.keySet()) { 84 if (key.equals(Suggest.DICT_KEY_WHITELIST)) continue; 85 final Dictionary dictionary = dictionaries.get(key); 86 if (null == dictionary) continue; 87 final int tempFreq = dictionary.getFrequency(word); 88 if (tempFreq >= maxFreq) { 89 maxFreq = tempFreq; 90 } 91 } 92 return maxFreq; 93 } 94 95 public static boolean allowsToBeAutoCorrected( 96 final ConcurrentHashMap<String, Dictionary> dictionaries, 97 final CharSequence word, final boolean ignoreCase) { 98 final WhitelistDictionary whitelistDictionary = 99 (WhitelistDictionary)dictionaries.get(Suggest.DICT_KEY_WHITELIST); 100 // If "word" is in the whitelist dictionary, it should not be auto corrected. 101 if (whitelistDictionary != null 102 && whitelistDictionary.shouldForciblyAutoCorrectFrom(word)) { 103 return true; 104 } 105 return !isValidWord(dictionaries, word, ignoreCase); 106 } 107 108 private static boolean hasAutoCorrectionForWhitelistedWord(CharSequence whiteListedWord) { 109 return whiteListedWord != null; 110 } 111 112 private static boolean hasAutoCorrectionForConsideredWord( 113 final ConcurrentHashMap<String, Dictionary> dictionaries, 114 final WordComposer wordComposer, final ArrayList<SuggestedWordInfo> suggestions, 115 final CharSequence consideredWord) { 116 if (TextUtils.isEmpty(consideredWord)) return false; 117 return wordComposer.size() > 1 && suggestions.size() > 0 118 && !allowsToBeAutoCorrected(dictionaries, consideredWord, false); 119 } 120 121 private static boolean hasAutoCorrectionForBinaryDictionary(WordComposer wordComposer, 122 ArrayList<SuggestedWordInfo> suggestions, 123 CharSequence consideredWord, float autoCorrectionThreshold) { 124 if (wordComposer.size() > 1 && suggestions.size() > 0) { 125 final SuggestedWordInfo autoCorrectionSuggestion = suggestions.get(0); 126 //final int autoCorrectionSuggestionScore = sortedScores[0]; 127 final int autoCorrectionSuggestionScore = autoCorrectionSuggestion.mScore; 128 // TODO: when the normalized score of the first suggestion is nearly equals to 129 // the normalized score of the second suggestion, behave less aggressive. 130 final float normalizedScore = BinaryDictionary.calcNormalizedScore( 131 consideredWord.toString(), autoCorrectionSuggestion.mWord.toString(), 132 autoCorrectionSuggestionScore); 133 if (DBG) { 134 Log.d(TAG, "Normalized " + consideredWord + "," + autoCorrectionSuggestion + "," 135 + autoCorrectionSuggestionScore + ", " + normalizedScore 136 + "(" + autoCorrectionThreshold + ")"); 137 } 138 if (normalizedScore >= autoCorrectionThreshold) { 139 if (DBG) { 140 Log.d(TAG, "Auto corrected by S-threshold."); 141 } 142 return true; 143 } 144 } 145 return false; 146 } 147 148 } 149