BinaryDictionary.java revision da50e1e98dadc3733c615dfb8d87fe8b4688c782
1/* 2 * Copyright (C) 2008 The Android Open Source Project 3 * 4 * Licensed under the Apache License, Version 2.0 (the "License"); you may not 5 * use this file except in compliance with the License. You may obtain a copy of 6 * 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, WITHOUT 12 * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the 13 * License for the specific language governing permissions and limitations under 14 * the License. 15 */ 16 17package com.android.inputmethod.latin; 18 19import android.content.Context; 20import android.content.res.AssetFileDescriptor; 21import android.util.Log; 22 23import java.util.Arrays; 24 25/** 26 * Implements a static, compacted, binary dictionary of standard words. 27 */ 28public class BinaryDictionary extends Dictionary { 29 30 /** 31 * There is difference between what java and native code can handle. 32 * This value should only be used in BinaryDictionary.java 33 * It is necessary to keep it at this value because some languages e.g. German have 34 * really long words. 35 */ 36 protected static final int MAX_WORD_LENGTH = 48; 37 38 private static final String TAG = "BinaryDictionary"; 39 private static final int MAX_ALTERNATIVES = 16; 40 private static final int MAX_WORDS = 18; 41 private static final int MAX_BIGRAMS = 60; 42 43 private static final int TYPED_LETTER_MULTIPLIER = 2; 44 45 private static final BinaryDictionary sInstance = new BinaryDictionary(); 46 private int mDicTypeId; 47 private int mNativeDict; 48 private long mDictLength; 49 private final int[] mInputCodes = new int[MAX_WORD_LENGTH * MAX_ALTERNATIVES]; 50 private final char[] mOutputChars = new char[MAX_WORD_LENGTH * MAX_WORDS]; 51 private final char[] mOutputChars_bigrams = new char[MAX_WORD_LENGTH * MAX_BIGRAMS]; 52 private final int[] mFrequencies = new int[MAX_WORDS]; 53 private final int[] mFrequencies_bigrams = new int[MAX_BIGRAMS]; 54 55 static { 56 try { 57 System.loadLibrary("jni_latinime"); 58 } catch (UnsatisfiedLinkError ule) { 59 Log.e(TAG, "Could not load native library jni_latinime"); 60 } 61 } 62 63 private BinaryDictionary() { 64 } 65 66 /** 67 * Initialize a dictionary from a raw resource file 68 * @param context application context for reading resources 69 * @param resId the resource containing the raw binary dictionary 70 * @return initialized instance of BinaryDictionary 71 */ 72 public static BinaryDictionary initDictionary(Context context, int resId, int dicTypeId) { 73 synchronized (sInstance) { 74 sInstance.closeInternal(); 75 if (resId != 0) { 76 sInstance.loadDictionary(context, resId); 77 sInstance.mDicTypeId = dicTypeId; 78 } 79 } 80 return sInstance; 81 } 82 83 private native int openNative(String sourceDir, long dictOffset, long dictSize, 84 int typedLetterMultiplier, int fullWordMultiplier, int maxWordLength, 85 int maxWords, int maxAlternatives); 86 private native void closeNative(int dict); 87 private native boolean isValidWordNative(int nativeData, char[] word, int wordLength); 88 private native int getSuggestionsNative(int dict, int[] inputCodes, int codesSize, 89 char[] outputChars, int[] frequencies, 90 int[] nextLettersFrequencies, int nextLettersSize); 91 private native int getBigramsNative(int dict, char[] prevWord, int prevWordLength, 92 int[] inputCodes, int inputCodesLength, char[] outputChars, int[] frequencies, 93 int maxWordLength, int maxBigrams, int maxAlternatives); 94 95 private final void loadDictionary(Context context, int resId) { 96 try { 97 final AssetFileDescriptor afd = context.getResources().openRawResourceFd(resId); 98 if (afd == null) { 99 Log.e(TAG, "Found the resource but it is compressed. resId=" + resId); 100 return; 101 } 102 mNativeDict = openNative(context.getApplicationInfo().sourceDir, 103 afd.getStartOffset(), afd.getLength(), 104 TYPED_LETTER_MULTIPLIER, FULL_WORD_FREQ_MULTIPLIER, 105 MAX_WORD_LENGTH, MAX_WORDS, MAX_ALTERNATIVES); 106 mDictLength = afd.getLength(); 107 } catch (android.content.res.Resources.NotFoundException e) { 108 Log.e(TAG, "Could not find the resource. resId=" + resId); 109 return; 110 } 111 } 112 113 @Override 114 public void getBigrams(final WordComposer codes, final CharSequence previousWord, 115 final WordCallback callback, int[] nextLettersFrequencies) { 116 if (mNativeDict == 0) return; 117 118 char[] chars = previousWord.toString().toCharArray(); 119 Arrays.fill(mOutputChars_bigrams, (char) 0); 120 Arrays.fill(mFrequencies_bigrams, 0); 121 122 int codesSize = codes.size(); 123 Arrays.fill(mInputCodes, -1); 124 int[] alternatives = codes.getCodesAt(0); 125 System.arraycopy(alternatives, 0, mInputCodes, 0, 126 Math.min(alternatives.length, MAX_ALTERNATIVES)); 127 128 int count = getBigramsNative(mNativeDict, chars, chars.length, mInputCodes, codesSize, 129 mOutputChars_bigrams, mFrequencies_bigrams, MAX_WORD_LENGTH, MAX_BIGRAMS, 130 MAX_ALTERNATIVES); 131 132 for (int j = 0; j < count; ++j) { 133 if (mFrequencies_bigrams[j] < 1) break; 134 final int start = j * MAX_WORD_LENGTH; 135 int len = 0; 136 while (len < MAX_WORD_LENGTH && mOutputChars_bigrams[start + len] != 0) { 137 ++len; 138 } 139 if (len > 0) { 140 callback.addWord(mOutputChars_bigrams, start, len, mFrequencies_bigrams[j], 141 mDicTypeId, DataType.BIGRAM); 142 } 143 } 144 } 145 146 @Override 147 public void getWords(final WordComposer codes, final WordCallback callback, 148 int[] nextLettersFrequencies) { 149 if (mNativeDict == 0) return; 150 151 final int codesSize = codes.size(); 152 // Won't deal with really long words. 153 if (codesSize > MAX_WORD_LENGTH - 1) return; 154 155 Arrays.fill(mInputCodes, -1); 156 for (int i = 0; i < codesSize; i++) { 157 int[] alternatives = codes.getCodesAt(i); 158 System.arraycopy(alternatives, 0, mInputCodes, i * MAX_ALTERNATIVES, 159 Math.min(alternatives.length, MAX_ALTERNATIVES)); 160 } 161 Arrays.fill(mOutputChars, (char) 0); 162 Arrays.fill(mFrequencies, 0); 163 164 int count = getSuggestionsNative(mNativeDict, mInputCodes, codesSize, mOutputChars, 165 mFrequencies, nextLettersFrequencies, 166 nextLettersFrequencies != null ? nextLettersFrequencies.length : 0); 167 168 for (int j = 0; j < count; ++j) { 169 if (mFrequencies[j] < 1) break; 170 final int start = j * MAX_WORD_LENGTH; 171 int len = 0; 172 while (len < MAX_WORD_LENGTH && mOutputChars[start + len] != 0) { 173 ++len; 174 } 175 if (len > 0) { 176 callback.addWord(mOutputChars, start, len, mFrequencies[j], mDicTypeId, 177 DataType.UNIGRAM); 178 } 179 } 180 } 181 182 @Override 183 public boolean isValidWord(CharSequence word) { 184 if (word == null) return false; 185 char[] chars = word.toString().toCharArray(); 186 return isValidWordNative(mNativeDict, chars, chars.length); 187 } 188 189 public long getSize() { 190 return mDictLength; // This value is initialized in loadDictionary() 191 } 192 193 @Override 194 public synchronized void close() { 195 closeInternal(); 196 } 197 198 private void closeInternal() { 199 if (mNativeDict != 0) { 200 closeNative(mNativeDict); 201 mNativeDict = 0; 202 mDictLength = 0; 203 } 204 } 205 206 @Override 207 protected void finalize() throws Throwable { 208 try { 209 closeInternal(); 210 } finally { 211 super.finalize(); 212 } 213 } 214} 215