BinaryDictionary.java revision 673cebf9e97289b3b0cd343ff7193dff69684a48 
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 com.android.inputmethod.keyboard.Keyboard;
20import com.android.inputmethod.keyboard.KeyboardSwitcher;
21import com.android.inputmethod.keyboard.ProximityInfo;
22
23import android.content.Context;
24
25import java.util.Arrays;
26
27/**
28 * Implements a static, compacted, binary dictionary of standard words.
29 */
30public class BinaryDictionary extends Dictionary {
31
32    public static final String DICTIONARY_PACK_AUTHORITY =
33            "com.android.inputmethod.latin.dictionarypack";
34
35    /**
36     * There is a difference between what java and native code can handle.
37     * This value should only be used in BinaryDictionary.java
38     * It is necessary to keep it at this value because some languages e.g. German have
39     * really long words.
40     */
41    public static final int MAX_WORD_LENGTH = 48;
42    public static final int MAX_WORDS = 18;
43
44    @SuppressWarnings("unused")
45    private static final String TAG = "BinaryDictionary";
46    private static final int MAX_PROXIMITY_CHARS_SIZE = ProximityInfo.MAX_PROXIMITY_CHARS_SIZE;
47    private static final int MAX_BIGRAMS = 60;
48
49    private static final int TYPED_LETTER_MULTIPLIER = 2;
50
51    private int mDicTypeId;
52    private int mNativeDict;
53    private final int[] mInputCodes = new int[MAX_WORD_LENGTH * MAX_PROXIMITY_CHARS_SIZE];
54    private final char[] mOutputChars = new char[MAX_WORD_LENGTH * MAX_WORDS];
55    private final char[] mOutputChars_bigrams = new char[MAX_WORD_LENGTH * MAX_BIGRAMS];
56    private final int[] mScores = new int[MAX_WORDS];
57    private final int[] mBigramScores = new int[MAX_BIGRAMS];
58
59    private final KeyboardSwitcher mKeyboardSwitcher = KeyboardSwitcher.getInstance();
60
61    public static final Flag FLAG_REQUIRES_GERMAN_UMLAUT_PROCESSING =
62            new Flag(R.bool.config_require_umlaut_processing, 0x1);
63
64    // FULL_EDIT_DISTANCE is a flag that forces the dictionary to use full words
65    // when computing edit distance, instead of the default behavior of stopping
66    // the evaluation at the size the user typed.
67    public static final Flag FLAG_USE_FULL_EDIT_DISTANCE = new Flag(0x2);
68
69    // Can create a new flag from extravalue :
70    // public static final Flag FLAG_MYFLAG =
71    //         new Flag("my_flag", 0x02);
72
73    // ALL_CONFIG_FLAGS is a collection of flags that enable reading all flags from configuration.
74    // This is but a mask - it does not mean the flags will be on, only that the configuration
75    // will be read for this particular flag.
76    public static final Flag[] ALL_CONFIG_FLAGS = {
77        // Here should reside all flags that trigger some special processing
78        // These *must* match the definition in UnigramDictionary enum in
79        // unigram_dictionary.h so please update both at the same time.
80        // Please note that flags created with a resource are of type CONFIG while flags
81        // created with a string are of type EXTRAVALUE. These behave like masks, and the
82        // actual value will be read from the configuration/extra value at run time for
83        // the configuration at dictionary creation time.
84        FLAG_REQUIRES_GERMAN_UMLAUT_PROCESSING,
85    };
86
87    private int mFlags = 0;
88
89    /**
90     * Constructor for the binary dictionary. This is supposed to be called from the
91     * dictionary factory.
92     * All implementations should pass null into flagArray, except for testing purposes.
93     * @param context the context to access the environment from.
94     * @param filename the name of the file to read through native code.
95     * @param offset the offset of the dictionary data within the file.
96     * @param length the length of the binary data.
97     * @param flagArray the flags to limit the dictionary to, or null for default.
98     */
99    public BinaryDictionary(final Context context,
100            final String filename, final long offset, final long length, Flag[] flagArray) {
101        // Note: at the moment a binary dictionary is always of the "main" type.
102        // Initializing this here will help transitioning out of the scheme where
103        // the Suggest class knows everything about every single dictionary.
104        mDicTypeId = Suggest.DIC_MAIN;
105        // TODO: Stop relying on the state of SubtypeSwitcher, get it as a parameter
106        mFlags = Flag.initFlags(null == flagArray ? ALL_CONFIG_FLAGS : flagArray, context,
107                SubtypeSwitcher.getInstance());
108        loadDictionary(filename, offset, length);
109    }
110
111    static {
112        Utils.loadNativeLibrary();
113    }
114
115    private native int openNative(String sourceDir, long dictOffset, long dictSize,
116            int typedLetterMultiplier, int fullWordMultiplier, int maxWordLength,
117            int maxWords, int maxAlternatives);
118    private native void closeNative(int dict);
119    private native boolean isValidWordNative(int nativeData, char[] word, int wordLength);
120    private native int getSuggestionsNative(int dict, int proximityInfo, int[] xCoordinates,
121            int[] yCoordinates, int[] inputCodes, int codesSize, int flags, char[] outputChars,
122            int[] scores);
123    private native int getBigramsNative(int dict, char[] prevWord, int prevWordLength,
124            int[] inputCodes, int inputCodesLength, char[] outputChars, int[] scores,
125            int maxWordLength, int maxBigrams, int maxAlternatives);
126
127    private final void loadDictionary(String path, long startOffset, long length) {
128        mNativeDict = openNative(path, startOffset, length,
129                    TYPED_LETTER_MULTIPLIER, FULL_WORD_SCORE_MULTIPLIER,
130                    MAX_WORD_LENGTH, MAX_WORDS, MAX_PROXIMITY_CHARS_SIZE);
131    }
132
133    @Override
134    public void getBigrams(final WordComposer codes, final CharSequence previousWord,
135            final WordCallback callback) {
136        if (mNativeDict == 0) return;
137
138        char[] chars = previousWord.toString().toCharArray();
139        Arrays.fill(mOutputChars_bigrams, (char) 0);
140        Arrays.fill(mBigramScores, 0);
141
142        int codesSize = codes.size();
143        if (codesSize <= 0) {
144            // Do not return bigrams from BinaryDictionary when nothing was typed.
145            // Only use user-history bigrams (or whatever other bigram dictionaries decide).
146            return;
147        }
148        Arrays.fill(mInputCodes, -1);
149        int[] alternatives = codes.getCodesAt(0);
150        System.arraycopy(alternatives, 0, mInputCodes, 0,
151                Math.min(alternatives.length, MAX_PROXIMITY_CHARS_SIZE));
152
153        int count = getBigramsNative(mNativeDict, chars, chars.length, mInputCodes, codesSize,
154                mOutputChars_bigrams, mBigramScores, MAX_WORD_LENGTH, MAX_BIGRAMS,
155                MAX_PROXIMITY_CHARS_SIZE);
156
157        for (int j = 0; j < count; ++j) {
158            if (mBigramScores[j] < 1) break;
159            final int start = j * MAX_WORD_LENGTH;
160            int len = 0;
161            while (len <  MAX_WORD_LENGTH && mOutputChars_bigrams[start + len] != 0) {
162                ++len;
163            }
164            if (len > 0) {
165                callback.addWord(mOutputChars_bigrams, start, len, mBigramScores[j],
166                        mDicTypeId, DataType.BIGRAM);
167            }
168        }
169    }
170
171    // proximityInfo may not be null.
172    @Override
173    public void getWords(final WordComposer codes, final WordCallback callback,
174            final ProximityInfo proximityInfo) {
175        final int count = getSuggestions(codes, proximityInfo, mOutputChars, mScores);
176
177        for (int j = 0; j < count; ++j) {
178            if (mScores[j] < 1) break;
179            final int start = j * MAX_WORD_LENGTH;
180            int len = 0;
181            while (len < MAX_WORD_LENGTH && mOutputChars[start + len] != 0) {
182                ++len;
183            }
184            if (len > 0) {
185                callback.addWord(mOutputChars, start, len, mScores[j], mDicTypeId,
186                        DataType.UNIGRAM);
187            }
188        }
189    }
190
191    /* package for test */ boolean isValidDictionary() {
192        return mNativeDict != 0;
193    }
194
195    // proximityInfo may not be null.
196    /* package for test */ int getSuggestions(final WordComposer codes,
197            final ProximityInfo proximityInfo, char[] outputChars, int[] scores) {
198        if (!isValidDictionary()) return -1;
199
200        final int codesSize = codes.size();
201        // Won't deal with really long words.
202        if (codesSize > MAX_WORD_LENGTH - 1) return -1;
203
204        Arrays.fill(mInputCodes, WordComposer.NOT_A_CODE);
205        for (int i = 0; i < codesSize; i++) {
206            int[] alternatives = codes.getCodesAt(i);
207            System.arraycopy(alternatives, 0, mInputCodes, i * MAX_PROXIMITY_CHARS_SIZE,
208                    Math.min(alternatives.length, MAX_PROXIMITY_CHARS_SIZE));
209        }
210        Arrays.fill(outputChars, (char) 0);
211        Arrays.fill(scores, 0);
212
213        return getSuggestionsNative(
214                mNativeDict, proximityInfo.getNativeProximityInfo(),
215                codes.getXCoordinates(), codes.getYCoordinates(), mInputCodes, codesSize,
216                mFlags, outputChars, scores);
217    }
218
219    @Override
220    public boolean isValidWord(CharSequence word) {
221        if (word == null) return false;
222        char[] chars = word.toString().toCharArray();
223        return isValidWordNative(mNativeDict, chars, chars.length);
224    }
225
226    @Override
227    public synchronized void close() {
228        closeInternal();
229    }
230
231    private void closeInternal() {
232        if (mNativeDict != 0) {
233            closeNative(mNativeDict);
234            mNativeDict = 0;
235        }
236    }
237
238    @Override
239    protected void finalize() throws Throwable {
240        try {
241            closeInternal();
242        } finally {
243            super.finalize();
244        }
245    }
246}
247