Merge "Fix dicttool build" am: a92d0f9262 am: 438a303732

am: 6aed05232b

Change-Id: I558fb5ff22a7428d1d82eac82b4caba469ed48d4
This commit is contained in:
Tadashi G. Takaoka 2018-10-31 09:24:08 -07:00 committed by android-build-merger
commit b6301306fa
12 changed files with 1193 additions and 61 deletions

View File

@ -174,6 +174,9 @@ public final class FormatSpec {
public static final int VERSION202 = 202; public static final int VERSION202 = 202;
// format version for Fava Dictionaries. // format version for Fava Dictionaries.
public static final int VERSION_DELIGHT3 = 86736212; public static final int VERSION_DELIGHT3 = 86736212;
public static final int MINIMUM_SUPPORTED_VERSION_OF_CODE_POINT_TABLE = VERSION201;
// Dictionary version used for testing.
public static final int VERSION4_ONLY_FOR_TESTING = 399;
public static final int VERSION402 = 402; public static final int VERSION402 = 402;
public static final int VERSION403 = 403; public static final int VERSION403 = 403;
public static final int VERSION4 = VERSION403; public static final int VERSION4 = VERSION403;

View File

@ -0,0 +1,677 @@
/*
* Copyright (C) 2012 The Android Open Source Project
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package com.android.inputmethod.latin.makedict;
import android.test.AndroidTestCase;
import android.test.suitebuilder.annotation.LargeTest;
import android.util.Log;
import android.util.Pair;
import android.util.SparseArray;
import com.android.inputmethod.latin.BinaryDictionary;
import com.android.inputmethod.latin.common.CodePointUtils;
import com.android.inputmethod.latin.makedict.BinaryDictDecoderUtils.CharEncoding;
import com.android.inputmethod.latin.makedict.BinaryDictDecoderUtils.DictBuffer;
import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions;
import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode;
import com.android.inputmethod.latin.makedict.FusionDictionary.PtNodeArray;
import com.android.inputmethod.latin.utils.BinaryDictionaryUtils;
import com.android.inputmethod.latin.utils.ByteArrayDictBuffer;
import java.io.File;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.HashMap;
import java.util.HashSet;
import java.util.List;
import java.util.Locale;
import java.util.Map.Entry;
import java.util.Random;
import java.util.Set;
import java.util.TreeMap;
/**
* Unit tests for BinaryDictDecoderUtils and BinaryDictEncoderUtils.
*/
@LargeTest
public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
private static final String TAG = BinaryDictDecoderEncoderTests.class.getSimpleName();
private static final int DEFAULT_MAX_UNIGRAMS = 300;
private static final int DEFAULT_CODE_POINT_SET_SIZE = 50;
private static final int LARGE_CODE_POINT_SET_SIZE = 300;
private static final int UNIGRAM_FREQ = 10;
private static final int BIGRAM_FREQ = 50;
private static final int TOLERANCE_OF_BIGRAM_FREQ = 5;
private static final ArrayList<String> sWords = new ArrayList<>();
private static final ArrayList<String> sWordsWithVariousCodePoints = new ArrayList<>();
private static final SparseArray<List<Integer>> sEmptyBigrams = new SparseArray<>();
private static final SparseArray<List<Integer>> sStarBigrams = new SparseArray<>();
private static final SparseArray<List<Integer>> sChainBigrams = new SparseArray<>();
final Random mRandom;
public BinaryDictDecoderEncoderTests() {
this(System.currentTimeMillis(), DEFAULT_MAX_UNIGRAMS);
}
public BinaryDictDecoderEncoderTests(final long seed, final int maxUnigrams) {
super();
BinaryDictionaryUtils.setCurrentTimeForTest(0);
Log.e(TAG, "Testing dictionary: seed is " + seed);
mRandom = new Random(seed);
sWords.clear();
sWordsWithVariousCodePoints.clear();
generateWords(maxUnigrams, mRandom);
for (int i = 0; i < sWords.size(); ++i) {
sChainBigrams.put(i, new ArrayList<Integer>());
if (i > 0) {
sChainBigrams.get(i - 1).add(i);
}
}
sStarBigrams.put(0, new ArrayList<Integer>());
// MAX - 1 because we added one above already
final int maxBigrams = Math.min(sWords.size(), FormatSpec.MAX_BIGRAMS_IN_A_PTNODE - 1);
for (int i = 1; i < maxBigrams; ++i) {
sStarBigrams.get(0).add(i);
}
}
@Override
protected void setUp() throws Exception {
super.setUp();
BinaryDictionaryUtils.setCurrentTimeForTest(0);
}
@Override
protected void tearDown() throws Exception {
// Quit test mode.
BinaryDictionaryUtils.setCurrentTimeForTest(-1);
super.tearDown();
}
private static void generateWords(final int number, final Random random) {
final int[] codePointSet = CodePointUtils.generateCodePointSet(DEFAULT_CODE_POINT_SET_SIZE,
random);
final Set<String> wordSet = new HashSet<>();
while (wordSet.size() < number) {
wordSet.add(CodePointUtils.generateWord(random, codePointSet));
}
sWords.addAll(wordSet);
final int[] largeCodePointSet = CodePointUtils.generateCodePointSet(
LARGE_CODE_POINT_SET_SIZE, random);
wordSet.clear();
while (wordSet.size() < number) {
wordSet.add(CodePointUtils.generateWord(random, largeCodePointSet));
}
sWordsWithVariousCodePoints.addAll(wordSet);
}
/**
* Adds unigrams to the dictionary.
*/
private static void addUnigrams(final int number, final FusionDictionary dict,
final List<String> words) {
for (int i = 0; i < number; ++i) {
final String word = words.get(i);
final ArrayList<WeightedString> shortcuts = new ArrayList<>();
dict.add(word, new ProbabilityInfo(UNIGRAM_FREQ), false /* isNotAWord */,
false /* isPossiblyOffensive */);
}
}
private static void addBigrams(final FusionDictionary dict,
final List<String> words,
final SparseArray<List<Integer>> bigrams) {
for (int i = 0; i < bigrams.size(); ++i) {
final int w1 = bigrams.keyAt(i);
for (int w2 : bigrams.valueAt(i)) {
dict.setBigram(words.get(w1), words.get(w2), new ProbabilityInfo(BIGRAM_FREQ));
}
}
}
// The following is useful to dump the dictionary into a textual file, but it can't compile
// on-device, so it's commented out.
// private void dumpToCombinedFileForDebug(final FusionDictionary dict, final String filename)
// throws IOException {
// com.android.inputmethod.latin.dicttool.CombinedInputOutput.writeDictionaryCombined(
// new java.io.FileWriter(new File(filename)), dict);
// }
private static long timeWritingDictToFile(final File file, final FusionDictionary dict,
final FormatSpec.FormatOptions formatOptions) {
long now = -1, diff = -1;
try {
final DictEncoder dictEncoder = BinaryDictUtils.getDictEncoder(file, formatOptions);
now = System.currentTimeMillis();
// If you need to dump the dict to a textual file, uncomment the line below and the
// function above
// dumpToCombinedFileForDebug(file, "/tmp/foo");
dictEncoder.writeDictionary(dict, formatOptions);
diff = System.currentTimeMillis() - now;
} catch (IOException e) {
Log.e(TAG, "IO exception while writing file", e);
} catch (UnsupportedFormatException e) {
Log.e(TAG, "UnsupportedFormatException", e);
}
return diff;
}
private static void checkDictionary(final FusionDictionary dict, final List<String> words,
final SparseArray<List<Integer>> bigrams) {
assertNotNull(dict);
// check unigram
for (final String word : words) {
final PtNode ptNode = FusionDictionary.findWordInTree(dict.mRootNodeArray, word);
assertNotNull(ptNode);
}
// check bigram
for (int i = 0; i < bigrams.size(); ++i) {
final int w1 = bigrams.keyAt(i);
for (final int w2 : bigrams.valueAt(i)) {
final PtNode ptNode = FusionDictionary.findWordInTree(dict.mRootNodeArray,
words.get(w1));
assertNotNull(words.get(w1) + "," + words.get(w2), ptNode.getBigram(words.get(w2)));
}
}
}
private static String outputOptions(final int bufferType,
final FormatSpec.FormatOptions formatOptions) {
final String result = " : buffer type = "
+ ((bufferType == BinaryDictUtils.USE_BYTE_BUFFER) ? "byte buffer" : "byte array");
return result + " : version = " + formatOptions.mVersion;
}
// Tests for readDictionaryBinary and writeDictionaryBinary
private static long timeReadingAndCheckDict(final File file, final List<String> words,
final SparseArray<List<Integer>> bigrams, final int bufferType) {
long now, diff = -1;
FusionDictionary dict = null;
try {
final DictDecoder dictDecoder = BinaryDictIOUtils.getDictDecoder(file, 0, file.length(),
bufferType);
now = System.currentTimeMillis();
dict = dictDecoder.readDictionaryBinary(false /* deleteDictIfBroken */);
diff = System.currentTimeMillis() - now;
} catch (IOException e) {
Log.e(TAG, "IOException while reading dictionary", e);
} catch (UnsupportedFormatException e) {
Log.e(TAG, "Unsupported format", e);
}
checkDictionary(dict, words, bigrams);
return diff;
}
// Tests for readDictionaryBinary and writeDictionaryBinary
private String runReadAndWrite(final List<String> words,
final SparseArray<List<Integer>> bigrams,
final int bufferType, final FormatSpec.FormatOptions formatOptions,
final String message) {
final String dictName = "runReadAndWrite";
final String dictVersion = Long.toString(System.currentTimeMillis());
final File file = BinaryDictUtils.getDictFile(dictName, dictVersion, formatOptions,
getContext().getCacheDir());
final FusionDictionary dict = new FusionDictionary(new PtNodeArray(),
BinaryDictUtils.makeDictionaryOptions(dictName, dictVersion, formatOptions));
addUnigrams(words.size(), dict, words);
addBigrams(dict, words, bigrams);
checkDictionary(dict, words, bigrams);
final long write = timeWritingDictToFile(file, dict, formatOptions);
final long read = timeReadingAndCheckDict(file, words, bigrams, bufferType);
return "PROF: read=" + read + "ms, write=" + write + "ms :" + message
+ " : " + outputOptions(bufferType, formatOptions);
}
private void runReadAndWriteTests(final List<String> results, final int bufferType,
final FormatSpec.FormatOptions formatOptions) {
results.add(runReadAndWrite(sWords, sEmptyBigrams, bufferType,
formatOptions, "unigram"));
results.add(runReadAndWrite(sWords, sChainBigrams, bufferType,
formatOptions, "chain"));
results.add(runReadAndWrite(sWords, sStarBigrams, bufferType,
formatOptions, "star"));
results.add(runReadAndWrite(sWords, sEmptyBigrams, bufferType, formatOptions,
"unigram with shortcuts"));
results.add(runReadAndWrite(sWords, sChainBigrams, bufferType, formatOptions,
"chain with shortcuts"));
results.add(runReadAndWrite(sWords, sStarBigrams, bufferType, formatOptions,
"star with shortcuts"));
results.add(runReadAndWrite(sWordsWithVariousCodePoints, sEmptyBigrams,
bufferType, formatOptions,
"unigram with various code points"));
}
public void testCharacterTableIsPresent() throws IOException, UnsupportedFormatException {
final String[] wordSource = {"words", "used", "for", "testing", "a", "code point", "table"};
final List<String> words = Arrays.asList(wordSource);
final String correctCodePointTable = "toesdrniawuplgfcb ";
final String dictName = "codePointTableTest";
final String dictVersion = Long.toString(System.currentTimeMillis());
final String codePointTableAttribute = DictionaryHeader.CODE_POINT_TABLE_KEY;
final File file = BinaryDictUtils.getDictFile(dictName, dictVersion,
BinaryDictUtils.STATIC_OPTIONS, getContext().getCacheDir());
// Write a test dictionary
final DictEncoder dictEncoder = new Ver2DictEncoder(file,
Ver2DictEncoder.CODE_POINT_TABLE_ON);
final FormatSpec.FormatOptions formatOptions =
new FormatSpec.FormatOptions(
FormatSpec.MINIMUM_SUPPORTED_STATIC_VERSION);
final FusionDictionary sourcedict = new FusionDictionary(new PtNodeArray(),
BinaryDictUtils.makeDictionaryOptions(dictName, dictVersion, formatOptions));
addUnigrams(words.size(), sourcedict, words);
dictEncoder.writeDictionary(sourcedict, formatOptions);
// Read the dictionary
final DictDecoder dictDecoder = BinaryDictIOUtils.getDictDecoder(file, 0, file.length(),
DictDecoder.USE_BYTEARRAY);
final DictionaryHeader fileHeader = dictDecoder.readHeader();
// Check if codePointTable is present
assertTrue("codePointTable is not present",
fileHeader.mDictionaryOptions.mAttributes.containsKey(codePointTableAttribute));
final String codePointTable =
fileHeader.mDictionaryOptions.mAttributes.get(codePointTableAttribute);
// Check if codePointTable is correct
assertEquals("codePointTable is incorrect", codePointTable, correctCodePointTable);
}
// Unit test for CharEncoding.readString and CharEncoding.writeString.
public void testCharEncoding() {
// the max length of a word in sWords is less than 50.
// See generateWords.
final byte[] buffer = new byte[50 * 3];
final DictBuffer dictBuffer = new ByteArrayDictBuffer(buffer);
for (final String word : sWords) {
Arrays.fill(buffer, (byte) 0);
CharEncoding.writeString(buffer, 0, word, null);
dictBuffer.position(0);
final String str = CharEncoding.readString(dictBuffer);
assertEquals(word, str);
}
}
public void testReadAndWriteWithByteBuffer() {
final List<String> results = new ArrayList<>();
runReadAndWriteTests(results, BinaryDictUtils.USE_BYTE_BUFFER,
BinaryDictUtils.STATIC_OPTIONS);
runReadAndWriteTests(results, BinaryDictUtils.USE_BYTE_BUFFER,
BinaryDictUtils.DYNAMIC_OPTIONS_WITHOUT_TIMESTAMP);
runReadAndWriteTests(results, BinaryDictUtils.USE_BYTE_BUFFER,
BinaryDictUtils.DYNAMIC_OPTIONS_WITH_TIMESTAMP);
for (final String result : results) {
Log.d(TAG, result);
}
}
public void testReadAndWriteWithByteArray() {
final List<String> results = new ArrayList<>();
runReadAndWriteTests(results, BinaryDictUtils.USE_BYTE_ARRAY,
BinaryDictUtils.STATIC_OPTIONS);
runReadAndWriteTests(results, BinaryDictUtils.USE_BYTE_ARRAY,
BinaryDictUtils.DYNAMIC_OPTIONS_WITHOUT_TIMESTAMP);
runReadAndWriteTests(results, BinaryDictUtils.USE_BYTE_ARRAY,
BinaryDictUtils.DYNAMIC_OPTIONS_WITH_TIMESTAMP);
for (final String result : results) {
Log.d(TAG, result);
}
}
// Tests for readUnigramsAndBigramsBinary
private static void checkWordMap(final List<String> expectedWords,
final SparseArray<List<Integer>> expectedBigrams,
final TreeMap<Integer, String> resultWords,
final TreeMap<Integer, Integer> resultFrequencies,
final TreeMap<Integer, ArrayList<PendingAttribute>> resultBigrams,
final boolean checkProbability) {
// check unigrams
final Set<String> actualWordsSet = new HashSet<>(resultWords.values());
final Set<String> expectedWordsSet = new HashSet<>(expectedWords);
assertEquals(actualWordsSet, expectedWordsSet);
if (checkProbability) {
for (int freq : resultFrequencies.values()) {
assertEquals(freq, UNIGRAM_FREQ);
}
}
// check bigrams
final HashMap<String, Set<String>> expBigrams = new HashMap<>();
for (int i = 0; i < expectedBigrams.size(); ++i) {
final String word1 = expectedWords.get(expectedBigrams.keyAt(i));
for (int w2 : expectedBigrams.valueAt(i)) {
if (expBigrams.get(word1) == null) {
expBigrams.put(word1, new HashSet<String>());
}
expBigrams.get(word1).add(expectedWords.get(w2));
}
}
final HashMap<String, Set<String>> actBigrams = new HashMap<>();
for (Entry<Integer, ArrayList<PendingAttribute>> entry : resultBigrams.entrySet()) {
final String word1 = resultWords.get(entry.getKey());
final int unigramFreq = resultFrequencies.get(entry.getKey());
for (PendingAttribute attr : entry.getValue()) {
final String word2 = resultWords.get(attr.mAddress);
if (actBigrams.get(word1) == null) {
actBigrams.put(word1, new HashSet<String>());
}
actBigrams.get(word1).add(word2);
if (checkProbability) {
final int bigramFreq = BinaryDictIOUtils.reconstructBigramFrequency(
unigramFreq, attr.mFrequency);
assertTrue(Math.abs(bigramFreq - BIGRAM_FREQ) < TOLERANCE_OF_BIGRAM_FREQ);
}
}
}
assertEquals(actBigrams, expBigrams);
}
private static long timeAndCheckReadUnigramsAndBigramsBinary(final File file,
final List<String> words, final SparseArray<List<Integer>> bigrams,
final int bufferType, final boolean checkProbability) {
final TreeMap<Integer, String> resultWords = new TreeMap<>();
final TreeMap<Integer, ArrayList<PendingAttribute>> resultBigrams = new TreeMap<>();
final TreeMap<Integer, Integer> resultFreqs = new TreeMap<>();
long now = -1, diff = -1;
try {
final DictDecoder dictDecoder = BinaryDictIOUtils.getDictDecoder(file, 0, file.length(),
bufferType);
now = System.currentTimeMillis();
dictDecoder.readUnigramsAndBigramsBinary(resultWords, resultFreqs, resultBigrams);
diff = System.currentTimeMillis() - now;
} catch (IOException e) {
Log.e(TAG, "IOException", e);
} catch (UnsupportedFormatException e) {
Log.e(TAG, "UnsupportedFormatException", e);
}
checkWordMap(words, bigrams, resultWords, resultFreqs, resultBigrams, checkProbability);
return diff;
}
private String runReadUnigramsAndBigramsBinary(final ArrayList<String> words,
final SparseArray<List<Integer>> bigrams, final int bufferType,
final FormatSpec.FormatOptions formatOptions, final String message) {
final String dictName = "runReadUnigrams";
final String dictVersion = Long.toString(System.currentTimeMillis());
final File file = BinaryDictUtils.getDictFile(dictName, dictVersion, formatOptions,
getContext().getCacheDir());
// making the dictionary from lists of words.
final FusionDictionary dict = new FusionDictionary(new PtNodeArray(),
BinaryDictUtils.makeDictionaryOptions(dictName, dictVersion, formatOptions));
addUnigrams(words.size(), dict, words);
addBigrams(dict, words, bigrams);
timeWritingDictToFile(file, dict, formatOptions);
// Caveat: Currently, the Java code to read a v4 dictionary doesn't calculate the
// probability when there's a timestamp for the entry.
// TODO: Abandon the Java code, and implement the v4 dictionary reading code in native.
long wordMap = timeAndCheckReadUnigramsAndBigramsBinary(file, words, bigrams, bufferType,
!formatOptions.mHasTimestamp /* checkProbability */);
long fullReading = timeReadingAndCheckDict(file, words, bigrams,
bufferType);
return "readDictionaryBinary=" + fullReading + ", readUnigramsAndBigramsBinary=" + wordMap
+ " : " + message + " : " + outputOptions(bufferType, formatOptions);
}
private void runReadUnigramsAndBigramsTests(final ArrayList<String> results,
final int bufferType, final FormatSpec.FormatOptions formatOptions) {
results.add(runReadUnigramsAndBigramsBinary(sWords, sEmptyBigrams, bufferType,
formatOptions, "unigram"));
results.add(runReadUnigramsAndBigramsBinary(sWords, sChainBigrams, bufferType,
formatOptions, "chain"));
results.add(runReadUnigramsAndBigramsBinary(sWords, sStarBigrams, bufferType,
formatOptions, "star"));
}
public void testReadUnigramsAndBigramsBinaryWithByteBuffer() {
final ArrayList<String> results = new ArrayList<>();
runReadUnigramsAndBigramsTests(results, BinaryDictUtils.USE_BYTE_BUFFER,
BinaryDictUtils.STATIC_OPTIONS);
for (final String result : results) {
Log.d(TAG, result);
}
}
public void testReadUnigramsAndBigramsBinaryWithByteArray() {
final ArrayList<String> results = new ArrayList<>();
runReadUnigramsAndBigramsTests(results, BinaryDictUtils.USE_BYTE_ARRAY,
BinaryDictUtils.STATIC_OPTIONS);
for (final String result : results) {
Log.d(TAG, result);
}
}
// Tests for getTerminalPosition
private static String getWordFromBinary(final DictDecoder dictDecoder, final int address) {
if (dictDecoder.getPosition() != 0) dictDecoder.setPosition(0);
DictionaryHeader fileHeader = null;
try {
fileHeader = dictDecoder.readHeader();
} catch (IOException e) {
return null;
} catch (UnsupportedFormatException e) {
return null;
}
if (fileHeader == null) return null;
return BinaryDictDecoderUtils.getWordAtPosition(dictDecoder, fileHeader.mBodyOffset,
address).mWord;
}
private static long checkGetTerminalPosition(final DictDecoder dictDecoder, final String word,
final boolean contained) {
long diff = -1;
int position = -1;
try {
final long now = System.nanoTime();
position = dictDecoder.getTerminalPosition(word);
diff = System.nanoTime() - now;
} catch (IOException e) {
Log.e(TAG, "IOException while getTerminalPosition", e);
} catch (UnsupportedFormatException e) {
Log.e(TAG, "UnsupportedFormatException while getTerminalPosition", e);
}
assertEquals(FormatSpec.NOT_VALID_WORD != position, contained);
if (contained) assertEquals(getWordFromBinary(dictDecoder, position), word);
return diff;
}
private void runGetTerminalPosition(final ArrayList<String> words,
final SparseArray<List<Integer>> bigrams, final int bufferType,
final FormatOptions formatOptions, final String message) {
final String dictName = "testGetTerminalPosition";
final String dictVersion = Long.toString(System.currentTimeMillis());
final File file = BinaryDictUtils.getDictFile(dictName, dictVersion, formatOptions,
getContext().getCacheDir());
final FusionDictionary dict = new FusionDictionary(new PtNodeArray(),
BinaryDictUtils.makeDictionaryOptions(dictName, dictVersion, formatOptions));
addUnigrams(sWords.size(), dict, sWords);
addBigrams(dict, words, bigrams);
timeWritingDictToFile(file, dict, formatOptions);
final DictDecoder dictDecoder = BinaryDictIOUtils.getDictDecoder(file, 0, file.length(),
DictDecoder.USE_BYTEARRAY);
try {
dictDecoder.openDictBuffer();
} catch (IOException e) {
Log.e(TAG, "IOException while opening the buffer", e);
} catch (UnsupportedFormatException e) {
Log.e(TAG, "IOException while opening the buffer", e);
}
assertTrue("Can't get the buffer", dictDecoder.isDictBufferOpen());
try {
// too long word
final String longWord = "abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyz";
assertEquals(FormatSpec.NOT_VALID_WORD, dictDecoder.getTerminalPosition(longWord));
// null
assertEquals(FormatSpec.NOT_VALID_WORD, dictDecoder.getTerminalPosition(null));
// empty string
assertEquals(FormatSpec.NOT_VALID_WORD, dictDecoder.getTerminalPosition(""));
} catch (IOException e) {
} catch (UnsupportedFormatException e) {
}
// Test a word that is contained within the dictionary.
long sum = 0;
for (int i = 0; i < sWords.size(); ++i) {
final long time = checkGetTerminalPosition(dictDecoder, sWords.get(i), true);
sum += time == -1 ? 0 : time;
}
Log.d(TAG, "per search : " + (((double)sum) / sWords.size() / 1000000) + " : " + message
+ " : " + outputOptions(bufferType, formatOptions));
// Test a word that isn't contained within the dictionary.
final int[] codePointSet = CodePointUtils.generateCodePointSet(DEFAULT_CODE_POINT_SET_SIZE,
mRandom);
for (int i = 0; i < 1000; ++i) {
final String word = CodePointUtils.generateWord(mRandom, codePointSet);
if (sWords.indexOf(word) != -1) continue;
checkGetTerminalPosition(dictDecoder, word, false);
}
}
private void runGetTerminalPositionTests(final int bufferType,
final FormatOptions formatOptions) {
runGetTerminalPosition(sWords, sEmptyBigrams, bufferType, formatOptions, "unigram");
}
public void testGetTerminalPosition() {
final ArrayList<String> results = new ArrayList<>();
runGetTerminalPositionTests(BinaryDictUtils.USE_BYTE_ARRAY,
BinaryDictUtils.STATIC_OPTIONS);
runGetTerminalPositionTests(BinaryDictUtils.USE_BYTE_BUFFER,
BinaryDictUtils.STATIC_OPTIONS);
for (final String result : results) {
Log.d(TAG, result);
}
}
public void testVer2DictGetWordProperty() {
final FormatOptions formatOptions = BinaryDictUtils.STATIC_OPTIONS;
final ArrayList<String> words = sWords;
final String dictName = "testGetWordProperty";
final String dictVersion = Long.toString(System.currentTimeMillis());
final FusionDictionary dict = new FusionDictionary(new PtNodeArray(),
BinaryDictUtils.makeDictionaryOptions(dictName, dictVersion, formatOptions));
addUnigrams(words.size(), dict, words);
addBigrams(dict, words, sEmptyBigrams);
final File file = BinaryDictUtils.getDictFile(dictName, dictVersion, formatOptions,
getContext().getCacheDir());
file.delete();
timeWritingDictToFile(file, dict, formatOptions);
final BinaryDictionary binaryDictionary = new BinaryDictionary(file.getAbsolutePath(),
0 /* offset */, file.length(), true /* useFullEditDistance */,
Locale.ENGLISH, dictName, false /* isUpdatable */);
for (final String word : words) {
final WordProperty wordProperty = binaryDictionary.getWordProperty(word,
false /* isBeginningOfSentence */);
assertEquals(word, wordProperty.mWord);
assertEquals(UNIGRAM_FREQ, wordProperty.getProbability());
}
}
public void testVer2DictIteration() {
final FormatOptions formatOptions = BinaryDictUtils.STATIC_OPTIONS;
final ArrayList<String> words = sWords;
final SparseArray<List<Integer>> bigrams = sEmptyBigrams;
final String dictName = "testGetWordProperty";
final String dictVersion = Long.toString(System.currentTimeMillis());
final FusionDictionary dict = new FusionDictionary(new PtNodeArray(),
BinaryDictUtils.makeDictionaryOptions(dictName, dictVersion, formatOptions));
addUnigrams(words.size(), dict, words);
addBigrams(dict, words, bigrams);
final File file = BinaryDictUtils.getDictFile(dictName, dictVersion, formatOptions,
getContext().getCacheDir());
timeWritingDictToFile(file, dict, formatOptions);
Log.d(TAG, file.getAbsolutePath());
final BinaryDictionary binaryDictionary = new BinaryDictionary(file.getAbsolutePath(),
0 /* offset */, file.length(), true /* useFullEditDistance */,
Locale.ENGLISH, dictName, false /* isUpdatable */);
final HashSet<String> wordSet = new HashSet<>(words);
final HashSet<Pair<String, String>> bigramSet = new HashSet<>();
for (int i = 0; i < words.size(); i++) {
final List<Integer> bigramList = bigrams.get(i);
if (bigramList != null) {
for (final Integer word1Index : bigramList) {
final String word1 = words.get(word1Index);
bigramSet.add(new Pair<>(words.get(i), word1));
}
}
}
int token = 0;
do {
final BinaryDictionary.GetNextWordPropertyResult result =
binaryDictionary.getNextWordProperty(token);
final WordProperty wordProperty = result.mWordProperty;
final String word0 = wordProperty.mWord;
assertEquals(UNIGRAM_FREQ, wordProperty.mProbabilityInfo.mProbability);
wordSet.remove(word0);
if (wordProperty.mHasNgrams) {
for (final WeightedString bigramTarget : wordProperty.getBigrams()) {
final String word1 = bigramTarget.mWord;
final Pair<String, String> bigram = new Pair<>(word0, word1);
assertTrue(bigramSet.contains(bigram));
bigramSet.remove(bigram);
}
}
token = result.mNextToken;
} while (token != 0);
assertTrue(wordSet.isEmpty());
assertTrue(bigramSet.isEmpty());
}
}

View File

@ -0,0 +1,80 @@
/*
* Copyright (C) 2013 The Android Open Source Project
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package com.android.inputmethod.latin.makedict;
import com.android.inputmethod.latin.makedict.FormatSpec.DictionaryOptions;
import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions;
import java.io.File;
import java.util.HashMap;
public class BinaryDictUtils {
public static final int USE_BYTE_ARRAY = 1;
public static final int USE_BYTE_BUFFER = 2;
public static final String TEST_DICT_FILE_EXTENSION = ".testDict";
public static final FormatSpec.FormatOptions STATIC_OPTIONS =
new FormatSpec.FormatOptions(FormatSpec.VERSION202);
public static final FormatSpec.FormatOptions DYNAMIC_OPTIONS_WITHOUT_TIMESTAMP =
new FormatSpec.FormatOptions(FormatSpec.VERSION4, false /* hasTimestamp */);
public static final FormatSpec.FormatOptions DYNAMIC_OPTIONS_WITH_TIMESTAMP =
new FormatSpec.FormatOptions(FormatSpec.VERSION4, true /* hasTimestamp */);
public static DictionaryOptions makeDictionaryOptions(final String id, final String version,
final FormatSpec.FormatOptions formatOptions) {
final DictionaryOptions options = new DictionaryOptions(new HashMap<String, String>());
options.mAttributes.put(DictionaryHeader.DICTIONARY_LOCALE_KEY, "en_US");
options.mAttributes.put(DictionaryHeader.DICTIONARY_ID_KEY, id);
options.mAttributes.put(DictionaryHeader.DICTIONARY_VERSION_KEY, version);
if (formatOptions.mHasTimestamp) {
options.mAttributes.put(DictionaryHeader.HAS_HISTORICAL_INFO_KEY,
DictionaryHeader.ATTRIBUTE_VALUE_TRUE);
options.mAttributes.put(DictionaryHeader.USES_FORGETTING_CURVE_KEY,
DictionaryHeader.ATTRIBUTE_VALUE_TRUE);
}
return options;
}
public static File getDictFile(final String name, final String version,
final FormatOptions formatOptions, final File directory) {
if (formatOptions.mVersion == FormatSpec.VERSION2
|| formatOptions.mVersion == FormatSpec.VERSION201
|| formatOptions.mVersion == FormatSpec.VERSION202) {
return new File(directory, name + "." + version + TEST_DICT_FILE_EXTENSION);
} else if (formatOptions.mVersion == FormatSpec.VERSION4) {
return new File(directory, name + "." + version);
} else {
throw new RuntimeException("the format option has a wrong version : "
+ formatOptions.mVersion);
}
}
public static DictEncoder getDictEncoder(final File file, final FormatOptions formatOptions) {
if (formatOptions.mVersion == FormatSpec.VERSION4) {
if (!file.isDirectory()) {
file.mkdir();
}
return new Ver4DictEncoder(file);
} else if (formatOptions.mVersion == FormatSpec.VERSION202) {
return new Ver2DictEncoder(file, Ver2DictEncoder.CODE_POINT_TABLE_OFF);
} else {
throw new RuntimeException("The format option has a wrong version : "
+ formatOptions.mVersion);
}
}
}

View File

@ -0,0 +1,279 @@
/*
* Copyright (C) 2013 The Android Open Source Project
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package com.android.inputmethod.latin.makedict;
import com.android.inputmethod.annotations.UsedForTesting;
import com.android.inputmethod.latin.makedict.BinaryDictDecoderUtils.CharEncoding;
import com.android.inputmethod.latin.makedict.BinaryDictEncoderUtils.CodePointTable;
import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions;
import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode;
import com.android.inputmethod.latin.makedict.FusionDictionary.PtNodeArray;
import java.io.File;
import java.io.FileNotFoundException;
import java.io.FileOutputStream;
import java.io.IOException;
import java.io.OutputStream;
import java.util.ArrayList;
import java.util.Collections;
import java.util.Comparator;
import java.util.HashMap;
import java.util.Iterator;
import java.util.Map.Entry;
/**
* An implementation of DictEncoder for version 2 binary dictionary.
*/
@UsedForTesting
public class Ver2DictEncoder implements DictEncoder {
private final File mDictFile;
private OutputStream mOutStream;
private byte[] mBuffer;
private int mPosition;
private final int mCodePointTableMode;
public static final int CODE_POINT_TABLE_OFF = 0;
public static final int CODE_POINT_TABLE_ON = 1;
@UsedForTesting
public Ver2DictEncoder(final File dictFile, final int codePointTableMode) {
mDictFile = dictFile;
mOutStream = null;
mBuffer = null;
mCodePointTableMode = codePointTableMode;
}
// This constructor is used only by BinaryDictOffdeviceUtilsTests.
// If you want to use this in the production code, you should consider keeping consistency of
// the interface of Ver3DictDecoder by using factory.
@UsedForTesting
public Ver2DictEncoder(final OutputStream outStream) {
mDictFile = null;
mOutStream = outStream;
mCodePointTableMode = CODE_POINT_TABLE_OFF;
}
private void openStream() throws FileNotFoundException {
mOutStream = new FileOutputStream(mDictFile);
}
private void close() throws IOException {
if (mOutStream != null) {
mOutStream.close();
mOutStream = null;
}
}
// Package for testing
static CodePointTable makeCodePointTable(final FusionDictionary dict) {
final HashMap<Integer, Integer> codePointOccurrenceCounts = new HashMap<>();
for (final WordProperty word : dict) {
// Store per code point occurrence
final String wordString = word.mWord;
for (int i = 0; i < wordString.length(); ++i) {
final int codePoint = Character.codePointAt(wordString, i);
if (codePointOccurrenceCounts.containsKey(codePoint)) {
codePointOccurrenceCounts.put(codePoint,
codePointOccurrenceCounts.get(codePoint) + 1);
} else {
codePointOccurrenceCounts.put(codePoint, 1);
}
}
}
final ArrayList<Entry<Integer, Integer>> codePointOccurrenceArray =
new ArrayList<>(codePointOccurrenceCounts.entrySet());
// Descending order sort by occurrence (value side)
Collections.sort(codePointOccurrenceArray, new Comparator<Entry<Integer, Integer>>() {
@Override
public int compare(final Entry<Integer, Integer> a, final Entry<Integer, Integer> b) {
if (a.getValue() != b.getValue()) {
return b.getValue().compareTo(a.getValue());
}
return b.getKey().compareTo(a.getKey());
}
});
int currentCodePointTableIndex = FormatSpec.MINIMAL_ONE_BYTE_CHARACTER_VALUE;
// Temporary map for writing of nodes
final HashMap<Integer, Integer> codePointToOneByteCodeMap = new HashMap<>();
for (final Entry<Integer, Integer> entry : codePointOccurrenceArray) {
// Put a relation from the original code point to the one byte code.
codePointToOneByteCodeMap.put(entry.getKey(), currentCodePointTableIndex);
if (FormatSpec.MAXIMAL_ONE_BYTE_CHARACTER_VALUE < ++currentCodePointTableIndex) {
break;
}
}
// codePointToOneByteCodeMap for writing the trie
// codePointOccurrenceArray for writing the header
return new CodePointTable(codePointToOneByteCodeMap, codePointOccurrenceArray);
}
@Override
public void writeDictionary(final FusionDictionary dict, final FormatOptions formatOptions)
throws IOException, UnsupportedFormatException {
// We no longer support anything but the latest version of v2.
if (formatOptions.mVersion != FormatSpec.VERSION202) {
throw new UnsupportedFormatException(
"The given format options has wrong version number : "
+ formatOptions.mVersion);
}
if (mOutStream == null) {
openStream();
}
// Make code point conversion table ordered by occurrence of code points
// Version 201 or later have codePointTable
final CodePointTable codePointTable;
if (mCodePointTableMode == CODE_POINT_TABLE_OFF || formatOptions.mVersion
< FormatSpec.MINIMUM_SUPPORTED_VERSION_OF_CODE_POINT_TABLE) {
codePointTable = new CodePointTable();
} else {
codePointTable = makeCodePointTable(dict);
}
BinaryDictEncoderUtils.writeDictionaryHeader(mOutStream, dict, formatOptions,
codePointTable.mCodePointOccurrenceArray);
// Addresses are limited to 3 bytes, but since addresses can be relative to each node
// array, the structure itself is not limited to 16MB. However, if it is over 16MB deciding
// the order of the PtNode arrays becomes a quite complicated problem, because though the
// dictionary itself does not have a size limit, each node array must still be within 16MB
// of all its children and parents. As long as this is ensured, the dictionary file may
// grow to any size.
// Leave the choice of the optimal node order to the flattenTree function.
MakedictLog.i("Flattening the tree...");
ArrayList<PtNodeArray> flatNodes = BinaryDictEncoderUtils.flattenTree(dict.mRootNodeArray);
MakedictLog.i("Computing addresses...");
BinaryDictEncoderUtils.computeAddresses(dict, flatNodes,
codePointTable.mCodePointToOneByteCodeMap);
MakedictLog.i("Checking PtNode array...");
if (MakedictLog.DBG) BinaryDictEncoderUtils.checkFlatPtNodeArrayList(flatNodes);
// Create a buffer that matches the final dictionary size.
final PtNodeArray lastNodeArray = flatNodes.get(flatNodes.size() - 1);
final int bufferSize = lastNodeArray.mCachedAddressAfterUpdate + lastNodeArray.mCachedSize;
mBuffer = new byte[bufferSize];
MakedictLog.i("Writing file...");
for (PtNodeArray nodeArray : flatNodes) {
BinaryDictEncoderUtils.writePlacedPtNodeArray(dict, this, nodeArray,
codePointTable.mCodePointToOneByteCodeMap);
}
if (MakedictLog.DBG) BinaryDictEncoderUtils.showStatistics(flatNodes);
mOutStream.write(mBuffer, 0, mPosition);
MakedictLog.i("Done");
close();
}
@Override
public void setPosition(final int position) {
if (mBuffer == null || position < 0 || position >= mBuffer.length) return;
mPosition = position;
}
@Override
public int getPosition() {
return mPosition;
}
@Override
public void writePtNodeCount(final int ptNodeCount) {
final int countSize = BinaryDictIOUtils.getPtNodeCountSize(ptNodeCount);
if (countSize != 1 && countSize != 2) {
throw new RuntimeException("Strange size from getGroupCountSize : " + countSize);
}
final int encodedPtNodeCount = (countSize == 2) ?
(ptNodeCount | FormatSpec.LARGE_PTNODE_ARRAY_SIZE_FIELD_SIZE_FLAG) : ptNodeCount;
mPosition = BinaryDictEncoderUtils.writeUIntToBuffer(mBuffer, mPosition, encodedPtNodeCount,
countSize);
}
private void writePtNodeFlags(final PtNode ptNode,
final HashMap<Integer, Integer> codePointToOneByteCodeMap) {
final int childrenPos = BinaryDictEncoderUtils.getChildrenPosition(ptNode,
codePointToOneByteCodeMap);
mPosition = BinaryDictEncoderUtils.writeUIntToBuffer(mBuffer, mPosition,
BinaryDictEncoderUtils.makePtNodeFlags(ptNode, childrenPos),
FormatSpec.PTNODE_FLAGS_SIZE);
}
private void writeCharacters(final int[] codePoints, final boolean hasSeveralChars,
final HashMap<Integer, Integer> codePointToOneByteCodeMap) {
mPosition = CharEncoding.writeCharArray(codePoints, mBuffer, mPosition,
codePointToOneByteCodeMap);
if (hasSeveralChars) {
mBuffer[mPosition++] = FormatSpec.PTNODE_CHARACTERS_TERMINATOR;
}
}
private void writeFrequency(final int frequency) {
if (frequency >= 0) {
mPosition = BinaryDictEncoderUtils.writeUIntToBuffer(mBuffer, mPosition, frequency,
FormatSpec.PTNODE_FREQUENCY_SIZE);
}
}
private void writeChildrenPosition(final PtNode ptNode,
final HashMap<Integer, Integer> codePointToOneByteCodeMap) {
final int childrenPos = BinaryDictEncoderUtils.getChildrenPosition(ptNode,
codePointToOneByteCodeMap);
mPosition += BinaryDictEncoderUtils.writeChildrenPosition(mBuffer, mPosition,
childrenPos);
}
/**
* Write a bigram attributes list to mBuffer.
*
* @param bigrams the bigram attributes list.
* @param dict the dictionary the node array is a part of (for relative offsets).
*/
private void writeBigrams(final ArrayList<WeightedString> bigrams,
final FusionDictionary dict) {
if (bigrams == null) return;
final Iterator<WeightedString> bigramIterator = bigrams.iterator();
while (bigramIterator.hasNext()) {
final WeightedString bigram = bigramIterator.next();
final PtNode target =
FusionDictionary.findWordInTree(dict.mRootNodeArray, bigram.mWord);
final int addressOfBigram = target.mCachedAddressAfterUpdate;
final int unigramFrequencyForThisWord = target.getProbability();
final int offset = addressOfBigram
- (mPosition + FormatSpec.PTNODE_ATTRIBUTE_FLAGS_SIZE);
final int bigramFlags = BinaryDictEncoderUtils.makeBigramFlags(bigramIterator.hasNext(),
offset, bigram.getProbability(), unigramFrequencyForThisWord, bigram.mWord);
mPosition = BinaryDictEncoderUtils.writeUIntToBuffer(mBuffer, mPosition, bigramFlags,
FormatSpec.PTNODE_ATTRIBUTE_FLAGS_SIZE);
mPosition += BinaryDictEncoderUtils.writeChildrenPosition(mBuffer, mPosition,
Math.abs(offset));
}
}
@Override
public void writePtNode(final PtNode ptNode, final FusionDictionary dict,
final HashMap<Integer, Integer> codePointToOneByteCodeMap) {
writePtNodeFlags(ptNode, codePointToOneByteCodeMap);
writeCharacters(ptNode.mChars, ptNode.hasSeveralChars(), codePointToOneByteCodeMap);
writeFrequency(ptNode.getProbability());
writeChildrenPosition(ptNode, codePointToOneByteCodeMap);
writeBigrams(ptNode.mBigrams, dict);
}
}

View File

@ -0,0 +1,133 @@
/*
* Copyright (C) 2013 The Android Open Source Project
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package com.android.inputmethod.latin.makedict;
import com.android.inputmethod.annotations.UsedForTesting;
import com.android.inputmethod.latin.BinaryDictionary;
import com.android.inputmethod.latin.Dictionary;
import com.android.inputmethod.latin.NgramContext;
import com.android.inputmethod.latin.common.LocaleUtils;
import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions;
import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode;
import com.android.inputmethod.latin.utils.BinaryDictionaryUtils;
import java.io.File;
import java.io.IOException;
import java.util.HashMap;
/**
* An implementation of DictEncoder for version 4 binary dictionary.
*/
@UsedForTesting
public class Ver4DictEncoder implements DictEncoder {
private final File mDictPlacedDir;
@UsedForTesting
public Ver4DictEncoder(final File dictPlacedDir) {
mDictPlacedDir = dictPlacedDir;
}
// TODO: This builds a FusionDictionary first and iterates it to add words to the binary
// dictionary. However, it is possible to just add words directly to the binary dictionary
// instead.
// In the long run, when we stop supporting version 2, FusionDictionary will become deprecated
// and we can remove it. Then we'll be able to just call BinaryDictionary directly.
@Override
public void writeDictionary(FusionDictionary dict, FormatOptions formatOptions)
throws IOException, UnsupportedFormatException {
if (formatOptions.mVersion != FormatSpec.VERSION4) {
throw new UnsupportedFormatException("File header has a wrong version number : "
+ formatOptions.mVersion);
}
if (!mDictPlacedDir.isDirectory()) {
throw new UnsupportedFormatException("Given path is not a directory.");
}
if (!BinaryDictionaryUtils.createEmptyDictFile(mDictPlacedDir.getAbsolutePath(),
FormatSpec.VERSION4, LocaleUtils.constructLocaleFromString(
dict.mOptions.mAttributes.get(DictionaryHeader.DICTIONARY_LOCALE_KEY)),
dict.mOptions.mAttributes)) {
throw new IOException("Cannot create dictionary file : "
+ mDictPlacedDir.getAbsolutePath());
}
final BinaryDictionary binaryDict = new BinaryDictionary(mDictPlacedDir.getAbsolutePath(),
0l, mDictPlacedDir.length(), true /* useFullEditDistance */,
LocaleUtils.constructLocaleFromString(dict.mOptions.mAttributes.get(
DictionaryHeader.DICTIONARY_LOCALE_KEY)),
Dictionary.TYPE_USER /* Dictionary type. Does not matter for us */,
true /* isUpdatable */);
if (!binaryDict.isValidDictionary()) {
// Somehow createEmptyDictFile returned true, but the file was not created correctly
throw new IOException("Cannot create dictionary file");
}
for (final WordProperty wordProperty : dict) {
if (!binaryDict.addUnigramEntry(wordProperty.mWord, wordProperty.getProbability(),
wordProperty.mIsBeginningOfSentence, wordProperty.mIsNotAWord,
wordProperty.mIsPossiblyOffensive, 0 /* timestamp */)) {
MakedictLog.e("Cannot add unigram entry for " + wordProperty.mWord);
}
if (binaryDict.needsToRunGC(true /* mindsBlockByGC */)) {
if (!binaryDict.flushWithGC()) {
MakedictLog.e("Cannot flush dict with GC.");
return;
}
}
}
for (final WordProperty word0Property : dict) {
if (!word0Property.mHasNgrams) continue;
// TODO: Support ngram.
for (final WeightedString word1 : word0Property.getBigrams()) {
final NgramContext ngramContext =
new NgramContext(new NgramContext.WordInfo(word0Property.mWord));
if (!binaryDict.addNgramEntry(ngramContext, word1.mWord,
word1.getProbability(), 0 /* timestamp */)) {
MakedictLog.e("Cannot add n-gram entry for "
+ ngramContext + " -> " + word1.mWord);
return;
}
if (binaryDict.needsToRunGC(true /* mindsBlockByGC */)) {
if (!binaryDict.flushWithGC()) {
MakedictLog.e("Cannot flush dict with GC.");
return;
}
}
}
}
if (!binaryDict.flushWithGC()) {
MakedictLog.e("Cannot flush dict with GC.");
return;
}
binaryDict.close();
}
@Override
public void setPosition(int position) {
}
@Override
public int getPosition() {
return 0;
}
@Override
public void writePtNodeCount(int ptNodeCount) {
}
@Override
public void writePtNode(PtNode ptNode, FusionDictionary dict,
HashMap<Integer, Integer> codePointToOneByteCodeMap) {
}
}

View File

@ -47,10 +47,14 @@ LATINIME_SRC_FILES_FOR_DICTTOOL := \
latin/utils/JniUtils.java latin/utils/JniUtils.java
LATINIME_OVERRIDABLE_SRC_FILES_FOR_DICTTOOL := \ LATINIME_OVERRIDABLE_SRC_FILES_FOR_DICTTOOL := \
latin/define/DebugFlags.java latin/define/DebugFlags.java \
latin/define/DecoderSpecificConstants.java
LATINIME_TEST_SRC_FILES_FOR_DICTTOOL := \ LATINIME_TEST_SRC_FILES_FOR_DICTTOOL := \
utils/ByteArrayDictBuffer.java utils/ByteArrayDictBuffer.java \
makedict/Ver2DictEncoder.java \
makedict/Ver4DictEncoder.java \
makedict/BinaryDictDecoderEncoderTests.java
USED_TARGETED_SRC_FILES := \ USED_TARGETED_SRC_FILES := \
$(addprefix $(LATINIME_BASE_SRC_DIR)/, $(LATINIME_SRC_FILES_FOR_DICTTOOL)) \ $(addprefix $(LATINIME_BASE_SRC_DIR)/, $(LATINIME_SRC_FILES_FOR_DICTTOOL)) \

View File

@ -106,8 +106,7 @@ public class CombinedInputOutput {
final String args[] = line.trim().split(","); final String args[] = line.trim().split(",");
if (args[0].matches(CombinedFormatUtils.WORD_TAG + "=.*")) { if (args[0].matches(CombinedFormatUtils.WORD_TAG + "=.*")) {
if (null != word) { if (null != word) {
dict.add(word, probabilityInfo, shortcuts.isEmpty() ? null : shortcuts, dict.add(word, probabilityInfo, isNotAWord, isPossiblyOffensive);
isNotAWord, isPossiblyOffensive);
for (WeightedString s : bigrams) { for (WeightedString s : bigrams) {
dict.setBigram(word, s.mWord, s.mProbabilityInfo); dict.setBigram(word, s.mWord, s.mProbabilityInfo);
} }
@ -148,25 +147,6 @@ public class CombinedInputOutput {
break; break;
} }
} }
} else if (args[0].matches(CombinedFormatUtils.SHORTCUT_TAG + "=.*")) {
String shortcut = null;
int shortcutFreq = 0;
for (String param : args) {
final String params[] = param.split("=", 2);
if (2 != params.length) throw new RuntimeException("Wrong format : " + line);
if (CombinedFormatUtils.SHORTCUT_TAG.equals(params[0])) {
shortcut = params[1];
} else if (CombinedFormatUtils.PROBABILITY_TAG.equals(params[0])) {
shortcutFreq = WHITELIST_TAG.equals(params[1])
? FormatSpec.SHORTCUT_WHITELIST_FREQUENCY
: Integer.parseInt(params[1]);
}
}
if (null != shortcut) {
shortcuts.add(new WeightedString(shortcut, shortcutFreq));
} else {
throw new RuntimeException("Wrong format : " + line);
}
} else if (args[0].matches(CombinedFormatUtils.BIGRAM_TAG + "=.*")) { } else if (args[0].matches(CombinedFormatUtils.BIGRAM_TAG + "=.*")) {
String secondWordOfBigram = null; String secondWordOfBigram = null;
ProbabilityInfo bigramProbabilityInfo = new ProbabilityInfo(0); ProbabilityInfo bigramProbabilityInfo = new ProbabilityInfo(0);
@ -200,8 +180,7 @@ public class CombinedInputOutput {
} }
} }
if (null != word) { if (null != word) {
dict.add(word, probabilityInfo, shortcuts.isEmpty() ? null : shortcuts, isNotAWord, dict.add(word, probabilityInfo, isNotAWord, isPossiblyOffensive);
isPossiblyOffensive);
for (WeightedString s : bigrams) { for (WeightedString s : bigrams) {
dict.setBigram(word, s.mWord, s.mProbabilityInfo); dict.setBigram(word, s.mWord, s.mProbabilityInfo);
} }

View File

@ -136,9 +136,6 @@ public class Diff extends Dicttool.Command {
} }
hasDifferences |= hasAttributesDifferencesAndPrintThemIfAny(word0Property.mWord, hasDifferences |= hasAttributesDifferencesAndPrintThemIfAny(word0Property.mWord,
"Bigram", word0Property.getBigrams(), word1PtNode.getBigrams()); "Bigram", word0Property.getBigrams(), word1PtNode.getBigrams());
hasDifferences |= hasAttributesDifferencesAndPrintThemIfAny(word0Property.mWord,
"Shortcut", word0Property.mShortcutTargets,
word1PtNode.getShortcutTargets());
} }
} }
for (final WordProperty word1Property : dict1) { for (final WordProperty word1Property : dict1) {

View File

@ -48,15 +48,6 @@ public class Info extends Dicttool.Command {
if (wordProperty.mHasNgrams) { if (wordProperty.mHasNgrams) {
bigramCount += wordProperty.mNgrams.size(); bigramCount += wordProperty.mNgrams.size();
} }
if (null != wordProperty.mShortcutTargets) {
shortcutCount += wordProperty.mShortcutTargets.size();
for (WeightedString shortcutTarget : wordProperty.mShortcutTargets) {
if (FormatSpec.SHORTCUT_WHITELIST_FREQUENCY
== shortcutTarget.getProbability()) {
++whitelistCount;
}
}
}
} }
System.out.println("Words in the dictionary : " + wordCount); System.out.println("Words in the dictionary : " + wordCount);
System.out.println("Bigram count : " + bigramCount); System.out.println("Bigram count : " + bigramCount);
@ -78,17 +69,6 @@ public class Info extends Dicttool.Command {
if (ptNode.getIsPossiblyOffensive()) { if (ptNode.getIsPossiblyOffensive()) {
System.out.println(" Is possibly offensive"); System.out.println(" Is possibly offensive");
} }
final ArrayList<WeightedString> shortcutTargets = ptNode.getShortcutTargets();
if (null == shortcutTargets || shortcutTargets.isEmpty()) {
System.out.println(" No shortcuts");
} else {
for (final WeightedString shortcutTarget : shortcutTargets) {
System.out.println(" Shortcut target: " + shortcutTarget.mWord + " ("
+ (FormatSpec.SHORTCUT_WHITELIST_FREQUENCY
== shortcutTarget.getProbability() ?
"whitelist" : shortcutTarget.getProbability()) + ")");
}
}
final ArrayList<WeightedString> bigrams = ptNode.getBigrams(); final ArrayList<WeightedString> bigrams = ptNode.getBigrams();
if (null == bigrams || bigrams.isEmpty()) { if (null == bigrams || bigrams.isEmpty()) {
System.out.println(" No bigrams"); System.out.println(" No bigrams");

View File

@ -85,15 +85,15 @@ public class BinaryDictOffdeviceUtilsTests extends TestCase {
testOptions.mAttributes.put(DictionaryHeader.DICTIONARY_LOCALE_KEY, LOCALE); testOptions.mAttributes.put(DictionaryHeader.DICTIONARY_LOCALE_KEY, LOCALE);
testOptions.mAttributes.put(DictionaryHeader.DICTIONARY_ID_KEY, ID); testOptions.mAttributes.put(DictionaryHeader.DICTIONARY_ID_KEY, ID);
final FusionDictionary dict = new FusionDictionary(new PtNodeArray(), testOptions); final FusionDictionary dict = new FusionDictionary(new PtNodeArray(), testOptions);
dict.add("foo", new ProbabilityInfo(TEST_FREQ), null, false /* isNotAWord */, dict.add("foo", new ProbabilityInfo(TEST_FREQ), false /* isNotAWord */,
false /* isPossiblyOffensive */); false /* isPossiblyOffensive */);
dict.add("fta", new ProbabilityInfo(1), null, false /* isNotAWord */, dict.add("fta", new ProbabilityInfo(1), false /* isNotAWord */,
false /* isPossiblyOffensive */); false /* isPossiblyOffensive */);
dict.add("ftb", new ProbabilityInfo(1), null, false /* isNotAWord */, dict.add("ftb", new ProbabilityInfo(1), false /* isNotAWord */,
false /* isPossiblyOffensive */); false /* isPossiblyOffensive */);
dict.add("bar", new ProbabilityInfo(1), null, false /* isNotAWord */, dict.add("bar", new ProbabilityInfo(1), false /* isNotAWord */,
false /* isPossiblyOffensive */); false /* isPossiblyOffensive */);
dict.add("fool", new ProbabilityInfo(1), null, false /* isNotAWord */, dict.add("fool", new ProbabilityInfo(1), false /* isNotAWord */,
false /* isPossiblyOffensive */); false /* isPossiblyOffensive */);
final File dst = File.createTempFile("testGetRawDict", ".tmp"); final File dst = File.createTempFile("testGetRawDict", ".tmp");
@ -171,8 +171,8 @@ public class BinaryDictOffdeviceUtilsTests extends TestCase {
for (int i = 0; i < sWords.size(); ++i) { for (int i = 0; i < sWords.size(); ++i) {
final String word = sWords.get(i); final String word = sWords.get(i);
dict.add(word, new ProbabilityInfo(TEST_FREQ), null /* shortcuts */, dict.add(word, new ProbabilityInfo(TEST_FREQ), false /* isNotAWord */,
false /* isNotAWord */, false /* isPossiblyOffensive */); false /* isPossiblyOffensive */);
} }
File file = File.createTempFile(dictName, ".tmp"); File file = File.createTempFile(dictName, ".tmp");

View File

@ -33,15 +33,15 @@ public class BinaryDictEncoderFlattenTreeTests extends TestCase {
public void testFlattenNodes() { public void testFlattenNodes() {
final FusionDictionary dict = new FusionDictionary(new PtNodeArray(), final FusionDictionary dict = new FusionDictionary(new PtNodeArray(),
new DictionaryOptions(new HashMap<String, String>())); new DictionaryOptions(new HashMap<String, String>()));
dict.add("foo", new ProbabilityInfo(1), null, false /* isNotAWord */, dict.add("foo", new ProbabilityInfo(1), false /* isNotAWord */,
false /* isPossiblyOffensive */); false /* isPossiblyOffensive */);
dict.add("fta", new ProbabilityInfo(1), null, false /* isNotAWord */, dict.add("fta", new ProbabilityInfo(1), false /* isNotAWord */,
false /* isPossiblyOffensive */); false /* isPossiblyOffensive */);
dict.add("ftb", new ProbabilityInfo(1), null, false /* isNotAWord */, dict.add("ftb", new ProbabilityInfo(1), false /* isNotAWord */,
false /* isPossiblyOffensive */); false /* isPossiblyOffensive */);
dict.add("bar", new ProbabilityInfo(1), null, false /* isNotAWord */, dict.add("bar", new ProbabilityInfo(1), false /* isNotAWord */,
false /* isPossiblyOffensive */); false /* isPossiblyOffensive */);
dict.add("fool", new ProbabilityInfo(1), null, false /* isNotAWord */, dict.add("fool", new ProbabilityInfo(1), false /* isNotAWord */,
false /* isPossiblyOffensive */); false /* isPossiblyOffensive */);
final ArrayList<PtNodeArray> result = final ArrayList<PtNodeArray> result =
BinaryDictEncoderUtils.flattenTree(dict.mRootNodeArray); BinaryDictEncoderUtils.flattenTree(dict.mRootNodeArray);

View File

@ -102,7 +102,7 @@ public class FusionDictionaryTest extends TestCase {
prepare(time); prepare(time);
for (int i = 0; i < sWords.size(); ++i) { for (int i = 0; i < sWords.size(); ++i) {
System.out.println("Adding in pos " + i + " : " + dumpWord(sWords.get(i))); System.out.println("Adding in pos " + i + " : " + dumpWord(sWords.get(i)));
dict.add(sWords.get(i), new ProbabilityInfo(180), null, false, dict.add(sWords.get(i), new ProbabilityInfo(180), false,
false /* isPossiblyOffensive */); false /* isPossiblyOffensive */);
dumpDict(dict); dumpDict(dict);
checkDictionary(dict, sWords, i); checkDictionary(dict, sWords, i);