nyancrimew/LatinIME
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

Merge "Use native logic to read Ver4 dict."

Browse source
This commit is contained in:
Keisuke Kuroyanagi 2014-02-06 13:00:09 +00:00 committed by Android (Google) Code Review
commit 6422f77e94
8 changed files with 106 additions and 1066 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

46
java/src/com/android/inputmethod/latin/makedict/AbstractDictDecoder.java
View file

@ -22,6 +22,7 @@ import com.android.inputmethod.latin.makedict.BinaryDictDecoderUtils.DictBuffer;
import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions;
import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString;
import java.io.FileNotFoundException;
import java.io.IOException;
import java.util.ArrayList;
import java.util.HashMap;
@ -223,4 +224,49 @@ public abstract class AbstractDictDecoder implements DictDecoder {
public boolean hasValidRawBinaryDictionary() {
return checkHeader() == SUCCESS;
}
// Placeholder implementations below. These are actually unused.
@Override
public void openDictBuffer() throws FileNotFoundException, IOException,
UnsupportedFormatException {
}
@Override
public boolean isDictBufferOpen() {
return false;
}
@Override
public PtNodeInfo readPtNode(final int ptNodePos, final FormatOptions options) {
return null;
}
@Override
public void setPosition(int newPos) {
}
@Override
public int getPosition() {
return 0;
}
@Override
public int readPtNodeCount() {
return 0;
}
@Override
public boolean readAndFollowForwardLink() {
return false;
}
@Override
public boolean hasNextPtNodeArray() {
return false;
}
@Override
@UsedForTesting
public void skipPtNode(final FormatOptions formatOptions) {
}
}

22
java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java
View file

@ -436,25 +436,25 @@ public final class FusionDictionary implements Iterable<WordProperty> {
/**
* Helper method to add a new bigram to the dictionary.
*
* @param word1 the previous word of the context
* @param word2 the next word of the context
* @param word0 the previous word of the context
* @param word1 the next word of the context
* @param frequency the bigram frequency
*/
public void setBigram(final String word1, final String word2, final int frequency) {
PtNode ptNode = findWordInTree(mRootNodeArray, word1);
if (ptNode != null) {
final PtNode ptNode2 = findWordInTree(mRootNodeArray, word2);
if (ptNode2 == null) {
add(getCodePoints(word2), 0, null, false /* isNotAWord */,
public void setBigram(final String word0, final String word1, final int frequency) {
PtNode ptNode0 = findWordInTree(mRootNodeArray, word0);
if (ptNode0 != null) {
final PtNode ptNode1 = findWordInTree(mRootNodeArray, word1);
if (ptNode1 == null) {
add(getCodePoints(word1), 0, null, false /* isNotAWord */,
false /* isBlacklistEntry */);
// The PtNode for the first word may have moved by the above insertion,
// if word1 and word2 share a common stem that happens not to have been
// a cutting point until now. In this case, we need to refresh ptNode.
ptNode = findWordInTree(mRootNodeArray, word1);
ptNode0 = findWordInTree(mRootNodeArray, word0);
}
ptNode.addBigram(word2, frequency);
ptNode0.addBigram(word1, frequency);
} else {
throw new RuntimeException("First word of bigram not found");
throw new RuntimeException("First word of bigram not found " + word0);
}
}

223
java/src/com/android/inputmethod/latin/makedict/SparseTable.java
View file

@ -1,223 +0,0 @@
/*
* 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.utils.CollectionUtils;
import java.io.File;
import java.io.FileInputStream;
import java.io.FileOutputStream;
import java.io.IOException;
import java.io.OutputStream;
import java.util.ArrayList;
import java.util.Collections;
/**
* SparseTable is an extensible map from integer to integer.
* This holds one value for every mBlockSize keys, so it uses 1/mBlockSize'th of the full index
* memory.
*/
@UsedForTesting
public class SparseTable {
/**
* mLookupTable is indexed by terminal ID, containing exactly one entry for every mBlockSize
* terminals.
* It contains at index i = j / mBlockSize the index in each ArrayList in mContentsTables where
* the values for terminals with IDs j to j + mBlockSize - 1 are stored as an mBlockSize-sized
* integer array.
*/
private final ArrayList<Integer> mLookupTable;
private final ArrayList<ArrayList<Integer>> mContentTables;
private final int mBlockSize;
private final int mContentTableCount;
public static final int NOT_EXIST = -1;
public static final int SIZE_OF_INT_IN_BYTES = 4;
@UsedForTesting
public SparseTable(final int initialCapacity, final int blockSize,
final int contentTableCount) {
mBlockSize = blockSize;
final int lookupTableSize = initialCapacity / mBlockSize
+ (initialCapacity % mBlockSize > 0 ? 1 : 0);
mLookupTable = new ArrayList<Integer>(Collections.nCopies(lookupTableSize, NOT_EXIST));
mContentTableCount = contentTableCount;
mContentTables = CollectionUtils.newArrayList();
for (int i = 0; i < mContentTableCount; ++i) {
mContentTables.add(new ArrayList<Integer>());
}
}
@UsedForTesting
public SparseTable(final ArrayList<Integer> lookupTable,
final ArrayList<ArrayList<Integer>> contentTables, final int blockSize) {
mBlockSize = blockSize;
mContentTableCount = contentTables.size();
mLookupTable = lookupTable;
mContentTables = contentTables;
}
/**
* Converts an byte array to an int array considering each set of 4 bytes is an int stored in
* big-endian.
* The length of byteArray must be a multiple of four.
* Otherwise, IndexOutOfBoundsException will be raised.
*/
@UsedForTesting
private static ArrayList<Integer> convertByteArrayToIntegerArray(final byte[] byteArray) {
final ArrayList<Integer> integerArray = new ArrayList<Integer>(byteArray.length / 4);
for (int i = 0; i < byteArray.length; i += 4) {
int value = 0;
for (int j = i; j < i + 4; ++j) {
value <<= 8;
value |= byteArray[j] & 0xFF;
}
integerArray.add(value);
}
return integerArray;
}
@UsedForTesting
public int get(final int contentTableIndex, final int index) {
if (!contains(index)) {
return NOT_EXIST;
}
return mContentTables.get(contentTableIndex).get(
mLookupTable.get(index / mBlockSize) + (index % mBlockSize));
}
@UsedForTesting
public ArrayList<Integer> getAll(final int index) {
final ArrayList<Integer> ret = CollectionUtils.newArrayList();
for (int i = 0; i < mContentTableCount; ++i) {
ret.add(get(i, index));
}
return ret;
}
@UsedForTesting
public void set(final int contentTableIndex, final int index, final int value) {
if (mLookupTable.get(index / mBlockSize) == NOT_EXIST) {
mLookupTable.set(index / mBlockSize, mContentTables.get(contentTableIndex).size());
for (int i = 0; i < mContentTableCount; ++i) {
for (int j = 0; j < mBlockSize; ++j) {
mContentTables.get(i).add(NOT_EXIST);
}
}
}
mContentTables.get(contentTableIndex).set(
mLookupTable.get(index / mBlockSize) + (index % mBlockSize), value);
}
public void remove(final int indexOfContent, final int index) {
set(indexOfContent, index, NOT_EXIST);
}
@UsedForTesting
public int size() {
return mLookupTable.size() * mBlockSize;
}
@UsedForTesting
/* package */ int getContentTableSize() {
// This class always has at least one content table.
return mContentTables.get(0).size();
}
@UsedForTesting
/* package */ int getLookupTableSize() {
return mLookupTable.size();
}
public boolean contains(final int index) {
if (index < 0 || index / mBlockSize >= mLookupTable.size()
|| mLookupTable.get(index / mBlockSize) == NOT_EXIST) {
return false;
}
return true;
}
@UsedForTesting
public void write(final OutputStream lookupOutStream, final OutputStream[] contentOutStreams)
throws IOException {
if (contentOutStreams.length != mContentTableCount) {
throw new RuntimeException(contentOutStreams.length + " streams are given, but the"
+ " table has " + mContentTableCount + " content tables.");
}
for (final int index : mLookupTable) {
BinaryDictEncoderUtils.writeUIntToStream(lookupOutStream, index, SIZE_OF_INT_IN_BYTES);
}
for (int i = 0; i < contentOutStreams.length; ++i) {
for (final int data : mContentTables.get(i)) {
BinaryDictEncoderUtils.writeUIntToStream(contentOutStreams[i], data,
SIZE_OF_INT_IN_BYTES);
}
}
}
@UsedForTesting
public void writeToFiles(final File lookupTableFile, final File[] contentFiles)
throws IOException {
FileOutputStream lookupTableOutStream = null;
final FileOutputStream[] contentTableOutStreams = new FileOutputStream[mContentTableCount];
try {
lookupTableOutStream = new FileOutputStream(lookupTableFile);
for (int i = 0; i < contentFiles.length; ++i) {
contentTableOutStreams[i] = new FileOutputStream(contentFiles[i]);
}
write(lookupTableOutStream, contentTableOutStreams);
} finally {
if (lookupTableOutStream != null) {
lookupTableOutStream.close();
}
for (int i = 0; i < contentTableOutStreams.length; ++i) {
if (contentTableOutStreams[i] != null) {
contentTableOutStreams[i].close();
}
}
}
}
private static byte[] readFileToByteArray(final File file) throws IOException {
final byte[] contents = new byte[(int) file.length()];
FileInputStream inStream = null;
try {
inStream = new FileInputStream(file);
inStream.read(contents);
} finally {
if (inStream != null) {
inStream.close();
}
}
return contents;
}
@UsedForTesting
public static SparseTable readFromFiles(final File lookupTableFile, final File[] contentFiles,
final int blockSize) throws IOException {
final ArrayList<ArrayList<Integer>> contentTables =
new ArrayList<ArrayList<Integer>>(contentFiles.length);
for (int i = 0; i < contentFiles.length; ++i) {
contentTables.add(convertByteArrayToIntegerArray(readFileToByteArray(contentFiles[i])));
}
return new SparseTable(convertByteArrayToIntegerArray(readFileToByteArray(lookupTableFile)),
contentTables, blockSize);
}
}

126
java/src/com/android/inputmethod/latin/makedict/SparseTableContentReader.java
View file

@ -1,126 +0,0 @@
/*
* 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.BinaryDictDecoderUtils.DictBuffer;
import com.android.inputmethod.latin.makedict.DictDecoder.DictionaryBufferFactory;
import java.io.File;
import java.io.FileNotFoundException;
import java.io.IOException;
/**
* An auxiliary class for reading SparseTable and data written by SparseTableContentWriter.
*/
public class SparseTableContentReader {
/**
* An interface of a function which is passed to SparseTableContentReader.read.
*/
public interface SparseTableContentReaderInterface {
/**
* Reads data.
*
* @param buffer the DictBuffer. The position of the buffer is set to the head of data.
*/
public void read(final DictBuffer buffer);
}
protected final int mContentCount;
protected final int mBlockSize;
protected final File mBaseDir;
protected final File mLookupTableFile;
protected final File[] mAddressTableFiles;
protected final File[] mContentFiles;
protected DictBuffer mLookupTableBuffer;
protected final DictBuffer[] mAddressTableBuffers;
private final DictBuffer[] mContentBuffers;
protected final DictionaryBufferFactory mFactory;
/**
* Sole constructor of SparseTableContentReader.
*
* @param name the name of SparseTable.
* @param blockSize the block size of the content table.
* @param baseDir the directory which contains the files of the content table.
* @param contentFilenames the file names of content files.
* @param contentSuffixes the ids of contents. These ids are used for a suffix of a name of
* address files and content files.
* @param factory the DictionaryBufferFactory which is used for opening the files.
*/
public SparseTableContentReader(final String name, final int blockSize, final File baseDir,
final String[] contentFilenames, final String[] contentSuffixes,
final DictionaryBufferFactory factory) {
if (contentFilenames.length != contentSuffixes.length) {
throw new RuntimeException("The length of contentFilenames and the length of"
+ " contentSuffixes are different " + contentFilenames.length + ", "
+ contentSuffixes.length);
}
mBlockSize = blockSize;
mBaseDir = baseDir;
mFactory = factory;
mContentCount = contentFilenames.length;
mLookupTableFile = new File(baseDir, name + FormatSpec.LOOKUP_TABLE_FILE_SUFFIX);
mAddressTableFiles = new File[mContentCount];
mContentFiles = new File[mContentCount];
for (int i = 0; i < mContentCount; ++i) {
mAddressTableFiles[i] = new File(mBaseDir,
name + FormatSpec.CONTENT_TABLE_FILE_SUFFIX + contentSuffixes[i]);
mContentFiles[i] = new File(mBaseDir, contentFilenames[i] + contentSuffixes[i]);
}
mAddressTableBuffers = new DictBuffer[mContentCount];
mContentBuffers = new DictBuffer[mContentCount];
}
public void openBuffers() throws FileNotFoundException, IOException {
mLookupTableBuffer = mFactory.getDictionaryBuffer(mLookupTableFile);
for (int i = 0; i < mContentCount; ++i) {
mAddressTableBuffers[i] = mFactory.getDictionaryBuffer(mAddressTableFiles[i]);
mContentBuffers[i] = mFactory.getDictionaryBuffer(mContentFiles[i]);
}
}
/**
* Calls the read() callback of the reader with the appropriate buffer appropriately positioned.
* @param contentNumber the index in the original contentFilenames[] array.
* @param terminalId the terminal ID to read.
* @param reader the reader on which to call the callback.
*/
protected void read(final int contentNumber, final int terminalId,
final SparseTableContentReaderInterface reader) {
if (terminalId < 0 || (terminalId / mBlockSize) * SparseTable.SIZE_OF_INT_IN_BYTES
>= mLookupTableBuffer.limit()) {
return;
}
mLookupTableBuffer.position((terminalId / mBlockSize) * SparseTable.SIZE_OF_INT_IN_BYTES);
final int indexInAddressTable = mLookupTableBuffer.readInt();
if (indexInAddressTable == SparseTable.NOT_EXIST) {
return;
}
mAddressTableBuffers[contentNumber].position(SparseTable.SIZE_OF_INT_IN_BYTES
* ((indexInAddressTable * mBlockSize) + (terminalId % mBlockSize)));
final int address = mAddressTableBuffers[contentNumber].readInt();
if (address == SparseTable.NOT_EXIST) {
return;
}
mContentBuffers[contentNumber].position(address);
reader.read(mContentBuffers[contentNumber]);
}
}

93
java/src/com/android/inputmethod/latin/makedict/SparseTableContentWriter.java
View file

@ -1,93 +0,0 @@
/*
* 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 java.io.File;
import java.io.FileNotFoundException;
import java.io.FileOutputStream;
import java.io.IOException;
import java.io.OutputStream;
/**
* An auxiliary class for writing data associated with SparseTable to files.
*/
public class SparseTableContentWriter {
public interface SparseTableContentWriterInterface {
public void write(final OutputStream outStream) throws IOException;
}
private final int mContentCount;
private final SparseTable mSparseTable;
private final File mLookupTableFile;
protected final File mBaseDir;
private final File[] mAddressTableFiles;
private final File[] mContentFiles;
protected final OutputStream[] mContentOutStreams;
/**
* Sole constructor of SparseTableContentWriter.
*
* @param name the name of SparseTable.
* @param initialCapacity the initial capacity of SparseTable.
* @param blockSize the block size of the content table.
* @param baseDir the directory which contains the files of the content table.
* @param contentFilenames the file names of content files.
* @param contentIds the ids of contents. These ids are used for a suffix of a name of address
* files and content files.
*/
public SparseTableContentWriter(final String name, final int initialCapacity,
final int blockSize, final File baseDir, final String[] contentFilenames,
final String[] contentIds) {
if (contentFilenames.length != contentIds.length) {
throw new RuntimeException("The length of contentFilenames and the length of"
+ " contentIds are different " + contentFilenames.length + ", "
+ contentIds.length);
}
mContentCount = contentFilenames.length;
mSparseTable = new SparseTable(initialCapacity, blockSize, mContentCount);
mLookupTableFile = new File(baseDir, name + FormatSpec.LOOKUP_TABLE_FILE_SUFFIX);
mAddressTableFiles = new File[mContentCount];
mContentFiles = new File[mContentCount];
mBaseDir = baseDir;
for (int i = 0; i < mContentCount; ++i) {
mAddressTableFiles[i] = new File(mBaseDir,
name + FormatSpec.CONTENT_TABLE_FILE_SUFFIX + contentIds[i]);
mContentFiles[i] = new File(mBaseDir, contentFilenames[i] + contentIds[i]);
}
mContentOutStreams = new OutputStream[mContentCount];
}
public void openStreams() throws FileNotFoundException {
for (int i = 0; i < mContentCount; ++i) {
mContentOutStreams[i] = new FileOutputStream(mContentFiles[i]);
}
}
protected void write(final int contentIndex, final int index,
final SparseTableContentWriterInterface writer) throws IOException {
mSparseTable.set(contentIndex, index, (int) mContentFiles[contentIndex].length());
writer.write(mContentOutStreams[contentIndex]);
mContentOutStreams[contentIndex].flush();
}
public void closeStreams() throws IOException {
mSparseTable.writeToFiles(mLookupTableFile, mAddressTableFiles);
for (int i = 0; i < mContentCount; ++i) {
mContentOutStreams[i].close();
}
}
}

450
java/src/com/android/inputmethod/latin/makedict/Ver4DictDecoder.java
View file

@ -17,20 +17,15 @@
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.BinaryDictDecoderUtils.DictBuffer;
import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions;
import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode;
import com.android.inputmethod.latin.BinaryDictionary;
import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString;
import com.android.inputmethod.latin.utils.CollectionUtils;
import android.util.Log;
import com.android.inputmethod.latin.utils.FileUtils;
import java.io.File;
import java.io.FileNotFoundException;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Arrays;
/**
* An implementation of binary dictionary decoder for version 4 binary dictionary.
@ -39,421 +34,74 @@ import java.util.Arrays;
public class Ver4DictDecoder extends AbstractDictDecoder {
private static final String TAG = Ver4DictDecoder.class.getSimpleName();
protected static final int FILETYPE_TRIE = 1;
protected static final int FILETYPE_FREQUENCY = 2;
protected static final int FILETYPE_TERMINAL_ADDRESS_TABLE = 3;
protected static final int FILETYPE_BIGRAM_FREQ = 4;
protected static final int FILETYPE_SHORTCUT = 5;
protected static final int FILETYPE_HEADER = 6;
protected final File mDictDirectory;
protected final DictionaryBufferFactory mBufferFactory;
protected DictBuffer mDictBuffer;
protected DictBuffer mHeaderBuffer;
protected DictBuffer mFrequencyBuffer;
protected DictBuffer mTerminalAddressTableBuffer;
private BigramContentReader mBigramReader;
private ShortcutContentReader mShortcutReader;
/**
* Raw PtNode info straight out of a trie file in version 4 dictionary.
*/
protected static final class Ver4PtNodeInfo {
public final int mFlags;
public final int[] mCharacters;
public final int mTerminalId;
public final int mChildrenPos;
public final int mParentPos;
public final int mNodeSize;
public int mStartIndexOfCharacters;
public int mEndIndexOfCharacters; // exclusive
public Ver4PtNodeInfo(final int flags, final int[] characters, final int terminalId,
final int childrenPos, final int parentPos, final int nodeSize) {
mFlags = flags;
mCharacters = characters;
mTerminalId = terminalId;
mChildrenPos = childrenPos;
mParentPos = parentPos;
mNodeSize = nodeSize;
mStartIndexOfCharacters = 0;
mEndIndexOfCharacters = characters.length;
}
}
final File mDictDirectory;
final BinaryDictionary mBinaryDictionary;
@UsedForTesting
/* package */ Ver4DictDecoder(final File dictDirectory, final int factoryFlag) {
mDictDirectory = dictDirectory;
mDictBuffer = mHeaderBuffer = mFrequencyBuffer = null;
if ((factoryFlag & MASK_DICTBUFFER) == USE_READONLY_BYTEBUFFER) {
mBufferFactory = new DictionaryBufferFromReadOnlyByteBufferFactory();
} else if ((factoryFlag & MASK_DICTBUFFER) == USE_BYTEARRAY) {
mBufferFactory = new DictionaryBufferFromByteArrayFactory();
} else if ((factoryFlag & MASK_DICTBUFFER) == USE_WRITABLE_BYTEBUFFER) {
mBufferFactory = new DictionaryBufferFromWritableByteBufferFactory();
} else {
mBufferFactory = new DictionaryBufferFromReadOnlyByteBufferFactory();
}
this(dictDirectory, null /* factory */);
}
@UsedForTesting
/* package */ Ver4DictDecoder(final File dictDirectory, final DictionaryBufferFactory factory) {
mDictDirectory = dictDirectory;
mBufferFactory = factory;
mDictBuffer = mHeaderBuffer = mFrequencyBuffer = null;
}
protected File getFile(final int fileType) throws UnsupportedFormatException {
if (fileType == FILETYPE_TRIE) {
return new File(mDictDirectory,
mDictDirectory.getName() + FormatSpec.TRIE_FILE_EXTENSION);
} else if (fileType == FILETYPE_HEADER) {
return new File(mDictDirectory,
mDictDirectory.getName() + FormatSpec.HEADER_FILE_EXTENSION);
} else if (fileType == FILETYPE_FREQUENCY) {
return new File(mDictDirectory,
mDictDirectory.getName() + FormatSpec.FREQ_FILE_EXTENSION);
} else if (fileType == FILETYPE_TERMINAL_ADDRESS_TABLE) {
return new File(mDictDirectory,
mDictDirectory.getName() + FormatSpec.TERMINAL_ADDRESS_TABLE_FILE_EXTENSION);
} else if (fileType == FILETYPE_BIGRAM_FREQ) {
return new File(mDictDirectory,
mDictDirectory.getName() + FormatSpec.BIGRAM_FILE_EXTENSION
+ FormatSpec.BIGRAM_FREQ_CONTENT_ID);
} else if (fileType == FILETYPE_SHORTCUT) {
return new File(mDictDirectory,
mDictDirectory.getName() + FormatSpec.SHORTCUT_FILE_EXTENSION
+ FormatSpec.SHORTCUT_CONTENT_ID);
} else {
throw new UnsupportedFormatException("Unsupported kind of file : " + fileType);
}
}
@Override
public void openDictBuffer() throws FileNotFoundException, IOException,
UnsupportedFormatException {
if (!mDictDirectory.isDirectory()) {
throw new UnsupportedFormatException("Format 4 dictionary needs a directory");
}
mHeaderBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_HEADER));
mDictBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_TRIE));
mFrequencyBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_FREQUENCY));
mTerminalAddressTableBuffer = mBufferFactory.getDictionaryBuffer(
getFile(FILETYPE_TERMINAL_ADDRESS_TABLE));
mBigramReader = new BigramContentReader(mDictDirectory.getName(),
mDictDirectory, mBufferFactory);
mBigramReader.openBuffers();
mShortcutReader = new ShortcutContentReader(mDictDirectory.getName(), mDictDirectory,
mBufferFactory);
mShortcutReader.openBuffers();
}
@Override
public boolean isDictBufferOpen() {
return mDictBuffer != null;
}
@UsedForTesting
/* package */ DictBuffer getHeaderBuffer() {
return mHeaderBuffer;
}
@UsedForTesting
/* package */ DictBuffer getDictBuffer() {
return mDictBuffer;
mBinaryDictionary = new BinaryDictionary(dictDirectory.getAbsolutePath(),
0 /* offset */, 0 /* length */, true /* useFullEditDistance */, null /* locale */,
"" /* dictType */, true /* isUpdatable */);
}
@Override
public DictionaryHeader readHeader() throws IOException, UnsupportedFormatException {
if (mHeaderBuffer == null) {
openDictBuffer();
}
mHeaderBuffer.position(0);
final DictionaryHeader header = super.readHeader(mHeaderBuffer);
final int version = header.mFormatOptions.mVersion;
if (version != FormatSpec.VERSION4) {
throw new UnsupportedFormatException("File header has a wrong version : " + version);
}
return header;
}
/**
* An auxiliary class for reading bigrams.
*/
protected static class BigramContentReader extends SparseTableContentReader {
public BigramContentReader(final String name, final File baseDir,
final DictionaryBufferFactory factory) {
super(name + FormatSpec.BIGRAM_FILE_EXTENSION,
FormatSpec.BIGRAM_ADDRESS_TABLE_BLOCK_SIZE, baseDir,
getContentFilenames(name), getContentIds(), factory);
}
// TODO: Consolidate this method and BigramContentWriter.getContentFilenames.
protected static String[] getContentFilenames(final String name) {
return new String[] { name + FormatSpec.BIGRAM_FILE_EXTENSION };
}
// TODO: Consolidate this method and BigramContentWriter.getContentIds.
protected static String[] getContentIds() {
return new String[] { FormatSpec.BIGRAM_FREQ_CONTENT_ID };
}
public ArrayList<PendingAttribute> readTargetsAndFrequencies(final int terminalId,
final DictBuffer terminalAddressTableBuffer, final FormatOptions options) {
final ArrayList<PendingAttribute> bigrams = CollectionUtils.newArrayList();
read(FormatSpec.BIGRAM_FREQ_CONTENT_INDEX, terminalId,
new SparseTableContentReaderInterface() {
@Override
public void read(final DictBuffer buffer) {
while (bigrams.size() < FormatSpec.MAX_BIGRAMS_IN_A_PTNODE) {
// If bigrams.size() reaches FormatSpec.MAX_BIGRAMS_IN_A_PTNODE,
// remaining bigram entries are ignored.
final int bigramFlags = buffer.readUnsignedByte();
final int probability;
if (options.mHasTimestamp) {
probability = buffer.readUnsignedByte();
// Skip timestamp
buffer.readInt();
// Skip level
buffer.readUnsignedByte();
// Skip count
buffer.readUnsignedByte();
} else {
probability = bigramFlags
& FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY;
}
final int targetTerminalId = buffer.readUnsignedInt24();
terminalAddressTableBuffer.position(targetTerminalId
* FormatSpec.TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE);
final int targetAddress =
terminalAddressTableBuffer.readUnsignedInt24();
bigrams.add(new PendingAttribute(probability, targetAddress));
if (0 == (bigramFlags
& FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_HAS_NEXT)) {
break;
}
}
if (bigrams.size() >= FormatSpec.MAX_BIGRAMS_IN_A_PTNODE) {
throw new RuntimeException("Too many bigrams in a PtNode ("
+ bigrams.size() + " but max is "
+ FormatSpec.MAX_BIGRAMS_IN_A_PTNODE + ")");
}
}
});
if (bigrams.isEmpty()) return null;
return bigrams;
}
}
/**
* An auxiliary class for reading shortcuts.
*/
protected static class ShortcutContentReader extends SparseTableContentReader {
public ShortcutContentReader(final String name, final File baseDir,
final DictionaryBufferFactory factory) {
super(name + FormatSpec.SHORTCUT_FILE_EXTENSION,
FormatSpec.SHORTCUT_ADDRESS_TABLE_BLOCK_SIZE, baseDir,
new String[] { name + FormatSpec.SHORTCUT_FILE_EXTENSION },
new String[] { FormatSpec.SHORTCUT_CONTENT_ID }, factory);
}
public ArrayList<WeightedString> readShortcuts(final int terminalId) {
final ArrayList<WeightedString> shortcuts = CollectionUtils.newArrayList();
read(FormatSpec.SHORTCUT_CONTENT_INDEX, terminalId,
new SparseTableContentReaderInterface() {
@Override
public void read(final DictBuffer buffer) {
while (true) {
final int flags = buffer.readUnsignedByte();
final String word = CharEncoding.readString(buffer);
shortcuts.add(new WeightedString(word,
flags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY));
if (0 == (flags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_HAS_NEXT)) {
break;
}
}
}
});
if (shortcuts.isEmpty()) return null;
return shortcuts;
}
}
protected static class PtNodeReader extends AbstractDictDecoder.PtNodeReader {
protected static int readFrequency(final DictBuffer frequencyBuffer, final int terminalId,
final FormatOptions formatOptions) {
final int readingPos;
if (formatOptions.mHasTimestamp) {
final int entrySize = FormatSpec.FREQUENCY_AND_FLAGS_SIZE
+ FormatSpec.UNIGRAM_TIMESTAMP_SIZE + FormatSpec.UNIGRAM_LEVEL_SIZE
+ FormatSpec.UNIGRAM_COUNTER_SIZE;
readingPos = terminalId * entrySize + FormatSpec.FLAGS_IN_FREQ_FILE_SIZE;
} else {
readingPos = terminalId * FormatSpec.FREQUENCY_AND_FLAGS_SIZE
+ FormatSpec.FLAGS_IN_FREQ_FILE_SIZE;
}
frequencyBuffer.position(readingPos);
return frequencyBuffer.readUnsignedByte();
}
protected static int readTerminalId(final DictBuffer dictBuffer) {
return dictBuffer.readInt();
}
}
private final int[] mCharacterBufferForReadingVer4PtNodeInfo
= new int[FormatSpec.MAX_WORD_LENGTH];
/**
* Reads PtNode from ptNodePos in the trie file and returns Ver4PtNodeInfo.
*
* @param ptNodePos the position of PtNode.
* @param options the format options.
* @return Ver4PtNodeInfo.
*/
// TODO: Make this buffer thread safe.
// TODO: Support words longer than FormatSpec.MAX_WORD_LENGTH.
protected Ver4PtNodeInfo readVer4PtNodeInfo(final int ptNodePos, final FormatOptions options) {
int readingPos = ptNodePos;
final int flags = PtNodeReader.readPtNodeOptionFlags(mDictBuffer);
readingPos += FormatSpec.PTNODE_FLAGS_SIZE;
final int parentPos = PtNodeReader.readParentAddress(mDictBuffer, options);
if (BinaryDictIOUtils.supportsDynamicUpdate(options)) {
readingPos += FormatSpec.PARENT_ADDRESS_SIZE;
}
final int characters[];
if (0 != (flags & FormatSpec.FLAG_HAS_MULTIPLE_CHARS)) {
int index = 0;
int character = CharEncoding.readChar(mDictBuffer);
readingPos += CharEncoding.getCharSize(character);
while (FormatSpec.INVALID_CHARACTER != character
&& index < FormatSpec.MAX_WORD_LENGTH) {
mCharacterBufferForReadingVer4PtNodeInfo[index++] = character;
character = CharEncoding.readChar(mDictBuffer);
readingPos += CharEncoding.getCharSize(character);
}
characters = Arrays.copyOfRange(mCharacterBufferForReadingVer4PtNodeInfo, 0, index);
} else {
final int character = CharEncoding.readChar(mDictBuffer);
readingPos += CharEncoding.getCharSize(character);
characters = new int[] { character };
}
final int terminalId;
if (0 != (FormatSpec.FLAG_IS_TERMINAL & flags)) {
terminalId = PtNodeReader.readTerminalId(mDictBuffer);
readingPos += FormatSpec.PTNODE_TERMINAL_ID_SIZE;
} else {
terminalId = PtNode.NOT_A_TERMINAL;
}
int childrenPos = PtNodeReader.readChildrenAddress(mDictBuffer, flags, options);
if (childrenPos != FormatSpec.NO_CHILDREN_ADDRESS) {
childrenPos += readingPos;
}
readingPos += BinaryDictIOUtils.getChildrenAddressSize(flags, options);
return new Ver4PtNodeInfo(flags, characters, terminalId, childrenPos, parentPos,
readingPos - ptNodePos);
}
@Override
public PtNodeInfo readPtNode(final int ptNodePos, final FormatOptions options) {
final Ver4PtNodeInfo nodeInfo = readVer4PtNodeInfo(ptNodePos, options);
final int frequency;
if (0 != (FormatSpec.FLAG_IS_TERMINAL & nodeInfo.mFlags)) {
frequency = PtNodeReader.readFrequency(mFrequencyBuffer, nodeInfo.mTerminalId, options);
} else {
frequency = PtNode.NOT_A_TERMINAL;
}
final ArrayList<WeightedString> shortcutTargets = mShortcutReader.readShortcuts(
nodeInfo.mTerminalId);
final ArrayList<PendingAttribute> bigrams = mBigramReader.readTargetsAndFrequencies(
nodeInfo.mTerminalId, mTerminalAddressTableBuffer, options);
return new PtNodeInfo(ptNodePos, ptNodePos + nodeInfo.mNodeSize, nodeInfo.mFlags,
nodeInfo.mCharacters, frequency, nodeInfo.mParentPos, nodeInfo.mChildrenPos,
shortcutTargets, bigrams);
}
private void deleteDictFiles() {
final File[] files = mDictDirectory.listFiles();
for (int i = 0; i < files.length; ++i) {
files[i].delete();
}
return mBinaryDictionary.getHeader();
}
@Override
public FusionDictionary readDictionaryBinary(final FusionDictionary dict,
final boolean deleteDictIfBroken)
throws FileNotFoundException, IOException, UnsupportedFormatException {
if (mDictBuffer == null) {
openDictBuffer();
}
try {
return BinaryDictDecoderUtils.readDictionaryBinary(this, dict);
} catch (IOException e) {
Log.e(TAG, "The dictionary " + mDictDirectory.getName() + " is broken.", e);
if (deleteDictIfBroken) {
deleteDictFiles();
final DictionaryHeader header = readHeader();
final FusionDictionary fusionDict = dict != null ? dict :
new FusionDictionary(new FusionDictionary.PtNodeArray(), header.mDictionaryOptions);
int token = 0;
final ArrayList<WordProperty> wordProperties = CollectionUtils.newArrayList();
do {
final BinaryDictionary.GetNextWordPropertyResult result =
mBinaryDictionary.getNextWordProperty(token);
final WordProperty wordProperty = result.mWordProperty;
if (wordProperty == null) {
if (deleteDictIfBroken) {
mBinaryDictionary.close();
FileUtils.deleteRecursively(mDictDirectory);
}
return null;
}
throw e;
} catch (UnsupportedFormatException e) {
Log.e(TAG, "The dictionary " + mDictDirectory.getName() + " is broken.", e);
if (deleteDictIfBroken) {
deleteDictFiles();
}
throw e;
}
}
wordProperties.add(wordProperty);
token = result.mNextToken;
} while (token != 0);
@Override
public void setPosition(int newPos) {
mDictBuffer.position(newPos);
}
@Override
public int getPosition() {
return mDictBuffer.position();
}
@Override
public int readPtNodeCount() {
return BinaryDictDecoderUtils.readPtNodeCount(mDictBuffer);
}
@Override
public boolean readAndFollowForwardLink() {
final int forwardLinkPos = mDictBuffer.position();
int nextRelativePos = BinaryDictDecoderUtils.readSInt24(mDictBuffer);
if (nextRelativePos != FormatSpec.NO_FORWARD_LINK_ADDRESS) {
final int nextPos = forwardLinkPos + nextRelativePos;
if (nextPos >= 0 && nextPos < mDictBuffer.limit()) {
mDictBuffer.position(nextPos);
return true;
// Insert unigrams to the fusion dictionary.
for (final WordProperty wordProperty : wordProperties) {
// TODO: Support probability that is -1.
final int probability = wordProperty.getProbability() < 0 ?
0 : wordProperty.getProbability();
if (wordProperty.mIsBlacklistEntry) {
fusionDict.addBlacklistEntry(wordProperty.mWord, wordProperty.mShortcutTargets,
wordProperty.mIsNotAWord);
} else {
fusionDict.add(wordProperty.mWord, probability,
wordProperty.mShortcutTargets, wordProperty.mIsNotAWord);
}
}
return false;
}
@Override
public boolean hasNextPtNodeArray() {
return mDictBuffer.position() != FormatSpec.NO_FORWARD_LINK_ADDRESS;
}
@Override
@UsedForTesting
public void skipPtNode(final FormatOptions formatOptions) {
final int flags = PtNodeReader.readPtNodeOptionFlags(mDictBuffer);
PtNodeReader.readParentAddress(mDictBuffer, formatOptions);
BinaryDictIOUtils.skipString(mDictBuffer,
(flags & FormatSpec.FLAG_HAS_MULTIPLE_CHARS) != 0);
if ((flags & FormatSpec.FLAG_IS_TERMINAL) != 0) PtNodeReader.readTerminalId(mDictBuffer);
PtNodeReader.readChildrenAddress(mDictBuffer, flags, formatOptions);
// Insert bigrams to the fusion dictionary.
for (final WordProperty wordProperty : wordProperties) {
if (wordProperty.mBigrams == null) {
continue;
}
final String word0 = wordProperty.mWord;
for (final WeightedString bigram : wordProperty.mBigrams) {
fusionDict.setBigram(word0, bigram.mWord, bigram.getProbability());
}
}
return fusionDict;
}
}

17
tests/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderEncoderTests.java
View file

@ -468,10 +468,6 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
runReadUnigramsAndBigramsTests(results, BinaryDictUtils.USE_BYTE_BUFFER,
BinaryDictUtils.VERSION2_OPTIONS);
runReadUnigramsAndBigramsTests(results, BinaryDictUtils.USE_BYTE_BUFFER,
BinaryDictUtils.VERSION4_OPTIONS_WITHOUT_TIMESTAMP);
runReadUnigramsAndBigramsTests(results, BinaryDictUtils.USE_BYTE_BUFFER,
BinaryDictUtils.VERSION4_OPTIONS_WITH_TIMESTAMP);
for (final String result : results) {
Log.d(TAG, result);
@ -483,10 +479,6 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
runReadUnigramsAndBigramsTests(results, BinaryDictUtils.USE_BYTE_ARRAY,
BinaryDictUtils.VERSION2_OPTIONS);
runReadUnigramsAndBigramsTests(results, BinaryDictUtils.USE_BYTE_ARRAY,
BinaryDictUtils.VERSION4_OPTIONS_WITHOUT_TIMESTAMP);
runReadUnigramsAndBigramsTests(results, BinaryDictUtils.USE_BYTE_ARRAY,
BinaryDictUtils.VERSION4_OPTIONS_WITH_TIMESTAMP);
for (final String result : results) {
Log.d(TAG, result);
@ -597,17 +589,8 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
runGetTerminalPositionTests(BinaryDictUtils.USE_BYTE_ARRAY,
BinaryDictUtils.VERSION2_OPTIONS);
runGetTerminalPositionTests(BinaryDictUtils.USE_BYTE_ARRAY,
BinaryDictUtils.VERSION4_OPTIONS_WITHOUT_TIMESTAMP);
runGetTerminalPositionTests(BinaryDictUtils.USE_BYTE_ARRAY,
BinaryDictUtils.VERSION4_OPTIONS_WITH_TIMESTAMP);
runGetTerminalPositionTests(BinaryDictUtils.USE_BYTE_BUFFER,
BinaryDictUtils.VERSION2_OPTIONS);
runGetTerminalPositionTests(BinaryDictUtils.USE_BYTE_BUFFER,
BinaryDictUtils.VERSION4_OPTIONS_WITHOUT_TIMESTAMP);
runGetTerminalPositionTests(BinaryDictUtils.USE_BYTE_BUFFER,
BinaryDictUtils.VERSION4_OPTIONS_WITH_TIMESTAMP);
for (final String result : results) {
Log.d(TAG, result);

195
tests/src/com/android/inputmethod/latin/makedict/SparseTableTests.java
View file

@ -1,195 +0,0 @@
/*
* 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 android.test.AndroidTestCase;
import android.test.suitebuilder.annotation.LargeTest;
import android.util.Log;
import java.io.File;
import java.io.FileOutputStream;
import java.io.IOException;
import java.io.OutputStream;
import java.util.ArrayList;
import java.util.Random;
/**
* Unit tests for SparseTable.
*/
@LargeTest
public class SparseTableTests extends AndroidTestCase {
private static final String TAG = SparseTableTests.class.getSimpleName();
private final Random mRandom;
private final ArrayList<Integer> mRandomIndex;
private static final int DEFAULT_SIZE = 10000;
private static final int BLOCK_SIZE = 8;
public SparseTableTests() {
this(System.currentTimeMillis(), DEFAULT_SIZE);
}
public SparseTableTests(final long seed, final int tableSize) {
super();
Log.d(TAG, "Seed for test is " + seed + ", size is " + tableSize);
mRandom = new Random(seed);
mRandomIndex = new ArrayList<Integer>(tableSize);
for (int i = 0; i < tableSize; ++i) {
mRandomIndex.add(SparseTable.NOT_EXIST);
}
}
public void testSet() {
final SparseTable table = new SparseTable(16, BLOCK_SIZE, 1);
table.set(0, 3, 6);
table.set(0, 8, 16);
for (int i = 0; i < 16; ++i) {
if (i == 3 || i == 8) {
assertEquals(i * 2, table.get(0, i));
} else {
assertEquals(SparseTable.NOT_EXIST, table.get(0, i));
}
}
}
private void generateRandomIndex(final int size, final int prop) {
for (int i = 0; i < size; ++i) {
if (mRandom.nextInt(100) < prop) {
mRandomIndex.set(i, mRandom.nextInt());
} else {
mRandomIndex.set(i, SparseTable.NOT_EXIST);
}
}
}
private void runTestRandomSet() {
final SparseTable table = new SparseTable(DEFAULT_SIZE, BLOCK_SIZE, 1);
int elementCount = 0;
for (int i = 0; i < DEFAULT_SIZE; ++i) {
if (mRandomIndex.get(i) != SparseTable.NOT_EXIST) {
table.set(0, i, mRandomIndex.get(i));
elementCount++;
}
}
Log.d(TAG, "table size = " + table.getLookupTableSize() + " + "
+ table.getContentTableSize());
Log.d(TAG, "the table has " + elementCount + " elements");
for (int i = 0; i < DEFAULT_SIZE; ++i) {
assertEquals(table.get(0, i), (int)mRandomIndex.get(i));
}
// flush and reload
OutputStream lookupOutStream = null;
OutputStream contentOutStream = null;
try {
final File lookupIndexFile = File.createTempFile("testRandomSet", ".small");
final File contentFile = File.createTempFile("testRandomSet", ".big");
lookupOutStream = new FileOutputStream(lookupIndexFile);
contentOutStream = new FileOutputStream(contentFile);
table.write(lookupOutStream, new OutputStream[] { contentOutStream });
lookupOutStream.flush();
contentOutStream.flush();
final SparseTable newTable = SparseTable.readFromFiles(lookupIndexFile,
new File[] { contentFile }, BLOCK_SIZE);
for (int i = 0; i < DEFAULT_SIZE; ++i) {
assertEquals(table.get(0, i), newTable.get(0, i));
}
} catch (IOException e) {
Log.d(TAG, "IOException while flushing and realoding", e);
} finally {
if (lookupOutStream != null) {
try {
lookupOutStream.close();
} catch (IOException e) {
Log.d(TAG, "IOException while closing the stream", e);
}
}
if (contentOutStream != null) {
try {
contentOutStream.close();
} catch (IOException e) {
Log.d(TAG, "IOException while closing contentStream.", e);
}
}
}
}
public void testRandomSet() {
for (int i = 0; i <= 100; i += 10) {
generateRandomIndex(DEFAULT_SIZE, i);
runTestRandomSet();
}
}
public void testMultipleContents() {
final int numOfContents = 5;
generateRandomIndex(DEFAULT_SIZE, 20);
final SparseTable table = new SparseTable(DEFAULT_SIZE, BLOCK_SIZE, numOfContents);
for (int i = 0; i < mRandomIndex.size(); ++i) {
if (mRandomIndex.get(i) != SparseTable.NOT_EXIST) {
for (int j = 0; j < numOfContents; ++j) {
table.set(j, i, mRandomIndex.get(i));
}
}
}
OutputStream lookupOutStream = null;
OutputStream[] contentsOutStream = new OutputStream[numOfContents];
try {
final File lookupIndexFile = File.createTempFile("testMultipleContents", "small");
lookupOutStream = new FileOutputStream(lookupIndexFile);
final File[] contentFiles = new File[numOfContents];
for (int i = 0; i < numOfContents; ++i) {
contentFiles[i] = File.createTempFile("testMultipleContents", "big" + i);
contentsOutStream[i] = new FileOutputStream(contentFiles[i]);
}
table.write(lookupOutStream, contentsOutStream);
lookupOutStream.flush();
for (int i = 0; i < numOfContents; ++i) {
contentsOutStream[i].flush();
}
final SparseTable newTable = SparseTable.readFromFiles(lookupIndexFile, contentFiles,
BLOCK_SIZE);
for (int i = 0; i < numOfContents; ++i) {
for (int j = 0; j < DEFAULT_SIZE; ++j) {
assertEquals(table.get(i, j), newTable.get(i, j));
}
}
} catch (IOException e) {
Log.d(TAG, "IOException while flushing and reloading", e);
} finally {
if (lookupOutStream != null) {
try {
lookupOutStream.close();
} catch (IOException e) {
Log.d(TAG, "IOException while closing the stream", e);
}
}
for (int i = 0; i < numOfContents; ++i) {
if (contentsOutStream[i] != null) {
try {
contentsOutStream[i].close();
} catch (IOException e) {
Log.d(TAG, "IOException while closing the stream.", e);
}
}
}
}
}
}