am af5511c8: am 25f47828: Merge "Separate bigram entries from the trie file."

* commit 'af5511c8309f6a5cfd12b8d4631b59ea0c43f66e':
  Separate bigram entries from the trie file.
This commit is contained in:
Ken Wakasa 2013-10-03 12:17:40 -07:00 committed by Android Git Automerger
commit 9019d51c0c
5 changed files with 142 additions and 30 deletions

View file

@ -385,6 +385,7 @@ public class BinaryDictEncoderUtils {
nodeSize + size, ptNode.mChildren));
}
nodeSize += getShortcutListSize(ptNode.mShortcutTargets);
if (formatOptions.mVersion < FormatSpec.FIRST_VERSION_WITH_TERMINAL_ID) {
if (null != ptNode.mBigrams) {
for (WeightedString bigram : ptNode.mBigrams) {
final int offset = getOffsetToTargetPtNodeDuringUpdate(ptNodeArray,
@ -393,6 +394,7 @@ public class BinaryDictEncoderUtils {
nodeSize += getByteSize(offset) + FormatSpec.PTNODE_ATTRIBUTE_FLAGS_SIZE;
}
}
}
ptNode.mCachedSize = nodeSize;
size += nodeSize;
}

View file

@ -265,8 +265,12 @@ public final class FormatSpec {
static final String FREQ_FILE_EXTENSION = ".freq";
// tat = Terminal Address Table
static final String TERMINAL_ADDRESS_TABLE_FILE_EXTENSION = ".tat";
static final String BIGRAM_FILE_EXTENSION = ".bigram";
static final String BIGRAM_LOOKUP_TABLE_FILE_EXTENSION = ".bigram_lookup";
static final String BIGRAM_ADDRESS_TABLE_FILE_EXTENSION = ".bigram_index";
static final int FREQUENCY_AND_FLAGS_SIZE = 2;
static final int TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE = 3;
static final int BIGRAM_ADDRESS_TABLE_BLOCK_SIZE = 4;
static final int NO_CHILDREN_ADDRESS = Integer.MIN_VALUE;
static final int NO_PARENT_ADDRESS = 0;

View file

@ -18,6 +18,9 @@ package com.android.inputmethod.latin.makedict;
import com.android.inputmethod.annotations.UsedForTesting;
import java.io.File;
import java.io.FileInputStream;
import java.io.FileOutputStream;
import java.io.IOException;
import java.io.OutputStream;
import java.util.ArrayList;
@ -147,4 +150,45 @@ public class SparseTable {
BinaryDictEncoderUtils.writeUIntToStream(contentOutStream, index, 4);
}
}
@UsedForTesting
public void writeToFiles(final File lookupTableFile, final File contentFile)
throws IOException {
FileOutputStream lookupTableOutStream = null;
FileOutputStream contentOutStream = null;
try {
lookupTableOutStream = new FileOutputStream(lookupTableFile);
contentOutStream = new FileOutputStream(contentFile);
write(lookupTableOutStream, contentOutStream);
} finally {
if (lookupTableOutStream != null) {
lookupTableOutStream.close();
}
if (contentOutStream != null) {
contentOutStream.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 contentFile,
final int blockSize) throws IOException {
final byte[] lookupTable = readFileToByteArray(lookupTableFile);
final byte[] content = readFileToByteArray(contentFile);
return new SparseTable(lookupTable, content, blockSize);
}
}

View file

@ -42,12 +42,15 @@ public class Ver4DictDecoder extends DictDecoder {
private static final int FILETYPE_TRIE = 1;
private static final int FILETYPE_FREQUENCY = 2;
private static final int FILETYPE_TERMINAL_ADDRESS_TABLE = 3;
private static final int FILETYPE_BIGRAM = 4;
private final File mDictDirectory;
private final DictionaryBufferFactory mBufferFactory;
private DictBuffer mDictBuffer;
private DictBuffer mFrequencyBuffer;
private DictBuffer mTerminalAddressTableBuffer;
private DictBuffer mBigramBuffer;
private SparseTable mBigramAddressTable;
@UsedForTesting
/* package */ Ver4DictDecoder(final File dictDirectory, final int factoryFlag) {
@ -82,6 +85,9 @@ public class Ver4DictDecoder extends DictDecoder {
} else if (fileType == FILETYPE_TERMINAL_ADDRESS_TABLE) {
return new File(mDictDirectory,
mDictDirectory.getName() + FormatSpec.TERMINAL_ADDRESS_TABLE_FILE_EXTENSION);
} else if (fileType == FILETYPE_BIGRAM) {
return new File(mDictDirectory,
mDictDirectory.getName() + FormatSpec.BIGRAM_FILE_EXTENSION);
} else {
throw new RuntimeException("Unsupported kind of file : " + fileType);
}
@ -94,6 +100,8 @@ public class Ver4DictDecoder extends DictDecoder {
mFrequencyBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_FREQUENCY));
mTerminalAddressTableBuffer = mBufferFactory.getDictionaryBuffer(
getFile(FILETYPE_TERMINAL_ADDRESS_TABLE));
mBigramBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_BIGRAM));
loadBigramAddressSparseTable();
}
@Override
@ -118,6 +126,15 @@ public class Ver4DictDecoder extends DictDecoder {
return header;
}
private void loadBigramAddressSparseTable() throws IOException {
final File lookupIndexFile = new File(mDictDirectory,
mDictDirectory.getName() + FormatSpec.BIGRAM_LOOKUP_TABLE_FILE_EXTENSION);
final File contentFile = new File(mDictDirectory,
mDictDirectory.getName() + FormatSpec.BIGRAM_ADDRESS_TABLE_FILE_EXTENSION);
mBigramAddressTable = SparseTable.readFromFiles(lookupIndexFile, contentFile,
FormatSpec.BIGRAM_ADDRESS_TABLE_BLOCK_SIZE);
}
protected static class PtNodeReader extends DictDecoder.PtNodeReader {
protected static int readFrequency(final DictBuffer frequencyBuffer, final int terminalId) {
frequencyBuffer.position(terminalId * FormatSpec.FREQUENCY_AND_FLAGS_SIZE + 1);
@ -191,8 +208,21 @@ public class Ver4DictDecoder extends DictDecoder {
final ArrayList<PendingAttribute> bigrams;
if (0 != (flags & FormatSpec.FLAG_HAS_BIGRAMS)) {
bigrams = new ArrayList<PendingAttribute>();
addressPointer += PtNodeReader.readBigramAddresses(mDictBuffer, bigrams,
addressPointer);
final int posOfBigrams = mBigramAddressTable.get(terminalId);
mBigramBuffer.position(posOfBigrams);
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 = mBigramBuffer.readUnsignedByte();
final int targetTerminalId = mBigramBuffer.readUnsignedInt24();
mTerminalAddressTableBuffer.position(
targetTerminalId * FormatSpec.TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE);
final int targetAddress = mTerminalAddressTableBuffer.readUnsignedInt24();
bigrams.add(new PendingAttribute(
bigramFlags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY,
targetAddress));
if (0 == (bigramFlags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_HAS_NEXT)) break;
}
if (bigrams.size() >= FormatSpec.MAX_BIGRAMS_IN_A_PTNODE) {
MakedictLog.d("too many bigrams in a node.");
}

View file

@ -26,6 +26,7 @@ import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode;
import com.android.inputmethod.latin.makedict.FusionDictionary.PtNodeArray;
import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString;
import java.io.ByteArrayOutputStream;
import java.io.File;
import java.io.FileNotFoundException;
import java.io.FileOutputStream;
@ -43,9 +44,13 @@ public class Ver4DictEncoder implements DictEncoder {
private byte[] mTrieBuf;
private int mTriePos;
private int mHeaderSize;
private SparseTable mBigramAddressTable;
private OutputStream mTrieOutStream;
private OutputStream mFreqOutStream;
private OutputStream mTerminalAddressTableOutStream;
private OutputStream mBigramOutStream;
private File mDictDir;
private String mBaseFilename;
@UsedForTesting
public Ver4DictEncoder(final File dictPlacedDir) {
@ -55,12 +60,14 @@ public class Ver4DictEncoder implements DictEncoder {
private void openStreams(final FormatOptions formatOptions, final DictionaryOptions dictOptions)
throws FileNotFoundException, IOException {
final FileHeader header = new FileHeader(0, dictOptions, formatOptions);
final String filename = header.getId() + "." + header.getVersion();
final File mDictDir = new File(mDictPlacedDir, filename);
final File trieFile = new File(mDictDir, filename + FormatSpec.TRIE_FILE_EXTENSION);
final File freqFile = new File(mDictDir, filename + FormatSpec.FREQ_FILE_EXTENSION);
mBaseFilename = header.getId() + "." + header.getVersion();
mDictDir = new File(mDictPlacedDir, mBaseFilename);
final File trieFile = new File(mDictDir, mBaseFilename + FormatSpec.TRIE_FILE_EXTENSION);
final File freqFile = new File(mDictDir, mBaseFilename + FormatSpec.FREQ_FILE_EXTENSION);
final File terminalAddressTableFile = new File(mDictDir,
filename + FormatSpec.TERMINAL_ADDRESS_TABLE_FILE_EXTENSION);
mBaseFilename + FormatSpec.TERMINAL_ADDRESS_TABLE_FILE_EXTENSION);
final File bigramFile = new File(mDictDir,
mBaseFilename + FormatSpec.BIGRAM_FILE_EXTENSION);
if (!mDictDir.isDirectory()) {
if (mDictDir.exists()) mDictDir.delete();
mDictDir.mkdirs();
@ -71,6 +78,7 @@ public class Ver4DictEncoder implements DictEncoder {
mTrieOutStream = new FileOutputStream(trieFile);
mFreqOutStream = new FileOutputStream(freqFile);
mTerminalAddressTableOutStream = new FileOutputStream(terminalAddressTableFile);
mBigramOutStream = new FileOutputStream(bigramFile);
}
private void close() throws IOException {
@ -84,10 +92,14 @@ public class Ver4DictEncoder implements DictEncoder {
if (mTerminalAddressTableOutStream != null) {
mTerminalAddressTableOutStream.close();
}
if (mBigramOutStream != null) {
mBigramOutStream.close();
}
} finally {
mTrieOutStream = null;
mFreqOutStream = null;
mTerminalAddressTableOutStream = null;
mBigramOutStream = null;
}
}
@ -123,6 +135,10 @@ public class Ver4DictEncoder implements DictEncoder {
if (MakedictLog.DBG) BinaryDictEncoderUtils.checkFlatPtNodeArrayList(flatNodes);
writeTerminalData(flatNodes, terminalCount);
mBigramAddressTable = new SparseTable(terminalCount,
FormatSpec.BIGRAM_ADDRESS_TABLE_BLOCK_SIZE);
writeBigrams(flatNodes, dict);
writeBigramAddressSparseTable();
final PtNodeArray lastNodeArray = flatNodes.get(flatNodes.size() - 1);
final int bufferSize = lastNodeArray.mCachedAddressAfterUpdate + lastNodeArray.mCachedSize;
@ -230,25 +246,42 @@ public class Ver4DictEncoder implements DictEncoder {
shortcutByteSize, FormatSpec.PTNODE_SHORTCUT_LIST_SIZE_SIZE);
}
private void writeBigrams(ArrayList<WeightedString> bigrams, FusionDictionary dict) {
if (bigrams == null) return;
private void writeBigrams(final ArrayList<PtNodeArray> flatNodes, final FusionDictionary dict)
throws IOException {
final ByteArrayOutputStream bigramBuffer = new ByteArrayOutputStream();
final Iterator<WeightedString> bigramIterator = bigrams.iterator();
for (final PtNodeArray nodeArray : flatNodes) {
for (final PtNode ptNode : nodeArray.mData) {
if (ptNode.mBigrams != null) {
final int startPos = bigramBuffer.size();
mBigramAddressTable.set(ptNode.mTerminalId, startPos);
final Iterator<WeightedString> bigramIterator = ptNode.mBigrams.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.mFrequency;
final int offset = addressOfBigram
- (mTriePos + FormatSpec.PTNODE_ATTRIBUTE_FLAGS_SIZE);
int bigramFlags = BinaryDictEncoderUtils.makeBigramFlags(bigramIterator.hasNext(),
offset, bigram.mFrequency, unigramFrequencyForThisWord, bigram.mWord);
mTrieBuf[mTriePos++] = (byte) bigramFlags;
mTriePos += BinaryDictEncoderUtils.writeChildrenPosition(mTrieBuf,
mTriePos, Math.abs(offset));
final int bigramFlags = BinaryDictEncoderUtils.makeBigramFlags(
bigramIterator.hasNext(), 0, bigram.mFrequency,
unigramFrequencyForThisWord, bigram.mWord);
BinaryDictEncoderUtils.writeUIntToStream(bigramBuffer, bigramFlags,
FormatSpec.PTNODE_ATTRIBUTE_FLAGS_SIZE);
BinaryDictEncoderUtils.writeUIntToStream(bigramBuffer, target.mTerminalId,
FormatSpec.PTNODE_ATTRIBUTE_MAX_ADDRESS_SIZE);
}
}
}
}
bigramBuffer.writeTo(mBigramOutStream);
}
private void writeBigramAddressSparseTable() throws IOException {
final File lookupIndexFile =
new File(mDictDir, mBaseFilename + FormatSpec.BIGRAM_LOOKUP_TABLE_FILE_EXTENSION);
final File contentFile =
new File(mDictDir, mBaseFilename + FormatSpec.BIGRAM_ADDRESS_TABLE_FILE_EXTENSION);
mBigramAddressTable.writeToFiles(lookupIndexFile, contentFile);
}
@Override
public void writeForwardLinkAddress(int forwardLinkAddress) {
@ -267,7 +300,6 @@ public class Ver4DictEncoder implements DictEncoder {
}
writeChildrenPosition(ptNode, formatOptions);
writeShortcuts(ptNode.mShortcutTargets);
writeBigrams(ptNode.mBigrams, dict);
}
private void writeTerminalData(final ArrayList<PtNodeArray> flatNodes,