Make Ver4DictEncoder write an address table of terminal nodes.

Bug: 10920165
Change-Id: I86017456ea9fa5d6e12b57172c34f5ed4f88ef94
main
Yuichiro Hanada 2013-09-12 20:53:44 +09:00
parent 46fdc6fd23
commit 22c5c450fe
4 changed files with 49 additions and 12 deletions

View File

@ -882,8 +882,9 @@ public class BinaryDictEncoderUtils {
* @param destination the stream to write the file header to.
* @param dict the dictionary to write.
* @param formatOptions file format options.
* @return the size of the header.
*/
/* package */ static void writeDictionaryHeader(final OutputStream destination,
/* package */ static int writeDictionaryHeader(final OutputStream destination,
final FusionDictionary dict, final FormatOptions formatOptions)
throws IOException, UnsupportedFormatException {
final int version = formatOptions.mVersion;
@ -932,5 +933,6 @@ public class BinaryDictEncoderUtils {
destination.write(bytes);
headerBuffer.close();
return size;
}
}

View File

@ -263,7 +263,10 @@ public final class FormatSpec {
// These values are used only by version 4 or later.
static final String TRIE_FILE_EXTENSION = ".trie";
static final String FREQ_FILE_EXTENSION = ".freq";
// tat = Terminal Address Table
static final String TERMINAL_ADDRESS_TABLE_FILE_EXTENSION = ".tat";
static final int FREQUENCY_AND_FLAGS_SIZE = 2;
static final int TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE = 3;
static final int NO_CHILDREN_ADDRESS = Integer.MIN_VALUE;
static final int NO_PARENT_ADDRESS = 0;

View File

@ -41,11 +41,13 @@ 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 final File mDictDirectory;
private final DictionaryBufferFactory mBufferFactory;
private DictBuffer mDictBuffer;
private DictBuffer mFrequencyBuffer;
private DictBuffer mTerminalAddressTableBuffer;
@UsedForTesting
/* package */ Ver4DictDecoder(final File dictDirectory, final int factoryFlag) {
@ -77,6 +79,9 @@ public class Ver4DictDecoder extends DictDecoder {
} 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 {
throw new RuntimeException("Unsupported kind of file : " + fileType);
}
@ -87,6 +92,8 @@ public class Ver4DictDecoder extends DictDecoder {
final String filename = mDictDirectory.getName();
mDictBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_TRIE));
mFrequencyBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_FREQUENCY));
mTerminalAddressTableBuffer = mBufferFactory.getDictionaryBuffer(
getFile(FILETYPE_TERMINAL_ADDRESS_TABLE));
}
@Override

View File

@ -41,10 +41,11 @@ import java.util.Iterator;
public class Ver4DictEncoder implements DictEncoder {
private final File mDictPlacedDir;
private byte[] mTrieBuf;
private byte[] mFreqBuf;
private int mTriePos;
private int mHeaderSize;
private OutputStream mTrieOutStream;
private OutputStream mFreqOutStream;
private OutputStream mTerminalAddressTableOutStream;
@UsedForTesting
public Ver4DictEncoder(final File dictPlacedDir) {
@ -58,14 +59,18 @@ public class Ver4DictEncoder implements DictEncoder {
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);
final File terminalAddressTableFile = new File(mDictDir,
filename + FormatSpec.TERMINAL_ADDRESS_TABLE_FILE_EXTENSION);
if (!mDictDir.isDirectory()) {
if (mDictDir.exists()) mDictDir.delete();
mDictDir.mkdirs();
}
if (!trieFile.exists()) trieFile.createNewFile();
if (!freqFile.exists()) freqFile.createNewFile();
if (!terminalAddressTableFile.exists()) terminalAddressTableFile.createNewFile();
mTrieOutStream = new FileOutputStream(trieFile);
mFreqOutStream = new FileOutputStream(freqFile);
mTerminalAddressTableOutStream = new FileOutputStream(terminalAddressTableFile);
}
private void close() throws IOException {
@ -76,9 +81,13 @@ public class Ver4DictEncoder implements DictEncoder {
if (mFreqOutStream != null) {
mFreqOutStream.close();
}
if (mTerminalAddressTableOutStream != null) {
mTerminalAddressTableOutStream.close();
}
} finally {
mTrieOutStream = null;
mFreqOutStream = null;
mTerminalAddressTableOutStream = null;
}
}
@ -97,7 +106,8 @@ public class Ver4DictEncoder implements DictEncoder {
openStreams(formatOptions, dict.mOptions);
}
BinaryDictEncoderUtils.writeDictionaryHeader(mTrieOutStream, dict, formatOptions);
mHeaderSize = BinaryDictEncoderUtils.writeDictionaryHeader(mTrieOutStream, dict,
formatOptions);
MakedictLog.i("Flattening the tree...");
ArrayList<PtNodeArray> flatNodes = BinaryDictEncoderUtils.flattenTree(dict.mRootNodeArray);
@ -112,10 +122,11 @@ public class Ver4DictEncoder implements DictEncoder {
BinaryDictEncoderUtils.computeAddresses(dict, flatNodes, formatOptions);
if (MakedictLog.DBG) BinaryDictEncoderUtils.checkFlatPtNodeArrayList(flatNodes);
writeTerminalData(flatNodes, terminalCount);
final PtNodeArray lastNodeArray = flatNodes.get(flatNodes.size() - 1);
final int bufferSize = lastNodeArray.mCachedAddressAfterUpdate + lastNodeArray.mCachedSize;
mTrieBuf = new byte[bufferSize];
mFreqBuf = new byte[terminalCount * FormatSpec.FREQUENCY_AND_FLAGS_SIZE];
MakedictLog.i("Writing file...");
for (PtNodeArray nodeArray : flatNodes) {
@ -126,7 +137,6 @@ public class Ver4DictEncoder implements DictEncoder {
MakedictLog.i("has " + terminalCount + " terminals.");
}
mTrieOutStream.write(mTrieBuf);
mFreqOutStream.write(mFreqBuf);
MakedictLog.i("Done");
close();
@ -185,12 +195,6 @@ public class Ver4DictEncoder implements DictEncoder {
FormatSpec.PTNODE_TERMINAL_ID_SIZE);
}
private void writeFrequency(final int frequency, final int terminalId) {
final int freqPos = terminalId * FormatSpec.FREQUENCY_AND_FLAGS_SIZE;
BinaryDictEncoderUtils.writeUIntToBuffer(mFreqBuf, freqPos, frequency,
FormatSpec.FREQUENCY_AND_FLAGS_SIZE);
}
private void writeChildrenPosition(PtNode ptNode, FormatOptions formatOptions) {
final int childrenPos = BinaryDictEncoderUtils.getChildrenPosition(ptNode, formatOptions);
if (formatOptions.mSupportsDynamicUpdate) {
@ -260,10 +264,31 @@ public class Ver4DictEncoder implements DictEncoder {
writeCharacters(ptNode.mChars, ptNode.hasSeveralChars());
if (ptNode.isTerminal()) {
writeTerminalId(ptNode.mTerminalId);
writeFrequency(ptNode.mFrequency, ptNode.mTerminalId);
}
writeChildrenPosition(ptNode, formatOptions);
writeShortcuts(ptNode.mShortcutTargets);
writeBigrams(ptNode.mBigrams, dict);
}
private void writeTerminalData(final ArrayList<PtNodeArray> flatNodes,
final int terminalCount) throws IOException {
final byte[] freqBuf = new byte[terminalCount * FormatSpec.FREQUENCY_AND_FLAGS_SIZE];
final byte[] terminalAddressTableBuf =
new byte[terminalCount * FormatSpec.TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE];
for (final PtNodeArray nodeArray : flatNodes) {
for (final PtNode ptNode : nodeArray.mData) {
if (ptNode.isTerminal()) {
BinaryDictEncoderUtils.writeUIntToBuffer(freqBuf,
ptNode.mTerminalId * FormatSpec.FREQUENCY_AND_FLAGS_SIZE,
ptNode.mFrequency, FormatSpec.FREQUENCY_AND_FLAGS_SIZE);
BinaryDictEncoderUtils.writeUIntToBuffer(terminalAddressTableBuf,
ptNode.mTerminalId * FormatSpec.TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE,
ptNode.mCachedAddressAfterUpdate + mHeaderSize,
FormatSpec.TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE);
}
}
}
mFreqOutStream.write(freqBuf);
mTerminalAddressTableOutStream.write(terminalAddressTableBuf);
}
}