Separate shortcut targets from the trie file.

Bug: 10920165
Change-Id: I340759eadbde7fb64cb3b9a3c619ee3a768cedf8
main
Yuichiro Hanada 2013-10-02 14:45:22 +09:00
parent 90fbc7e221
commit 73b9d3b879
5 changed files with 107 additions and 37 deletions

View File

@ -23,11 +23,11 @@ import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode;
import com.android.inputmethod.latin.makedict.FusionDictionary.PtNodeArray; import com.android.inputmethod.latin.makedict.FusionDictionary.PtNodeArray;
import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString;
import java.io.ByteArrayOutputStream;
import java.io.File; import java.io.File;
import java.io.FileInputStream; import java.io.FileInputStream;
import java.io.FileNotFoundException; import java.io.FileNotFoundException;
import java.io.IOException; import java.io.IOException;
import java.io.OutputStream;
import java.nio.ByteBuffer; import java.nio.ByteBuffer;
import java.nio.channels.FileChannel; import java.nio.channels.FileChannel;
import java.util.ArrayList; import java.util.ArrayList;
@ -219,14 +219,14 @@ public final class BinaryDictDecoderUtils {
} }
/** /**
* Writes a string with our character format to a ByteArrayOutputStream. * Writes a string with our character format to an OutputStream.
* *
* This will also write the terminator byte. * This will also write the terminator byte.
* *
* @param buffer the ByteArrayOutputStream to write to. * @param buffer the OutputStream to write to.
* @param word the string to write. * @param word the string to write.
*/ */
static void writeString(final ByteArrayOutputStream buffer, final String word) { static void writeString(final OutputStream buffer, final String word) throws IOException {
final int length = word.length(); final int length = word.length();
for (int i = 0; i < length; i = word.offsetByCodePoints(i, 1)) { for (int i = 0; i < length; i = word.offsetByCodePoints(i, 1)) {
final int codePoint = word.codePointAt(i); final int codePoint = word.codePointAt(i);

View File

@ -383,8 +383,8 @@ public class BinaryDictEncoderUtils {
nodeSize += getByteSize(getOffsetToTargetNodeArrayDuringUpdate(ptNodeArray, nodeSize += getByteSize(getOffsetToTargetNodeArrayDuringUpdate(ptNodeArray,
nodeSize + size, ptNode.mChildren)); nodeSize + size, ptNode.mChildren));
} }
nodeSize += getShortcutListSize(ptNode.mShortcutTargets);
if (formatOptions.mVersion < FormatSpec.FIRST_VERSION_WITH_TERMINAL_ID) { if (formatOptions.mVersion < FormatSpec.FIRST_VERSION_WITH_TERMINAL_ID) {
nodeSize += getShortcutListSize(ptNode.mShortcutTargets);
if (null != ptNode.mBigrams) { if (null != ptNode.mBigrams) {
for (WeightedString bigram : ptNode.mBigrams) { for (WeightedString bigram : ptNode.mBigrams) {
final int offset = getOffsetToTargetPtNodeDuringUpdate(ptNodeArray, final int offset = getOffsetToTargetPtNodeDuringUpdate(ptNodeArray,

View File

@ -266,15 +266,28 @@ public final class FormatSpec {
// tat = Terminal Address Table // tat = Terminal Address Table
static final String TERMINAL_ADDRESS_TABLE_FILE_EXTENSION = ".tat"; static final String TERMINAL_ADDRESS_TABLE_FILE_EXTENSION = ".tat";
static final String BIGRAM_FILE_EXTENSION = ".bigram"; static final String BIGRAM_FILE_EXTENSION = ".bigram";
static final String SHORTCUT_FILE_EXTENSION = ".shortcut";
static final String LOOKUP_TABLE_FILE_SUFFIX = "_lookup"; static final String LOOKUP_TABLE_FILE_SUFFIX = "_lookup";
static final String CONTENT_TABLE_FILE_SUFFIX = "_index"; static final String CONTENT_TABLE_FILE_SUFFIX = "_index";
static final int FREQUENCY_AND_FLAGS_SIZE = 2; static final int FREQUENCY_AND_FLAGS_SIZE = 2;
static final int TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE = 3; static final int TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE = 3;
// With the English main dictionary as of October 2013, the size of bigram address table is
// is 584KB with the block size being 4.
// This is 91% of that of full address table.
static final int BIGRAM_ADDRESS_TABLE_BLOCK_SIZE = 4; static final int BIGRAM_ADDRESS_TABLE_BLOCK_SIZE = 4;
static final int BIGRAM_CONTENT_COUNT = 1; static final int BIGRAM_CONTENT_COUNT = 1;
static final int BIGRAM_FREQ_CONTENT_INDEX = 0; static final int BIGRAM_FREQ_CONTENT_INDEX = 0;
static final String BIGRAM_FREQ_CONTENT_ID = "_freq"; static final String BIGRAM_FREQ_CONTENT_ID = "_freq";
static final int SHORTCUT_CONTENT_COUNT = 1;
static final int SHORTCUT_CONTENT_INDEX = 0;
// With the English main dictionary as of October 2013, the size of shortcut address table is
// 29KB with the block size being 64.
// This is only 4.4% of that of full address table.
static final int SHORTCUT_ADDRESS_TABLE_BLOCK_SIZE = 64;
static final String SHORTCUT_CONTENT_ID = "_shortcut";
static final int NO_CHILDREN_ADDRESS = Integer.MIN_VALUE; static final int NO_CHILDREN_ADDRESS = Integer.MIN_VALUE;
static final int NO_PARENT_ADDRESS = 0; static final int NO_PARENT_ADDRESS = 0;
static final int NO_FORWARD_LINK_ADDRESS = 0; static final int NO_FORWARD_LINK_ADDRESS = 0;

View File

@ -23,6 +23,7 @@ import com.android.inputmethod.latin.makedict.FormatSpec.FileHeader;
import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions; import com.android.inputmethod.latin.makedict.FormatSpec.FormatOptions;
import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode; import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode;
import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString;
import com.android.inputmethod.latin.utils.CollectionUtils;
import android.util.Log; import android.util.Log;
@ -43,6 +44,7 @@ public class Ver4DictDecoder extends DictDecoder {
private static final int FILETYPE_FREQUENCY = 2; private static final int FILETYPE_FREQUENCY = 2;
private static final int FILETYPE_TERMINAL_ADDRESS_TABLE = 3; private static final int FILETYPE_TERMINAL_ADDRESS_TABLE = 3;
private static final int FILETYPE_BIGRAM_FREQ = 4; private static final int FILETYPE_BIGRAM_FREQ = 4;
private static final int FILETYPE_SHORTCUT = 5;
private final File mDictDirectory; private final File mDictDirectory;
private final DictionaryBufferFactory mBufferFactory; private final DictionaryBufferFactory mBufferFactory;
@ -50,7 +52,9 @@ public class Ver4DictDecoder extends DictDecoder {
private DictBuffer mFrequencyBuffer; private DictBuffer mFrequencyBuffer;
private DictBuffer mTerminalAddressTableBuffer; private DictBuffer mTerminalAddressTableBuffer;
private DictBuffer mBigramBuffer; private DictBuffer mBigramBuffer;
private DictBuffer mShortcutBuffer;
private SparseTable mBigramAddressTable; private SparseTable mBigramAddressTable;
private SparseTable mShortcutAddressTable;
@UsedForTesting @UsedForTesting
/* package */ Ver4DictDecoder(final File dictDirectory, final int factoryFlag) { /* package */ Ver4DictDecoder(final File dictDirectory, final int factoryFlag) {
@ -89,6 +93,10 @@ public class Ver4DictDecoder extends DictDecoder {
return new File(mDictDirectory, return new File(mDictDirectory,
mDictDirectory.getName() + FormatSpec.BIGRAM_FILE_EXTENSION mDictDirectory.getName() + FormatSpec.BIGRAM_FILE_EXTENSION
+ FormatSpec.BIGRAM_FREQ_CONTENT_ID); + 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 { } else {
throw new RuntimeException("Unsupported kind of file : " + fileType); throw new RuntimeException("Unsupported kind of file : " + fileType);
} }
@ -102,6 +110,8 @@ public class Ver4DictDecoder extends DictDecoder {
getFile(FILETYPE_TERMINAL_ADDRESS_TABLE)); getFile(FILETYPE_TERMINAL_ADDRESS_TABLE));
mBigramBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_BIGRAM_FREQ)); mBigramBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_BIGRAM_FREQ));
loadBigramAddressSparseTable(); loadBigramAddressSparseTable();
mShortcutBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_SHORTCUT));
loadShortcutAddressSparseTable();
} }
@Override @Override
@ -136,6 +146,18 @@ public class Ver4DictDecoder extends DictDecoder {
FormatSpec.BIGRAM_ADDRESS_TABLE_BLOCK_SIZE); FormatSpec.BIGRAM_ADDRESS_TABLE_BLOCK_SIZE);
} }
// TODO: Let's have something like SparseTableContentsReader in this class.
private void loadShortcutAddressSparseTable() throws IOException {
final File lookupIndexFile = new File(mDictDirectory, mDictDirectory.getName()
+ FormatSpec.SHORTCUT_FILE_EXTENSION + FormatSpec.LOOKUP_TABLE_FILE_SUFFIX);
final File contentFile = new File(mDictDirectory, mDictDirectory.getName()
+ FormatSpec.SHORTCUT_FILE_EXTENSION + FormatSpec.CONTENT_TABLE_FILE_SUFFIX
+ FormatSpec.SHORTCUT_CONTENT_ID);
mShortcutAddressTable = SparseTable.readFromFiles(lookupIndexFile,
new File[] { contentFile }, FormatSpec.SHORTCUT_ADDRESS_TABLE_BLOCK_SIZE);
}
protected static class PtNodeReader extends DictDecoder.PtNodeReader { protected static class PtNodeReader extends DictDecoder.PtNodeReader {
protected static int readFrequency(final DictBuffer frequencyBuffer, final int terminalId) { protected static int readFrequency(final DictBuffer frequencyBuffer, final int terminalId) {
frequencyBuffer.position(terminalId * FormatSpec.FREQUENCY_AND_FLAGS_SIZE + 1); frequencyBuffer.position(terminalId * FormatSpec.FREQUENCY_AND_FLAGS_SIZE + 1);
@ -147,6 +169,23 @@ public class Ver4DictDecoder extends DictDecoder {
} }
} }
private ArrayList<WeightedString> readShortcuts(final int terminalId) {
if (mShortcutAddressTable.get(0, terminalId) == SparseTable.NOT_EXIST) return null;
final ArrayList<WeightedString> ret = CollectionUtils.newArrayList();
final int posOfShortcuts = mShortcutAddressTable.get(FormatSpec.SHORTCUT_CONTENT_INDEX,
terminalId);
mShortcutBuffer.position(posOfShortcuts);
while (true) {
final int flags = mShortcutBuffer.readUnsignedByte();
final String word = CharEncoding.readString(mShortcutBuffer);
ret.add(new WeightedString(word,
flags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY));
if (0 == (flags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_HAS_NEXT)) break;
}
return ret;
}
// TODO: Make this buffer thread safe. // TODO: Make this buffer thread safe.
// TODO: Support words longer than FormatSpec.MAX_WORD_LENGTH. // TODO: Support words longer than FormatSpec.MAX_WORD_LENGTH.
private final int[] mCharacterBuffer = new int[FormatSpec.MAX_WORD_LENGTH]; private final int[] mCharacterBuffer = new int[FormatSpec.MAX_WORD_LENGTH];
@ -197,14 +236,7 @@ public class Ver4DictDecoder extends DictDecoder {
childrenAddress += addressPointer; childrenAddress += addressPointer;
} }
addressPointer += BinaryDictIOUtils.getChildrenAddressSize(flags, options); addressPointer += BinaryDictIOUtils.getChildrenAddressSize(flags, options);
final ArrayList<WeightedString> shortcutTargets; final ArrayList<WeightedString> shortcutTargets = readShortcuts(terminalId);
if (0 != (flags & FormatSpec.FLAG_HAS_SHORTCUT_TARGETS)) {
// readShortcut will add shortcuts to shortcutTargets.
shortcutTargets = new ArrayList<WeightedString>();
addressPointer += PtNodeReader.readShortcut(mDictBuffer, shortcutTargets);
} else {
shortcutTargets = null;
}
final ArrayList<PendingAttribute> bigrams; final ArrayList<PendingAttribute> bigrams;
if (0 != (flags & FormatSpec.FLAG_HAS_BIGRAMS)) { if (0 != (flags & FormatSpec.FLAG_HAS_BIGRAMS)) {

View File

@ -49,6 +49,7 @@ public class Ver4DictEncoder implements DictEncoder {
private File mDictDir; private File mDictDir;
private String mBaseFilename; private String mBaseFilename;
private BigramContentWriter mBigramWriter; private BigramContentWriter mBigramWriter;
private ShortcutContentWriter mShortcutWriter;
@UsedForTesting @UsedForTesting
public Ver4DictEncoder(final File dictPlacedDir) { public Ver4DictEncoder(final File dictPlacedDir) {
@ -152,6 +153,39 @@ public class Ver4DictEncoder implements DictEncoder {
} }
} }
private static class ShortcutContentWriter extends SparseTableContentWriter {
public ShortcutContentWriter(final String name, final int initialCapacity,
final File baseDir) {
super(name + FormatSpec.SHORTCUT_FILE_EXTENSION, FormatSpec.SHORTCUT_CONTENT_COUNT,
initialCapacity, FormatSpec.SHORTCUT_ADDRESS_TABLE_BLOCK_SIZE, baseDir,
new String[] { name + FormatSpec.SHORTCUT_FILE_EXTENSION },
new String[] { FormatSpec.SHORTCUT_CONTENT_ID });
}
public void writeShortcutForOneWord(final int terminalId,
final Iterator<WeightedString> shortcutIterator) throws IOException {
write(FormatSpec.SHORTCUT_CONTENT_INDEX, terminalId,
new SparseTableContentWriterInterface() {
@Override
public void write(final OutputStream outStream) throws IOException {
writeShortcutForOneWordInternal(outStream, shortcutIterator);
}
});
}
private void writeShortcutForOneWordInternal(final OutputStream outStream,
final Iterator<WeightedString> shortcutIterator) throws IOException {
while (shortcutIterator.hasNext()) {
final WeightedString target = shortcutIterator.next();
final int shortcutFlags = BinaryDictEncoderUtils.makeShortcutFlags(
shortcutIterator.hasNext(), target.mFrequency);
BinaryDictEncoderUtils.writeUIntToStream(outStream, shortcutFlags,
FormatSpec.PTNODE_ATTRIBUTE_FLAGS_SIZE);
CharEncoding.writeString(outStream, target.mWord);
}
}
}
private void openStreams(final FormatOptions formatOptions, final DictionaryOptions dictOptions) private void openStreams(final FormatOptions formatOptions, final DictionaryOptions dictOptions)
throws FileNotFoundException, IOException { throws FileNotFoundException, IOException {
final FileHeader header = new FileHeader(0, dictOptions, formatOptions); final FileHeader header = new FileHeader(0, dictOptions, formatOptions);
@ -225,6 +259,8 @@ public class Ver4DictEncoder implements DictEncoder {
writeTerminalData(flatNodes, terminalCount); writeTerminalData(flatNodes, terminalCount);
mBigramWriter = new BigramContentWriter(mBaseFilename, terminalCount, mDictDir); mBigramWriter = new BigramContentWriter(mBaseFilename, terminalCount, mDictDir);
writeBigrams(flatNodes, dict); writeBigrams(flatNodes, dict);
mShortcutWriter = new ShortcutContentWriter(mBaseFilename, terminalCount, mDictDir);
writeShortcuts(flatNodes);
final PtNodeArray lastNodeArray = flatNodes.get(flatNodes.size() - 1); final PtNodeArray lastNodeArray = flatNodes.get(flatNodes.size() - 1);
final int bufferSize = lastNodeArray.mCachedAddressAfterUpdate + lastNodeArray.mCachedSize; final int bufferSize = lastNodeArray.mCachedAddressAfterUpdate + lastNodeArray.mCachedSize;
@ -306,29 +342,6 @@ public class Ver4DictEncoder implements DictEncoder {
} }
} }
private void writeShortcuts(ArrayList<WeightedString> shortcuts) {
if (null == shortcuts || shortcuts.isEmpty()) return;
final int indexOfShortcutByteSize = mTriePos;
mTriePos += FormatSpec.PTNODE_SHORTCUT_LIST_SIZE_SIZE;
final Iterator<WeightedString> shortcutIterator = shortcuts.iterator();
while (shortcutIterator.hasNext()) {
final WeightedString target = shortcutIterator.next();
final int shortcutFlags = BinaryDictEncoderUtils.makeShortcutFlags(
shortcutIterator.hasNext(), target.mFrequency);
mTrieBuf[mTriePos++] = (byte)shortcutFlags;
final int shortcutShift = CharEncoding.writeString(mTrieBuf, mTriePos,
target.mWord);
mTriePos += shortcutShift;
}
final int shortcutByteSize = mTriePos - indexOfShortcutByteSize;
if (shortcutByteSize > FormatSpec.MAX_SHORTCUT_LIST_SIZE_IN_A_PTNODE) {
throw new RuntimeException("Shortcut list too large : " + shortcutByteSize);
}
BinaryDictEncoderUtils.writeUIntToBuffer(mTrieBuf, indexOfShortcutByteSize,
shortcutByteSize, FormatSpec.PTNODE_SHORTCUT_LIST_SIZE_SIZE);
}
private void writeBigrams(final ArrayList<PtNodeArray> flatNodes, final FusionDictionary dict) private void writeBigrams(final ArrayList<PtNodeArray> flatNodes, final FusionDictionary dict)
throws IOException { throws IOException {
mBigramWriter.openStreams(); mBigramWriter.openStreams();
@ -343,6 +356,19 @@ public class Ver4DictEncoder implements DictEncoder {
mBigramWriter.closeStreams(); mBigramWriter.closeStreams();
} }
private void writeShortcuts(final ArrayList<PtNodeArray> flatNodes) throws IOException {
mShortcutWriter.openStreams();
for (final PtNodeArray nodeArray : flatNodes) {
for (final PtNode ptNode : nodeArray.mData) {
if (ptNode.mShortcutTargets != null && !ptNode.mShortcutTargets.isEmpty()) {
mShortcutWriter.writeShortcutForOneWord(ptNode.mTerminalId,
ptNode.mShortcutTargets.iterator());
}
}
}
mShortcutWriter.closeStreams();
}
@Override @Override
public void writeForwardLinkAddress(int forwardLinkAddress) { public void writeForwardLinkAddress(int forwardLinkAddress) {
mTriePos = BinaryDictEncoderUtils.writeUIntToBuffer(mTrieBuf, mTriePos, mTriePos = BinaryDictEncoderUtils.writeUIntToBuffer(mTrieBuf, mTriePos,
@ -359,7 +385,6 @@ public class Ver4DictEncoder implements DictEncoder {
writeTerminalId(ptNode.mTerminalId); writeTerminalId(ptNode.mTerminalId);
} }
writeChildrenPosition(ptNode, formatOptions); writeChildrenPosition(ptNode, formatOptions);
writeShortcuts(ptNode.mShortcutTargets);
} }
private void writeTerminalData(final ArrayList<PtNodeArray> flatNodes, private void writeTerminalData(final ArrayList<PtNodeArray> flatNodes,