445 lines
16 KiB
Java
445 lines
16 KiB
Java
/*
|
|
* Copyright (C) 2009 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.tools.dict;
|
|
|
|
import org.xml.sax.Attributes;
|
|
import org.xml.sax.helpers.DefaultHandler;
|
|
|
|
import java.io.File;
|
|
import java.io.FileOutputStream;
|
|
import java.io.IOException;
|
|
import java.util.ArrayList;
|
|
import java.util.Arrays;
|
|
import java.util.HashMap;
|
|
import java.util.List;
|
|
import java.util.Map;
|
|
|
|
import javax.xml.parsers.SAXParser;
|
|
import javax.xml.parsers.SAXParserFactory;
|
|
|
|
/**
|
|
* Compresses a list of words, frequencies, and bigram data
|
|
* into a tree structured binary dictionary.
|
|
* Dictionary Version: 200 (may contain bigrams)
|
|
* Version number started from 200 rather than 1 because we wanted to prevent number of roots in
|
|
* any old dictionaries being mistaken as the version number. There is not a chance that there
|
|
* will be more than 200 roots. Version number should be increased when there is structural change
|
|
* in the data. There is no need to increase the version when only the words in the data changes.
|
|
*/
|
|
public class MakeBinaryDictionary {
|
|
private static final int VERSION_NUM = 200;
|
|
|
|
private static final String TAG_WORD = "w";
|
|
private static final String ATTR_FREQ = "f";
|
|
|
|
private static final int FLAG_ADDRESS_MASK = 0x400000;
|
|
private static final int FLAG_TERMINAL_MASK = 0x800000;
|
|
private static final int ADDRESS_MASK = 0x3FFFFF;
|
|
|
|
private static final int INITIAL_STRING_BUILDER_CAPACITY = 48;
|
|
|
|
/**
|
|
* Unit for this variable is in bytes
|
|
* If destination file name is main.dict and file limit causes dictionary to be separated into
|
|
* multiple file, it will generate main0.dict, main1.dict, and so forth.
|
|
*/
|
|
private static int sOutputFileSize;
|
|
private static boolean sSplitOutput;
|
|
|
|
private static final CharNode EMPTY_NODE = new CharNode();
|
|
|
|
private List<CharNode> mRoots;
|
|
private Map<String, Integer> mDictionary;
|
|
private int mWordCount;
|
|
|
|
private BigramDictionary mBigramDict;
|
|
|
|
private static class CharNode {
|
|
char data;
|
|
int freq;
|
|
boolean terminal;
|
|
List<CharNode> children;
|
|
static int sNodes;
|
|
|
|
public CharNode() {
|
|
sNodes++;
|
|
}
|
|
}
|
|
|
|
private static void usage() {
|
|
System.err.println("Usage: makedict -s <src_dict.xml> [-b <src_bigram.xml>] "
|
|
+ "-d <dest.dict> [--size filesize]");
|
|
System.exit(-1);
|
|
}
|
|
|
|
public static void main(String[] args) {
|
|
int checkSource = -1;
|
|
int checkBigram = -1;
|
|
int checkDest = -1;
|
|
int checkFileSize = -1;
|
|
for (int i = 0; i < args.length; i+=2) {
|
|
if (args[i].equals("-s")) checkSource = (i + 1);
|
|
if (args[i].equals("-b")) checkBigram = (i + 1);
|
|
if (args[i].equals("-d")) checkDest = (i + 1);
|
|
if (args[i].equals("--size")) checkFileSize = (i + 1);
|
|
}
|
|
if (checkFileSize >= 0) {
|
|
sSplitOutput = true;
|
|
sOutputFileSize = Integer.parseInt(args[checkFileSize]);
|
|
} else {
|
|
sSplitOutput = false;
|
|
}
|
|
if (checkDest >= 0 && !args[checkDest].endsWith(".dict")) {
|
|
System.err.println("Error: Dictionary output file extension should be \".dict\"");
|
|
usage();
|
|
} else if (checkSource >= 0 && checkBigram >= 0 && checkDest >= 0 &&
|
|
((!sSplitOutput && args.length == 6) || (sSplitOutput && args.length == 8))) {
|
|
new MakeBinaryDictionary(args[checkSource], args[checkBigram], args[checkDest]);
|
|
} else if (checkSource >= 0 && checkDest >= 0 &&
|
|
((!sSplitOutput && args.length == 4) || (sSplitOutput && args.length == 6))) {
|
|
new MakeBinaryDictionary(args[checkSource], null, args[checkDest]);
|
|
} else {
|
|
usage();
|
|
}
|
|
}
|
|
|
|
private MakeBinaryDictionary(String srcFilename, String bigramSrcFilename,
|
|
String destFilename) {
|
|
System.out.println("Generating dictionary version " + VERSION_NUM);
|
|
mBigramDict = new BigramDictionary(bigramSrcFilename, (bigramSrcFilename != null));
|
|
populateDictionary(srcFilename);
|
|
writeToDict(destFilename);
|
|
|
|
// Enable the code below to verify that the generated tree is traversable
|
|
// and bigram data is stored correctly.
|
|
if (false) {
|
|
mBigramDict.reverseLookupAll(mDictionary, mDict);
|
|
traverseDict(2, new char[32], 0);
|
|
}
|
|
}
|
|
|
|
private void populateDictionary(String filename) {
|
|
mRoots = new ArrayList<CharNode>();
|
|
mDictionary = new HashMap<String, Integer>();
|
|
try {
|
|
SAXParser parser = SAXParserFactory.newInstance().newSAXParser();
|
|
parser.parse(new File(filename), new DefaultHandler() {
|
|
boolean inWord;
|
|
int freq;
|
|
StringBuilder wordBuilder = new StringBuilder(INITIAL_STRING_BUILDER_CAPACITY);
|
|
|
|
@Override
|
|
public void startElement(String uri, String localName,
|
|
String qName, Attributes attributes) {
|
|
if (qName.equals(TAG_WORD)) {
|
|
inWord = true;
|
|
freq = Integer.parseInt(attributes.getValue(ATTR_FREQ));
|
|
wordBuilder.setLength(0);
|
|
}
|
|
}
|
|
|
|
@Override
|
|
public void characters(char[] data, int offset, int length) {
|
|
// Ignore other whitespace
|
|
if (!inWord) return;
|
|
wordBuilder.append(data, offset, length);
|
|
}
|
|
|
|
@Override
|
|
public void endElement(String uri, String localName,
|
|
String qName) {
|
|
if (qName.equals(TAG_WORD)) {
|
|
if (wordBuilder.length() >= 1) {
|
|
addWordTop(wordBuilder.toString(), freq);
|
|
mWordCount++;
|
|
}
|
|
inWord = false;
|
|
}
|
|
}
|
|
});
|
|
} catch (Exception ioe) {
|
|
System.err.println("Exception in parsing\n" + ioe);
|
|
ioe.printStackTrace();
|
|
}
|
|
System.out.println("Nodes = " + CharNode.sNodes);
|
|
}
|
|
|
|
private static int indexOf(List<CharNode> children, char c) {
|
|
if (children == null) {
|
|
return -1;
|
|
}
|
|
for (int i = 0; i < children.size(); i++) {
|
|
if (children.get(i).data == c) {
|
|
return i;
|
|
}
|
|
}
|
|
return -1;
|
|
}
|
|
|
|
private void addWordTop(String word, int freq) {
|
|
if (freq < 0) {
|
|
freq = 0;
|
|
} else if (freq > 255) {
|
|
freq = 255;
|
|
}
|
|
char firstChar = word.charAt(0);
|
|
int index = indexOf(mRoots, firstChar);
|
|
if (index == -1) {
|
|
CharNode newNode = new CharNode();
|
|
newNode.data = firstChar;
|
|
index = mRoots.size();
|
|
mRoots.add(newNode);
|
|
}
|
|
final CharNode node = mRoots.get(index);
|
|
if (word.length() > 1) {
|
|
addWordRec(node, word, 1, freq);
|
|
} else {
|
|
node.terminal = true;
|
|
node.freq = freq;
|
|
}
|
|
}
|
|
|
|
private void addWordRec(CharNode parent, String word, int charAt, int freq) {
|
|
CharNode child = null;
|
|
char data = word.charAt(charAt);
|
|
if (parent.children == null) {
|
|
parent.children = new ArrayList<CharNode>();
|
|
} else {
|
|
for (int i = 0; i < parent.children.size(); i++) {
|
|
CharNode node = parent.children.get(i);
|
|
if (node.data == data) {
|
|
child = node;
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
if (child == null) {
|
|
child = new CharNode();
|
|
parent.children.add(child);
|
|
}
|
|
child.data = data;
|
|
if (word.length() > charAt + 1) {
|
|
addWordRec(child, word, charAt + 1, freq);
|
|
} else {
|
|
child.terminal = true;
|
|
child.freq = freq;
|
|
}
|
|
}
|
|
|
|
private byte[] mDict;
|
|
private int mDictSize;
|
|
private static final int CHAR_WIDTH = 8;
|
|
private static final int FLAGS_WIDTH = 1; // Terminal flag (word end)
|
|
private static final int ADDR_WIDTH = 23; // Offset to children
|
|
private static final int FREQ_WIDTH_BYTES = 1;
|
|
private static final int COUNT_WIDTH_BYTES = 1;
|
|
|
|
private void addCount(int count) {
|
|
mDict[mDictSize++] = (byte) (0xFF & count);
|
|
}
|
|
|
|
private void addNode(CharNode node, String word1) {
|
|
if (node.terminal) { // store address of each word1 for bigram dic generation
|
|
mDictionary.put(word1, mDictSize);
|
|
}
|
|
|
|
int charData = 0xFFFF & node.data;
|
|
if (charData > 254) {
|
|
mDict[mDictSize++] = (byte) 255;
|
|
mDict[mDictSize++] = (byte) ((node.data >> 8) & 0xFF);
|
|
mDict[mDictSize++] = (byte) (node.data & 0xFF);
|
|
} else {
|
|
mDict[mDictSize++] = (byte) (0xFF & node.data);
|
|
}
|
|
if (node.children != null) {
|
|
mDictSize += 3; // Space for children address
|
|
} else {
|
|
mDictSize += 1; // Space for just the terminal/address flags
|
|
}
|
|
if ((0xFFFFFF & node.freq) > 255) {
|
|
node.freq = 255;
|
|
}
|
|
if (node.terminal) {
|
|
byte freq = (byte) (0xFF & node.freq);
|
|
mDict[mDictSize++] = freq;
|
|
// bigram
|
|
if (mBigramDict.mBi.containsKey(word1)) {
|
|
int count = mBigramDict.mBi.get(word1).count;
|
|
mBigramDict.mBigramToFill.add(word1);
|
|
mBigramDict.mBigramToFillAddress.add(mDictSize);
|
|
mDictSize += (4 * count);
|
|
} else {
|
|
mDict[mDictSize++] = (byte) (0x00);
|
|
}
|
|
}
|
|
}
|
|
|
|
private int mNullChildrenCount = 0;
|
|
private int mNotTerminalCount = 0;
|
|
|
|
private void updateNodeAddress(int nodeAddress, CharNode node,
|
|
int childrenAddress) {
|
|
if ((mDict[nodeAddress] & 0xFF) == 0xFF) { // 3 byte character
|
|
nodeAddress += 2;
|
|
}
|
|
childrenAddress = ADDRESS_MASK & childrenAddress;
|
|
if (childrenAddress == 0) {
|
|
mNullChildrenCount++;
|
|
} else {
|
|
childrenAddress |= FLAG_ADDRESS_MASK;
|
|
}
|
|
if (node.terminal) {
|
|
childrenAddress |= FLAG_TERMINAL_MASK;
|
|
} else {
|
|
mNotTerminalCount++;
|
|
}
|
|
mDict[nodeAddress + 1] = (byte) (childrenAddress >> 16);
|
|
if ((childrenAddress & FLAG_ADDRESS_MASK) != 0) {
|
|
mDict[nodeAddress + 2] = (byte) ((childrenAddress & 0xFF00) >> 8);
|
|
mDict[nodeAddress + 3] = (byte) ((childrenAddress & 0xFF));
|
|
}
|
|
}
|
|
|
|
private void writeWordsRec(List<CharNode> children, StringBuilder word) {
|
|
if (children == null || children.size() == 0) {
|
|
return;
|
|
}
|
|
final int childCount = children.size();
|
|
addCount(childCount);
|
|
int[] childrenAddresses = new int[childCount];
|
|
for (int j = 0; j < childCount; j++) {
|
|
CharNode child = children.get(j);
|
|
childrenAddresses[j] = mDictSize;
|
|
word.append(child.data);
|
|
addNode(child, word.toString());
|
|
word.setLength(word.length() - 1);
|
|
}
|
|
for (int j = 0; j < childCount; j++) {
|
|
CharNode child = children.get(j);
|
|
int nodeAddress = childrenAddresses[j];
|
|
int cacheDictSize = mDictSize;
|
|
word.append(child.data);
|
|
writeWordsRec(child.children, word);
|
|
word.setLength(word.length() - 1);
|
|
updateNodeAddress(nodeAddress, child, child.children != null ? cacheDictSize : 0);
|
|
}
|
|
}
|
|
|
|
private void writeToDict(String dictFilename) {
|
|
// 4MB max, 22-bit offsets
|
|
mDict = new byte[4 * 1024 * 1024]; // 4MB upper limit. Actual is probably
|
|
// < 1MB in most cases, as there is a limit in the
|
|
// resource size in apks.
|
|
mDictSize = 0;
|
|
|
|
mDict[mDictSize++] = (byte) (0xFF & VERSION_NUM); // version info
|
|
mDict[mDictSize++] = (byte) (0xFF & (mBigramDict.mHasBigram ? 1 : 0));
|
|
|
|
final StringBuilder word = new StringBuilder(INITIAL_STRING_BUILDER_CAPACITY);
|
|
writeWordsRec(mRoots, word);
|
|
mDict = mBigramDict.writeBigrams(mDict, mDictionary);
|
|
System.out.println("Dict Size = " + mDictSize);
|
|
if (!sSplitOutput) {
|
|
sOutputFileSize = mDictSize;
|
|
}
|
|
try {
|
|
int currentLoc = 0;
|
|
int i = 0;
|
|
int extension = dictFilename.indexOf(".dict");
|
|
String filename = dictFilename.substring(0, extension);
|
|
while (mDictSize > 0) {
|
|
FileOutputStream fos;
|
|
if (sSplitOutput) {
|
|
fos = new FileOutputStream(filename + i + ".dict");
|
|
} else {
|
|
fos = new FileOutputStream(filename + ".dict");
|
|
}
|
|
if (mDictSize > sOutputFileSize) {
|
|
fos.write(mDict, currentLoc, sOutputFileSize);
|
|
mDictSize -= sOutputFileSize;
|
|
currentLoc += sOutputFileSize;
|
|
} else {
|
|
fos.write(mDict, currentLoc, mDictSize);
|
|
mDictSize = 0;
|
|
}
|
|
fos.close();
|
|
i++;
|
|
}
|
|
} catch (IOException ioe) {
|
|
System.err.println("Error writing dict file:" + ioe);
|
|
}
|
|
}
|
|
|
|
private void traverseDict(int pos, char[] word, int depth) {
|
|
int count = mDict[pos++] & 0xFF;
|
|
for (int i = 0; i < count; i++) {
|
|
char c = (char) (mDict[pos++] & 0xFF);
|
|
if (c == 0xFF) { // two byte character
|
|
c = (char) (((mDict[pos] & 0xFF) << 8) | (mDict[pos+1] & 0xFF));
|
|
pos += 2;
|
|
}
|
|
word[depth] = c;
|
|
boolean terminal = getFirstBitOfByte(pos, mDict);
|
|
int address = 0;
|
|
if ((mDict[pos] & (FLAG_ADDRESS_MASK >> 16)) > 0) { // address check
|
|
address = get22BitAddress(pos, mDict);
|
|
pos += 3;
|
|
} else {
|
|
pos += 1;
|
|
}
|
|
if (terminal) {
|
|
showWord(word, depth + 1, mDict[pos] & 0xFF);
|
|
pos++;
|
|
|
|
int bigramExist = (mDict[pos] & mBigramDict.FLAG_BIGRAM_READ);
|
|
if (bigramExist > 0) {
|
|
int nextBigramExist = 1;
|
|
while (nextBigramExist > 0) {
|
|
int bigramAddress = get22BitAddress(pos, mDict);
|
|
pos += 3;
|
|
int frequency = (mBigramDict.FLAG_BIGRAM_FREQ & mDict[pos]);
|
|
mBigramDict.searchForTerminalNode(bigramAddress, frequency, mDict);
|
|
nextBigramExist = (mDict[pos++] & mBigramDict.FLAG_BIGRAM_CONTINUED);
|
|
}
|
|
} else {
|
|
pos++;
|
|
}
|
|
}
|
|
if (address != 0) {
|
|
traverseDict(address, word, depth + 1);
|
|
}
|
|
}
|
|
}
|
|
|
|
private static void showWord(char[] word, int size, int freq) {
|
|
System.out.print(new String(word, 0, size) + " " + freq + "\n");
|
|
}
|
|
|
|
/* package */ static int get22BitAddress(int pos, byte[] dict) {
|
|
return ((dict[pos + 0] & 0x3F) << 16)
|
|
| ((dict[pos + 1] & 0xFF) << 8)
|
|
| ((dict[pos + 2] & 0xFF));
|
|
}
|
|
|
|
/* package */ static boolean getFirstBitOfByte(int pos, byte[] dict) {
|
|
return (dict[pos] & 0x80) > 0;
|
|
}
|
|
|
|
/* package */ static boolean getSecondBitOfByte(int pos, byte[] dict) {
|
|
return (dict[pos] & 0x40) > 0;
|
|
}
|
|
}
|