Merge "add readUnigramsAndBigramsBinary." into jb-mr1-dev

main
Ken Wakasa 2012-08-30 22:55:11 -07:00 committed by Android (Google) Code Review
commit 48643b08a9
2 changed files with 281 additions and 29 deletions

View File

@ -34,6 +34,7 @@ import java.util.Arrays;
import java.util.HashMap;
import java.util.Iterator;
import java.util.Map;
import java.util.Stack;
import java.util.TreeMap;
/**
@ -197,20 +198,21 @@ public class BinaryDictInputOutput {
public void position(int newPosition);
}
private static final class ByteBufferWrapper implements FusionDictionaryBufferInterface {
private ByteBuffer buffer;
ByteBufferWrapper(final ByteBuffer buffer) {
this.buffer = buffer;
public static final class ByteBufferWrapper implements FusionDictionaryBufferInterface {
private ByteBuffer mBuffer;
public ByteBufferWrapper(final ByteBuffer buffer) {
mBuffer = buffer;
}
@Override
public int readUnsignedByte() {
return ((int)buffer.get()) & 0xFF;
return ((int)mBuffer.get()) & 0xFF;
}
@Override
public int readUnsignedShort() {
return ((int)buffer.getShort()) & 0xFFFF;
return ((int)mBuffer.getShort()) & 0xFFFF;
}
@Override
@ -221,17 +223,17 @@ public class BinaryDictInputOutput {
@Override
public int readInt() {
return buffer.getInt();
return mBuffer.getInt();
}
@Override
public int position() {
return buffer.position();
return mBuffer.position();
}
@Override
public void position(int newPos) {
buffer.position(newPos);
mBuffer.position(newPos);
return;
}
}
@ -1367,6 +1369,109 @@ public class BinaryDictInputOutput {
return node;
}
// TODO: move these methods (readUnigramsAndBigramsBinary(|Inner)) and an inner class (Position)
// out of this class.
private static class Position {
public static final int NOT_READ_GROUPCOUNT = -1;
public int mAddress;
public int mNumOfCharGroup;
public int mPosition;
public int mLength;
public Position(int address, int length) {
mAddress = address;
mLength = length;
mNumOfCharGroup = NOT_READ_GROUPCOUNT;
}
}
/**
* Tours all node without recursive call.
*/
private static void readUnigramsAndBigramsBinaryInner(
final FusionDictionaryBufferInterface buffer, final int headerSize,
final Map<Integer, String> words, final Map<Integer, Integer> frequencies,
final Map<Integer, ArrayList<PendingAttribute>> bigrams) {
int[] pushedChars = new int[MAX_WORD_LENGTH + 1];
Stack<Position> stack = new Stack<Position>();
int index = 0;
Position initPos = new Position(headerSize, 0);
stack.push(initPos);
while (!stack.empty()) {
Position p = stack.peek();
if (DBG) {
MakedictLog.d("read: address=" + p.mAddress + ", numOfCharGroup=" +
p.mNumOfCharGroup + ", position=" + p.mPosition + ", length=" + p.mLength);
}
if (buffer.position() != p.mAddress) buffer.position(p.mAddress);
if (index != p.mLength) index = p.mLength;
if (p.mNumOfCharGroup == Position.NOT_READ_GROUPCOUNT) {
p.mNumOfCharGroup = readCharGroupCount(buffer);
p.mAddress += getGroupCountSize(p.mNumOfCharGroup);
p.mPosition = 0;
}
CharGroupInfo info = readCharGroup(buffer, p.mAddress - headerSize);
for (int i = 0; i < info.mCharacters.length; ++i) {
pushedChars[index++] = info.mCharacters[i];
}
p.mPosition++;
if (info.mFrequency != FusionDictionary.CharGroup.NOT_A_TERMINAL) { // found word
words.put(info.mOriginalAddress, new String(pushedChars, 0, index));
frequencies.put(info.mOriginalAddress, info.mFrequency);
if (info.mBigrams != null) bigrams.put(info.mOriginalAddress, info.mBigrams);
}
if (p.mPosition == p.mNumOfCharGroup) {
stack.pop();
} else {
// the node has more groups.
p.mAddress = buffer.position();
}
if (hasChildrenAddress(info.mChildrenAddress)) {
Position childrenPos = new Position(info.mChildrenAddress + headerSize, index);
stack.push(childrenPos);
}
}
return;
}
/**
* Reads unigrams and bigrams from the binary file.
* Doesn't make the memory representation of the dictionary.
*
* @param buffer the buffer to read.
* @param words the map to store the address as a key and the word as a value.
* @param frequencies the map to store the address as a key and the frequency as a value.
* @param bigrams the map to store the address as a key and the list of address as a value.
* @throws IOException
* @throws UnsupportedFormatException
*/
public static void readUnigramsAndBigramsBinary(final FusionDictionaryBufferInterface buffer,
final Map<Integer, String> words, final Map<Integer, Integer> frequencies,
final Map<Integer, ArrayList<PendingAttribute>> bigrams) throws IOException,
UnsupportedFormatException {
// Read header
final int version = checkFormatVersion(buffer);
final int optionsFlags = buffer.readUnsignedShort();
final HashMap<String, String> options = new HashMap<String, String>();
final int headerSize = readHeader(buffer, options, version);
readUnigramsAndBigramsBinaryInner(buffer, headerSize, words, frequencies, bigrams);
}
/**
* Helper function to get the binary format version from the header.
* @throws IOException

View File

@ -20,6 +20,7 @@ import com.android.inputmethod.latin.makedict.BinaryDictInputOutput;
import com.android.inputmethod.latin.makedict.FusionDictionary;
import com.android.inputmethod.latin.makedict.FusionDictionary.CharGroup;
import com.android.inputmethod.latin.makedict.FusionDictionary.Node;
import com.android.inputmethod.latin.makedict.PendingAttribute;
import com.android.inputmethod.latin.makedict.UnsupportedFormatException;
import android.test.AndroidTestCase;
@ -34,7 +35,10 @@ import java.io.FileInputStream;
import java.io.IOException;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.Map.Entry;
import java.util.Random;
import java.util.Set;
@ -46,6 +50,7 @@ public class BinaryDictIOTests extends AndroidTestCase {
private static final int MAX_UNIGRAMS = 1000;
private static final int UNIGRAM_FREQ = 10;
private static final int BIGRAM_FREQ = 50;
private static final int TOLERANCE_OF_BIGRAM_FREQ = 5;
private static final String[] CHARACTERS =
{
@ -53,6 +58,7 @@ public class BinaryDictIOTests extends AndroidTestCase {
"n", "o", "p", "q", "r", "s", "t", "u", "v", "w", "x", "y", "z"
};
// Utilities for test
/**
* Generates a random word.
*/
@ -75,6 +81,9 @@ public class BinaryDictIOTests extends AndroidTestCase {
return new ArrayList<String>(wordSet);
}
/**
* Adds unigrams to the dictionary.
*/
private void addUnigrams(final int number,
final FusionDictionary dict,
final List<String> words) {
@ -86,19 +95,17 @@ public class BinaryDictIOTests extends AndroidTestCase {
private void addBigrams(final FusionDictionary dict,
final List<String> words,
final SparseArray<List<Integer>> sparseArray) {
for (int i = 0; i < sparseArray.size(); ++i) {
final int w1 = sparseArray.keyAt(i);
for (int w2 : sparseArray.valueAt(i)) {
final SparseArray<List<Integer>> bigrams) {
for (int i = 0; i < bigrams.size(); ++i) {
final int w1 = bigrams.keyAt(i);
for (int w2 : bigrams.valueAt(i)) {
dict.setBigram(words.get(w1), words.get(w2), BIGRAM_FREQ);
}
}
}
private long timeWritingDictToFile(final String fileName,
final FusionDictionary dict) {
private long timeWritingDictToFile(final File file, final FusionDictionary dict) {
final File file = new File(getContext().getFilesDir(), fileName);
long now = -1, diff = -1;
try {
@ -140,15 +147,16 @@ public class BinaryDictIOTests extends AndroidTestCase {
}
}
private long timeReadingAndCheckDict(final String fileName,
final List<String> words,
// Tests for readDictionaryBinary and writeDictionaryBinary
private long timeReadingAndCheckDict(final File file, final List<String> words,
final SparseArray<List<Integer>> bigrams) {
long now, diff = -1;
FileInputStream inStream = null;
try {
final File file = new File(getContext().getFilesDir(), fileName);
final FileInputStream inStream = new FileInputStream(file);
inStream = new FileInputStream(file);
final ByteBuffer buffer = inStream.getChannel().map(
FileChannel.MapMode.READ_ONLY, 0, file.length());
@ -166,6 +174,14 @@ public class BinaryDictIOTests extends AndroidTestCase {
Log.e(TAG, "raise IOException while reading file " + e);
} catch (UnsupportedFormatException e) {
Log.e(TAG, "Unsupported format: " + e);
} finally {
if (inStream != null) {
try {
inStream.close();
} catch (IOException e) {
// do nothing
}
}
}
return diff;
@ -178,25 +194,26 @@ public class BinaryDictIOTests extends AndroidTestCase {
new FusionDictionary.DictionaryOptions(
new HashMap<String,String>(), false, false));
final String fileName = generateWord((int)System.currentTimeMillis()) + ".dict";
File file = null;
try {
file = File.createTempFile("runReadAndWrite", ".dict");
} catch (IOException e) {
Log.e(TAG, "IOException: " + e);
}
assertNotNull(file);
addUnigrams(words.size(), dict, words);
addBigrams(dict, words, bigrams);
// check original dictionary
checkDictionary(dict, words, bigrams);
final long write = timeWritingDictToFile(fileName, dict);
final long read = timeReadingAndCheckDict(fileName, words, bigrams);
deleteFile(fileName);
final long write = timeWritingDictToFile(file, dict);
final long read = timeReadingAndCheckDict(file, words, bigrams);
return "PROF: read=" + read + "ms, write=" + write + "ms :" + message;
}
private void deleteFile(final String fileName) {
final File file = new File(getContext().getFilesDir(), fileName);
file.delete();
}
public void testReadAndWrite() {
final List<String> results = new ArrayList<String>();
@ -221,4 +238,134 @@ public class BinaryDictIOTests extends AndroidTestCase {
Log.d(TAG, result);
}
}
// Tests for readUnigramsAndBigramsBinary
private void checkWordMap(final List<String> expectedWords,
final SparseArray<List<Integer>> expectedBigrams,
final Map<Integer, String> resultWords,
final Map<Integer, Integer> resultFrequencies,
final Map<Integer, ArrayList<PendingAttribute>> resultBigrams) {
// check unigrams
final Set<String> actualWordsSet = new HashSet<String>(resultWords.values());
final Set<String> expectedWordsSet = new HashSet<String>(expectedWords);
assertEquals(actualWordsSet, expectedWordsSet);
for (int freq : resultFrequencies.values()) {
assertEquals(freq, UNIGRAM_FREQ);
}
// check bigrams
final Map<String, List<String>> expBigrams = new HashMap<String, List<String>>();
for (int i = 0; i < expectedBigrams.size(); ++i) {
final String word1 = expectedWords.get(expectedBigrams.keyAt(i));
for (int w2 : expectedBigrams.valueAt(i)) {
if (expBigrams.get(word1) == null) {
expBigrams.put(word1, new ArrayList<String>());
}
expBigrams.get(word1).add(expectedWords.get(w2));
}
}
final Map<String, List<String>> actBigrams = new HashMap<String, List<String>>();
for (Entry<Integer, ArrayList<PendingAttribute>> entry : resultBigrams.entrySet()) {
final String word1 = resultWords.get(entry.getKey());
final int unigramFreq = resultFrequencies.get(entry.getKey());
for (PendingAttribute attr : entry.getValue()) {
final String word2 = resultWords.get(attr.mAddress);
if (actBigrams.get(word1) == null) {
actBigrams.put(word1, new ArrayList<String>());
}
actBigrams.get(word1).add(word2);
final int bigramFreq = BinaryDictInputOutput.reconstructBigramFrequency(
unigramFreq, attr.mFrequency);
assertTrue(Math.abs(bigramFreq - BIGRAM_FREQ) < TOLERANCE_OF_BIGRAM_FREQ);
}
}
assertEquals(actBigrams, expBigrams);
}
private long timeAndCheckReadUnigramsAndBigramsBinary(final File file, final List<String> words,
final SparseArray<List<Integer>> bigrams) {
FileInputStream inStream = null;
final Map<Integer, String> resultWords = CollectionUtils.newTreeMap();
final Map<Integer, ArrayList<PendingAttribute>> resultBigrams =
CollectionUtils.newTreeMap();
final Map<Integer, Integer> resultFreqs = CollectionUtils.newTreeMap();
long now = -1, diff = -1;
try {
inStream = new FileInputStream(file);
final ByteBuffer buffer = inStream.getChannel().map(
FileChannel.MapMode.READ_ONLY, 0, file.length());
now = System.currentTimeMillis();
BinaryDictInputOutput.readUnigramsAndBigramsBinary(
new BinaryDictInputOutput.ByteBufferWrapper(buffer), resultWords, resultFreqs,
resultBigrams);
diff = System.currentTimeMillis() - now;
checkWordMap(words, bigrams, resultWords, resultFreqs, resultBigrams);
} catch (IOException e) {
Log.e(TAG, "IOException " + e);
} catch (UnsupportedFormatException e) {
Log.e(TAG, "UnsupportedFormatException: " + e);
} finally {
if (inStream != null) {
try {
inStream.close();
} catch (IOException e) {
// do nothing
}
}
}
return diff;
}
private void runReadUnigramsAndBigramsBinary(final List<String> words,
final SparseArray<List<Integer>> bigrams) {
// making the dictionary from lists of words.
final FusionDictionary dict = new FusionDictionary(new Node(),
new FusionDictionary.DictionaryOptions(
new HashMap<String, String>(), false, false));
File file = null;
try {
file = File.createTempFile("runReadUnigrams", ".dict");
} catch (IOException e) {
Log.e(TAG, "IOException: " + e);
}
assertNotNull(file);
addUnigrams(words.size(), dict, words);
addBigrams(dict, words, bigrams);
timeWritingDictToFile(file, dict);
long wordMap = timeAndCheckReadUnigramsAndBigramsBinary(file, words, bigrams);
long fullReading = timeReadingAndCheckDict(file, words, bigrams);
Log.d(TAG, "read=" + fullReading + ", bytearray=" + wordMap);
}
public void testReadUnigramsAndBigramsBinary() {
final List<String> results = new ArrayList<String>();
final Random random = new Random(123456);
final List<String> words = generateWords(MAX_UNIGRAMS, random);
final SparseArray<List<Integer>> emptyArray = CollectionUtils.newSparseArray();
runReadUnigramsAndBigramsBinary(words, emptyArray);
final SparseArray<List<Integer>> star = CollectionUtils.newSparseArray();
for (int i = 1; i < words.size(); ++i) {
star.put(i-1, new ArrayList<Integer>());
star.get(i-1).add(i);
}
runReadUnigramsAndBigramsBinary(words, star);
}
}