Merge "Correctly read the header of APK-embedded dicts"
commit
6e224a236a
|
@ -229,7 +229,7 @@ final public class BinaryDictionaryGetter {
|
||||||
private static boolean hackCanUseDictionaryFile(final Locale locale, final File f) {
|
private static boolean hackCanUseDictionaryFile(final Locale locale, final File f) {
|
||||||
try {
|
try {
|
||||||
// Read the version of the file
|
// Read the version of the file
|
||||||
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(f);
|
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(f, 0, f.length());
|
||||||
final DictionaryHeader header = dictDecoder.readHeader();
|
final DictionaryHeader header = dictDecoder.readHeader();
|
||||||
|
|
||||||
final String version = header.mDictionaryOptions.mAttributes.get(VERSION_KEY);
|
final String version = header.mDictionaryOptions.mAttributes.get(VERSION_KEY);
|
||||||
|
|
|
@ -357,7 +357,7 @@ public final class BinaryDictDecoderUtils {
|
||||||
* @return true if it's a binary dictionary, false otherwise
|
* @return true if it's a binary dictionary, false otherwise
|
||||||
*/
|
*/
|
||||||
public static boolean isBinaryDictionary(final File file) {
|
public static boolean isBinaryDictionary(final File file) {
|
||||||
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(file);
|
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(file, 0, file.length());
|
||||||
if (dictDecoder == null) {
|
if (dictDecoder == null) {
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
|
|
|
@ -237,7 +237,7 @@ public final class BinaryDictIOUtils {
|
||||||
final File file, final long offset, final long length)
|
final File file, final long offset, final long length)
|
||||||
throws FileNotFoundException, IOException, UnsupportedFormatException {
|
throws FileNotFoundException, IOException, UnsupportedFormatException {
|
||||||
final byte[] buffer = new byte[HEADER_READING_BUFFER_SIZE];
|
final byte[] buffer = new byte[HEADER_READING_BUFFER_SIZE];
|
||||||
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(file,
|
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(file, offset, length,
|
||||||
new DictDecoder.DictionaryBufferFactory() {
|
new DictDecoder.DictionaryBufferFactory() {
|
||||||
@Override
|
@Override
|
||||||
public DictBuffer getDictionaryBuffer(File file)
|
public DictBuffer getDictionaryBuffer(File file)
|
||||||
|
@ -251,8 +251,7 @@ public final class BinaryDictIOUtils {
|
||||||
inStream.close();
|
inStream.close();
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
});
|
||||||
);
|
|
||||||
if (dictDecoder == null) {
|
if (dictDecoder == null) {
|
||||||
return null;
|
return null;
|
||||||
}
|
}
|
||||||
|
|
|
@ -326,30 +326,34 @@ public final class FormatSpec {
|
||||||
* Returns new dictionary decoder.
|
* Returns new dictionary decoder.
|
||||||
*
|
*
|
||||||
* @param dictFile the dictionary file.
|
* @param dictFile the dictionary file.
|
||||||
|
* @param offset the offset in the file.
|
||||||
|
* @param length the length of the file, in bytes.
|
||||||
* @param bufferType The type of buffer, as one of USE_* in DictDecoder.
|
* @param bufferType The type of buffer, as one of USE_* in DictDecoder.
|
||||||
* @return new dictionary decoder if the dictionary file exists, otherwise null.
|
* @return new dictionary decoder if the dictionary file exists, otherwise null.
|
||||||
*/
|
*/
|
||||||
public static DictDecoder getDictDecoder(final File dictFile, final int bufferType) {
|
public static DictDecoder getDictDecoder(final File dictFile, final long offset,
|
||||||
|
final long length, final int bufferType) {
|
||||||
if (dictFile.isDirectory()) {
|
if (dictFile.isDirectory()) {
|
||||||
return new Ver4DictDecoder(dictFile, bufferType);
|
return new Ver4DictDecoder(dictFile, bufferType);
|
||||||
} else if (dictFile.isFile()) {
|
} else if (dictFile.isFile()) {
|
||||||
return new Ver2DictDecoder(dictFile, bufferType);
|
return new Ver2DictDecoder(dictFile, offset, length, bufferType);
|
||||||
}
|
}
|
||||||
return null;
|
return null;
|
||||||
}
|
}
|
||||||
|
|
||||||
public static DictDecoder getDictDecoder(final File dictFile,
|
public static DictDecoder getDictDecoder(final File dictFile, final long offset,
|
||||||
final DictionaryBufferFactory factory) {
|
final long length, final DictionaryBufferFactory factory) {
|
||||||
if (dictFile.isDirectory()) {
|
if (dictFile.isDirectory()) {
|
||||||
return new Ver4DictDecoder(dictFile, factory);
|
return new Ver4DictDecoder(dictFile, factory);
|
||||||
} else if (dictFile.isFile()) {
|
} else if (dictFile.isFile()) {
|
||||||
return new Ver2DictDecoder(dictFile, factory);
|
return new Ver2DictDecoder(dictFile, offset, length, factory);
|
||||||
}
|
}
|
||||||
return null;
|
return null;
|
||||||
}
|
}
|
||||||
|
|
||||||
public static DictDecoder getDictDecoder(final File dictFile) {
|
public static DictDecoder getDictDecoder(final File dictFile, final long offset,
|
||||||
return getDictDecoder(dictFile, DictDecoder.USE_READONLY_BYTEBUFFER);
|
final long length) {
|
||||||
|
return getDictDecoder(dictFile, offset, length, DictDecoder.USE_READONLY_BYTEBUFFER);
|
||||||
}
|
}
|
||||||
|
|
||||||
private FormatSpec() {
|
private FormatSpec() {
|
||||||
|
|
|
@ -116,13 +116,18 @@ public class Ver2DictDecoder extends AbstractDictDecoder {
|
||||||
}
|
}
|
||||||
|
|
||||||
protected final File mDictionaryBinaryFile;
|
protected final File mDictionaryBinaryFile;
|
||||||
|
protected final long mOffset;
|
||||||
|
protected final long mLength;
|
||||||
// TODO: Remove mBufferFactory and mDictBuffer from this class members because they are now
|
// TODO: Remove mBufferFactory and mDictBuffer from this class members because they are now
|
||||||
// used only for testing.
|
// used only for testing.
|
||||||
private final DictionaryBufferFactory mBufferFactory;
|
private final DictionaryBufferFactory mBufferFactory;
|
||||||
protected DictBuffer mDictBuffer;
|
protected DictBuffer mDictBuffer;
|
||||||
|
|
||||||
/* package */ Ver2DictDecoder(final File file, final int factoryFlag) {
|
/* package */ Ver2DictDecoder(final File file, final long offset, final long length,
|
||||||
|
final int factoryFlag) {
|
||||||
mDictionaryBinaryFile = file;
|
mDictionaryBinaryFile = file;
|
||||||
|
mOffset = offset;
|
||||||
|
mLength = length;
|
||||||
mDictBuffer = null;
|
mDictBuffer = null;
|
||||||
if ((factoryFlag & MASK_DICTBUFFER) == USE_READONLY_BYTEBUFFER) {
|
if ((factoryFlag & MASK_DICTBUFFER) == USE_READONLY_BYTEBUFFER) {
|
||||||
mBufferFactory = new DictionaryBufferFromReadOnlyByteBufferFactory();
|
mBufferFactory = new DictionaryBufferFromReadOnlyByteBufferFactory();
|
||||||
|
@ -135,8 +140,11 @@ public class Ver2DictDecoder extends AbstractDictDecoder {
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
/* package */ Ver2DictDecoder(final File file, final DictionaryBufferFactory factory) {
|
/* package */ Ver2DictDecoder(final File file, final long offset, final long length,
|
||||||
|
final DictionaryBufferFactory factory) {
|
||||||
mDictionaryBinaryFile = file;
|
mDictionaryBinaryFile = file;
|
||||||
|
mOffset = offset;
|
||||||
|
mLength = length;
|
||||||
mBufferFactory = factory;
|
mBufferFactory = factory;
|
||||||
}
|
}
|
||||||
|
|
||||||
|
@ -164,9 +172,9 @@ public class Ver2DictDecoder extends AbstractDictDecoder {
|
||||||
public DictionaryHeader readHeader() throws IOException, UnsupportedFormatException {
|
public DictionaryHeader readHeader() throws IOException, UnsupportedFormatException {
|
||||||
// dictType is not being used in dicttool. Passing an empty string.
|
// dictType is not being used in dicttool. Passing an empty string.
|
||||||
final BinaryDictionary binaryDictionary = new BinaryDictionary(
|
final BinaryDictionary binaryDictionary = new BinaryDictionary(
|
||||||
mDictionaryBinaryFile.getAbsolutePath(), 0 /* offset */,
|
mDictionaryBinaryFile.getAbsolutePath(), mOffset, mLength,
|
||||||
mDictionaryBinaryFile.length() /* length */, true /* useFullEditDistance */,
|
true /* useFullEditDistance */, null /* locale */, "" /* dictType */,
|
||||||
null /* locale */, "" /* dictType */, false /* isUpdatable */);
|
false /* isUpdatable */);
|
||||||
final DictionaryHeader header = binaryDictionary.getHeader();
|
final DictionaryHeader header = binaryDictionary.getHeader();
|
||||||
binaryDictionary.close();
|
binaryDictionary.close();
|
||||||
if (header == null) {
|
if (header == null) {
|
||||||
|
|
|
@ -150,7 +150,7 @@ public class BinaryDictionaryDecayingTests extends AndroidTestCase {
|
||||||
binaryDictionary.flushWithGC();
|
binaryDictionary.flushWithGC();
|
||||||
binaryDictionary.close();
|
binaryDictionary.close();
|
||||||
|
|
||||||
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(dictFile);
|
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(dictFile, 0, dictFile.length());
|
||||||
try {
|
try {
|
||||||
final FusionDictionary dict =
|
final FusionDictionary dict =
|
||||||
dictDecoder.readDictionaryBinary(false /* deleteDictIfBroken */);
|
dictDecoder.readDictionaryBinary(false /* deleteDictIfBroken */);
|
||||||
|
|
|
@ -251,7 +251,8 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
|
||||||
|
|
||||||
FusionDictionary dict = null;
|
FusionDictionary dict = null;
|
||||||
try {
|
try {
|
||||||
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(file, bufferType);
|
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(file, 0, file.length(),
|
||||||
|
bufferType);
|
||||||
now = System.currentTimeMillis();
|
now = System.currentTimeMillis();
|
||||||
dict = dictDecoder.readDictionaryBinary(false /* deleteDictIfBroken */);
|
dict = dictDecoder.readDictionaryBinary(false /* deleteDictIfBroken */);
|
||||||
diff = System.currentTimeMillis() - now;
|
diff = System.currentTimeMillis() - now;
|
||||||
|
@ -413,7 +414,8 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
|
||||||
|
|
||||||
long now = -1, diff = -1;
|
long now = -1, diff = -1;
|
||||||
try {
|
try {
|
||||||
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(file, bufferType);
|
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(file, 0, file.length(),
|
||||||
|
bufferType);
|
||||||
now = System.currentTimeMillis();
|
now = System.currentTimeMillis();
|
||||||
dictDecoder.readUnigramsAndBigramsBinary(resultWords, resultFreqs, resultBigrams);
|
dictDecoder.readUnigramsAndBigramsBinary(resultWords, resultFreqs, resultBigrams);
|
||||||
diff = System.currentTimeMillis() - now;
|
diff = System.currentTimeMillis() - now;
|
||||||
|
@ -537,7 +539,8 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
|
||||||
addBigrams(dict, words, bigrams);
|
addBigrams(dict, words, bigrams);
|
||||||
timeWritingDictToFile(file, dict, formatOptions);
|
timeWritingDictToFile(file, dict, formatOptions);
|
||||||
|
|
||||||
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(file, DictDecoder.USE_BYTEARRAY);
|
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(file, 0, file.length(),
|
||||||
|
DictDecoder.USE_BYTEARRAY);
|
||||||
try {
|
try {
|
||||||
dictDecoder.openDictBuffer();
|
dictDecoder.openDictBuffer();
|
||||||
} catch (IOException e) {
|
} catch (IOException e) {
|
||||||
|
|
|
@ -68,7 +68,8 @@ public class Ver2DictDecoderTests extends AndroidTestCase {
|
||||||
}
|
}
|
||||||
|
|
||||||
assertNotNull(testFile);
|
assertNotNull(testFile);
|
||||||
final Ver2DictDecoder dictDecoder = new Ver2DictDecoder(testFile, factory);
|
final Ver2DictDecoder dictDecoder = new Ver2DictDecoder(testFile, 0, testFile.length(),
|
||||||
|
factory);
|
||||||
try {
|
try {
|
||||||
dictDecoder.openDictBuffer();
|
dictDecoder.openDictBuffer();
|
||||||
} catch (Exception e) {
|
} catch (Exception e) {
|
||||||
|
@ -110,7 +111,8 @@ public class Ver2DictDecoderTests extends AndroidTestCase {
|
||||||
Log.e(TAG, "IOException while the creating temporary file", e);
|
Log.e(TAG, "IOException while the creating temporary file", e);
|
||||||
}
|
}
|
||||||
|
|
||||||
final Ver2DictDecoder dictDecoder = new Ver2DictDecoder(testFile, factory);
|
final Ver2DictDecoder dictDecoder = new Ver2DictDecoder(testFile, 0, testFile.length(),
|
||||||
|
factory);
|
||||||
|
|
||||||
// the default return value of getBuffer() must be null.
|
// the default return value of getBuffer() must be null.
|
||||||
assertNull("the default return value of getBuffer() is not null",
|
assertNull("the default return value of getBuffer() is not null",
|
||||||
|
|
|
@ -192,7 +192,7 @@ public final class BinaryDictOffdeviceUtils {
|
||||||
new BufferedInputStream(new FileInputStream(decodedSpec.mFile)));
|
new BufferedInputStream(new FileInputStream(decodedSpec.mFile)));
|
||||||
} else {
|
} else {
|
||||||
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(decodedSpec.mFile,
|
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(decodedSpec.mFile,
|
||||||
DictDecoder.USE_BYTEARRAY);
|
0, decodedSpec.mFile.length(), DictDecoder.USE_BYTEARRAY);
|
||||||
if (report) {
|
if (report) {
|
||||||
System.out.println("Format : Binary dictionary format");
|
System.out.println("Format : Binary dictionary format");
|
||||||
System.out.println("Packaging : " + decodedSpec.describeChain());
|
System.out.println("Packaging : " + decodedSpec.describeChain());
|
||||||
|
|
|
@ -264,7 +264,7 @@ public class DictionaryMaker {
|
||||||
private static FusionDictionary readBinaryFile(final String binaryFilename)
|
private static FusionDictionary readBinaryFile(final String binaryFilename)
|
||||||
throws FileNotFoundException, IOException, UnsupportedFormatException {
|
throws FileNotFoundException, IOException, UnsupportedFormatException {
|
||||||
final File file = new File(binaryFilename);
|
final File file = new File(binaryFilename);
|
||||||
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(file);
|
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(file, 0, file.length());
|
||||||
return dictDecoder.readDictionaryBinary(false /* deleteDictIfBroken */);
|
return dictDecoder.readDictionaryBinary(false /* deleteDictIfBroken */);
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
|
@ -77,7 +77,8 @@ public class BinaryDictOffdeviceUtilsTests extends TestCase {
|
||||||
assertEquals("Wrong decode spec", BinaryDictOffdeviceUtils.COMPRESSION, step);
|
assertEquals("Wrong decode spec", BinaryDictOffdeviceUtils.COMPRESSION, step);
|
||||||
}
|
}
|
||||||
assertEquals("Wrong decode spec", 3, decodeSpec.mDecoderSpec.size());
|
assertEquals("Wrong decode spec", 3, decodeSpec.mDecoderSpec.size());
|
||||||
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(decodeSpec.mFile);
|
final DictDecoder dictDecoder = FormatSpec.getDictDecoder(decodeSpec.mFile, 0,
|
||||||
|
decodeSpec.mFile.length());
|
||||||
final FusionDictionary resultDict =
|
final FusionDictionary resultDict =
|
||||||
dictDecoder.readDictionaryBinary(false /* deleteDictIfBroken */);
|
dictDecoder.readDictionaryBinary(false /* deleteDictIfBroken */);
|
||||||
assertEquals("Wrong version attribute", VERSION, resultDict.mOptions.mAttributes.get(
|
assertEquals("Wrong version attribute", VERSION, resultDict.mOptions.mAttributes.get(
|
||||||
|
|
Loading…
Reference in New Issue