diff options
Diffstat (limited to 'java/src')
7 files changed, 57 insertions, 29 deletions
diff --git a/java/src/com/android/inputmethod/latin/makedict/AbstractDictDecoder.java b/java/src/com/android/inputmethod/latin/makedict/AbstractDictDecoder.java index b75fc37d9..bbbb8e461 100644 --- a/java/src/com/android/inputmethod/latin/makedict/AbstractDictDecoder.java +++ b/java/src/com/android/inputmethod/latin/makedict/AbstractDictDecoder.java @@ -36,27 +36,25 @@ public abstract class AbstractDictDecoder implements DictDecoder { private static final int ERROR_CANNOT_READ = 1; private static final int ERROR_WRONG_FORMAT = 2; - protected FileHeader readHeader(final DictBuffer dictBuffer) + protected FileHeader readHeader(final DictBuffer headerBuffer) throws IOException, UnsupportedFormatException { - if (dictBuffer == null) { + if (headerBuffer == null) { openDictBuffer(); } - final int version = HeaderReader.readVersion(dictBuffer); + final int version = HeaderReader.readVersion(headerBuffer); if (version < FormatSpec.MINIMUM_SUPPORTED_VERSION || version > FormatSpec.MAXIMUM_SUPPORTED_VERSION) { throw new UnsupportedFormatException("Unsupported version : " + version); } // TODO: Remove this field. - final int optionsFlags = HeaderReader.readOptionFlags(dictBuffer); - - final int headerSize = HeaderReader.readHeaderSize(dictBuffer); - + final int optionsFlags = HeaderReader.readOptionFlags(headerBuffer); + final int headerSize = HeaderReader.readHeaderSize(headerBuffer); if (headerSize < 0) { throw new UnsupportedFormatException("header size can't be negative."); } - final HashMap<String, String> attributes = HeaderReader.readAttributes(dictBuffer, + final HashMap<String, String> attributes = HeaderReader.readAttributes(headerBuffer, headerSize); final FileHeader header = new FileHeader(headerSize, diff --git a/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java b/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java index 17c609fbe..83ee7d685 100644 --- a/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java +++ b/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java @@ -600,7 +600,7 @@ public final class BinaryDictDecoderUtils { Map<Integer, PtNodeArray> reverseNodeArrayMapping = new TreeMap<Integer, PtNodeArray>(); Map<Integer, PtNode> reversePtNodeMapping = new TreeMap<Integer, PtNode>(); - final PtNodeArray root = readNodeArray(dictDecoder, fileHeader.mHeaderSize, + final PtNodeArray root = readNodeArray(dictDecoder, fileHeader.mBodyOffset, reverseNodeArrayMapping, reversePtNodeMapping, fileHeader.mFormatOptions); FusionDictionary newDict = new FusionDictionary(root, fileHeader.mDictionaryOptions); diff --git a/java/src/com/android/inputmethod/latin/makedict/BinaryDictIOUtils.java b/java/src/com/android/inputmethod/latin/makedict/BinaryDictIOUtils.java index 62d9ad86d..86ebf5844 100644 --- a/java/src/com/android/inputmethod/latin/makedict/BinaryDictIOUtils.java +++ b/java/src/com/android/inputmethod/latin/makedict/BinaryDictIOUtils.java @@ -62,7 +62,7 @@ public final class BinaryDictIOUtils { * Retrieves all node arrays without recursive call. */ private static void readUnigramsAndBigramsBinaryInner(final DictDecoder dictDecoder, - final int headerSize, final Map<Integer, String> words, + final int bodyOffset, final Map<Integer, String> words, final Map<Integer, Integer> frequencies, final Map<Integer, ArrayList<PendingAttribute>> bigrams, final FormatOptions formatOptions) { @@ -71,7 +71,7 @@ public final class BinaryDictIOUtils { Stack<Position> stack = new Stack<Position>(); int index = 0; - Position initPos = new Position(headerSize, 0); + Position initPos = new Position(bodyOffset, 0); stack.push(initPos); while (!stack.empty()) { @@ -154,7 +154,7 @@ public final class BinaryDictIOUtils { UnsupportedFormatException { // Read header final FileHeader header = dictDecoder.readHeader(); - readUnigramsAndBigramsBinaryInner(dictDecoder, header.mHeaderSize, words, + readUnigramsAndBigramsBinaryInner(dictDecoder, header.mBodyOffset, words, frequencies, bigrams, header.mFormatOptions); } diff --git a/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java b/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java index b81c8d8e5..846aacf11 100644 --- a/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java +++ b/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java @@ -199,13 +199,19 @@ public final class FormatSpec { */ public static final int MAGIC_NUMBER = 0x9BC13AFE; - static final int MINIMUM_SUPPORTED_VERSION = 2; - static final int MAXIMUM_SUPPORTED_VERSION = 4; static final int NOT_A_VERSION_NUMBER = -1; static final int FIRST_VERSION_WITH_DYNAMIC_UPDATE = 3; static final int FIRST_VERSION_WITH_TERMINAL_ID = 4; + + // These MUST have the same values as the relevant constants in format_utils.h. + // From version 4 on, we use version * 100 + revision as a version number. That allows + // us to change the format during development while having testing devices remove + // older files with each upgrade, while still having a readable versioning scheme. + public static final int VERSION2 = 2; public static final int VERSION3 = 3; - public static final int VERSION4 = 4; + public static final int VERSION4 = 400; + static final int MINIMUM_SUPPORTED_VERSION = VERSION2; + static final int MAXIMUM_SUPPORTED_VERSION = VERSION4; // These options need to be the same numeric values as the one in the native reading code. static final int GERMAN_UMLAUT_PROCESSING_FLAG = 0x1; @@ -263,8 +269,10 @@ public final class FormatSpec { static final int PTNODE_ATTRIBUTE_MAX_ADDRESS_SIZE = 3; static final int PTNODE_SHORTCUT_LIST_SIZE_SIZE = 2; - // These values are used only by version 4 or later. - public static final String TRIE_FILE_EXTENSION = ".trie"; + // These values are used only by version 4 or later. They MUST match the definitions in + // ver4_dict_constants.cpp. + static final String TRIE_FILE_EXTENSION = ".trie"; + public static final String HEADER_FILE_EXTENSION = ".header"; static final String FREQ_FILE_EXTENSION = ".freq"; static final String UNIGRAM_TIMESTAMP_FILE_EXTENSION = ".timestamp"; // tat = Terminal Address Table @@ -361,7 +369,7 @@ public final class FormatSpec { * Class representing file header. */ public static final class FileHeader { - public final int mHeaderSize; + public final int mBodyOffset; public final DictionaryOptions mDictionaryOptions; public final FormatOptions mFormatOptions; // Note that these are corresponding definitions in native code in latinime::HeaderPolicy @@ -377,9 +385,9 @@ public final class FormatSpec { private static final String DICTIONARY_DESCRIPTION_ATTRIBUTE = "description"; public FileHeader(final int headerSize, final DictionaryOptions dictionaryOptions, final FormatOptions formatOptions) { - mHeaderSize = headerSize; mDictionaryOptions = dictionaryOptions; mFormatOptions = formatOptions; + mBodyOffset = formatOptions.mVersion < VERSION4 ? headerSize : 0; if (null == getLocaleString()) { throw new RuntimeException("Cannot create a FileHeader without a locale"); } diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver4DictDecoder.java b/java/src/com/android/inputmethod/latin/makedict/Ver4DictDecoder.java index 2cbec7cec..07522b54b 100644 --- a/java/src/com/android/inputmethod/latin/makedict/Ver4DictDecoder.java +++ b/java/src/com/android/inputmethod/latin/makedict/Ver4DictDecoder.java @@ -45,10 +45,12 @@ public class Ver4DictDecoder extends AbstractDictDecoder { protected static final int FILETYPE_TERMINAL_ADDRESS_TABLE = 3; protected static final int FILETYPE_BIGRAM_FREQ = 4; protected static final int FILETYPE_SHORTCUT = 5; + protected static final int FILETYPE_HEADER = 6; protected final File mDictDirectory; protected final DictionaryBufferFactory mBufferFactory; protected DictBuffer mDictBuffer; + protected DictBuffer mHeaderBuffer; protected DictBuffer mFrequencyBuffer; protected DictBuffer mTerminalAddressTableBuffer; private BigramContentReader mBigramReader; @@ -83,7 +85,7 @@ public class Ver4DictDecoder extends AbstractDictDecoder { @UsedForTesting /* package */ Ver4DictDecoder(final File dictDirectory, final int factoryFlag) { mDictDirectory = dictDirectory; - mDictBuffer = mFrequencyBuffer = null; + mDictBuffer = mHeaderBuffer = mFrequencyBuffer = null; if ((factoryFlag & MASK_DICTBUFFER) == USE_READONLY_BYTEBUFFER) { mBufferFactory = new DictionaryBufferFromReadOnlyByteBufferFactory(); @@ -100,13 +102,16 @@ public class Ver4DictDecoder extends AbstractDictDecoder { /* package */ Ver4DictDecoder(final File dictDirectory, final DictionaryBufferFactory factory) { mDictDirectory = dictDirectory; mBufferFactory = factory; - mDictBuffer = mFrequencyBuffer = null; + mDictBuffer = mHeaderBuffer = mFrequencyBuffer = null; } protected File getFile(final int fileType) throws UnsupportedFormatException { if (fileType == FILETYPE_TRIE) { return new File(mDictDirectory, mDictDirectory.getName() + FormatSpec.TRIE_FILE_EXTENSION); + } else if (fileType == FILETYPE_HEADER) { + return new File(mDictDirectory, + mDictDirectory.getName() + FormatSpec.HEADER_FILE_EXTENSION); } else if (fileType == FILETYPE_FREQUENCY) { return new File(mDictDirectory, mDictDirectory.getName() + FormatSpec.FREQ_FILE_EXTENSION); @@ -132,6 +137,7 @@ public class Ver4DictDecoder extends AbstractDictDecoder { if (!mDictDirectory.isDirectory()) { throw new UnsupportedFormatException("Format 4 dictionary needs a directory"); } + mHeaderBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_HEADER)); mDictBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_TRIE)); mFrequencyBuffer = mBufferFactory.getDictionaryBuffer(getFile(FILETYPE_FREQUENCY)); mTerminalAddressTableBuffer = mBufferFactory.getDictionaryBuffer( @@ -150,16 +156,22 @@ public class Ver4DictDecoder extends AbstractDictDecoder { } @UsedForTesting + /* package */ DictBuffer getHeaderBuffer() { + return mHeaderBuffer; + } + + @UsedForTesting /* package */ DictBuffer getDictBuffer() { return mDictBuffer; } @Override public FileHeader readHeader() throws IOException, UnsupportedFormatException { - if (mDictBuffer == null) { + if (mHeaderBuffer == null) { openDictBuffer(); } - final FileHeader header = super.readHeader(mDictBuffer); + mHeaderBuffer.position(0); + final FileHeader header = super.readHeader(mHeaderBuffer); final int version = header.mFormatOptions.mVersion; if (version != FormatSpec.VERSION4) { throw new UnsupportedFormatException("File header has a wrong version : " + version); diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver4DictEncoder.java b/java/src/com/android/inputmethod/latin/makedict/Ver4DictEncoder.java index 6d045c88a..1a245b6db 100644 --- a/java/src/com/android/inputmethod/latin/makedict/Ver4DictEncoder.java +++ b/java/src/com/android/inputmethod/latin/makedict/Ver4DictEncoder.java @@ -26,6 +26,7 @@ import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode; import com.android.inputmethod.latin.makedict.FusionDictionary.PtNodeArray; import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; import com.android.inputmethod.latin.utils.CollectionUtils; +import com.android.inputmethod.latin.utils.FileUtils; import java.io.File; import java.io.FileNotFoundException; @@ -45,8 +46,8 @@ public class Ver4DictEncoder implements DictEncoder { private final File mDictPlacedDir; private byte[] mTrieBuf; private int mTriePos; - private int mHeaderSize; private OutputStream mTrieOutStream; + private OutputStream mHeaderOutStream; private OutputStream mFreqOutStream; private OutputStream mUnigramTimestampOutStream; private OutputStream mTerminalAddressTableOutStream; @@ -185,16 +186,21 @@ public class Ver4DictEncoder implements DictEncoder { mBaseFilename = header.getId() + "." + header.getVersion(); mDictDir = new File(mDictPlacedDir, mBaseFilename); final File trieFile = new File(mDictDir, mBaseFilename + FormatSpec.TRIE_FILE_EXTENSION); + final File headerFile = new File(mDictDir, + mBaseFilename + FormatSpec.HEADER_FILE_EXTENSION); final File freqFile = new File(mDictDir, mBaseFilename + FormatSpec.FREQ_FILE_EXTENSION); final File timestampFile = new File(mDictDir, mBaseFilename + FormatSpec.UNIGRAM_TIMESTAMP_FILE_EXTENSION); final File terminalAddressTableFile = new File(mDictDir, mBaseFilename + FormatSpec.TERMINAL_ADDRESS_TABLE_FILE_EXTENSION); if (!mDictDir.isDirectory()) { - if (mDictDir.exists()) mDictDir.delete(); + if (mDictDir.exists()) { + FileUtils.deleteRecursively(mDictDir); + } mDictDir.mkdirs(); } mTrieOutStream = new FileOutputStream(trieFile); + mHeaderOutStream = new FileOutputStream(headerFile); mFreqOutStream = new FileOutputStream(freqFile); mTerminalAddressTableOutStream = new FileOutputStream(terminalAddressTableFile); if (formatOptions.mHasTimestamp) { @@ -207,6 +213,9 @@ public class Ver4DictEncoder implements DictEncoder { if (mTrieOutStream != null) { mTrieOutStream.close(); } + if (mHeaderOutStream != null) { + mHeaderOutStream.close(); + } if (mFreqOutStream != null) { mFreqOutStream.close(); } @@ -218,6 +227,7 @@ public class Ver4DictEncoder implements DictEncoder { } } finally { mTrieOutStream = null; + mHeaderOutStream = null; mFreqOutStream = null; mTerminalAddressTableOutStream = null; } @@ -238,8 +248,7 @@ public class Ver4DictEncoder implements DictEncoder { openStreams(formatOptions, dict.mOptions); } - mHeaderSize = BinaryDictEncoderUtils.writeDictionaryHeader(mTrieOutStream, dict, - formatOptions); + BinaryDictEncoderUtils.writeDictionaryHeader(mHeaderOutStream, dict, formatOptions); MakedictLog.i("Flattening the tree..."); ArrayList<PtNodeArray> flatNodes = BinaryDictEncoderUtils.flattenTree(dict.mRootNodeArray); @@ -423,7 +432,7 @@ public class Ver4DictEncoder implements DictEncoder { ptNode.mFrequency, FormatSpec.FREQUENCY_AND_FLAGS_SIZE); BinaryDictEncoderUtils.writeUIntToBuffer(terminalAddressTableBuf, ptNode.mTerminalId * FormatSpec.TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE, - ptNode.mCachedAddressAfterUpdate + mHeaderSize, + ptNode.mCachedAddressAfterUpdate, FormatSpec.TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE); } } diff --git a/java/src/com/android/inputmethod/latin/personalization/DecayingExpandableBinaryDictionaryBase.java b/java/src/com/android/inputmethod/latin/personalization/DecayingExpandableBinaryDictionaryBase.java index f9733f725..02b09fb0c 100644 --- a/java/src/com/android/inputmethod/latin/personalization/DecayingExpandableBinaryDictionaryBase.java +++ b/java/src/com/android/inputmethod/latin/personalization/DecayingExpandableBinaryDictionaryBase.java @@ -122,7 +122,8 @@ public abstract class DecayingExpandableBinaryDictionaryBase extends ExpandableB @Override protected String getFileNameExtentionToOpenDict() { - return "/" + FormatSpec.TRIE_FILE_EXTENSION; + // TODO: pass the directory name instead + return "/" + FormatSpec.HEADER_FILE_EXTENSION; } public void addMultipleDictionaryEntriesToDictionary( |