diff options
Diffstat (limited to 'java/src/com/android/inputmethod/latin/makedict/Ver3DictDecoder.java')
| -rw-r--r-- | java/src/com/android/inputmethod/latin/makedict/Ver3DictDecoder.java | 178 |
1 files changed, 15 insertions, 163 deletions
diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver3DictDecoder.java b/java/src/com/android/inputmethod/latin/makedict/Ver3DictDecoder.java index 1a90a4b98..848277cd4 100644 --- a/java/src/com/android/inputmethod/latin/makedict/Ver3DictDecoder.java +++ b/java/src/com/android/inputmethod/latin/makedict/Ver3DictDecoder.java @@ -32,14 +32,12 @@ import java.io.FileNotFoundException; import java.io.IOException; import java.util.ArrayList; import java.util.Arrays; -import java.util.HashMap; -import java.util.TreeMap; /** * An implementation of DictDecoder for version 3 binary dictionary. */ @UsedForTesting -public class Ver3DictDecoder implements DictDecoder { +public class Ver3DictDecoder extends DictDecoder { private static final String TAG = Ver3DictDecoder.class.getSimpleName(); static { @@ -49,124 +47,10 @@ public class Ver3DictDecoder implements DictDecoder { // TODO: implement something sensical instead of just a phony method private static native int doNothing(); - private final static class HeaderReader { - protected static int readVersion(final DictBuffer dictBuffer) - throws IOException, UnsupportedFormatException { - return BinaryDictDecoderUtils.checkFormatVersion(dictBuffer); - } - - protected static int readOptionFlags(final DictBuffer dictBuffer) { - return dictBuffer.readUnsignedShort(); - } - - protected static int readHeaderSize(final DictBuffer dictBuffer) { - return dictBuffer.readInt(); - } - - protected static HashMap<String, String> readAttributes(final DictBuffer dictBuffer, - final int headerSize) { - final HashMap<String, String> attributes = new HashMap<String, String>(); - while (dictBuffer.position() < headerSize) { - // We can avoid an infinite loop here since dictBuffer.position() is always - // increased by calling CharEncoding.readString. - final String key = CharEncoding.readString(dictBuffer); - final String value = CharEncoding.readString(dictBuffer); - attributes.put(key, value); - } - dictBuffer.position(headerSize); - return attributes; - } - } - - private final static class PtNodeReader { - protected static int readPtNodeOptionFlags(final DictBuffer dictBuffer) { + protected static class PtNodeReader extends DictDecoder.PtNodeReader { + private static int readFrequency(final DictBuffer dictBuffer) { return dictBuffer.readUnsignedByte(); } - - protected static int readParentAddress(final DictBuffer dictBuffer, - final FormatOptions formatOptions) { - if (BinaryDictIOUtils.supportsDynamicUpdate(formatOptions)) { - return BinaryDictDecoderUtils.readSInt24(dictBuffer); - } else { - return FormatSpec.NO_PARENT_ADDRESS; - } - } - - protected static int readFrequency(final DictBuffer dictBuffer) { - return dictBuffer.readUnsignedByte(); - } - - protected static int readChildrenAddress(final DictBuffer dictBuffer, final int optionFlags, - final FormatOptions formatOptions) { - if (BinaryDictIOUtils.supportsDynamicUpdate(formatOptions)) { - final int address = BinaryDictDecoderUtils.readSInt24(dictBuffer); - if (address == 0) return FormatSpec.NO_CHILDREN_ADDRESS; - return address; - } else { - switch (optionFlags & FormatSpec.MASK_CHILDREN_ADDRESS_TYPE) { - case FormatSpec.FLAG_CHILDREN_ADDRESS_TYPE_ONEBYTE: - return dictBuffer.readUnsignedByte(); - case FormatSpec.FLAG_CHILDREN_ADDRESS_TYPE_TWOBYTES: - return dictBuffer.readUnsignedShort(); - case FormatSpec.FLAG_CHILDREN_ADDRESS_TYPE_THREEBYTES: - return dictBuffer.readUnsignedInt24(); - case FormatSpec.FLAG_CHILDREN_ADDRESS_TYPE_NOADDRESS: - default: - return FormatSpec.NO_CHILDREN_ADDRESS; - } - } - } - - // Reads shortcuts and returns the read length. - protected static int readShortcut(final DictBuffer dictBuffer, - final ArrayList<WeightedString> shortcutTargets) { - final int pointerBefore = dictBuffer.position(); - dictBuffer.readUnsignedShort(); // skip the size - while (true) { - final int targetFlags = dictBuffer.readUnsignedByte(); - final String word = CharEncoding.readString(dictBuffer); - shortcutTargets.add(new WeightedString(word, - targetFlags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY)); - if (0 == (targetFlags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_HAS_NEXT)) break; - } - return dictBuffer.position() - pointerBefore; - } - - protected static int readBigrams(final DictBuffer dictBuffer, - final ArrayList<PendingAttribute> bigrams, final int baseAddress) { - int readLength = 0; - int bigramCount = 0; - while (bigramCount++ < FormatSpec.MAX_BIGRAMS_IN_A_PTNODE) { - final int bigramFlags = dictBuffer.readUnsignedByte(); - ++readLength; - final int sign = 0 == (bigramFlags & FormatSpec.FLAG_BIGRAM_ATTR_OFFSET_NEGATIVE) - ? 1 : -1; - int bigramAddress = baseAddress + readLength; - switch (bigramFlags & FormatSpec.MASK_BIGRAM_ATTR_ADDRESS_TYPE) { - case FormatSpec.FLAG_BIGRAM_ATTR_ADDRESS_TYPE_ONEBYTE: - bigramAddress += sign * dictBuffer.readUnsignedByte(); - readLength += 1; - break; - case FormatSpec.FLAG_BIGRAM_ATTR_ADDRESS_TYPE_TWOBYTES: - bigramAddress += sign * dictBuffer.readUnsignedShort(); - readLength += 2; - break; - case FormatSpec.FLAG_BIGRAM_ATTR_ADDRESS_TYPE_THREEBYTES: - final int offset = (dictBuffer.readUnsignedByte() << 16) - + dictBuffer.readUnsignedShort(); - bigramAddress += sign * offset; - readLength += 3; - break; - default: - throw new RuntimeException("Has bigrams with no address"); - } - bigrams.add(new PendingAttribute( - bigramFlags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY, - bigramAddress)); - if (0 == (bigramFlags & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_HAS_NEXT)) break; - } - return readLength; - } } private final File mDictionaryBinaryFile; @@ -199,7 +83,7 @@ public class Ver3DictDecoder implements DictDecoder { } @Override - public boolean isOpenedDictBuffer() { + public boolean isDictBufferOpen() { return mDictBuffer != null; } @@ -218,25 +102,11 @@ public class Ver3DictDecoder implements DictDecoder { if (mDictBuffer == null) { openDictBuffer(); } - - final int version = HeaderReader.readVersion(mDictBuffer); - final int optionsFlags = HeaderReader.readOptionFlags(mDictBuffer); - - final int headerSize = HeaderReader.readHeaderSize(mDictBuffer); - - if (headerSize < 0) { - throw new UnsupportedFormatException("header size can't be negative."); + final FileHeader header = super.readHeader(mDictBuffer); + final int version = header.mFormatOptions.mVersion; + if (!(version >= 2 && version <= 3)) { + throw new UnsupportedFormatException("File header has a wrong version : " + version); } - - final HashMap<String, String> attributes = HeaderReader.readAttributes(mDictBuffer, - headerSize); - - final FileHeader header = new FileHeader(headerSize, - new FusionDictionary.DictionaryOptions(attributes, - 0 != (optionsFlags & FormatSpec.GERMAN_UMLAUT_PROCESSING_FLAG), - 0 != (optionsFlags & FormatSpec.FRENCH_LIGATURE_PROCESSING_FLAG)), - new FormatOptions(version, - 0 != (optionsFlags & FormatSpec.SUPPORTS_DYNAMIC_UPDATE))); return header; } @@ -246,11 +116,11 @@ public class Ver3DictDecoder implements DictDecoder { public PtNodeInfo readPtNode(final int ptNodePos, final FormatOptions options) { int addressPointer = ptNodePos; final int flags = PtNodeReader.readPtNodeOptionFlags(mDictBuffer); - ++addressPointer; + addressPointer += FormatSpec.PTNODE_FLAGS_SIZE; final int parentAddress = PtNodeReader.readParentAddress(mDictBuffer, options); if (BinaryDictIOUtils.supportsDynamicUpdate(options)) { - addressPointer += 3; + addressPointer += FormatSpec.PARENT_ADDRESS_SIZE; } final int characters[]; @@ -258,7 +128,7 @@ public class Ver3DictDecoder implements DictDecoder { int index = 0; int character = CharEncoding.readChar(mDictBuffer); addressPointer += CharEncoding.getCharSize(character); - while (-1 != character) { + while (FormatSpec.INVALID_CHARACTER != character) { // FusionDictionary is making sure that the length of the word is smaller than // MAX_WORD_LENGTH. // So we'll never write past the end of mCharacterBuffer. @@ -274,8 +144,8 @@ public class Ver3DictDecoder implements DictDecoder { } final int frequency; if (0 != (FormatSpec.FLAG_IS_TERMINAL & flags)) { - ++addressPointer; frequency = PtNodeReader.readFrequency(mDictBuffer); + addressPointer += FormatSpec.PTNODE_FREQUENCY_SIZE; } else { frequency = PtNode.NOT_A_TERMINAL; } @@ -296,7 +166,8 @@ public class Ver3DictDecoder implements DictDecoder { final ArrayList<PendingAttribute> bigrams; if (0 != (flags & FormatSpec.FLAG_HAS_BIGRAMS)) { bigrams = new ArrayList<PendingAttribute>(); - addressPointer += PtNodeReader.readBigrams(mDictBuffer, bigrams, addressPointer); + addressPointer += PtNodeReader.readBigramAddresses(mDictBuffer, bigrams, + addressPointer); if (bigrams.size() >= FormatSpec.MAX_BIGRAMS_IN_A_PTNODE) { MakedictLog.d("too many bigrams in a PtNode."); } @@ -332,25 +203,6 @@ public class Ver3DictDecoder implements DictDecoder { } @Override - public int getTerminalPosition(String word) throws IOException, UnsupportedFormatException { - if (mDictBuffer == null) { - openDictBuffer(); - } - return BinaryDictIOUtils.getTerminalPosition(this, word); - } - - @Override - public void readUnigramsAndBigramsBinary(final TreeMap<Integer, String> words, - final TreeMap<Integer, Integer> frequencies, - final TreeMap<Integer, ArrayList<PendingAttribute>> bigrams) - throws IOException, UnsupportedFormatException { - if (mDictBuffer == null) { - openDictBuffer(); - } - BinaryDictIOUtils.readUnigramsAndBigramsBinary(this, words, frequencies, bigrams); - } - - @Override public void setPosition(int newPos) { mDictBuffer.position(newPos); } @@ -366,7 +218,7 @@ public class Ver3DictDecoder implements DictDecoder { } @Override - public boolean readForwardLinkAndAdvancePosition() { + public boolean readAndFollowForwardLink() { final int nextAddress = mDictBuffer.readUnsignedInt24(); if (nextAddress >= 0 && nextAddress < mDictBuffer.limit()) { mDictBuffer.position(nextAddress); |