am 3a1a94c3
: Merge "Add shortcut support to the in-memory format of makedict (B3)"
* commit '3a1a94c3cb4801dfe3985ab2b171e9a54b2d2ad2': Add shortcut support to the in-memory format of makedict (B3)
This commit is contained in:
commit
bfe8ed28cd
4 changed files with 56 additions and 27 deletions
|
@ -44,8 +44,9 @@ public class BinaryDictInputOutput {
|
|||
* a | 11 = 3 bytes : FLAG_GROUP_ADDRESS_TYPE_THREEBYTES
|
||||
* g | has several chars ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_MULTIPLE_CHARS
|
||||
* s | has a terminal ? 1 bit, 1 = yes, 0 = no : FLAG_IS_TERMINAL
|
||||
* | reserved 1 bit, 1 = yes, 0 = no
|
||||
* | has shortcut targets ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_SHORTCUT_TARGETS
|
||||
* | has bigrams ? 1 bit, 1 = yes, 0 = no : FLAG_HAS_BIGRAMS
|
||||
* | is shortcut only ? 1 bit, 1 = yes, 0 = no : FLAG_IS_SHORTCUT_ONLY
|
||||
*
|
||||
* c | IF FLAG_HAS_MULTIPLE_CHARS
|
||||
* h | char, char, char, char n * (1 or 3 bytes) : use CharGroupInfo for i/o helpers
|
||||
|
@ -71,6 +72,8 @@ public class BinaryDictInputOutput {
|
|||
* d
|
||||
* dress
|
||||
*
|
||||
* | IF FLAG_IS_TERMINAL && FLAG_HAS_SHORTCUT_TARGETS
|
||||
* | shortcut targets address list
|
||||
* | IF FLAG_IS_TERMINAL && FLAG_HAS_BIGRAMS
|
||||
* | bigrams address list
|
||||
*
|
||||
|
@ -126,7 +129,9 @@ public class BinaryDictInputOutput {
|
|||
private static final int FLAG_HAS_MULTIPLE_CHARS = 0x20;
|
||||
|
||||
private static final int FLAG_IS_TERMINAL = 0x10;
|
||||
private static final int FLAG_HAS_SHORTCUT_TARGETS = 0x08;
|
||||
private static final int FLAG_HAS_BIGRAMS = 0x04;
|
||||
private static final int FLAG_IS_SHORTCUT_ONLY = 0x02;
|
||||
|
||||
private static final int FLAG_ATTRIBUTE_HAS_NEXT = 0x80;
|
||||
private static final int FLAG_ATTRIBUTE_OFFSET_NEGATIVE = 0x40;
|
||||
|
@ -942,11 +947,13 @@ public class BinaryDictInputOutput {
|
|||
source.seek(currentPosition);
|
||||
}
|
||||
nodeContents.add(
|
||||
new CharGroup(info.mCharacters, bigrams, info.mFrequency,
|
||||
// TODO: read and pass the shortcut targets
|
||||
new CharGroup(info.mCharacters, null, bigrams, info.mFrequency,
|
||||
children));
|
||||
} else {
|
||||
// TODO: read and pass the shortcut targets
|
||||
nodeContents.add(
|
||||
new CharGroup(info.mCharacters, bigrams, info.mFrequency));
|
||||
new CharGroup(info.mCharacters, null, bigrams, info.mFrequency));
|
||||
}
|
||||
groupOffset = info.mEndAddress;
|
||||
}
|
||||
|
@ -996,7 +1003,8 @@ public class BinaryDictInputOutput {
|
|||
new FusionDictionary.DictionaryOptions());
|
||||
if (null != dict) {
|
||||
for (Word w : dict) {
|
||||
newDict.add(w.mWord, w.mFrequency, w.mBigrams);
|
||||
// TODO: pass the shortcut targets
|
||||
newDict.add(w.mWord, w.mFrequency, null, w.mBigrams);
|
||||
}
|
||||
}
|
||||
|
||||
|
|
|
@ -68,7 +68,7 @@ public class FusionDictionary implements Iterable<Word> {
|
|||
}
|
||||
|
||||
/**
|
||||
* A group of characters, with a frequency, shortcuts, bigrams, and children.
|
||||
* A group of characters, with a frequency, shortcut targets, bigrams, and children.
|
||||
*
|
||||
* This is the central class of the in-memory representation. A CharGroup is what can
|
||||
* be seen as a traditional "trie node", except it can hold several characters at the
|
||||
|
@ -82,6 +82,7 @@ public class FusionDictionary implements Iterable<Word> {
|
|||
public static class CharGroup {
|
||||
public static final int NOT_A_TERMINAL = -1;
|
||||
final int mChars[];
|
||||
final ArrayList<WeightedString> mShortcutTargets;
|
||||
final ArrayList<WeightedString> mBigrams;
|
||||
final int mFrequency; // NOT_A_TERMINAL == mFrequency indicates this is not a terminal.
|
||||
Node mChildren;
|
||||
|
@ -89,18 +90,20 @@ public class FusionDictionary implements Iterable<Word> {
|
|||
int mCachedSize;
|
||||
int mCachedAddress;
|
||||
|
||||
public CharGroup(final int[] chars,
|
||||
public CharGroup(final int[] chars, final ArrayList<WeightedString> shortcutTargets,
|
||||
final ArrayList<WeightedString> bigrams, final int frequency) {
|
||||
mChars = chars;
|
||||
mFrequency = frequency;
|
||||
mShortcutTargets = shortcutTargets;
|
||||
mBigrams = bigrams;
|
||||
mChildren = null;
|
||||
}
|
||||
|
||||
public CharGroup(final int[] chars,
|
||||
public CharGroup(final int[] chars, final ArrayList<WeightedString> shortcutTargets,
|
||||
final ArrayList<WeightedString> bigrams, final int frequency, final Node children) {
|
||||
mChars = chars;
|
||||
mFrequency = frequency;
|
||||
mShortcutTargets = shortcutTargets;
|
||||
mBigrams = bigrams;
|
||||
mChildren = children;
|
||||
}
|
||||
|
@ -165,18 +168,29 @@ public class FusionDictionary implements Iterable<Word> {
|
|||
*
|
||||
* @param word the word to add.
|
||||
* @param frequency the frequency of the word, in the range [0..255].
|
||||
* @param shortcutTargets a list of shortcut targets for this word, or null.
|
||||
* @param bigrams a list of bigrams, or null.
|
||||
*/
|
||||
public void add(String word, int frequency, ArrayList<WeightedString> bigrams) {
|
||||
public void add(final String word, final int frequency,
|
||||
final ArrayList<WeightedString> shortcutTargets,
|
||||
final ArrayList<WeightedString> bigrams) {
|
||||
if (null != shortcutTargets) {
|
||||
for (WeightedString target : shortcutTargets) {
|
||||
final CharGroup t = findWordInTree(mRoot, target.mWord);
|
||||
if (null == t) {
|
||||
add(getCodePoints(target.mWord), 0, null, null);
|
||||
}
|
||||
}
|
||||
}
|
||||
if (null != bigrams) {
|
||||
for (WeightedString bigram : bigrams) {
|
||||
final CharGroup t = findWordInTree(mRoot, bigram.mWord);
|
||||
if (null == t) {
|
||||
add(getCodePoints(bigram.mWord), 0, null);
|
||||
add(getCodePoints(bigram.mWord), 0, null, null);
|
||||
}
|
||||
}
|
||||
}
|
||||
add(getCodePoints(word), frequency, bigrams);
|
||||
add(getCodePoints(word), frequency, shortcutTargets, bigrams);
|
||||
}
|
||||
|
||||
/**
|
||||
|
@ -200,14 +214,17 @@ public class FusionDictionary implements Iterable<Word> {
|
|||
/**
|
||||
* Add a word to this dictionary.
|
||||
*
|
||||
* The bigrams, if any, have to be in the dictionary already. If they aren't,
|
||||
* The shortcuts and bigrams, if any, have to be in the dictionary already. If they aren't,
|
||||
* an exception is thrown.
|
||||
*
|
||||
* @param word the word, as an int array.
|
||||
* @param frequency the frequency of the word, in the range [0..255].
|
||||
* @param shortcutTargets an optional list of shortcut targets for this word (null if none).
|
||||
* @param bigrams an optional list of bigrams for this word (null if none).
|
||||
*/
|
||||
private void add(int[] word, int frequency, ArrayList<WeightedString> bigrams) {
|
||||
private void add(final int[] word, final int frequency,
|
||||
final ArrayList<WeightedString> shortcutTargets,
|
||||
final ArrayList<WeightedString> bigrams) {
|
||||
assert(frequency >= 0 && frequency <= 255);
|
||||
Node currentNode = mRoot;
|
||||
int charIndex = 0;
|
||||
|
@ -231,7 +248,8 @@ public class FusionDictionary implements Iterable<Word> {
|
|||
// No node at this point to accept the word. Create one.
|
||||
final int insertionIndex = findInsertionIndex(currentNode, word[charIndex]);
|
||||
final CharGroup newGroup = new CharGroup(
|
||||
Arrays.copyOfRange(word, charIndex, word.length), bigrams, frequency);
|
||||
Arrays.copyOfRange(word, charIndex, word.length),
|
||||
shortcutTargets, bigrams, frequency);
|
||||
currentNode.mData.add(insertionIndex, newGroup);
|
||||
checkStack(currentNode);
|
||||
} else {
|
||||
|
@ -245,7 +263,7 @@ public class FusionDictionary implements Iterable<Word> {
|
|||
+ new String(word, 0, word.length));
|
||||
} else {
|
||||
final CharGroup newNode = new CharGroup(currentGroup.mChars,
|
||||
bigrams, frequency, currentGroup.mChildren);
|
||||
shortcutTargets, bigrams, frequency, currentGroup.mChildren);
|
||||
currentNode.mData.set(nodeIndex, newNode);
|
||||
checkStack(currentNode);
|
||||
}
|
||||
|
@ -254,7 +272,7 @@ public class FusionDictionary implements Iterable<Word> {
|
|||
// We only have to create a new node and add it to the end of this.
|
||||
final CharGroup newNode = new CharGroup(
|
||||
Arrays.copyOfRange(word, charIndex + differentCharIndex, word.length),
|
||||
bigrams, frequency);
|
||||
shortcutTargets, bigrams, frequency);
|
||||
currentGroup.mChildren = new Node();
|
||||
currentGroup.mChildren.mData.add(newNode);
|
||||
}
|
||||
|
@ -268,7 +286,8 @@ public class FusionDictionary implements Iterable<Word> {
|
|||
+ new String(word, 0, word.length));
|
||||
}
|
||||
final CharGroup newGroup = new CharGroup(word,
|
||||
currentGroup.mBigrams, frequency, currentGroup.mChildren);
|
||||
currentGroup.mShortcutTargets, currentGroup.mBigrams,
|
||||
frequency, currentGroup.mChildren);
|
||||
currentNode.mData.set(nodeIndex, newGroup);
|
||||
}
|
||||
} else {
|
||||
|
@ -277,7 +296,7 @@ public class FusionDictionary implements Iterable<Word> {
|
|||
Node newChildren = new Node();
|
||||
final CharGroup newOldWord = new CharGroup(
|
||||
Arrays.copyOfRange(currentGroup.mChars, differentCharIndex,
|
||||
currentGroup.mChars.length),
|
||||
currentGroup.mChars.length), currentGroup.mShortcutTargets,
|
||||
currentGroup.mBigrams, currentGroup.mFrequency, currentGroup.mChildren);
|
||||
newChildren.mData.add(newOldWord);
|
||||
|
||||
|
@ -285,14 +304,14 @@ public class FusionDictionary implements Iterable<Word> {
|
|||
if (charIndex + differentCharIndex >= word.length) {
|
||||
newParent = new CharGroup(
|
||||
Arrays.copyOfRange(currentGroup.mChars, 0, differentCharIndex),
|
||||
bigrams, frequency, newChildren);
|
||||
shortcutTargets, bigrams, frequency, newChildren);
|
||||
} else {
|
||||
newParent = new CharGroup(
|
||||
Arrays.copyOfRange(currentGroup.mChars, 0, differentCharIndex),
|
||||
null, -1, newChildren);
|
||||
null, null, -1, newChildren);
|
||||
final CharGroup newWord = new CharGroup(
|
||||
Arrays.copyOfRange(word, charIndex + differentCharIndex,
|
||||
word.length), bigrams, frequency);
|
||||
word.length), shortcutTargets, bigrams, frequency);
|
||||
final int addIndex = word[charIndex + differentCharIndex]
|
||||
> currentGroup.mChars[differentCharIndex] ? 1 : 0;
|
||||
newChildren.mData.add(addIndex, newWord);
|
||||
|
@ -355,7 +374,7 @@ public class FusionDictionary implements Iterable<Word> {
|
|||
*/
|
||||
private static int findInsertionIndex(final Node node, int character) {
|
||||
final List data = node.mData;
|
||||
final CharGroup reference = new CharGroup(new int[] { character }, null, 0);
|
||||
final CharGroup reference = new CharGroup(new int[] { character }, null, null, 0);
|
||||
int result = Collections.binarySearch(data, reference, CHARGROUP_COMPARATOR);
|
||||
return result >= 0 ? result : -result - 1;
|
||||
}
|
||||
|
@ -573,6 +592,7 @@ public class FusionDictionary implements Iterable<Word> {
|
|||
}
|
||||
if (currentGroup.mFrequency >= 0)
|
||||
return new Word(mCurrentString.toString(), currentGroup.mFrequency,
|
||||
// TODO: pass the shortcut targets here
|
||||
currentGroup.mBigrams);
|
||||
} else {
|
||||
mPositions.removeLast();
|
||||
|
|
|
@ -107,7 +107,8 @@ public class XmlDictInputOutput {
|
|||
@Override
|
||||
public void endElement(String uri, String localName, String qName) {
|
||||
if (WORD == mState) {
|
||||
mDictionary.add(mWord, mFreq, mBigramsMap.get(mWord));
|
||||
// TODO: pass the shortcut targets
|
||||
mDictionary.add(mWord, mFreq, null, mBigramsMap.get(mWord));
|
||||
mState = START;
|
||||
}
|
||||
}
|
||||
|
|
|
@ -39,11 +39,11 @@ public class BinaryDictInputOutputTest extends TestCase {
|
|||
// that it does not contain any duplicates.
|
||||
public void testFlattenNodes() {
|
||||
final FusionDictionary dict = new FusionDictionary();
|
||||
dict.add("foo", 1, null);
|
||||
dict.add("fta", 1, null);
|
||||
dict.add("ftb", 1, null);
|
||||
dict.add("bar", 1, null);
|
||||
dict.add("fool", 1, null);
|
||||
dict.add("foo", 1, null, null);
|
||||
dict.add("fta", 1, null, null);
|
||||
dict.add("ftb", 1, null, null);
|
||||
dict.add("bar", 1, null, null);
|
||||
dict.add("fool", 1, null, null);
|
||||
final ArrayList<Node> result = BinaryDictInputOutput.flattenTree(dict.mRoot);
|
||||
assertEquals(4, result.size());
|
||||
while (!result.isEmpty()) {
|
||||
|
|
Loading…
Reference in a new issue