diff options
Diffstat (limited to 'java/src/com/android/inputmethod')
6 files changed, 51 insertions, 23 deletions
diff --git a/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java b/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java index 369184573..9f2345962 100644 --- a/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java +++ b/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java @@ -511,7 +511,7 @@ public final class BinaryDictDecoderUtils { final WeightedString word = getWordAtPosition(dictDecoder, headerSize, bigram.mAddress, options); final int reconstructedFrequency = - BinaryDictIOUtils.reconstructBigramFrequency(word.mFrequency, + BinaryDictIOUtils.reconstructBigramFrequency(word.getProbability(), bigram.mFrequency); bigrams.add(new WeightedString(word.mWord, reconstructedFrequency)); } @@ -618,7 +618,7 @@ public final class BinaryDictDecoderUtils { // words that are not also registered as unigrams so we don't have to avoid // them explicitly here. for (final WeightedString bigram : w.mBigrams) { - newDict.setBigram(w.mWord, bigram.mWord, bigram.mFrequency); + newDict.setBigram(w.mWord, bigram.mWord, bigram.getProbability()); } } } diff --git a/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java b/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java index 5b0e8399a..ef23acb71 100644 --- a/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java +++ b/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java @@ -67,29 +67,40 @@ public final class FusionDictionary implements Iterable<Word> { } /** - * A string with a frequency. + * A string with a probability. * * This represents an "attribute", that is either a bigram or a shortcut. */ public static final class WeightedString { public final String mWord; - public int mFrequency; - public WeightedString(String word, int frequency) { + public ProbabilityInfo mProbabilityInfo; + + public WeightedString(final String word, final int probability) { mWord = word; - mFrequency = frequency; + mProbabilityInfo = new ProbabilityInfo(probability); + } + + public int getProbability() { + return mProbabilityInfo.mProbability; + } + + public void setProbability(final int probability) { + mProbabilityInfo = new ProbabilityInfo(probability); } @Override public int hashCode() { - return Arrays.hashCode(new Object[] { mWord, mFrequency }); + return Arrays.hashCode(new Object[] { mWord, mProbabilityInfo.mProbability, + mProbabilityInfo.mTimestamp, mProbabilityInfo.mLevel, + mProbabilityInfo.mCount }); } @Override public boolean equals(Object o) { if (o == this) return true; if (!(o instanceof WeightedString)) return false; - WeightedString w = (WeightedString)o; - return mWord.equals(w.mWord) && mFrequency == w.mFrequency; + final WeightedString w = (WeightedString)o; + return mWord.equals(w.mWord) && mProbabilityInfo.equals(w.mProbabilityInfo); } } @@ -200,18 +211,18 @@ public final class FusionDictionary implements Iterable<Word> { } /** - * Adds a word to the bigram list. Updates the frequency if the word already + * Adds a word to the bigram list. Updates the probability if the word already * exists. */ - public void addBigram(final String word, final int frequency) { + public void addBigram(final String word, final int probability) { if (mBigrams == null) { mBigrams = new ArrayList<WeightedString>(); } WeightedString bigram = getBigram(word); if (bigram != null) { - bigram.mFrequency = frequency; + bigram.setProbability(probability); } else { - bigram = new WeightedString(word, frequency); + bigram = new WeightedString(word, probability); mBigrams.add(bigram); } } @@ -273,8 +284,8 @@ public final class FusionDictionary implements Iterable<Word> { final WeightedString existingShortcut = getShortcut(shortcut.mWord); if (existingShortcut == null) { mShortcutTargets.add(shortcut); - } else if (existingShortcut.mFrequency < shortcut.mFrequency) { - existingShortcut.mFrequency = shortcut.mFrequency; + } else if (existingShortcut.getProbability() < shortcut.getProbability()) { + existingShortcut.setProbability(shortcut.getProbability()); } } } @@ -289,8 +300,8 @@ public final class FusionDictionary implements Iterable<Word> { final WeightedString existingBigram = getBigram(bigram.mWord); if (existingBigram == null) { mBigrams.add(bigram); - } else if (existingBigram.mFrequency < bigram.mFrequency) { - existingBigram.mFrequency = bigram.mFrequency; + } else if (existingBigram.getProbability() < bigram.getProbability()) { + existingBigram.setProbability(bigram.getProbability()); } } } diff --git a/java/src/com/android/inputmethod/latin/makedict/ProbabilityInfo.java b/java/src/com/android/inputmethod/latin/makedict/ProbabilityInfo.java index c1a43cedf..79f924cc6 100644 --- a/java/src/com/android/inputmethod/latin/makedict/ProbabilityInfo.java +++ b/java/src/com/android/inputmethod/latin/makedict/ProbabilityInfo.java @@ -17,6 +17,7 @@ package com.android.inputmethod.latin.makedict; import com.android.inputmethod.latin.BinaryDictionary; +import com.android.inputmethod.latin.makedict.FusionDictionary.WeightedString; public final class ProbabilityInfo { public final int mProbability; @@ -39,8 +40,24 @@ public final class ProbabilityInfo { mCount = count; } + public boolean hasHistoricalInfo() { + return mTimestamp != BinaryDictionary.NOT_A_VALID_TIMESTAMP; + } + @Override public String toString() { return mTimestamp + ":" + mLevel + ":" + mCount; } + + @Override + public boolean equals(Object o) { + if (o == this) return true; + if (!(o instanceof ProbabilityInfo)) return false; + final ProbabilityInfo p = (ProbabilityInfo)o; + if (!hasHistoricalInfo() && !p.hasHistoricalInfo()) { + return mProbability == p.mProbability; + } + return mProbability == p.mProbability && mTimestamp == p.mTimestamp && mLevel == p.mLevel + && mCount == p.mCount; + } }
\ No newline at end of file diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver2DictEncoder.java b/java/src/com/android/inputmethod/latin/makedict/Ver2DictEncoder.java index a3a6c2c34..3de083ef3 100644 --- a/java/src/com/android/inputmethod/latin/makedict/Ver2DictEncoder.java +++ b/java/src/com/android/inputmethod/latin/makedict/Ver2DictEncoder.java @@ -197,7 +197,7 @@ public class Ver2DictEncoder implements DictEncoder { final WeightedString target = shortcutIterator.next(); final int shortcutFlags = BinaryDictEncoderUtils.makeShortcutFlags( shortcutIterator.hasNext(), - target.mFrequency); + target.getProbability()); mPosition = BinaryDictEncoderUtils.writeUIntToBuffer(mBuffer, mPosition, shortcutFlags, FormatSpec.PTNODE_ATTRIBUTE_FLAGS_SIZE); final int shortcutShift = CharEncoding.writeString(mBuffer, mPosition, target.mWord); @@ -231,7 +231,7 @@ public class Ver2DictEncoder implements DictEncoder { final int offset = addressOfBigram - (mPosition + FormatSpec.PTNODE_ATTRIBUTE_FLAGS_SIZE); final int bigramFlags = BinaryDictEncoderUtils.makeBigramFlags(bigramIterator.hasNext(), - offset, bigram.mFrequency, unigramFrequencyForThisWord, bigram.mWord); + offset, bigram.getProbability(), unigramFrequencyForThisWord, bigram.mWord); mPosition = BinaryDictEncoderUtils.writeUIntToBuffer(mBuffer, mPosition, bigramFlags, FormatSpec.PTNODE_ATTRIBUTE_FLAGS_SIZE); mPosition += BinaryDictEncoderUtils.writeChildrenPosition(mBuffer, mPosition, diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver4DictEncoder.java b/java/src/com/android/inputmethod/latin/makedict/Ver4DictEncoder.java index 4b0efbad0..a5b013586 100644 --- a/java/src/com/android/inputmethod/latin/makedict/Ver4DictEncoder.java +++ b/java/src/com/android/inputmethod/latin/makedict/Ver4DictEncoder.java @@ -80,7 +80,7 @@ public class Ver4DictEncoder implements DictEncoder { } else { for (final WeightedString shortcutTarget : word.mShortcutTargets) { binaryDict.addUnigramWord(word.mWord, word.mFrequency, - shortcutTarget.mWord, shortcutTarget.mFrequency, + shortcutTarget.mWord, shortcutTarget.getProbability(), word.mIsNotAWord, word.mIsBlacklistEntry, 0 /* timestamp */); } } @@ -91,7 +91,7 @@ public class Ver4DictEncoder implements DictEncoder { for (final Word word0 : dict) { if (null == word0.mBigrams) continue; for (final WeightedString word1 : word0.mBigrams) { - binaryDict.addBigramWords(word0.mWord, word1.mWord, word1.mFrequency, + binaryDict.addBigramWords(word0.mWord, word1.mWord, word1.getProbability(), 0 /* timestamp */); if (binaryDict.needsToRunGC(true /* mindsBlockByGC */)) { binaryDict.flushWithGC(); diff --git a/java/src/com/android/inputmethod/latin/utils/WordProperty.java b/java/src/com/android/inputmethod/latin/utils/WordProperty.java index da56b213f..37d1102e3 100644 --- a/java/src/com/android/inputmethod/latin/utils/WordProperty.java +++ b/java/src/com/android/inputmethod/latin/utils/WordProperty.java @@ -108,7 +108,7 @@ public class WordProperty { for (int i = 0; i < mBigramTargets.size(); i++) { builder.append(" bigram=" + mBigramTargets.get(i).mWord); builder.append(","); - builder.append("f=" + mBigramTargets.get(i).mFrequency); + builder.append("f=" + mBigramTargets.get(i).getProbability()); if (mBigramProbabilityInfo.get(i).mTimestamp != BinaryDictionary.NOT_A_VALID_TIMESTAMP) { builder.append(","); @@ -119,7 +119,7 @@ public class WordProperty { for (int i = 0; i < mShortcutTargets.size(); i++) { builder.append(" shortcut=" + mShortcutTargets.get(i).mWord); builder.append(","); - builder.append("f=" + mShortcutTargets.get(i).mFrequency); + builder.append("f=" + mShortcutTargets.get(i).getProbability()); builder.append("\n"); } return builder.toString(); |