Reading dictionary containing timestamps in Java Side.

Just skipping historical information fields.

Bug: 11281877
Change-Id: I43d2adaa576b7da11ed3ca54990265dbb6f53b08

3 files changed, 84 insertions, 16 deletions

diff --git a/tests/src/com/android/inputmethod/latin/BinaryDictionaryDecayingTests.java b/tests/src/com/android/inputmethod/latin/BinaryDictionaryDecayingTests.java
index 1dc1f5a1c..c42765633 100644
--- a/tests/src/com/android/inputmethod/latin/BinaryDictionaryDecayingTests.java
+++ b/tests/src/com/android/inputmethod/latin/BinaryDictionaryDecayingTests.java
@@ -21,7 +21,11 @@ import android.test.suitebuilder.annotation.LargeTest;
 import android.util.Pair;
 
 import com.android.inputmethod.latin.makedict.CodePointUtils;
+import com.android.inputmethod.latin.makedict.DictDecoder;
 import com.android.inputmethod.latin.makedict.FormatSpec;
+import com.android.inputmethod.latin.makedict.FusionDictionary;
+import com.android.inputmethod.latin.makedict.FusionDictionary.PtNode;
+import com.android.inputmethod.latin.makedict.UnsupportedFormatException;
 import com.android.inputmethod.latin.utils.FileUtils;
 
 import java.io.File;
@@ -98,6 +102,10 @@ public class BinaryDictionaryDecayingTests extends AndroidTestCase {
                 getContext().getCacheDir());
         FileUtils.deleteRecursively(file);
         Map<String, String> attributeMap = new HashMap<String, String>();
+        attributeMap.put(FormatSpec.FileHeader.DICTIONARY_ID_ATTRIBUTE, dictId);
+        attributeMap.put(FormatSpec.FileHeader.DICTIONARY_LOCALE_ATTRIBUTE, dictId);
+        attributeMap.put(FormatSpec.FileHeader.DICTIONARY_VERSION_ATTRIBUTE,
+                String.valueOf(TimeUnit.MILLISECONDS.toSeconds(System.currentTimeMillis())));
         attributeMap.put(FormatSpec.FileHeader.USES_FORGETTING_CURVE_ATTRIBUTE,
                 FormatSpec.FileHeader.ATTRIBUTE_VALUE_TRUE);
         attributeMap.put(FormatSpec.FileHeader.HAS_HISTORICAL_INFO_ATTRIBUTE,
@@ -119,6 +127,50 @@ public class BinaryDictionaryDecayingTests extends AndroidTestCase {
         return BinaryDictionary.setCurrentTimeForTest(-1);
     }
 
+    public void testReadDictInJavaSide() {
+        testReadDictInJavaSide(FormatSpec.VERSION4);
+    }
+
+    private void testReadDictInJavaSide(final int formatVersion) {
+        setCurrentTimeForTestMode(mCurrentTime);
+        File dictFile = null;
+        try {
+            dictFile = createEmptyDictionaryAndGetFile("TestBinaryDictionary", formatVersion);
+        } catch (IOException e) {
+            fail("IOException while writing an initial dictionary : " + e);
+        }
+        BinaryDictionary binaryDictionary = new BinaryDictionary(dictFile.getAbsolutePath(),
+                0 /* offset */, dictFile.length(), true /* useFullEditDistance */,
+                Locale.getDefault(), TEST_LOCALE, true /* isUpdatable */);
+        addUnigramWord(binaryDictionary, "a", DUMMY_PROBABILITY);
+        addUnigramWord(binaryDictionary, "ab", DUMMY_PROBABILITY);
+        addUnigramWord(binaryDictionary, "aaa", DUMMY_PROBABILITY);
+        addBigramWords(binaryDictionary, "a", "aaa", DUMMY_PROBABILITY);
+        binaryDictionary.flushWithGC();
+        binaryDictionary.close();
+
+        final DictDecoder dictDecoder = FormatSpec.getDictDecoder(dictFile);
+        try {
+            final FusionDictionary dict = dictDecoder.readDictionaryBinary(null,
+                    false /* deleteDictIfBroken */);
+            PtNode ptNode = FusionDictionary.findWordInTree(dict.mRootNodeArray, "a");
+            assertNotNull(ptNode);
+            assertTrue(ptNode.isTerminal());
+            assertNotNull(ptNode.getBigram("aaa"));
+            ptNode = FusionDictionary.findWordInTree(dict.mRootNodeArray, "ab");
+            assertNotNull(ptNode);
+            assertTrue(ptNode.isTerminal());
+            ptNode = FusionDictionary.findWordInTree(dict.mRootNodeArray, "aaa");
+            assertNotNull(ptNode);
+            assertTrue(ptNode.isTerminal());
+        } catch (IOException e) {
+            fail("IOException while reading dictionary: " + e);
+        } catch (UnsupportedFormatException e) {
+            fail("Unsupported format: " + e);
+        }
+        dictFile.delete();
+    }
+
     public void testControlCurrentTime() {
         testControlCurrentTime(FormatSpec.VERSION4);
     }
diff --git a/tests/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderEncoderTests.java b/tests/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderEncoderTests.java
index b5a71f0bf..8a1ac5233 100644
--- a/tests/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderEncoderTests.java
+++ b/tests/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderEncoderTests.java
@@ -17,11 +17,11 @@
 package com.android.inputmethod.latin.makedict;
 
 import android.test.AndroidTestCase;
-import android.test.MoreAsserts;
 import android.test.suitebuilder.annotation.LargeTest;
 import android.util.Log;
 import android.util.SparseArray;
 
+import com.android.inputmethod.latin.BinaryDictionary;
 import com.android.inputmethod.latin.makedict.BinaryDictDecoderUtils.CharEncoding;
 import com.android.inputmethod.latin.makedict.BinaryDictDecoderUtils.DictBuffer;
 import com.android.inputmethod.latin.makedict.FormatSpec.FileHeader;
@@ -76,6 +76,7 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
 
     public BinaryDictDecoderEncoderTests(final long seed, final int maxUnigrams) {
         super();
+        BinaryDictionary.setCurrentTimeForTest(0);
         Log.e(TAG, "Testing dictionary: seed is " + seed);
         final Random random = new Random(seed);
         sWords.clear();
@@ -262,7 +263,7 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
                 getContext().getCacheDir());
 
         final FusionDictionary dict = new FusionDictionary(new PtNodeArray(),
-                BinaryDictUtils.makeDictionaryOptions(dictName, dictVersion));
+                BinaryDictUtils.makeDictionaryOptions(dictName, dictVersion, formatOptions));
         addUnigrams(words.size(), dict, words, shortcuts);
         addBigrams(dict, words, bigrams);
         checkDictionary(dict, words, bigrams, shortcuts);
@@ -317,7 +318,6 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
                 BinaryDictUtils.VERSION4_OPTIONS_WITHOUT_TIMESTAMP);
         runReadAndWriteTests(results, BinaryDictUtils.USE_BYTE_BUFFER,
                 BinaryDictUtils.VERSION4_OPTIONS_WITH_TIMESTAMP);
-
         for (final String result : results) {
             Log.d(TAG, result);
         }
@@ -344,14 +344,16 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
             final SparseArray<List<Integer>> expectedBigrams,
             final TreeMap<Integer, String> resultWords,
             final TreeMap<Integer, Integer> resultFrequencies,
-            final TreeMap<Integer, ArrayList<PendingAttribute>> resultBigrams) {
+            final TreeMap<Integer, ArrayList<PendingAttribute>> resultBigrams,
+            final boolean checkProbability) {
         // check unigrams
         final Set<String> actualWordsSet = new HashSet<String>(resultWords.values());
         final Set<String> expectedWordsSet = new HashSet<String>(expectedWords);
         assertEquals(actualWordsSet, expectedWordsSet);
-
-        for (int freq : resultFrequencies.values()) {
-            assertEquals(freq, UNIGRAM_FREQ);
+        if (checkProbability) {
+            for (int freq : resultFrequencies.values()) {
+                assertEquals(freq, UNIGRAM_FREQ);
+            }
         }
 
         // check bigrams
@@ -377,16 +379,19 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
                 }
                 actBigrams.get(word1).add(word2);
 
-                final int bigramFreq = BinaryDictIOUtils.reconstructBigramFrequency(
-                        unigramFreq, attr.mFrequency);
-                assertTrue(Math.abs(bigramFreq - BIGRAM_FREQ) < TOLERANCE_OF_BIGRAM_FREQ);
+                if (checkProbability) {
+                    final int bigramFreq = BinaryDictIOUtils.reconstructBigramFrequency(
+                            unigramFreq, attr.mFrequency);
+                    assertTrue(Math.abs(bigramFreq - BIGRAM_FREQ) < TOLERANCE_OF_BIGRAM_FREQ);
+                }
             }
         }
         assertEquals(actBigrams, expBigrams);
     }
 
     private long timeAndCheckReadUnigramsAndBigramsBinary(final File file, final List<String> words,
-            final SparseArray<List<Integer>> bigrams, final int bufferType) {
+            final SparseArray<List<Integer>> bigrams, final int bufferType,
+            final boolean checkProbability) {
         final TreeMap<Integer, String> resultWords = CollectionUtils.newTreeMap();
         final TreeMap<Integer, ArrayList<PendingAttribute>> resultBigrams =
                 CollectionUtils.newTreeMap();
@@ -404,7 +409,7 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
             Log.e(TAG, "UnsupportedFormatException", e);
         }
 
-        checkWordMap(words, bigrams, resultWords, resultFreqs, resultBigrams);
+        checkWordMap(words, bigrams, resultWords, resultFreqs, resultBigrams, checkProbability);
         return diff;
     }
 
@@ -418,13 +423,17 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
 
         // making the dictionary from lists of words.
         final FusionDictionary dict = new FusionDictionary(new PtNodeArray(),
-                BinaryDictUtils.makeDictionaryOptions(dictName, dictVersion));
+                BinaryDictUtils.makeDictionaryOptions(dictName, dictVersion, formatOptions));
         addUnigrams(words.size(), dict, words, null /* shortcutMap */);
         addBigrams(dict, words, bigrams);
 
         timeWritingDictToFile(file, dict, formatOptions);
 
-        long wordMap = timeAndCheckReadUnigramsAndBigramsBinary(file, words, bigrams, bufferType);
+        // Caveat: Currently, the Java code to read a v4 dictionary doesn't calculate the
+        // probability when there's a timestamp for the entry.
+        // TODO: Abandon the Java code, and implement the v4 dictionary reading code in native.
+        long wordMap = timeAndCheckReadUnigramsAndBigramsBinary(file, words, bigrams, bufferType,
+                !formatOptions.mHasTimestamp /* checkProbability */);
         long fullReading = timeReadingAndCheckDict(file, words, bigrams, null /* shortcutMap */,
                 bufferType);
 
@@ -517,7 +526,7 @@ public class BinaryDictDecoderEncoderTests extends AndroidTestCase {
                 getContext().getCacheDir());
 
         final FusionDictionary dict = new FusionDictionary(new PtNodeArray(),
-                BinaryDictUtils.makeDictionaryOptions(dictName, dictVersion));
+                BinaryDictUtils.makeDictionaryOptions(dictName, dictVersion, formatOptions));
         addUnigrams(sWords.size(), dict, sWords, null /* shortcutMap */);
         addBigrams(dict, words, bigrams);
         timeWritingDictToFile(file, dict, formatOptions);
diff --git a/tests/src/com/android/inputmethod/latin/makedict/BinaryDictUtils.java b/tests/src/com/android/inputmethod/latin/makedict/BinaryDictUtils.java
index f7a808c1e..f17596865 100644
--- a/tests/src/com/android/inputmethod/latin/makedict/BinaryDictUtils.java
+++ b/tests/src/com/android/inputmethod/latin/makedict/BinaryDictUtils.java
@@ -36,11 +36,18 @@ public class BinaryDictUtils {
     public static final FormatSpec.FormatOptions VERSION4_OPTIONS_WITH_TIMESTAMP =
             new FormatSpec.FormatOptions(FormatSpec.VERSION4, true /* hasTimestamp */);
 
-    public static DictionaryOptions makeDictionaryOptions(final String id, final String version) {
+    public static DictionaryOptions makeDictionaryOptions(final String id, final String version,
+            final FormatSpec.FormatOptions formatOptions) {
         final DictionaryOptions options = new DictionaryOptions(new HashMap<String, String>());
         options.mAttributes.put(FileHeader.DICTIONARY_LOCALE_ATTRIBUTE, "en_US");
         options.mAttributes.put(FileHeader.DICTIONARY_ID_ATTRIBUTE, id);
         options.mAttributes.put(FileHeader.DICTIONARY_VERSION_ATTRIBUTE, version);
+        if (formatOptions.mHasTimestamp) {
+            options.mAttributes.put(FileHeader.HAS_HISTORICAL_INFO_ATTRIBUTE,
+                    FileHeader.ATTRIBUTE_VALUE_TRUE);
+            options.mAttributes.put(FileHeader.USES_FORGETTING_CURVE_ATTRIBUTE,
+                    FileHeader.ATTRIBUTE_VALUE_TRUE);
+        }
         return options;
     }