diff options
author | 2014-02-03 22:22:35 +0900 | |
---|---|---|
committer | 2014-02-04 12:42:14 +0900 | |
commit | 941734695b9eeb59135db737e4b153c45e88247a (patch) | |
tree | c905b75a36d35f7e0bc45882babaccd7923a6566 /tests/src/com/android/inputmethod/latin/BinaryDictionaryTests.java | |
parent | 0251c60b4434f702623803b454ac59e8ff2efc4a (diff) | |
download | latinime-941734695b9eeb59135db737e4b153c45e88247a.tar.gz latinime-941734695b9eeb59135db737e4b153c45e88247a.tar.xz latinime-941734695b9eeb59135db737e4b153c45e88247a.zip |
Implement Ver4PatriciaTriePolicy::getNextWordAndNextToken.
Bug: 12810574
Change-Id: Idea44f03c477964f58e65fbf2b55e3fcd77a2934
Diffstat (limited to 'tests/src/com/android/inputmethod/latin/BinaryDictionaryTests.java')
-rw-r--r-- | tests/src/com/android/inputmethod/latin/BinaryDictionaryTests.java | 93 |
1 files changed, 93 insertions, 0 deletions
diff --git a/tests/src/com/android/inputmethod/latin/BinaryDictionaryTests.java b/tests/src/com/android/inputmethod/latin/BinaryDictionaryTests.java index e39b46f94..bab86e546 100644 --- a/tests/src/com/android/inputmethod/latin/BinaryDictionaryTests.java +++ b/tests/src/com/android/inputmethod/latin/BinaryDictionaryTests.java @@ -971,6 +971,99 @@ public class BinaryDictionaryTests extends AndroidTestCase { } } + public void testIterateAllWords() { + testIterateAllWords(FormatSpec.VERSION4); + } + + private void testIterateAllWords(final int formatVersion) { + final long seed = System.currentTimeMillis(); + final Random random = new Random(seed); + final int UNIGRAM_COUNT = 1000; + final int BIGRAM_COUNT = 1000; + final int codePointSetSize = 20; + final int[] codePointSet = CodePointUtils.generateCodePointSet(codePointSetSize, random); + + File dictFile = null; + try { + dictFile = createEmptyDictionaryAndGetFile("TestBinaryDictionary", formatVersion); + } catch (IOException e) { + fail("IOException while writing an initial dictionary : " + e); + } + final BinaryDictionary binaryDictionary = new BinaryDictionary(dictFile.getAbsolutePath(), + 0 /* offset */, dictFile.length(), true /* useFullEditDistance */, + Locale.getDefault(), TEST_LOCALE, true /* isUpdatable */); + + final WordProperty invalidWordProperty = binaryDictionary.getWordProperty("dummyWord"); + assertFalse(invalidWordProperty.isValid()); + + final ArrayList<String> words = new ArrayList<String>(); + final HashMap<String, Integer> wordProbabilitiesToCheckLater = + new HashMap<String, Integer>(); + final HashMap<String, HashSet<String>> bigrams = new HashMap<String, HashSet<String>>(); + final HashMap<Pair<String, String>, Integer> bigramProbabilitiesToCheckLater = + new HashMap<Pair<String, String>, Integer>(); + + for (int i = 0; i < UNIGRAM_COUNT; i++) { + final String word = CodePointUtils.generateWord(random, codePointSet); + final int unigramProbability = random.nextInt(0xFF); + addUnigramWord(binaryDictionary, word, unigramProbability); + if (binaryDictionary.needsToRunGC(false /* mindsBlockByGC */)) { + binaryDictionary.flushWithGC(); + } + words.add(word); + wordProbabilitiesToCheckLater.put(word, unigramProbability); + } + + for (int i = 0; i < BIGRAM_COUNT; i++) { + final int word0Index = random.nextInt(wordProbabilitiesToCheckLater.size()); + final int word1Index = random.nextInt(wordProbabilitiesToCheckLater.size()); + if (word0Index == word1Index) { + continue; + } + final String word0 = words.get(word0Index); + final String word1 = words.get(word1Index); + final int bigramProbability = random.nextInt(0xF); + binaryDictionary.addBigramWords(word0, word1, bigramProbability, + BinaryDictionary.NOT_A_VALID_TIMESTAMP); + if (binaryDictionary.needsToRunGC(false /* mindsBlockByGC */)) { + binaryDictionary.flushWithGC(); + } + if (!bigrams.containsKey(word0)) { + final HashSet<String> bigramWord1s = new HashSet<String>(); + bigrams.put(word0, bigramWord1s); + } + bigrams.get(word0).add(word1); + bigramProbabilitiesToCheckLater.put( + new Pair<String, String>(word0, word1), bigramProbability); + } + + final HashSet<String> wordSet = new HashSet<String>(words); + final HashSet<Pair<String, String>> bigramSet = + new HashSet<Pair<String,String>>(bigramProbabilitiesToCheckLater.keySet()); + int token = 0; + do { + final BinaryDictionary.GetNextWordPropertyResult result = + binaryDictionary.getNextWordProperty(token); + final WordProperty wordProperty = result.mWordProperty; + final String word0 = wordProperty.mCodePoints; + assertEquals((int)wordProbabilitiesToCheckLater.get(word0), + wordProperty.mProbabilityInfo.mProbability); + wordSet.remove(word0); + final HashSet<String> bigramWord1s = bigrams.get(word0); + for (int j = 0; j < wordProperty.mBigramTargets.size(); j++) { + final String word1 = wordProperty.mBigramTargets.get(j).mWord; + assertTrue(bigramWord1s.contains(word1)); + final int probability = wordProperty.mBigramTargets.get(j).mFrequency; + final Pair<String, String> bigram = new Pair<String, String>(word0, word1); + assertEquals((int)bigramProbabilitiesToCheckLater.get(bigram), probability); + bigramSet.remove(bigram); + } + token = result.mNextToken; + } while (token != 0); + assertTrue(wordSet.isEmpty()); + assertTrue(bigramSet.isEmpty()); + } + public void testAddShortcuts() { testAddShortcuts(FormatSpec.VERSION4); } |