private Tokenizer MakeTokenizer(string userDictionaryEntry) { using (var stream = MakeUserDictionaryStream(userDictionaryEntry)) { var builder = new Tokenizer.Builder(TestUtils.AbsoluteIpadicResourcePath); builder.LoadUserDictionary(stream); return(new Tokenizer(builder)); } }
private Tokenizer MakeTokenizer() { using (var file = File.OpenRead(Locations.ABS_DICT_COMPILED_PATH + System.IO.Path.DirectorySeparatorChar + "userDict.txt")) { var builder = new Tokenizer.Builder(Locations.ABS_DICT_COMPILED_PATH); builder.LoadUserDictionary(file); builder.IsSplitOnNakaguro = true; return(new Tokenizer(builder)); } }
public void TestFeatureLengths() { string userDictionary = "" + "gsf,gsf,ジーエスーエフ,カスタム名詞\n"; using (var stream = new MemoryStream(Encoding.UTF8.GetBytes(userDictionary))) { var builder = new Tokenizer.Builder(TestUtils.DictResourcedPath); builder.LoadUserDictionary(stream); using (Tokenizer tokenizer = new Tokenizer(builder)) TestUtils.AssertEqualTokenFeatureLengths("ahgsfdajhgsfdこの丘はアクロポリスと呼ばれている。", tokenizer); } }
public void TestMultiThreadedUserDictionary() { var filePath = "./Core/Resource/userdict.txt"; using (var stream = File.OpenRead(filePath)) { using (var builder = new Tokenizer.Builder(TestUtils.AbsoluteIpadicResourcePath)) { builder.LoadUserDictionary(stream); TestUtils.AssertMultiThreadedTokenizedStreamEquals( 5, 10, TestUtils.AbsoluteIpadicResourcePath + "jawikisentences-ipadic-features.txt", TestUtils.AbsoluteIpadicResourcePath + "jawikisentences.txt", new Tokenizer(builder) ); } } }