OLD | NEW |
---|---|
1 // Copyright 2014 The Chromium Authors. All rights reserved. | 1 // Copyright 2014 The Chromium Authors. All rights reserved. |
2 // Use of this source code is governed by a BSD-style license that can be | 2 // Use of this source code is governed by a BSD-style license that can be |
3 // found in the LICENSE file. | 3 // found in the LICENSE file. |
4 | 4 |
5 package org.chromium.distiller; | 5 package org.chromium.distiller; |
6 | 6 |
7 import com.google.gwt.regexp.shared.RegExp; | 7 import com.google.gwt.regexp.shared.RegExp; |
8 | 8 |
9 public class StringUtilTest extends JsTestCase { | 9 public class StringUtilTest extends JsTestCase { |
10 public void testCountWords() { | 10 public void testCountWords() { |
11 assertEquals(0, StringUtil.countWords("")); | 11 assertEquals(0, StringUtil.countWords("")); |
12 assertEquals(0, StringUtil.countWords(" -@# ';]")); | 12 assertEquals(0, StringUtil.countWords(" -@# ';]")); |
13 assertEquals(1, StringUtil.countWords("word")); | 13 assertEquals(1, StringUtil.countWords("word")); |
14 assertEquals(1, StringUtil.countWords("b'fore")); | 14 assertEquals(1, StringUtil.countWords("b'fore")); |
15 assertEquals(1, StringUtil.countWords(" _word.under_score_ ")); | 15 assertEquals(1, StringUtil.countWords(" _word.under_score_ ")); |
16 assertEquals(2, StringUtil.countWords(" \ttwo\nwords")); | 16 assertEquals(2, StringUtil.countWords(" \ttwo\nwords")); |
17 assertEquals(2, StringUtil.countWords(" \ttwo @^@^&(@#$([][;;\nwords")); | 17 assertEquals(2, StringUtil.countWords(" \ttwo @^@^&(@#$([][;;\nwords")); |
18 assertEquals(5, StringUtil.countWords("dør når på svært dårlig")); | 18 assertEquals(5, StringUtil.countWords("dør når på svært dårlig")); |
19 assertEquals(5, StringUtil.countWords("svært få dør av blåbærsyltetøy")) ; | 19 assertEquals(5, StringUtil.countWords("svært få dør av blåbærsyltetøy")) ; |
20 | |
21 // One Chinese sentence, or a series of Japanese glyphs should not be tr eated | |
22 // as a single word. | |
23 assertTrue(StringUtil.countWords("一個中文句子不應該當成一個字") > 1); | |
24 assertTrue(StringUtil.countWords("ファイナルファンタジー") > 1); | |
25 // However, treating each Chinese/Japanese glyph as a word is also wrong . | |
26 assertTrue(StringUtil.countWords("一個中文句子不應該當成一個字") < 14); | |
27 assertTrue(StringUtil.countWords("ファイナルファンタジー") < 11); | |
kuan
2015/05/15 01:45:07
this is a katakana sentence, how about also adding
wychen
2015/05/18 18:49:20
All added.
kuan
2015/05/21 17:46:15
thanks!
| |
28 // Even if they are separated by spaces. | |
29 assertTrue(StringUtil.countWords("一 個 中 文 句 子 不 應 該 當 成 一 個 字") < 14); | |
30 assertTrue(StringUtil.countWords("フ ァ イ ナ ル フ ァ ン タ ジ ー") < 11); | |
31 | |
32 assertEquals(1, StringUtil.countWords("字")); | |
33 assertEquals(1, StringUtil.countWords("が")); | |
34 | |
35 // Mixing ASCII words and Chinese/Japanese glyphs | |
36 assertEquals(2, StringUtil.countWords("word字")); | |
37 assertEquals(2, StringUtil.countWords("word 字")); | |
38 | |
39 // Hangul uses space as word delimiter like English. | |
40 assertEquals(1, StringUtil.countWords("어")); | |
41 assertEquals(2, StringUtil.countWords("한국어 단어")); | |
42 assertEquals(5, StringUtil.countWords("한 국 어 단 어")); | |
43 assertEquals(8, StringUtil.countWords( | |
44 "예비군 훈련장 총기 난사범 최모씨의 군복에서 발견된 유서.")); | |
20 } | 45 } |
21 | 46 |
22 public void testIsWhitespace() { | 47 public void testIsWhitespace() { |
23 assertTrue(StringUtil.isWhitespace(' ')); | 48 assertTrue(StringUtil.isWhitespace(' ')); |
24 assertTrue(StringUtil.isWhitespace('\t')); | 49 assertTrue(StringUtil.isWhitespace('\t')); |
25 assertTrue(StringUtil.isWhitespace('\n')); | 50 assertTrue(StringUtil.isWhitespace('\n')); |
26 assertTrue(StringUtil.isWhitespace('\u00a0')); | 51 assertTrue(StringUtil.isWhitespace('\u00a0')); |
27 assertFalse(StringUtil.isWhitespace('a')); | 52 assertFalse(StringUtil.isWhitespace('a')); |
28 assertFalse(StringUtil.isWhitespace('$')); | 53 assertFalse(StringUtil.isWhitespace('$')); |
29 assertFalse(StringUtil.isWhitespace('_')); | 54 assertFalse(StringUtil.isWhitespace('_')); |
(...skipping 77 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
107 assertEquals(-1, StringUtil.toNumber("'8_")); | 132 assertEquals(-1, StringUtil.toNumber("'8_")); |
108 assertEquals(-1, StringUtil.toNumber("")); | 133 assertEquals(-1, StringUtil.toNumber("")); |
109 assertEquals(-1, StringUtil.toNumber(" ")); | 134 assertEquals(-1, StringUtil.toNumber(" ")); |
110 assertEquals(-1, StringUtil.toNumber("\u00a0\u0460")); | 135 assertEquals(-1, StringUtil.toNumber("\u00a0\u0460")); |
111 assertEquals(-1, StringUtil.toNumber("abc")); | 136 assertEquals(-1, StringUtil.toNumber("abc")); |
112 assertEquals(-1, StringUtil.toNumber("$")); | 137 assertEquals(-1, StringUtil.toNumber("$")); |
113 assertEquals(-1, StringUtil.toNumber("_")); | 138 assertEquals(-1, StringUtil.toNumber("_")); |
114 } | 139 } |
115 | 140 |
116 } | 141 } |
OLD | NEW |