1 /* 2 ******************************************************************************* 3 * Copyright (C) 2008-2015, International Business Machines Corporation and * 4 * others. All Rights Reserved. * 5 ******************************************************************************* 6 */ 7 package com.ibm.icu.dev.test.localespi; 8 9 import java.text.BreakIterator; 10 import java.util.Locale; 11 12 import com.ibm.icu.dev.test.TestFmwk; 13 14 public class BreakIteratorTest extends TestFmwk { main(String[] args)15 public static void main(String[] args) throws Exception { 16 new BreakIteratorTest().run(args); 17 } 18 19 private static final int CHARACTER_BRK = 0; 20 private static final int WORD_BRK = 1; 21 private static final int LINE_BRK = 2; 22 private static final int SENTENCE_BRK = 3; 23 24 /* 25 * Check if getInstance returns the ICU implementation. 26 */ TestGetInstance()27 public void TestGetInstance() { 28 for (Locale loc : BreakIterator.getAvailableLocales()) { 29 if (TestUtil.isExcluded(loc)) { 30 logln("Skipped " + loc); 31 continue; 32 } 33 checkGetInstance(CHARACTER_BRK, loc); 34 checkGetInstance(WORD_BRK, loc); 35 checkGetInstance(LINE_BRK, loc); 36 checkGetInstance(SENTENCE_BRK, loc); 37 } 38 } 39 checkGetInstance(int type, Locale loc)40 private void checkGetInstance(int type, Locale loc) { 41 BreakIterator brkitr = null; 42 String method = null; 43 switch (type) { 44 case CHARACTER_BRK: 45 brkitr = BreakIterator.getCharacterInstance(loc); 46 method = "getCharacterInstance"; 47 break; 48 case WORD_BRK: 49 brkitr = BreakIterator.getWordInstance(loc); 50 method = "getWordInstance"; 51 break; 52 case LINE_BRK: 53 brkitr = BreakIterator.getLineInstance(loc); 54 method = "getLineInstance"; 55 break; 56 case SENTENCE_BRK: 57 brkitr = BreakIterator.getSentenceInstance(loc); 58 method = "getSentenceInstance"; 59 break; 60 default: 61 errln("FAIL: Unknown break iterator type"); 62 return; 63 } 64 65 boolean isIcuImpl = (brkitr instanceof com.ibm.icu.impl.jdkadapter.BreakIteratorICU); 66 67 if (TestUtil.isICUExtendedLocale(loc)) { 68 if (!isIcuImpl) { 69 errln("FAIL: " + method + " returned JDK BreakIterator for locale " + loc); 70 } 71 } else { 72 if (isIcuImpl) { 73 logln("INFO: " + method + " returned ICU BreakIterator for locale " + loc); 74 } 75 BreakIterator brkitrIcu = null; 76 Locale iculoc = TestUtil.toICUExtendedLocale(loc); 77 switch (type) { 78 case CHARACTER_BRK: 79 brkitrIcu = BreakIterator.getCharacterInstance(iculoc); 80 break; 81 case WORD_BRK: 82 brkitrIcu = BreakIterator.getWordInstance(iculoc); 83 break; 84 case LINE_BRK: 85 brkitrIcu = BreakIterator.getLineInstance(iculoc); 86 break; 87 case SENTENCE_BRK: 88 brkitrIcu = BreakIterator.getSentenceInstance(iculoc); 89 break; 90 } 91 if (isIcuImpl) { 92 if (!brkitr.equals(brkitrIcu)) { 93 // BreakIterator.getXXXInstance returns a cached BreakIterator instance. 94 // BreakIterator does not override Object#equals, so the result may not be 95 // consistent. 96 // logln("INFO: " + method + " returned ICU BreakIterator for locale " + loc 97 // + ", but different from the one for locale " + iculoc); 98 } 99 } else { 100 if (!(brkitrIcu instanceof com.ibm.icu.impl.jdkadapter.BreakIteratorICU)) { 101 errln("FAIL: " + method + " returned JDK BreakIterator for locale " + iculoc); 102 } 103 } 104 } 105 } 106 107 /* 108 * Testing the behavior of text break between ICU instance and its 109 * equivalent created via the Locale SPI framework. 110 */ TestICUEquivalent()111 public void TestICUEquivalent() { 112 Locale[] TEST_LOCALES = { 113 new Locale("en", "US"), 114 new Locale("fr", "FR"), 115 new Locale("th", "TH"), 116 new Locale("zh", "CN"), 117 }; 118 119 String[] TEST_DATA = { 120 "International Components for Unicode (ICU) is an open source project of mature " 121 + "C/C++ and Java libraries for Unicode support, software internationalization and " 122 + "software globalization. ICU is widely portable to many operating systems and " 123 + "environments. It gives applications the same results on all platforms and between " 124 + "C/C++ and Java software. The ICU project is an open source development project " 125 + "that is sponsored, supported and used by IBM and many other companies.", 126 127 "L'International Components for Unicode (ICU) est un projet open source qui fourni " 128 + "des biblioth\u00e8ques pour les langages informatique C/C++ et Java pour supporter " 129 + "Unicode, l'internationalisation et la mondialisation des logiciels. ICU est largement " 130 + "portable vers beaucoup de syst\u00e8mes d'exploitations et d'environnements. Il " 131 + "donne aux applications les m\u00eames comportements et r\u00e9sultats sur toutes " 132 + "les plateformes et entre les logiciels C/C++ et Java. Le projet ICU est un projet " 133 + "dont les code sources sont disponibles qui est sponsoris\u00e9, support\u00e9 et " 134 + "utilis\u00e9 par IBM et beaucoup d'autres entreprises.", 135 136 "\u5728IBM\u7b49\u4f01\u696d\u4e2d\uff0c\u56fd\u9645\u5316\u7ecf\u5e38\u7b80\u5199" 137 + "\u4e3aI18N (\u6216i18n\u6216I18n)\uff0c\u5176\u4e2d18\u4ee3\u8868\u4e86\u4e2d\u95f4" 138 + "\u7701\u7565\u768418\u4e2a\u5b57\u6bcd\uff1b\u800c\u201c\u672c\u5730\u5316\u201d" 139 + "\u540c\u53ef\u7b80\u5199\u4e3al10n\u3002\u9019\u4e24\u4e2a\u6982\u5ff5\u6709\u65f6" 140 + "\u5408\u79f0\u5168\u7403\u5316\uff08g11n\uff09\uff0c\u4f46\u662f\u5168\u7403\u5316" 141 + "\u7684\u6db5\u4e49\u66f4\u4e3a\u4e00\u822c\u5316\u3002\u53e6\u5916\u5076\u5c14\u4f1a" 142 + "\u51fa\u73b0\u201cp13n\u201d\uff0c\u4ee3\u8868\u4e2a\u4eba\u5316\uff08personalization" 143 + "\uff09\u3002", 144 145 "\u0e01\u0e23\u0e38\u0e07\u0e40\u0e17\u0e1e\u0e21\u0e2b\u0e32\u0e19\u0e04\u0e23" 146 + "\u0e43\u0e19\u0e1b\u0e31\u0e08\u0e08\u0e38\u0e1a\u0e31\u0e19\u0e40\u0e1b\u0e47" 147 + "\u0e19\u0e28\u0e39\u0e19\u0e22\u0e4c\u0e01\u0e25\u0e32\u0e07\u0e01\u0e32\u0e23" 148 + "\u0e1b\u0e01\u0e04\u0e23\u0e2d\u0e07 \u0e01\u0e32\u0e23\u0e28\u0e36\u0e01\u0e29" 149 + "\u0e32 \u0e01\u0e32\u0e23\u0e04\u0e21\u0e19\u0e32\u0e04\u0e21\u0e02\u0e19\u0e2a" 150 + "\u0e48\u0e07 \u0e01\u0e32\u0e23\u0e40\u0e07\u0e34\u0e19\u0e01\u0e32\u0e23\u0e18" 151 + "\u0e19\u0e32\u0e04\u0e32\u0e23 \u0e01\u0e32\u0e23\u0e1e\u0e32\u0e13\u0e34\u0e0a" 152 + "\u0e22\u0e4c \u0e01\u0e32\u0e23\u0e2a\u0e37\u0e48\u0e2d\u0e2a\u0e32\u0e23 \u0e2f" 153 + "\u0e25\u0e2f \u0e42\u0e14\u0e22\u0e21\u0e35\u0e1e\u0e37\u0e49\u0e19\u0e17\u0e35" 154 + "\u0e48\u0e17\u0e31\u0e49\u0e07\u0e2b\u0e21\u0e14 1,562.2 \u0e15\u0e32\u0e23\u0e32" 155 + "\u0e07\u0e01\u0e34\u0e42\u0e25\u0e40\u0e21\u0e15\u0e23 \u0e1e\u0e34\u0e01\u0e31" 156 + "\u0e14\u0e17\u0e32\u0e07\u0e20\u0e39\u0e21\u0e34\u0e28\u0e32\u0e2a\u0e15\u0e23" 157 + "\u0e4c\u0e04\u0e37\u0e2d \u0e25\u0e30\u0e15\u0e34\u0e08\u0e39\u0e14 13\u00b0 45" 158 + "\u2019 \u0e40\u0e2b\u0e19\u0e37\u0e2d \u0e25\u0e2d\u0e07\u0e08\u0e34\u0e08\u0e39" 159 + "\u0e14 100\u00b0 31\u2019 \u0e15\u0e30\u0e27\u0e31\u0e19\u0e2d\u0e2d\u0e01" 160 }; 161 162 BreakIterator[] jdkBrkItrs = new BreakIterator[4]; 163 com.ibm.icu.text.BreakIterator[] icuBrkItrs = new com.ibm.icu.text.BreakIterator[4]; 164 165 for (Locale loc : TEST_LOCALES) { 166 Locale iculoc = TestUtil.toICUExtendedLocale(loc); 167 168 jdkBrkItrs[0] = BreakIterator.getCharacterInstance(iculoc); 169 jdkBrkItrs[1] = BreakIterator.getWordInstance(iculoc); 170 jdkBrkItrs[2] = BreakIterator.getLineInstance(iculoc); 171 jdkBrkItrs[3] = BreakIterator.getSentenceInstance(iculoc); 172 173 icuBrkItrs[0] = com.ibm.icu.text.BreakIterator.getCharacterInstance(iculoc); 174 icuBrkItrs[1] = com.ibm.icu.text.BreakIterator.getWordInstance(iculoc); 175 icuBrkItrs[2] = com.ibm.icu.text.BreakIterator.getLineInstance(iculoc); 176 icuBrkItrs[3] = com.ibm.icu.text.BreakIterator.getSentenceInstance(iculoc); 177 178 for (String text : TEST_DATA) { 179 for (int i = 0; i < 4; i++) { 180 compareBreaks(text, jdkBrkItrs[i], icuBrkItrs[i]); 181 } 182 } 183 } 184 } 185 compareBreaks(String text, BreakIterator jdkBrk, com.ibm.icu.text.BreakIterator icuBrk)186 private void compareBreaks(String text, BreakIterator jdkBrk, com.ibm.icu.text.BreakIterator icuBrk) { 187 jdkBrk.setText(text); 188 icuBrk.setText(text); 189 190 // Forward 191 int jidx = jdkBrk.first(); 192 int iidx = icuBrk.first(); 193 if (jidx != iidx) { 194 errln("FAIL: Different first boundaries (jdk=" + jidx + ",icu=" + iidx + ") for text:\n" + text); 195 } 196 while (true) { 197 jidx = jdkBrk.next(); 198 iidx = icuBrk.next(); 199 if (jidx != iidx) { 200 errln("FAIL: Different boundaries (jdk=" + jidx + ",icu=" + iidx + "direction=forward) for text:\n" + text); 201 } 202 if (jidx == BreakIterator.DONE) { 203 break; 204 } 205 } 206 207 // Backward 208 jidx = jdkBrk.last(); 209 iidx = jdkBrk.last(); 210 if (jidx != iidx) { 211 errln("FAIL: Different last boundaries (jdk=" + jidx + ",icu=" + iidx + ") for text:\n" + text); 212 } 213 while (true) { 214 jidx = jdkBrk.previous(); 215 iidx = icuBrk.previous(); 216 if (jidx != iidx) { 217 errln("FAIL: Different boundaries (jdk=" + jidx + ",icu=" + iidx + "direction=backward) for text:\n" + text); 218 } 219 if (jidx == BreakIterator.DONE) { 220 break; 221 } 222 } 223 } 224 } 225