]> git.saurik.com Git - apple/icu.git/blame - icuSources/test/intltest/jamotest.cpp
ICU-57132.0.1.tar.gz
[apple/icu.git] / icuSources / test / intltest / jamotest.cpp
CommitLineData
b75a7d8f
A
1/********************************************************************
2 * COPYRIGHT:
2ca993e8 3 * Copyright (c) 1997-2016, International Business Machines Corporation and
b75a7d8f
A
4 * others. All Rights Reserved.
5 ********************************************************************
6************************************************************************
7* Date Name Description
8* 02/28/2001 aliu Creation
9* 03/01/2001 George port to HP/UX
10************************************************************************/
11
12#include "unicode/utypes.h"
13
14#if !UCONFIG_NO_TRANSLITERATION
15
16#include "jamotest.h"
17#include "unicode/utypes.h"
18#include "unicode/translit.h"
2ca993e8 19#include "cmemory.h"
b75a7d8f
A
20#include "cpdtrans.h"
21
22// SEP is the disambiguation separator used by Latin-Jamo and Jamo-Latin
46f4442e 23#define SEP "-"
b75a7d8f
A
24
25JamoTest::JamoTest()
26{
46f4442e 27 UParseError parseError;
b75a7d8f 28 UErrorCode status = U_ZERO_ERROR;
46f4442e
A
29 NAME_JAMO = Transliterator::createFromRules("Name-Jamo",
30 UnicodeString(JAMO_NAMES_RULES, -1, US_INV),
31 UTRANS_FORWARD, parseError, status);
b75a7d8f
A
32
33 if (U_FAILURE(status)) {
34 delete NAME_JAMO;
35 NAME_JAMO = NULL;
36 }
37 status = U_ZERO_ERROR;
46f4442e
A
38 JAMO_NAME = Transliterator::createFromRules("Jamo-Name",
39 UnicodeString(JAMO_NAMES_RULES, -1, US_INV),
40 UTRANS_REVERSE, parseError, status);
b75a7d8f
A
41 if (U_FAILURE(status)) {
42 delete JAMO_NAME;
43 JAMO_NAME = NULL;
44 }
45}
46
47JamoTest::~JamoTest()
48{
49 delete NAME_JAMO;
50 delete JAMO_NAME;
51}
52
53void
54JamoTest::runIndexedTest(int32_t index, UBool exec,
55 const char* &name, char* /*par*/) {
56 switch (index) {
57 TESTCASE(0,TestJamo);
58 TESTCASE(1,TestRealText);
59 TESTCASE(2,TestPiecemeal);
60 default: name = ""; break;
61 }
62}
63
64void
65JamoTest::TestJamo() {
66 UParseError parseError;
67 UErrorCode status = U_ZERO_ERROR;
68 Transliterator* latinJamo = Transliterator::createInstance("Latin-Jamo", UTRANS_FORWARD, parseError, status);
69
70 if (latinJamo == 0 || U_FAILURE(status)) {
729e4ab9 71 dataerrln("FAIL: createInstance() returned 0 - %s", u_errorName(status));
b75a7d8f
A
72 return;
73 }
74
75 Transliterator* jamoLatin = latinJamo->createInverse(status);
76
77 if (jamoLatin == 0) {
78 delete latinJamo;
79 errln("FAIL: createInverse() returned 0");
80 return;
81 }
82
83 static const char* CASE[] = {
84 // Column 1 is the latin text L1 to be fed to Latin-Jamo
85 // to yield output J.
86
87 // Column 2 is expected value of J. J is fed to
88 // Jamo-Latin to yield output L2.
89
90 // Column 3 is expected value of L2. If the expected
91 // value of L2 is L1, then L2 is NULL.
46f4442e
A
92
93 // add tests for the update to fix problems where it didn't follow the standard
94 // see also http://www.unicode.org/cldr/data/charts/transforms/Latin-Hangul.html
95 "gach", "(Gi)(A)(Cf)", NULL,
96 "geumhui", "(Gi)(EU)(Mf)(Hi)(YI)", NULL,
97 "choe", "(Ci)(OE)", NULL,
98 "wo", "(IEUNG)(WEO)", NULL,
99 "Wonpil", "(IEUNG)(WEO)(Nf)(Pi)(I)(L)", "wonpil",
100 "GIPPEUM", "(Gi)(I)(BB)(EU)(Mf)", "gippeum",
101 "EUTTEUM", "(IEUNG)(EU)(DD)(EU)(Mf)", "eutteum",
102 "KKOTNAE", "(GGi)(O)(Tf)(Ni)(AE)", "kkotnae",
103 "gaga", "(Gi)(A)(Gi)(A)", NULL,
104 "gag-a", "(Gi)(A)(Gf)(IEUNG)(A)", NULL,
105 "gak-ka", "(Gi)(A)(Kf)(Ki)(A)", NULL,
106 "gakka", "(Gi)(A)(GGi)(A)", NULL,
107 "gakk-a", "(Gi)(A)(GGf)(IEUNG)(A)", NULL,
108 "gakkka", "(Gi)(A)(GGf)(Ki)(A)", NULL,
109 "gak-kka", "(Gi)(A)(Kf)(GGi)(A)", NULL,
110
b75a7d8f 111 "bab", "(Bi)(A)(Bf)", NULL,
46f4442e
A
112 "babb", "(Bi)(A)(Bf)(Bi)(EU)", "babbeu",
113 "babbba", "(Bi)(A)(Bf)(Bi)(EU)(Bi)(A)", "babbeuba",
114 "bagg", "(Bi)(A)(Gf)(Gi)(EU)", "baggeu",
115 "baggga", "(Bi)(A)(Gf)(Gi)(EU)(Gi)(A)", "baggeuga",
116 //"bag" SEP "gga", "(Bi)(A)(Gf)" SEP "(Gi)(EU)(Gi)(A)", "bag" SEP "geuga",
b75a7d8f
A
117 "kabsa", "(Ki)(A)(Bf)(Si)(A)", NULL,
118 "kabska", "(Ki)(A)(BS)(Ki)(A)", NULL,
119 "gabsbka", "(Gi)(A)(BS)(Bi)(EU)(Ki)(A)", "gabsbeuka", // not (Kf)
46f4442e 120 "gga", "(Gi)(EU)(Gi)(A)", "geuga",
b75a7d8f 121 "bsa", "(Bi)(EU)(Si)(A)", "beusa",
46f4442e
A
122 "agg", "(IEUNG)(A)(Gf)(Gi)(EU)", "aggeu",
123 "agga", "(IEUNG)(A)(Gf)(Gi)(A)", NULL,
124 "la", "(R)(A)", NULL,
b75a7d8f 125 "bs", "(Bi)(EU)(Sf)", "beus",
46f4442e 126 "kalgga", "(Ki)(A)(L)(Gi)(EU)(Gi)(A)", "kalgeuga",
b75a7d8f
A
127
128 // 'r' in a final position is treated like 'l'
129 "karka", "(Ki)(A)(L)(Ki)(A)", "kalka",
130 };
131
2ca993e8 132 enum { CASE_length = UPRV_LENGTHOF(CASE) };
b75a7d8f
A
133
134 int32_t i;
135 for (i=0; i<CASE_length; i+=3) {
136 UnicodeString jamo = nameToJamo(CASE[i+1]);
137 if (CASE[i+2] == NULL) {
138 expect(*latinJamo, CASE[i], jamo, *jamoLatin);
139 } else {
140 // Handle case where round-trip is expected to fail
141 expect(*latinJamo, CASE[i], jamo);
142 expect(*jamoLatin, jamo, CASE[i+2]);
143 }
144 }
145
146 delete latinJamo;
147 delete jamoLatin;
148}
149
150/**
151 * Test various step-at-a-time transformation of hangul to jamo to
152 * latin and back.
153 */
154void JamoTest::TestPiecemeal(void) {
155 UnicodeString hangul; hangul.append((UChar)0xBC0F);
156 UnicodeString jamo = nameToJamo("(Mi)(I)(Cf)");
157 UnicodeString latin("mic");
46f4442e 158 UnicodeString latin2("mich");
b75a7d8f
A
159
160 Transliterator *t = NULL;
161 UErrorCode status = U_ZERO_ERROR;
162
163 t = Transliterator::createInstance("NFD", UTRANS_FORWARD, status); // was Hangul-Jamo
164 if (U_FAILURE(status) || t == 0) {
729e4ab9 165 dataerrln("FAIL: createInstance failed");
b75a7d8f
A
166 return;
167 }
168 expect(*t, hangul, jamo);
169 delete t;
170
171 t = Transliterator::createInstance("NFC", UTRANS_FORWARD, status); // was Jamo-Hangul
172 if (U_FAILURE(status) || t == 0) {
173 errln("FAIL: createInstance failed");
174 return;
175 }
176 expect(*t, jamo, hangul);
177 delete t;
178
179 t = Transliterator::createInstance("Latin-Jamo", UTRANS_FORWARD, status);
180 if (U_FAILURE(status) || t == 0) {
729e4ab9 181 dataerrln("FAIL: createInstance failed - %s", u_errorName(status));
b75a7d8f
A
182 return;
183 }
184 expect(*t, latin, jamo);
185 delete t;
186
187 t = Transliterator::createInstance("Jamo-Latin", UTRANS_FORWARD, status);
188 if (U_FAILURE(status) || t == 0) {
189 errln("FAIL: createInstance failed");
190 return;
191 }
46f4442e 192 expect(*t, jamo, latin2);
b75a7d8f
A
193 delete t;
194
195 t = Transliterator::createInstance("Hangul-Latin", UTRANS_FORWARD, status);
196 if (U_FAILURE(status) || t == 0) {
197 errln("FAIL: createInstance failed");
198 return;
199 }
46f4442e 200 expect(*t, hangul, latin2);
b75a7d8f
A
201 delete t;
202
203 t = Transliterator::createInstance("Latin-Hangul", UTRANS_FORWARD, status);
204 if (U_FAILURE(status) || t == 0) {
205 errln("FAIL: createInstance failed");
206 return;
207 }
208 expect(*t, latin, hangul);
209 delete t;
210
211 t = Transliterator::createInstance("Hangul-Latin; Latin-Jamo", UTRANS_FORWARD, status);
212 if (U_FAILURE(status) || t == 0) {
213 errln("FAIL: createInstance failed");
214 return;
215 }
216 expect(*t, hangul, jamo);
217 delete t;
218
219 t = Transliterator::createInstance("Jamo-Latin; Latin-Hangul", UTRANS_FORWARD, status);
220 if (U_FAILURE(status) || t == 0) {
221 errln("FAIL: createInstance failed");
222 return;
223 }
224 expect(*t, jamo, hangul);
225 delete t;
226
227 t = Transliterator::createInstance("Hangul-Latin; Latin-Hangul", UTRANS_FORWARD, status);
228 if (U_FAILURE(status) || t == 0) {
229 errln("FAIL: createInstance failed");
230 return;
231 }
232 expect(*t, hangul, hangul);
233 delete t;
234}
235
236void
237JamoTest::TestRealText() {
238 // Test text taken from the Unicode web site
46f4442e 239 static const char* const WHAT_IS_UNICODE[] = {
b75a7d8f
A
240 "\\uc720\\ub2c8\\ucf54\\ub4dc\\uc5d0", "\\ub300\\ud574", "?",
241
242 "\\uc5b4\\ub5a4", "\\ud50c\\ub7ab\\ud3fc,", "\\uc5b4\\ub5a4",
243 "\\ud504\\ub85c\\uadf8\\ub7a8,", "\\uc5b4\\ub5a4", "\\uc5b8\\uc5b4\\uc5d0\\ub3c4",
244 "\\uc0c1\\uad00\\uc5c6\\uc774", "\\uc720\\ub2c8\\ucf54\\ub4dc\\ub294", "\\ubaa8\\ub4e0",
245 "\\ubb38\\uc790\\uc5d0", "\\ub300\\ud574", "\\uace0\\uc720", "\\ubc88\\ud638\\ub97c",
246 "\\uc81c\\uacf5\\ud569\\ub2c8\\ub2e4.",
247
248 "\\uae30\\ubcf8\\uc801\\uc73c\\ub85c", "\\ucef4\\ud4e8\\ud130\\ub294",
249 "\\uc22b\\uc790\\ub9cc", "\\ucc98\\ub9ac\\ud569\\ub2c8\\ub2e4.", "\\uae00\\uc790\\ub098",
250 "\\ub2e4\\ub978", "\\ubb38\\uc790\\uc5d0\\ub3c4", "\\uc22b\\uc790\\ub97c",
251 "\\uc9c0\\uc815\\ud558\\uc5ec",
252 "\\uc800\\uc7a5\\ud569\\ub2c8\\ub2e4.", "\\uc720\\ub2c8\\ucf54\\ub4dc\\uac00",
253 "\\uac1c\\ubc1c\\ub418\\uae30", "\\uc804\\uc5d0\\ub294", "\\uc774\\ub7ec\\ud55c",
254 "\\uc22b\\uc790\\ub97c", "\\uc9c0\\uc815\\ud558\\uae30", "\\uc704\\ud574", "\\uc218\\ubc31",
255 "\\uac00\\uc9c0\\uc758", "\\ub2e4\\ub978", "\\uae30\\ud638\\ud654",
256 "\\uc2dc\\uc2a4\\ud15c\\uc744",
257 "\\uc0ac\\uc6a9\\ud588\\uc2b5\\ub2c8\\ub2e4.", "\\ub2e8\\uc77c", "\\uae30\\ud638\\ud654",
258 "\\ubc29\\ubc95\\uc73c\\ub85c\\ub294", "\\ubaa8\\ub4e0", "\\ubb38\\uc790\\ub97c",
259 "\\ud3ec\\ud568\\ud560", "\\uc218", "\\uc5c6\\uc5c8\\uc2b5\\ub2c8\\ub2e4.", "\\uc608\\ub97c",
260 "\\ub4e4\\uc5b4", "\\uc720\\ub7fd", "\\uc5f0\\ud569\\uc5d0\\uc11c\\ub9cc",
261 "\\ubcf4\\ub354\\ub77c\\ub3c4", "\\ubaa8\\ub4e0", "\\uac01", "\\ub098\\ub77c\\ubcc4",
262 "\\uc5b8\\uc5b4\\ub97c", "\\ucc98\\ub9ac\\ud558\\ub824\\uba74", "\\uc5ec\\ub7ec",
263 "\\uac1c\\uc758", "\\ub2e4\\ub978", "\\uae30\\ud638\\ud654", "\\ubc29\\ubc95\\uc774",
264 "\\ud544\\uc694\\ud569\\ub2c8\\ub2e4.", "\\uc601\\uc5b4\\uc640", "\\uac19\\uc740",
265 "\\ub2e8\\uc77c", "\\uc5b8\\uc5b4\\uc758", "\\uacbd\\uc6b0\\ub3c4",
266 "\\uacf5\\ud1b5\\uc801\\uc73c\\ub85c", "\\uc0ac\\uc6a9\\ub418\\ub294", "\\ubaa8\\ub4e0",
267 "\\uae00\\uc790,", "\\ubb38\\uc7a5", "\\ubd80\\ud638", "\\ubc0f",
268 "\\ud14c\\ud06c\\ub2c8\\uceec", "\\uae30\\ud638\\uc5d0", "\\ub9de\\ub294", "\\ub2e8\\uc77c",
269 "\\uae30\\ud638\\ud654", "\\ubc29\\ubc95\\uc744", "\\uac16\\uace0", "\\uc788\\uc9c0",
270 "\\ubabb\\ud558\\uc600\\uc2b5\\ub2c8\\ub2e4.",
271
272 "\\uc774\\ub7ec\\ud55c", "\\uae30\\ud638\\ud654", "\\uc2dc\\uc2a4\\ud15c\\uc740",
273 "\\ub610\\ud55c", "\\ub2e4\\ub978", "\\uae30\\ud638\\ud654", "\\uc2dc\\uc2a4\\ud15c\\uacfc",
274 "\\ucda9\\ub3cc\\ud569\\ub2c8\\ub2e4.", "\\uc989", "\\ub450", "\\uac00\\uc9c0",
275 "\\uae30\\ud638\\ud654", "\\ubc29\\ubc95\\uc774", "\\ub450", "\\uac1c\\uc758", "\\ub2e4\\ub978",
276 "\\ubb38\\uc790\\uc5d0", "\\ub300\\ud574", "\\uac19\\uc740", "\\ubc88\\ud638\\ub97c",
277 "\\uc0ac\\uc6a9\\ud558\\uac70\\ub098", "\\uac19\\uc740", "\\ubb38\\uc790\\uc5d0",
278 "\\ub300\\ud574", "\\ub2e4\\ub978", "\\ubc88\\ud638\\ub97c", "\\uc0ac\\uc6a9\\ud560", "\\uc218",
279 "\\uc788\\uc2b5\\ub2c8\\ub2e4.", "\\uc8fc\\uc5b4\\uc9c4", "\\ubaa8\\ub4e0",
280 "\\ucef4\\ud4e8\\ud130(\\ud2b9\\ud788", "\\uc11c\\ubc84)\\ub294", "\\uc11c\\ub85c",
281 "\\ub2e4\\ub978", "\\uc5ec\\ub7ec", "\\uac00\\uc9c0", "\\uae30\\ud638\\ud654",
282 "\\ubc29\\ubc95\\uc744", "\\uc9c0\\uc6d0\\ud574\\uc57c",
283 "\\ud569\\ub2c8\\ub2e4.", "\\uadf8\\ub7ec\\ub098,", "\\ub370\\uc774\\ud130\\ub97c",
284 "\\uc11c\\ub85c", "\\ub2e4\\ub978", "\\uae30\\ud638\\ud654", "\\ubc29\\ubc95\\uc774\\ub098",
285 "\\ud50c\\ub7ab\\ud3fc", "\\uac04\\uc5d0", "\\uc804\\ub2ec\\ud560", "\\ub54c\\ub9c8\\ub2e4",
286 "\\uadf8", "\\ub370\\uc774\\ud130\\ub294", "\\ud56d\\uc0c1", "\\uc190\\uc0c1\\uc758",
287 "\\uc704\\ud5d8\\uc744", "\\uacaa\\uac8c", "\\ub429\\ub2c8\\ub2e4.",
288
289 "\\uc720\\ub2c8\\ucf54\\ub4dc\\ub85c", "\\ubaa8\\ub4e0", "\\uac83\\uc744",
290 "\\ud574\\uacb0\\ud560", "\\uc218", "\\uc788\\uc2b5\\ub2c8\\ub2e4!",
291 "\\uc720\\ub2c8\\ucf54\\ub4dc\\ub294", "\\uc0ac\\uc6a9", "\\uc911\\uc778",
292 "\\ud50c\\ub7ab\\ud3fc,", "\\ud504\\ub85c\\uadf8\\ub7a8,", "\\uc5b8\\uc5b4\\uc5d0",
293 "\\uad00\\uacc4\\uc5c6\\uc774", "\\ubb38\\uc790\\ub9c8\\ub2e4", "\\uace0\\uc720\\ud55c",
294 "\\uc22b\\uc790\\ub97c",
295 "\\uc81c\\uacf5\\ud569\\ub2c8\\ub2e4.", "\\uc720\\ub2c8\\ucf54\\ub4dc",
296 "\\ud45c\\uc900\\uc740", // "Apple, HP, IBM, JustSystem, Microsoft, Oracle, SAP, "
297 // "Sun, Sybase, Unisys "
298 "\\ubc0f", "\\uae30\\ud0c0", "\\uc5ec\\ub7ec",
299 "\\ud68c\\uc0ac\\uc640", "\\uac19\\uc740", "\\uc5c5\\uacc4",
300 "\\uc120\\ub450\\uc8fc\\uc790\\uc5d0", "\\uc758\\ud574",
301 "\\ucc44\\ud0dd\\ub418\\uc5c8\\uc2b5\\ub2c8\\ub2e4.", "\\uc720\\ub2c8\\ucf54\\ub4dc\\ub294",
302 // "XML, Java, ECMAScript(JavaScript), LDAP, CORBA 3.0, WML "
303 "\\ub4f1\\uacfc",
304 "\\uac19\\uc774", "\\ud604\\uc7ac", "\\ub110\\ub9ac", "\\uc0ac\\uc6a9\\ub418\\ub294",
305 "\\ud45c\\uc900\\uc5d0\\uc11c", "\\ud544\\uc694\\ud558\\uba70", "\\uc774\\ub294", // "ISO/IEC",
306 "10646\\uc744", "\\uad6c\\ud604\\ud558\\ub294", "\\uacf5\\uc2dd\\uc801\\uc778",
307 "\\ubc29\\ubc95\\uc785\\ub2c8\\ub2e4.", "\\uc774\\ub294", "\\ub9ce\\uc740", "\\uc6b4\\uc601",
308 "\\uccb4\\uc81c,", "\\uc694\\uc998", "\\uc0ac\\uc6a9\\ub418\\ub294", "\\ubaa8\\ub4e0",
309 "\\ube0c\\ub77c\\uc6b0\\uc800", "\\ubc0f", "\\uae30\\ud0c0", "\\ub9ce\\uc740",
310 "\\uc81c\\ud488\\uc5d0\\uc11c",
311 "\\uc9c0\\uc6d0\\ub429\\ub2c8\\ub2e4.", "\\uc720\\ub2c8\\ucf54\\ub4dc",
312 "\\ud45c\\uc900\\uc758", "\\ubd80\\uc0c1\\uacfc", "\\uc774\\ub97c",
313 "\\uc9c0\\uc6d0\\ud558\\ub294", "\\ub3c4\\uad6c\\uc758", "\\uac00\\uc6a9\\uc131\\uc740",
314 "\\ucd5c\\uadfc", "\\uc804", "\\uc138\\uacc4\\uc5d0", "\\ubd88\\uace0", "\\uc788\\ub294",
315 "\\uae30\\uc220", "\\uacbd\\ud5a5\\uc5d0\\uc11c", "\\uac00\\uc7a5", "\\uc911\\uc694\\ud55c",
316 "\\ubd80\\ubd84\\uc744", "\\ucc28\\uc9c0\\ud558\\uace0", "\\uc788\\uc2b5\\ub2c8\\ub2e4.",
317
318 "\\uc720\\ub2c8\\ucf54\\ub4dc\\ub97c",
46f4442e
A
319 // Replaced a hyphen with a space to make the test case work with CLDR1.5
320 //"\\ud074\\ub77c\\uc774\\uc5b8\\ud2b8-\\uc11c\\ubc84", "\\ub610\\ub294",
321 "\\ud074\\ub77c\\uc774\\uc5b8\\ud2b8 \\uc11c\\ubc84", "\\ub610\\ub294",
322 // Replaced a hyphen with a space.
323 //"\\ub2e4\\uc911-\\uc5f0\\uacb0", "\\uc751\\uc6a9", "\\ud504\\ub85c\\uadf8\\ub7a8\\uacfc",
324 "\\ub2e4\\uc911 \\uc5f0\\uacb0", "\\uc751\\uc6a9", "\\ud504\\ub85c\\uadf8\\ub7a8\\uacfc",
b75a7d8f
A
325 "\\uc6f9", "\\uc0ac\\uc774\\ud2b8\\uc5d0", "\\ud1b5\\ud569\\ud558\\uba74",
326 "\\ub808\\uac70\\uc2dc", "\\ubb38\\uc790", "\\uc138\\ud2b8", "\\uc0ac\\uc6a9\\uc5d0",
327 "\\uc788\\uc5b4\\uc11c", "\\uc0c1\\ub2f9\\ud55c", "\\ube44\\uc6a9", "\\uc808\\uac10",
328 "\\ud6a8\\uacfc\\uac00",
329 "\\ub098\\ud0c0\\ub0a9\\ub2c8\\ub2e4.", "\\uc720\\ub2c8\\ucf54\\ub4dc\\ub97c",
330 "\\ud1b5\\ud574", "\\ub9ac\\uc5d4\\uc9c0\\ub2c8\\uc5b4\\ub9c1", "\\uc5c6\\uc774",
331 "\\ub2e4\\uc911", "\\ud50c\\ub7ab\\ud3fc,", "\\uc5b8\\uc5b4", "\\ubc0f", "\\uad6d\\uac00",
332 "\\uac04\\uc5d0", "\\ub2e8\\uc77c", "\\uc18c\\ud504\\ud2b8\\uc6e8\\uc5b4",
333 "\\ud50c\\ub7ab\\ud3fc", "\\ub610\\ub294", "\\ub2e8\\uc77c", "\\uc6f9",
334 "\\uc0ac\\uc774\\ud2b8\\ub97c", "\\ubaa9\\ud45c\\ub85c", "\\uc0bc\\uc744", "\\uc218",
335 "\\uc788\\uc2b5\\ub2c8\\ub2e4.", "\\uc774\\ub97c", "\\uc0ac\\uc6a9\\ud558\\uba74",
336 "\\ub370\\uc774\\ud130\\ub97c", "\\uc190\\uc0c1", "\\uc5c6\\uc774", "\\uc5ec\\ub7ec",
337 "\\uc2dc\\uc2a4\\ud15c\\uc744", "\\ud1b5\\ud574", "\\uc804\\uc1a1\\ud560", "\\uc218",
338 "\\uc788\\uc2b5\\ub2c8\\ub2e4.",
339
340 "\\uc720\\ub2c8\\ucf54\\ub4dc", "\\ucf58\\uc18c\\uc2dc\\uc5c4\\uc5d0", "\\ub300\\ud574",
341 "\\uc720\\ub2c8\\ucf54\\ub4dc", "\\ucf58\\uc18c\\uc2dc\\uc5c4\\uc740",
342 "\\ube44\\uc601\\ub9ac", "\\uc870\\uc9c1\\uc73c\\ub85c\\uc11c", "\\ud604\\ub300",
343 "\\uc18c\\ud504\\ud2b8\\uc6e8\\uc5b4", "\\uc81c\\ud488\\uacfc",
344 "\\ud45c\\uc900\\uc5d0\\uc11c", "\\ud14d\\uc2a4\\ud2b8\\uc758", "\\ud45c\\ud604\\uc744",
345 "\\uc9c0\\uc815\\ud558\\ub294", "\\uc720\\ub2c8\\ucf54\\ub4dc", "\\ud45c\\uc900\\uc758",
346 "\\uc0ac\\uc6a9\\uc744", "\\uac1c\\ubc1c\\ud558\\uace0", "\\ud655\\uc7a5\\ud558\\uba70",
347 "\\uc7a5\\ub824\\ud558\\uae30", "\\uc704\\ud574",
348 "\\uc138\\uc6cc\\uc84c\\uc2b5\\ub2c8\\ub2e4.", "\\ucf58\\uc18c\\uc2dc\\uc5c4",
349 "\\uba64\\ubc84\\uc27d\\uc740", "\\ucef4\\ud4e8\\ud130\\uc640", "\\uc815\\ubcf4",
350 "\\ucc98\\ub9ac", "\\uc0b0\\uc5c5\\uc5d0", "\\uc885\\uc0ac\\ud558\\uace0", "\\uc788\\ub294",
351 "\\uad11\\ubc94\\uc704\\ud55c", "\\ud68c\\uc0ac", "\\ubc0f", "\\uc870\\uc9c1\\uc758",
352 "\\ubc94\\uc704\\ub97c",
353 "\\ub098\\ud0c0\\ub0c5\\ub2c8\\ub2e4.", "\\ucf58\\uc18c\\uc2dc\\uc5c4\\uc758",
354 "\\uc7ac\\uc815\\uc740", "\\uc804\\uc801\\uc73c\\ub85c", "\\ud68c\\ube44\\uc5d0",
355 "\\uc758\\ud574", "\\ucda9\\ub2f9\\ub429\\ub2c8\\ub2e4.", "\\uc720\\ub2c8\\ucf54\\ub4dc",
356 "\\ucee8\\uc18c\\uc2dc\\uc5c4\\uc5d0\\uc11c\\uc758", "\\uba64\\ubc84\\uc27d\\uc740",
357 "\\uc804", "\\uc138\\uacc4", "\\uc5b4\\ub290", "\\uacf3\\uc5d0\\uc11c\\ub098",
358 "\\uc720\\ub2c8\\ucf54\\ub4dc", "\\ud45c\\uc900\\uc744", "\\uc9c0\\uc6d0\\ud558\\uace0",
359 "\\uadf8", "\\ud655\\uc7a5\\uacfc", "\\uad6c\\ud604\\uc744",
360 "\\uc9c0\\uc6d0\\ud558\\uace0\\uc790\\ud558\\ub294", "\\uc870\\uc9c1\\uacfc",
361 "\\uac1c\\uc778\\uc5d0\\uac8c", "\\uac1c\\ubc29\\ub418\\uc5b4",
362 "\\uc788\\uc2b5\\ub2c8\\ub2e4.",
363
364 "\\ub354", "\\uc790\\uc138\\ud55c", "\\ub0b4\\uc6a9\\uc740", "\\uc6a9\\uc5b4\\uc9d1,",
365 "\\uc608\\uc81c", "\\uc720\\ub2c8\\ucf54\\ub4dc", "\\uc0ac\\uc6a9", "\\uac00\\ub2a5",
366 "\\uc81c\\ud488,", "\\uae30\\uc220", "\\uc815\\ubcf4", "\\ubc0f", "\\uae30\\ud0c0",
367 "\\uc720\\uc6a9\\ud55c", "\\uc815\\ubcf4\\ub97c",
368 "\\ucc38\\uc870\\ud558\\uc2ed\\uc2dc\\uc624."
369 };
370
2ca993e8 371 enum { WHAT_IS_UNICODE_length = UPRV_LENGTHOF(WHAT_IS_UNICODE) };
b75a7d8f
A
372
373 UParseError parseError;
374 UErrorCode status = U_ZERO_ERROR;
375 Transliterator* latinJamo = Transliterator::createInstance("Latin-Jamo", UTRANS_FORWARD, parseError, status);
376 Transliterator* jamoHangul = Transliterator::createInstance("NFC(NFD)", UTRANS_FORWARD, parseError, status);
377 if (latinJamo == 0 || jamoHangul == 0 || U_FAILURE(status)) {
378 delete latinJamo;
379 delete jamoHangul;
729e4ab9 380 dataerrln("FAIL: createInstance returned NULL - %s", u_errorName(status));
b75a7d8f
A
381 return;
382 }
383 Transliterator* jamoLatin = latinJamo->createInverse(status);
384 Transliterator* hangulJamo = jamoHangul->createInverse(status);
385 if (jamoLatin == 0 || hangulJamo == 0) {
386 errln("FAIL: createInverse returned NULL");
387 delete latinJamo;
388 delete jamoLatin;
389 delete jamoHangul;
390 delete hangulJamo;
391 return;
392 }
393
394 Transliterator* tarray[4] =
395 { hangulJamo, jamoLatin, latinJamo, jamoHangul };
396 CompoundTransliterator rt(tarray, 4);
397
398 UnicodeString buf;
399 int32_t total = 0;
400 int32_t errors = 0;
401 int32_t i;
402 for (i=0; i < WHAT_IS_UNICODE_length; ++i) {
403 ++total;
46f4442e 404 UnicodeString hangul = UnicodeString(WHAT_IS_UNICODE[i], -1, US_INV);
b75a7d8f
A
405 hangul = hangul.unescape(); // Parse backslash-u escapes
406 UnicodeString hangulX = hangul;
407 rt.transliterate(hangulX);
408 if (hangul != hangulX) {
409 ++errors;
410 UnicodeString jamo = hangul; hangulJamo->transliterate(jamo);
411 UnicodeString latin = jamo; jamoLatin->transliterate(latin);
412 UnicodeString jamo2 = latin; latinJamo->transliterate(jamo2);
413 UnicodeString hangul2 = jamo2; jamoHangul->transliterate(hangul2);
414
415 buf.remove(0);
416 buf.append("FAIL: ");
417 if (hangul2 != hangulX) {
418 buf.append((UnicodeString)"(Weird: " + hangulX + " != " + hangul2 + ")");
419 }
420 // The Hangul-Jamo conversion is not usually the
421 // bug here, so we hide it from display.
422 // Uncomment lines to see the Hangul.
423 buf.append(//hangul + " => " +
424 jamoToName(jamo) + " => " +
425 latin + " => " + jamoToName(jamo2)
426 //+ " => " + hangul2
427 );
428 errln(prettify(buf));
429 }
430 }
431 if (errors != 0) {
432 errln((UnicodeString)"Test word failures: " + errors + " out of " + total);
433 } else {
434 logln((UnicodeString)"All " + total + " test words passed");
435 }
436
437 delete latinJamo;
438 delete jamoLatin;
439 delete jamoHangul;
440 delete hangulJamo;
441}
442
443// Override TransliteratorTest
444void
445JamoTest::expectAux(const UnicodeString& tag,
446 const UnicodeString& summary, UBool pass,
447 const UnicodeString& expectedResult) {
448 UnicodeString jsum = jamoToName(summary);
449 UnicodeString jexp = jamoToName(expectedResult);
450 TransliteratorTest::expectAux(tag, jsum, pass, jexp);
451}
452
453const char* JamoTest::JAMO_NAMES_RULES =
454 "'(Gi)' <> \\u1100;"
455 "'(GGi)' <> \\u1101;"
456 "'(Ni)' <> \\u1102;"
457 "'(Di)' <> \\u1103;"
458 "'(DD)' <> \\u1104;"
459 "'(R)' <> \\u1105;"
460 "'(Mi)' <> \\u1106;"
461 "'(Bi)' <> \\u1107;"
462 "'(BB)' <> \\u1108;"
463 "'(Si)' <> \\u1109;"
464 "'(SSi)' <> \\u110A;"
465 "'(IEUNG)' <> \\u110B;"
466 "'(Ji)' <> \\u110C;"
467 "'(JJ)' <> \\u110D;"
468 "'(Ci)' <> \\u110E;"
469 "'(Ki)' <> \\u110F;"
470 "'(Ti)' <> \\u1110;"
471 "'(Pi)' <> \\u1111;"
472 "'(Hi)' <> \\u1112;"
473
474 "'(A)' <> \\u1161;"
475 "'(AE)' <> \\u1162;"
476 "'(YA)' <> \\u1163;"
477 "'(YAE)' <> \\u1164;"
478 "'(EO)' <> \\u1165;"
479 "'(E)' <> \\u1166;"
480 "'(YEO)' <> \\u1167;"
481 "'(YE)' <> \\u1168;"
482 "'(O)' <> \\u1169;"
483 "'(WA)' <> \\u116A;"
484 "'(WAE)' <> \\u116B;"
485 "'(OE)' <> \\u116C;"
486 "'(YO)' <> \\u116D;"
487 "'(U)' <> \\u116E;"
488 "'(WEO)' <> \\u116F;"
489 "'(WE)' <> \\u1170;"
490 "'(WI)' <> \\u1171;"
491 "'(YU)' <> \\u1172;"
492 "'(EU)' <> \\u1173;"
493 "'(YI)' <> \\u1174;"
494 "'(I)' <> \\u1175;"
495
496 "'(Gf)' <> \\u11A8;"
497 "'(GGf)' <> \\u11A9;"
498 "'(GS)' <> \\u11AA;"
499 "'(Nf)' <> \\u11AB;"
500 "'(NJ)' <> \\u11AC;"
501 "'(NH)' <> \\u11AD;"
502 "'(Df)' <> \\u11AE;"
503 "'(L)' <> \\u11AF;"
504 "'(LG)' <> \\u11B0;"
505 "'(LM)' <> \\u11B1;"
506 "'(LB)' <> \\u11B2;"
507 "'(LS)' <> \\u11B3;"
508 "'(LT)' <> \\u11B4;"
509 "'(LP)' <> \\u11B5;"
510 "'(LH)' <> \\u11B6;"
511 "'(Mf)' <> \\u11B7;"
512 "'(Bf)' <> \\u11B8;"
513 "'(BS)' <> \\u11B9;"
514 "'(Sf)' <> \\u11BA;"
515 "'(SSf)' <> \\u11BB;"
516 "'(NG)' <> \\u11BC;"
517 "'(Jf)' <> \\u11BD;"
518 "'(Cf)' <> \\u11BE;"
519 "'(Kf)' <> \\u11BF;"
520 "'(Tf)' <> \\u11C0;"
521 "'(Pf)' <> \\u11C1;"
522 "'(Hf)' <> \\u11C2;";
523
524/**
525 * Convert short names to actual jamo. E.g., "x(LG)y" returns
526 * "x\u11B0y". See JAMO_NAMES for table of names.
527 */
528UnicodeString
529JamoTest::nameToJamo(const UnicodeString& input) {
530 if (NAME_JAMO == 0) {
531 errln("Failed to create NAME_JAMO");
532 return input; /* failure! */
533 }
534 UnicodeString result(input);
535 NAME_JAMO->transliterate(result);
536 return result;
537}
538
539/**
540 * Convert jamo to short names. E.g., "x\u11B0y" returns
541 * "x(LG)y". See JAMO_NAMES for table of names.
542 */
543UnicodeString
544JamoTest::jamoToName(const UnicodeString& input) {
545 if (NAME_JAMO == 0) {
546 errln("Failed to create NAME_JAMO");
547 return input; /* failure! */
548 }
549 UnicodeString result(input);
550 JAMO_NAME->transliterate(result);
551 return result;
552}
553
554#endif /* #if !UCONFIG_NO_TRANSLITERATION */