1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /********************************************************************
4 * COPYRIGHT:
5 * Copyright (c) 1997-2016, International Business Machines Corporation and
6 * others. All Rights Reserved.
7 ********************************************************************/
8
9 #include "unicode/utypes.h"
10
11 #if !UCONFIG_NO_NORMALIZATION
12
13 #include "unicode/uchar.h"
14 #include "unicode/errorcode.h"
15 #include "unicode/normlzr.h"
16 #include "unicode/stringoptions.h"
17 #include "unicode/stringpiece.h"
18 #include "unicode/uniset.h"
19 #include "unicode/usetiter.h"
20 #include "unicode/schriter.h"
21 #include "unicode/utf16.h"
22 #include "cmemory.h"
23 #include "cstring.h"
24 #include "normalizer2impl.h"
25 #include "testutil.h"
26 #include "tstnorm.h"
27
28 #define ARRAY_LENGTH(array) UPRV_LENGTHOF(array)
29
runIndexedTest(int32_t index,UBool exec,const char * & name,char *)30 void BasicNormalizerTest::runIndexedTest(int32_t index, UBool exec,
31 const char* &name, char* /*par*/) {
32 if(exec) {
33 logln("TestSuite BasicNormalizerTest: ");
34 }
35 TESTCASE_AUTO_BEGIN;
36 TESTCASE_AUTO(TestDecomp);
37 TESTCASE_AUTO(TestCompatDecomp);
38 TESTCASE_AUTO(TestCanonCompose);
39 TESTCASE_AUTO(TestCompatCompose);
40 TESTCASE_AUTO(TestPrevious);
41 TESTCASE_AUTO(TestHangulDecomp);
42 TESTCASE_AUTO(TestHangulCompose);
43 TESTCASE_AUTO(TestTibetan);
44 TESTCASE_AUTO(TestCompositionExclusion);
45 TESTCASE_AUTO(TestZeroIndex);
46 TESTCASE_AUTO(TestVerisign);
47 TESTCASE_AUTO(TestPreviousNext);
48 TESTCASE_AUTO(TestNormalizerAPI);
49 TESTCASE_AUTO(TestConcatenate);
50 TESTCASE_AUTO(FindFoldFCDExceptions);
51 TESTCASE_AUTO(TestCompare);
52 TESTCASE_AUTO(TestSkippable);
53 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
54 TESTCASE_AUTO(TestCustomComp);
55 TESTCASE_AUTO(TestCustomFCC);
56 #endif
57 TESTCASE_AUTO(TestFilteredNormalizer2Coverage);
58 TESTCASE_AUTO(TestComposeUTF8WithEdits);
59 TESTCASE_AUTO(TestDecomposeUTF8WithEdits);
60 TESTCASE_AUTO(TestLowMappingToEmpty_D);
61 TESTCASE_AUTO(TestLowMappingToEmpty_FCD);
62 TESTCASE_AUTO(TestNormalizeIllFormedText);
63 TESTCASE_AUTO(TestComposeJamoTBase);
64 TESTCASE_AUTO(TestComposeBoundaryAfter);
65 TESTCASE_AUTO(TestNFKC_SCF);
66 TESTCASE_AUTO_END;
67 }
68
69 /**
70 * Convert Java-style strings with \u Unicode escapes into UnicodeString objects
71 */
str(const char * input)72 static UnicodeString str(const char *input)
73 {
74 UnicodeString str(input, ""); // Invariant conversion
75 return str.unescape();
76 }
77
78
BasicNormalizerTest()79 BasicNormalizerTest::BasicNormalizerTest()
80 {
81 // canonTest
82 // Input Decomposed Composed
83
84 canonTests[0][0] = str("cat"); canonTests[0][1] = str("cat"); canonTests[0][2] = str("cat");
85
86 canonTests[1][0] = str("\\u00e0ardvark"); canonTests[1][1] = str("a\\u0300ardvark"); canonTests[1][2] = str("\\u00e0ardvark");
87
88 canonTests[2][0] = str("\\u1e0a"); canonTests[2][1] = str("D\\u0307"); canonTests[2][2] = str("\\u1e0a"); // D-dot_above
89
90 canonTests[3][0] = str("D\\u0307"); canonTests[3][1] = str("D\\u0307"); canonTests[3][2] = str("\\u1e0a"); // D dot_above
91
92 canonTests[4][0] = str("\\u1e0c\\u0307"); canonTests[4][1] = str("D\\u0323\\u0307"); canonTests[4][2] = str("\\u1e0c\\u0307"); // D-dot_below dot_above
93
94 canonTests[5][0] = str("\\u1e0a\\u0323"); canonTests[5][1] = str("D\\u0323\\u0307"); canonTests[5][2] = str("\\u1e0c\\u0307"); // D-dot_above dot_below
95
96 canonTests[6][0] = str("D\\u0307\\u0323"); canonTests[6][1] = str("D\\u0323\\u0307"); canonTests[6][2] = str("\\u1e0c\\u0307"); // D dot_below dot_above
97
98 canonTests[7][0] = str("\\u1e10\\u0307\\u0323"); canonTests[7][1] = str("D\\u0327\\u0323\\u0307"); canonTests[7][2] = str("\\u1e10\\u0323\\u0307"); // D dot_below cedilla dot_above
99
100 canonTests[8][0] = str("D\\u0307\\u0328\\u0323"); canonTests[8][1] = str("D\\u0328\\u0323\\u0307"); canonTests[8][2] = str("\\u1e0c\\u0328\\u0307"); // D dot_above ogonek dot_below
101
102 canonTests[9][0] = str("\\u1E14"); canonTests[9][1] = str("E\\u0304\\u0300"); canonTests[9][2] = str("\\u1E14"); // E-macron-grave
103
104 canonTests[10][0] = str("\\u0112\\u0300"); canonTests[10][1] = str("E\\u0304\\u0300"); canonTests[10][2] = str("\\u1E14"); // E-macron + grave
105
106 canonTests[11][0] = str("\\u00c8\\u0304"); canonTests[11][1] = str("E\\u0300\\u0304"); canonTests[11][2] = str("\\u00c8\\u0304"); // E-grave + macron
107
108 canonTests[12][0] = str("\\u212b"); canonTests[12][1] = str("A\\u030a"); canonTests[12][2] = str("\\u00c5"); // angstrom_sign
109
110 canonTests[13][0] = str("\\u00c5"); canonTests[13][1] = str("A\\u030a"); canonTests[13][2] = str("\\u00c5"); // A-ring
111
112 canonTests[14][0] = str("\\u00C4ffin"); canonTests[14][1] = str("A\\u0308ffin"); canonTests[14][2] = str("\\u00C4ffin");
113
114 canonTests[15][0] = str("\\u00C4\\uFB03n"); canonTests[15][1] = str("A\\u0308\\uFB03n"); canonTests[15][2] = str("\\u00C4\\uFB03n");
115
116 canonTests[16][0] = str("Henry IV"); canonTests[16][1] = str("Henry IV"); canonTests[16][2] = str("Henry IV");
117
118 canonTests[17][0] = str("Henry \\u2163"); canonTests[17][1] = str("Henry \\u2163"); canonTests[17][2] = str("Henry \\u2163");
119
120 canonTests[18][0] = str("\\u30AC"); canonTests[18][1] = str("\\u30AB\\u3099"); canonTests[18][2] = str("\\u30AC"); // ga (Katakana)
121
122 canonTests[19][0] = str("\\u30AB\\u3099"); canonTests[19][1] = str("\\u30AB\\u3099"); canonTests[19][2] = str("\\u30AC"); // ka + ten
123
124 canonTests[20][0] = str("\\uFF76\\uFF9E"); canonTests[20][1] = str("\\uFF76\\uFF9E"); canonTests[20][2] = str("\\uFF76\\uFF9E"); // hw_ka + hw_ten
125
126 canonTests[21][0] = str("\\u30AB\\uFF9E"); canonTests[21][1] = str("\\u30AB\\uFF9E"); canonTests[21][2] = str("\\u30AB\\uFF9E"); // ka + hw_ten
127
128 canonTests[22][0] = str("\\uFF76\\u3099"); canonTests[22][1] = str("\\uFF76\\u3099"); canonTests[22][2] = str("\\uFF76\\u3099"); // hw_ka + ten
129
130 canonTests[23][0] = str("A\\u0300\\u0316"); canonTests[23][1] = str("A\\u0316\\u0300"); canonTests[23][2] = str("\\u00C0\\u0316");
131
132 /* compatTest */
133 // Input Decomposed Composed
134 compatTests[0][0] = str("cat"); compatTests[0][1] = str("cat"); compatTests[0][2] = str("cat") ;
135
136 compatTests[1][0] = str("\\uFB4f"); compatTests[1][1] = str("\\u05D0\\u05DC"); compatTests[1][2] = str("\\u05D0\\u05DC"); // Alef-Lamed vs. Alef, Lamed
137
138 compatTests[2][0] = str("\\u00C4ffin"); compatTests[2][1] = str("A\\u0308ffin"); compatTests[2][2] = str("\\u00C4ffin") ;
139
140 compatTests[3][0] = str("\\u00C4\\uFB03n"); compatTests[3][1] = str("A\\u0308ffin"); compatTests[3][2] = str("\\u00C4ffin") ; // ffi ligature -> f + f + i
141
142 compatTests[4][0] = str("Henry IV"); compatTests[4][1] = str("Henry IV"); compatTests[4][2] = str("Henry IV") ;
143
144 compatTests[5][0] = str("Henry \\u2163"); compatTests[5][1] = str("Henry IV"); compatTests[5][2] = str("Henry IV") ;
145
146 compatTests[6][0] = str("\\u30AC"); compatTests[6][1] = str("\\u30AB\\u3099"); compatTests[6][2] = str("\\u30AC") ; // ga (Katakana)
147
148 compatTests[7][0] = str("\\u30AB\\u3099"); compatTests[7][1] = str("\\u30AB\\u3099"); compatTests[7][2] = str("\\u30AC") ; // ka + ten
149
150 compatTests[8][0] = str("\\uFF76\\u3099"); compatTests[8][1] = str("\\u30AB\\u3099"); compatTests[8][2] = str("\\u30AC") ; // hw_ka + ten
151
152 /* These two are broken in Unicode 2.1.2 but fixed in 2.1.5 and later */
153 compatTests[9][0] = str("\\uFF76\\uFF9E"); compatTests[9][1] = str("\\u30AB\\u3099"); compatTests[9][2] = str("\\u30AC") ; // hw_ka + hw_ten
154
155 compatTests[10][0] = str("\\u30AB\\uFF9E"); compatTests[10][1] = str("\\u30AB\\u3099"); compatTests[10][2] = str("\\u30AC") ; // ka + hw_ten
156
157 /* Hangul Canonical */
158 // Input Decomposed Composed
159 hangulCanon[0][0] = str("\\ud4db"); hangulCanon[0][1] = str("\\u1111\\u1171\\u11b6"); hangulCanon[0][2] = str("\\ud4db") ;
160
161 hangulCanon[1][0] = str("\\u1111\\u1171\\u11b6"), hangulCanon[1][1] = str("\\u1111\\u1171\\u11b6"), hangulCanon[1][2] = str("\\ud4db");
162 }
163
~BasicNormalizerTest()164 BasicNormalizerTest::~BasicNormalizerTest()
165 {
166 }
167
TestPrevious()168 void BasicNormalizerTest::TestPrevious()
169 {
170 Normalizer* norm = new Normalizer("", UNORM_NFD);
171
172 logln("testing decomp...");
173 uint32_t i;
174 for (i = 0; i < ARRAY_LENGTH(canonTests); i++) {
175 backAndForth(norm, canonTests[i][0]);
176 }
177
178 logln("testing compose...");
179 norm->setMode(UNORM_NFC);
180 for (i = 0; i < ARRAY_LENGTH(canonTests); i++) {
181 backAndForth(norm, canonTests[i][0]);
182 }
183
184 delete norm;
185 }
186
TestDecomp()187 void BasicNormalizerTest::TestDecomp()
188 {
189 Normalizer* norm = new Normalizer("", UNORM_NFD);
190 iterateTest(norm, canonTests, ARRAY_LENGTH(canonTests), 1);
191 staticTest(UNORM_NFD, 0, canonTests, ARRAY_LENGTH(canonTests), 1);
192 delete norm;
193 }
194
TestCompatDecomp()195 void BasicNormalizerTest::TestCompatDecomp()
196 {
197 Normalizer* norm = new Normalizer("", UNORM_NFKD);
198 iterateTest(norm, compatTests, ARRAY_LENGTH(compatTests), 1);
199
200 staticTest(UNORM_NFKD, 0,
201 compatTests, ARRAY_LENGTH(compatTests), 1);
202 delete norm;
203 }
204
TestCanonCompose()205 void BasicNormalizerTest::TestCanonCompose()
206 {
207 Normalizer* norm = new Normalizer("", UNORM_NFC);
208 iterateTest(norm, canonTests, ARRAY_LENGTH(canonTests), 2);
209
210 staticTest(UNORM_NFC, 0, canonTests,
211 ARRAY_LENGTH(canonTests), 2);
212 delete norm;
213 }
214
TestCompatCompose()215 void BasicNormalizerTest::TestCompatCompose()
216 {
217 Normalizer* norm = new Normalizer("", UNORM_NFKC);
218 iterateTest(norm, compatTests, ARRAY_LENGTH(compatTests), 2);
219
220 staticTest(UNORM_NFKC, 0,
221 compatTests, ARRAY_LENGTH(compatTests), 2);
222 delete norm;
223 }
224
225
226 //-------------------------------------------------------------------------------
227
TestHangulCompose()228 void BasicNormalizerTest::TestHangulCompose()
229 {
230 // Make sure that the static composition methods work
231 logln("Canonical composition...");
232 staticTest(UNORM_NFC, 0, hangulCanon, ARRAY_LENGTH(hangulCanon), 2);
233 logln("Compatibility composition...");
234
235 // Now try iterative composition....
236 logln("Static composition...");
237 Normalizer* norm = new Normalizer("", UNORM_NFC);
238 iterateTest(norm, hangulCanon, ARRAY_LENGTH(hangulCanon), 2);
239 norm->setMode(UNORM_NFKC);
240
241 // And finally, make sure you can do it in reverse too
242 logln("Reverse iteration...");
243 norm->setMode(UNORM_NFC);
244 for (uint32_t i = 0; i < ARRAY_LENGTH(hangulCanon); i++) {
245 backAndForth(norm, hangulCanon[i][0]);
246 }
247 delete norm;
248 }
249
TestHangulDecomp()250 void BasicNormalizerTest::TestHangulDecomp()
251 {
252 // Make sure that the static decomposition methods work
253 logln("Canonical decomposition...");
254 staticTest(UNORM_NFD, 0, hangulCanon, ARRAY_LENGTH(hangulCanon), 1);
255 logln("Compatibility decomposition...");
256
257 // Now the iterative decomposition methods...
258 logln("Iterative decomposition...");
259 Normalizer* norm = new Normalizer("", UNORM_NFD);
260 iterateTest(norm, hangulCanon, ARRAY_LENGTH(hangulCanon), 1);
261 norm->setMode(UNORM_NFKD);
262
263 // And finally, make sure you can do it in reverse too
264 logln("Reverse iteration...");
265 norm->setMode(UNORM_NFD);
266 for (uint32_t i = 0; i < ARRAY_LENGTH(hangulCanon); i++) {
267 backAndForth(norm, hangulCanon[i][0]);
268 }
269 delete norm;
270 }
271
272 /**
273 * The Tibetan vowel sign AA, 0f71, was messed up prior to Unicode version 2.1.9.
274 */
TestTibetan()275 void BasicNormalizerTest::TestTibetan() {
276 UnicodeString decomp[1][3];
277 decomp[0][0] = str("\\u0f77");
278 decomp[0][1] = str("\\u0f77");
279 decomp[0][2] = str("\\u0fb2\\u0f71\\u0f80");
280
281 UnicodeString compose[1][3];
282 compose[0][0] = str("\\u0fb2\\u0f71\\u0f80");
283 compose[0][1] = str("\\u0fb2\\u0f71\\u0f80");
284 compose[0][2] = str("\\u0fb2\\u0f71\\u0f80");
285
286 staticTest(UNORM_NFD, 0, decomp, ARRAY_LENGTH(decomp), 1);
287 staticTest(UNORM_NFKD, 0, decomp, ARRAY_LENGTH(decomp), 2);
288 staticTest(UNORM_NFC, 0, compose, ARRAY_LENGTH(compose), 1);
289 staticTest(UNORM_NFKC, 0, compose, ARRAY_LENGTH(compose), 2);
290 }
291
292 /**
293 * Make sure characters in the CompositionExclusion.txt list do not get
294 * composed to.
295 */
TestCompositionExclusion()296 void BasicNormalizerTest::TestCompositionExclusion() {
297 // This list is generated from CompositionExclusion.txt.
298 // Update whenever the normalizer tables are updated. Note
299 // that we test all characters listed, even those that can be
300 // derived from the Unicode DB and are therefore commented
301 // out.
302 // ### TODO read composition exclusion from source/data/unidata file
303 // and test against that
304 UnicodeString EXCLUDED = str(
305 "\\u0340\\u0341\\u0343\\u0344\\u0374\\u037E\\u0387\\u0958"
306 "\\u0959\\u095A\\u095B\\u095C\\u095D\\u095E\\u095F\\u09DC"
307 "\\u09DD\\u09DF\\u0A33\\u0A36\\u0A59\\u0A5A\\u0A5B\\u0A5E"
308 "\\u0B5C\\u0B5D\\u0F43\\u0F4D\\u0F52\\u0F57\\u0F5C\\u0F69"
309 "\\u0F73\\u0F75\\u0F76\\u0F78\\u0F81\\u0F93\\u0F9D\\u0FA2"
310 "\\u0FA7\\u0FAC\\u0FB9\\u1F71\\u1F73\\u1F75\\u1F77\\u1F79"
311 "\\u1F7B\\u1F7D\\u1FBB\\u1FBE\\u1FC9\\u1FCB\\u1FD3\\u1FDB"
312 "\\u1FE3\\u1FEB\\u1FEE\\u1FEF\\u1FF9\\u1FFB\\u1FFD\\u2000"
313 "\\u2001\\u2126\\u212A\\u212B\\u2329\\u232A\\uF900\\uFA10"
314 "\\uFA12\\uFA15\\uFA20\\uFA22\\uFA25\\uFA26\\uFA2A\\uFB1F"
315 "\\uFB2A\\uFB2B\\uFB2C\\uFB2D\\uFB2E\\uFB2F\\uFB30\\uFB31"
316 "\\uFB32\\uFB33\\uFB34\\uFB35\\uFB36\\uFB38\\uFB39\\uFB3A"
317 "\\uFB3B\\uFB3C\\uFB3E\\uFB40\\uFB41\\uFB43\\uFB44\\uFB46"
318 "\\uFB47\\uFB48\\uFB49\\uFB4A\\uFB4B\\uFB4C\\uFB4D\\uFB4E"
319 );
320 UErrorCode status = U_ZERO_ERROR;
321 for (int32_t i=0; i<EXCLUDED.length(); ++i) {
322 UnicodeString a(EXCLUDED.charAt(i));
323 UnicodeString b;
324 UnicodeString c;
325 Normalizer::normalize(a, UNORM_NFKD, 0, b, status);
326 Normalizer::normalize(b, UNORM_NFC, 0, c, status);
327 if (c == a) {
328 errln("FAIL: " + hex(a) + " x DECOMP_COMPAT => " +
329 hex(b) + " x COMPOSE => " +
330 hex(c));
331 } else if (verbose) {
332 logln("Ok: " + hex(a) + " x DECOMP_COMPAT => " +
333 hex(b) + " x COMPOSE => " +
334 hex(c));
335 }
336 }
337 }
338
339 /**
340 * Test for a problem that showed up just before ICU 1.6 release
341 * having to do with combining characters with an index of zero.
342 * Such characters do not participate in any canonical
343 * decompositions. However, having an index of zero means that
344 * they all share one typeMask[] entry, that is, they all have to
345 * map to the same canonical class, which is not the case, in
346 * reality.
347 */
TestZeroIndex()348 void BasicNormalizerTest::TestZeroIndex() {
349 const char* DATA[] = {
350 // Expect col1 x COMPOSE_COMPAT => col2
351 // Expect col2 x DECOMP => col3
352 "A\\u0316\\u0300", "\\u00C0\\u0316", "A\\u0316\\u0300",
353 "A\\u0300\\u0316", "\\u00C0\\u0316", "A\\u0316\\u0300",
354 "A\\u0327\\u0300", "\\u00C0\\u0327", "A\\u0327\\u0300",
355 "c\\u0321\\u0327", "c\\u0321\\u0327", "c\\u0321\\u0327",
356 "c\\u0327\\u0321", "\\u00E7\\u0321", "c\\u0327\\u0321",
357 };
358 int32_t DATA_length = UPRV_LENGTHOF(DATA);
359
360 for (int32_t i=0; i<DATA_length; i+=3) {
361 UErrorCode status = U_ZERO_ERROR;
362 UnicodeString a(DATA[i], "");
363 a = a.unescape();
364 UnicodeString b;
365 Normalizer::normalize(a, UNORM_NFKC, 0, b, status);
366 if (U_FAILURE(status)) {
367 dataerrln("Error calling normalize UNORM_NFKC: %s", u_errorName(status));
368 } else {
369 UnicodeString exp(DATA[i+1], "");
370 exp = exp.unescape();
371 if (b == exp) {
372 logln(UnicodeString("Ok: ") + hex(a) + " x COMPOSE_COMPAT => " + hex(b));
373 } else {
374 errln(UnicodeString("FAIL: ") + hex(a) + " x COMPOSE_COMPAT => " + hex(b) +
375 ", expect " + hex(exp));
376 }
377 }
378 Normalizer::normalize(b, UNORM_NFD, 0, a, status);
379 if (U_FAILURE(status)) {
380 dataerrln("Error calling normalize UNORM_NFD: %s", u_errorName(status));
381 } else {
382 UnicodeString exp = UnicodeString(DATA[i+2], "").unescape();
383 if (a == exp) {
384 logln(UnicodeString("Ok: ") + hex(b) + " x DECOMP => " + hex(a));
385 } else {
386 errln(UnicodeString("FAIL: ") + hex(b) + " x DECOMP => " + hex(a) +
387 ", expect " + hex(exp));
388 }
389 }
390 }
391 }
392
393 /**
394 * Run a few specific cases that are failing for Verisign.
395 */
TestVerisign()396 void BasicNormalizerTest::TestVerisign() {
397 /*
398 > Their input:
399 > 05B8 05B9 05B1 0591 05C3 05B0 05AC 059F
400 > Their output (supposedly from ICU):
401 > 05B8 05B1 05B9 0591 05C3 05B0 05AC 059F
402 > My output from charlint:
403 > 05B1 05B8 05B9 0591 05C3 05B0 05AC 059F
404
405 05B8 05B9 05B1 0591 05C3 05B0 05AC 059F => 05B1 05B8 05B9 0591 05C3 05B0
406 05AC 059F
407
408 U+05B8 18 E HEBREW POINT QAMATS
409 U+05B9 19 F HEBREW POINT HOLAM
410 U+05B1 11 HEBREW POINT HATAF SEGOL
411 U+0591 220 HEBREW ACCENT ETNAHTA
412 U+05C3 0 HEBREW PUNCTUATION SOF PASUQ
413 U+05B0 10 HEBREW POINT SHEVA
414 U+05AC 230 HEBREW ACCENT ILUY
415 U+059F 230 HEBREW ACCENT QARNEY PARA
416
417 U+05B1 11 HEBREW POINT HATAF SEGOL
418 U+05B8 18 HEBREW POINT QAMATS
419 U+05B9 19 HEBREW POINT HOLAM
420 U+0591 220 HEBREW ACCENT ETNAHTA
421 U+05C3 0 HEBREW PUNCTUATION SOF PASUQ
422 U+05B0 10 HEBREW POINT SHEVA
423 U+05AC 230 HEBREW ACCENT ILUY
424 U+059F 230 HEBREW ACCENT QARNEY PARA
425
426 Wrong result:
427 U+05B8 18 HEBREW POINT QAMATS
428 U+05B1 11 HEBREW POINT HATAF SEGOL
429 U+05B9 19 HEBREW POINT HOLAM
430 U+0591 220 HEBREW ACCENT ETNAHTA
431 U+05C3 0 HEBREW PUNCTUATION SOF PASUQ
432 U+05B0 10 HEBREW POINT SHEVA
433 U+05AC 230 HEBREW ACCENT ILUY
434 U+059F 230 HEBREW ACCENT QARNEY PARA
435
436
437 > Their input:
438 >0592 05B7 05BC 05A5 05B0 05C0 05C4 05AD
439 >Their output (supposedly from ICU):
440 >0592 05B0 05B7 05BC 05A5 05C0 05AD 05C4
441 >My output from charlint:
442 >05B0 05B7 05BC 05A5 0592 05C0 05AD 05C4
443
444 0592 05B7 05BC 05A5 05B0 05C0 05C4 05AD => 05B0 05B7 05BC 05A5 0592 05C0
445 05AD 05C4
446
447 U+0592 230 HEBREW ACCENT SEGOL
448 U+05B7 17 HEBREW POINT PATAH
449 U+05BC 21 HEBREW POINT DAGESH OR MAPIQ
450 U+05A5 220 HEBREW ACCENT MERKHA
451 U+05B0 10 HEBREW POINT SHEVA
452 U+05C0 0 HEBREW PUNCTUATION PASEQ
453 U+05C4 230 HEBREW MARK UPPER DOT
454 U+05AD 222 HEBREW ACCENT DEHI
455
456 U+05B0 10 HEBREW POINT SHEVA
457 U+05B7 17 HEBREW POINT PATAH
458 U+05BC 21 HEBREW POINT DAGESH OR MAPIQ
459 U+05A5 220 HEBREW ACCENT MERKHA
460 U+0592 230 HEBREW ACCENT SEGOL
461 U+05C0 0 HEBREW PUNCTUATION PASEQ
462 U+05AD 222 HEBREW ACCENT DEHI
463 U+05C4 230 HEBREW MARK UPPER DOT
464
465 Wrong result:
466 U+0592 230 HEBREW ACCENT SEGOL
467 U+05B0 10 HEBREW POINT SHEVA
468 U+05B7 17 HEBREW POINT PATAH
469 U+05BC 21 HEBREW POINT DAGESH OR MAPIQ
470 U+05A5 220 HEBREW ACCENT MERKHA
471 U+05C0 0 HEBREW PUNCTUATION PASEQ
472 U+05AD 222 HEBREW ACCENT DEHI
473 U+05C4 230 HEBREW MARK UPPER DOT
474 */
475 UnicodeString data[2][3];
476 data[0][0] = str("\\u05B8\\u05B9\\u05B1\\u0591\\u05C3\\u05B0\\u05AC\\u059F");
477 data[0][1] = str("\\u05B1\\u05B8\\u05B9\\u0591\\u05C3\\u05B0\\u05AC\\u059F");
478 data[0][2] = str("");
479 data[1][0] = str("\\u0592\\u05B7\\u05BC\\u05A5\\u05B0\\u05C0\\u05C4\\u05AD");
480 data[1][1] = str("\\u05B0\\u05B7\\u05BC\\u05A5\\u0592\\u05C0\\u05AD\\u05C4");
481 data[1][2] = str("");
482
483 staticTest(UNORM_NFD, 0, data, ARRAY_LENGTH(data), 1);
484 staticTest(UNORM_NFC, 0, data, ARRAY_LENGTH(data), 1);
485 }
486
487 //------------------------------------------------------------------------
488 // Internal utilities
489 //
490
hex(char16_t ch)491 UnicodeString BasicNormalizerTest::hex(char16_t ch) {
492 UnicodeString result;
493 return appendHex(ch, 4, result);
494 }
495
hex(const UnicodeString & s)496 UnicodeString BasicNormalizerTest::hex(const UnicodeString& s) {
497 UnicodeString result;
498 for (int i = 0; i < s.length(); ++i) {
499 if (i != 0) result += static_cast<char16_t>(0x2c)/*,*/;
500 appendHex(s[i], 4, result);
501 }
502 return result;
503 }
504
505
insert(UnicodeString & dest,int pos,UChar32 ch)506 inline static void insert(UnicodeString& dest, int pos, UChar32 ch)
507 {
508 dest.replace(pos, 0, ch);
509 }
510
backAndForth(Normalizer * iter,const UnicodeString & input)511 void BasicNormalizerTest::backAndForth(Normalizer* iter, const UnicodeString& input)
512 {
513 UChar32 ch;
514 UErrorCode status = U_ZERO_ERROR;
515 iter->setText(input, status);
516
517 // Run through the iterator forwards and stick it into a StringBuffer
518 UnicodeString forward;
519 for (ch = iter->first(); ch != iter->DONE; ch = iter->next()) {
520 forward += ch;
521 }
522
523 // Now do it backwards
524 UnicodeString reverse;
525 for (ch = iter->last(); ch != iter->DONE; ch = iter->previous()) {
526 insert(reverse, 0, ch);
527 }
528
529 if (forward != reverse) {
530 errln("Forward/reverse mismatch for input " + hex(input)
531 + ", forward: " + hex(forward) + ", backward: " + hex(reverse));
532 }
533 }
534
staticTest(UNormalizationMode mode,int options,UnicodeString tests[][3],int length,int outCol)535 void BasicNormalizerTest::staticTest(UNormalizationMode mode, int options,
536 UnicodeString tests[][3], int length,
537 int outCol)
538 {
539 UErrorCode status = U_ZERO_ERROR;
540 for (int i = 0; i < length; i++)
541 {
542 UnicodeString& input = tests[i][0];
543 UnicodeString& expect = tests[i][outCol];
544
545 logln("Normalizing '" + input + "' (" + hex(input) + ")" );
546
547 UnicodeString output;
548 Normalizer::normalize(input, mode, options, output, status);
549
550 if (output != expect) {
551 dataerrln(UnicodeString("ERROR: case ") + i + " normalized " + hex(input) + "\n"
552 + " expected " + hex(expect) + "\n"
553 + " static got " + hex(output) );
554 }
555 }
556 }
557
iterateTest(Normalizer * iter,UnicodeString tests[][3],int length,int outCol)558 void BasicNormalizerTest::iterateTest(Normalizer* iter,
559 UnicodeString tests[][3], int length,
560 int outCol)
561 {
562 UErrorCode status = U_ZERO_ERROR;
563 for (int i = 0; i < length; i++)
564 {
565 UnicodeString& input = tests[i][0];
566 UnicodeString& expect = tests[i][outCol];
567
568 logln("Normalizing '" + input + "' (" + hex(input) + ")" );
569
570 iter->setText(input, status);
571 assertEqual(input, expect, iter, UnicodeString("ERROR: case ") + i + " ");
572 }
573 }
574
assertEqual(const UnicodeString & input,const UnicodeString & expected,Normalizer * iter,const UnicodeString & errPrefix)575 void BasicNormalizerTest::assertEqual(const UnicodeString& input,
576 const UnicodeString& expected,
577 Normalizer* iter,
578 const UnicodeString& errPrefix)
579 {
580 UnicodeString result;
581
582 for (UChar32 ch = iter->first(); ch != iter->DONE; ch = iter->next()) {
583 result += ch;
584 }
585 if (result != expected) {
586 dataerrln(errPrefix + "normalized " + hex(input) + "\n"
587 + " expected " + hex(expected) + "\n"
588 + " iterate got " + hex(result) );
589 }
590 }
591
592 // helper class for TestPreviousNext()
593 // simple UTF-32 character iterator
594 class UChar32Iterator {
595 public:
UChar32Iterator(const UChar32 * text,int32_t len,int32_t index)596 UChar32Iterator(const UChar32 *text, int32_t len, int32_t index) :
597 s(text), length(len), i(index) {}
598
current()599 UChar32 current() {
600 if(i<length) {
601 return s[i];
602 } else {
603 return 0xffff;
604 }
605 }
606
next()607 UChar32 next() {
608 if(i<length) {
609 return s[i++];
610 } else {
611 return 0xffff;
612 }
613 }
614
previous()615 UChar32 previous() {
616 if(i>0) {
617 return s[--i];
618 } else {
619 return 0xffff;
620 }
621 }
622
getIndex()623 int32_t getIndex() {
624 return i;
625 }
626 private:
627 const UChar32 *s;
628 int32_t length, i;
629 };
630
631 void
TestPreviousNext(const char16_t * src,int32_t srcLength,const UChar32 * expect,int32_t expectLength,const int32_t * expectIndex,int32_t srcMiddle,int32_t expectMiddle,const char * moves,UNormalizationMode mode,const char * name)632 BasicNormalizerTest::TestPreviousNext(const char16_t *src, int32_t srcLength,
633 const UChar32 *expect, int32_t expectLength,
634 const int32_t *expectIndex, // its length=expectLength+1
635 int32_t srcMiddle, int32_t expectMiddle,
636 const char *moves,
637 UNormalizationMode mode,
638 const char *name) {
639 // Sanity check non-iterative normalization.
640 {
641 IcuTestErrorCode errorCode(*this, "TestPreviousNext");
642 UnicodeString result;
643 Normalizer::normalize(UnicodeString(src, srcLength), mode, 0, result, errorCode);
644 if (errorCode.isFailure()) {
645 dataerrln("error: non-iterative normalization of %s failed: %s",
646 name, errorCode.errorName());
647 errorCode.reset();
648 return;
649 }
650 // UnicodeString::fromUTF32(expect, expectLength)
651 // would turn unpaired surrogates into U+FFFD.
652 for (int32_t i = 0, j = 0; i < result.length(); ++j) {
653 UChar32 c = result.char32At(i);
654 if (c != expect[j]) {
655 errln("error: non-iterative normalization of %s did not yield the expected result",
656 name);
657 }
658 i += U16_LENGTH(c);
659 }
660 }
661
662 // iterators
663 Normalizer iter(src, srcLength, mode);
664
665 // test getStaticClassID and getDynamicClassID
666 if(iter.getDynamicClassID() != Normalizer::getStaticClassID()) {
667 errln("getStaticClassID != getDynamicClassID for Normalizer.");
668 }
669
670 UChar32Iterator iter32(expect, expectLength, expectMiddle);
671
672 UChar32 c1, c2;
673 char m;
674
675 // initially set the indexes into the middle of the strings
676 iter.setIndexOnly(srcMiddle);
677
678 // move around and compare the iteration code points with
679 // the expected ones
680 const char *move=moves;
681 while((m=*move++)!=0) {
682 if(m=='-') {
683 c1=iter.previous();
684 c2=iter32.previous();
685 } else if(m=='0') {
686 c1=iter.current();
687 c2=iter32.current();
688 } else /* m=='+' */ {
689 c1=iter.next();
690 c2=iter32.next();
691 }
692
693 // compare results
694 if(c1!=c2) {
695 // copy the moves until the current (m) move, and terminate
696 char history[64];
697 uprv_strcpy(history, moves);
698 history[move-moves]=0;
699 dataerrln("error: mismatch in Normalizer iteration (%s) at %s: "
700 "got c1=U+%04lx != expected c2=U+%04lx",
701 name, history, c1, c2);
702 break;
703 }
704
705 // compare indexes
706 if(iter.getIndex()!=expectIndex[iter32.getIndex()]) {
707 // copy the moves until the current (m) move, and terminate
708 char history[64];
709 uprv_strcpy(history, moves);
710 history[move-moves]=0;
711 errln("error: index mismatch in Normalizer iteration (%s) at %s: "
712 "Normalizer index %ld expected %ld\n",
713 name, history, iter.getIndex(), expectIndex[iter32.getIndex()]);
714 break;
715 }
716 }
717 }
718
719 void
TestPreviousNext()720 BasicNormalizerTest::TestPreviousNext() {
721 // src and expect strings
722 static const char16_t src[]={
723 U16_LEAD(0x2f999), U16_TRAIL(0x2f999),
724 U16_LEAD(0x1d15f), U16_TRAIL(0x1d15f),
725 0xc4,
726 0x1ed0
727 };
728 static const UChar32 expect[]={
729 0x831d,
730 0x1d158, 0x1d165,
731 0x41, 0x308,
732 0x4f, 0x302, 0x301
733 };
734
735 // expected src indexes corresponding to expect indexes
736 static const int32_t expectIndex[]={
737 0,
738 2, 2,
739 4, 4,
740 5, 5, 5,
741 6 // behind last character
742 };
743
744 // src and expect strings for regression test for j2911
745 static const char16_t src_j2911[]={
746 U16_LEAD(0x2f999), U16_TRAIL(0x2f999),
747 0xdd00, 0xd900, // unpaired surrogates - regression test for j2911
748 0xc4,
749 0x4f, 0x302, 0x301
750 };
751 static const UChar32 expect_j2911[]={
752 0x831d,
753 0xdd00, 0xd900, // unpaired surrogates - regression test for j2911
754 0xc4,
755 0x1ed0
756 };
757
758 // expected src indexes corresponding to expect indexes
759 static const int32_t expectIndex_j2911[]={
760 0,
761 2, 3,
762 4,
763 5,
764 8 // behind last character
765 };
766
767 // initial indexes into the src and expect strings
768 // for both sets of test data
769 enum {
770 SRC_MIDDLE=4,
771 EXPECT_MIDDLE=3,
772 SRC_MIDDLE_2=2,
773 EXPECT_MIDDLE_2=1
774 };
775
776 // movement vector
777 // - for previous(), 0 for current(), + for next()
778 // for both sets of test data
779 static const char *const moves="0+0+0--0-0-+++0--+++++++0--------";
780
781 TestPreviousNext(src, UPRV_LENGTHOF(src),
782 expect, UPRV_LENGTHOF(expect),
783 expectIndex,
784 SRC_MIDDLE, EXPECT_MIDDLE,
785 moves, UNORM_NFD, "basic");
786
787 TestPreviousNext(src_j2911, UPRV_LENGTHOF(src_j2911),
788 expect_j2911, UPRV_LENGTHOF(expect_j2911),
789 expectIndex_j2911,
790 SRC_MIDDLE, EXPECT_MIDDLE,
791 moves, UNORM_NFKC, "j2911");
792
793 // try again from different "middle" indexes
794 TestPreviousNext(src, UPRV_LENGTHOF(src),
795 expect, UPRV_LENGTHOF(expect),
796 expectIndex,
797 SRC_MIDDLE_2, EXPECT_MIDDLE_2,
798 moves, UNORM_NFD, "basic_2");
799
800 TestPreviousNext(src_j2911, UPRV_LENGTHOF(src_j2911),
801 expect_j2911, UPRV_LENGTHOF(expect_j2911),
802 expectIndex_j2911,
803 SRC_MIDDLE_2, EXPECT_MIDDLE_2,
804 moves, UNORM_NFKC, "j2911_2");
805 }
806
TestConcatenate()807 void BasicNormalizerTest::TestConcatenate() {
808 static const char *const
809 cases[][4]={
810 /* mode, left, right, result */
811 {
812 "C",
813 "re",
814 "\\u0301sum\\u00e9",
815 "r\\u00e9sum\\u00e9"
816 },
817 {
818 "C",
819 "a\\u1100",
820 "\\u1161bcdefghijk",
821 "a\\uac00bcdefghijk"
822 },
823 /* ### TODO: add more interesting cases */
824 {
825 "D",
826 "\\u03B1\\u0345",
827 "\\u0C4D\\U000110BA\\U0001D169",
828 "\\u03B1\\U0001D169\\U000110BA\\u0C4D\\u0345"
829 }
830 };
831
832 UnicodeString left, right, expect, result, r;
833 UErrorCode errorCode;
834 UNormalizationMode mode;
835 int32_t i;
836
837 /* test concatenation */
838 for(i=0; i<UPRV_LENGTHOF(cases); ++i) {
839 switch(*cases[i][0]) {
840 case 'C': mode=UNORM_NFC; break;
841 case 'D': mode=UNORM_NFD; break;
842 case 'c': mode=UNORM_NFKC; break;
843 case 'd': mode=UNORM_NFKD; break;
844 default: mode=UNORM_NONE; break;
845 }
846
847 left=UnicodeString(cases[i][1], "").unescape();
848 right=UnicodeString(cases[i][2], "").unescape();
849 expect=UnicodeString(cases[i][3], "").unescape();
850
851 //result=r=UnicodeString();
852 errorCode=U_ZERO_ERROR;
853
854 r=Normalizer::concatenate(left, right, result, mode, 0, errorCode);
855 if(U_FAILURE(errorCode) || /*result!=r ||*/ result!=expect) {
856 dataerrln("error in Normalizer::concatenate(), cases[] fails with "+
857 UnicodeString(u_errorName(errorCode))+", result==expect: expected: "+
858 hex(expect)+" =========> got: " + hex(result));
859 }
860 }
861
862 /* test error cases */
863
864 /* left.getBuffer()==result.getBuffer() */
865 result=r=expect=UnicodeString("zz", "");
866 errorCode=U_UNEXPECTED_TOKEN;
867 r=Normalizer::concatenate(left, right, result, mode, 0, errorCode);
868 if(errorCode!=U_UNEXPECTED_TOKEN || result!=r || !result.isBogus()) {
869 errln("error in Normalizer::concatenate(), violates UErrorCode protocol");
870 }
871
872 left.setToBogus();
873 errorCode=U_ZERO_ERROR;
874 r=Normalizer::concatenate(left, right, result, mode, 0, errorCode);
875 if(errorCode!=U_ILLEGAL_ARGUMENT_ERROR || result!=r || !result.isBogus()) {
876 errln("error in Normalizer::concatenate(), does not detect left.isBogus()");
877 }
878 }
879
880 // reference implementation of Normalizer::compare
881 static int32_t
ref_norm_compare(const UnicodeString & s1,const UnicodeString & s2,uint32_t options,UErrorCode & errorCode)882 ref_norm_compare(const UnicodeString &s1, const UnicodeString &s2, uint32_t options, UErrorCode &errorCode) {
883 UnicodeString r1, r2, t1, t2;
884 int32_t normOptions = static_cast<int32_t>(options >> UNORM_COMPARE_NORM_OPTIONS_SHIFT);
885
886 if(options&U_COMPARE_IGNORE_CASE) {
887 Normalizer::decompose(s1, false, normOptions, r1, errorCode);
888 Normalizer::decompose(s2, false, normOptions, r2, errorCode);
889
890 r1.foldCase(options);
891 r2.foldCase(options);
892 } else {
893 r1=s1;
894 r2=s2;
895 }
896
897 Normalizer::decompose(r1, false, normOptions, t1, errorCode);
898 Normalizer::decompose(r2, false, normOptions, t2, errorCode);
899
900 if(options&U_COMPARE_CODE_POINT_ORDER) {
901 return t1.compareCodePointOrder(t2);
902 } else {
903 return t1.compare(t2);
904 }
905 }
906
907 // test wrapper for Normalizer::compare, sets UNORM_INPUT_IS_FCD appropriately
908 static int32_t
_norm_compare(const UnicodeString & s1,const UnicodeString & s2,uint32_t options,UErrorCode & errorCode)909 _norm_compare(const UnicodeString &s1, const UnicodeString &s2, uint32_t options, UErrorCode &errorCode) {
910 int32_t normOptions = static_cast<int32_t>(options >> UNORM_COMPARE_NORM_OPTIONS_SHIFT);
911
912 if( UNORM_YES==Normalizer::quickCheck(s1, UNORM_FCD, normOptions, errorCode) &&
913 UNORM_YES==Normalizer::quickCheck(s2, UNORM_FCD, normOptions, errorCode)) {
914 options|=UNORM_INPUT_IS_FCD;
915 }
916
917 return Normalizer::compare(s1, s2, options, errorCode);
918 }
919
920 // reference implementation of UnicodeString::caseCompare
921 static int32_t
ref_case_compare(const UnicodeString & s1,const UnicodeString & s2,uint32_t options)922 ref_case_compare(const UnicodeString &s1, const UnicodeString &s2, uint32_t options) {
923 UnicodeString t1, t2;
924
925 t1=s1;
926 t2=s2;
927
928 t1.foldCase(options);
929 t2.foldCase(options);
930
931 if(options&U_COMPARE_CODE_POINT_ORDER) {
932 return t1.compareCodePointOrder(t2);
933 } else {
934 return t1.compare(t2);
935 }
936 }
937
938 // reduce an integer to -1/0/1
939 static inline int32_t
_sign(int32_t value)940 _sign(int32_t value) {
941 if(value==0) {
942 return 0;
943 } else {
944 return (value>>31)|1;
945 }
946 }
947
948 static const char *
_signString(int32_t value)949 _signString(int32_t value) {
950 if(value<0) {
951 return "<0";
952 } else if(value==0) {
953 return "=0";
954 } else /* value>0 */ {
955 return ">0";
956 }
957 }
958
959 void
TestCompare()960 BasicNormalizerTest::TestCompare() {
961 // test Normalizer::compare and unorm_compare (thinly wrapped by the former)
962 // by comparing it with its semantic equivalent
963 // since we trust the pieces, this is sufficient
964
965 // test each string with itself and each other
966 // each time with all options
967 static const char *const
968 strings[]={
969 // some cases from NormalizationTest.txt
970 // 0..3
971 "D\\u031B\\u0307\\u0323",
972 "\\u1E0C\\u031B\\u0307",
973 "D\\u031B\\u0323\\u0307",
974 "d\\u031B\\u0323\\u0307",
975
976 // 4..6
977 "\\u00E4",
978 "a\\u0308",
979 "A\\u0308",
980
981 // Angstrom sign = A ring
982 // 7..10
983 "\\u212B",
984 "\\u00C5",
985 "A\\u030A",
986 "a\\u030A",
987
988 // 11.14
989 "a\\u059A\\u0316\\u302A\\u032Fb",
990 "a\\u302A\\u0316\\u032F\\u059Ab",
991 "a\\u302A\\u0316\\u032F\\u059Ab",
992 "A\\u059A\\u0316\\u302A\\u032Fb",
993
994 // from ICU case folding tests
995 // 15..20
996 "A\\u00df\\u00b5\\ufb03\\U0001040c\\u0131",
997 "ass\\u03bcffi\\U00010434i",
998 "\\u0061\\u0042\\u0131\\u03a3\\u00df\\ufb03\\ud93f\\udfff",
999 "\\u0041\\u0062\\u0069\\u03c3\\u0073\\u0053\\u0046\\u0066\\u0049\\ud93f\\udfff",
1000 "\\u0041\\u0062\\u0131\\u03c3\\u0053\\u0073\\u0066\\u0046\\u0069\\ud93f\\udfff",
1001 "\\u0041\\u0062\\u0069\\u03c3\\u0073\\u0053\\u0046\\u0066\\u0049\\ud93f\\udffd",
1002
1003 // U+d800 U+10001 see implementation comment in unorm_cmpEquivFold
1004 // vs. U+10000 at bottom - code point order
1005 // 21..22
1006 "\\ud800\\ud800\\udc01",
1007 "\\ud800\\udc00",
1008
1009 // other code point order tests from ustrtest.cpp
1010 // 23..31
1011 "\\u20ac\\ud801",
1012 "\\u20ac\\ud800\\udc00",
1013 "\\ud800",
1014 "\\ud800\\uff61",
1015 "\\udfff",
1016 "\\uff61\\udfff",
1017 "\\uff61\\ud800\\udc02",
1018 "\\ud800\\udc02",
1019 "\\ud84d\\udc56",
1020
1021 // long strings, see cnormtst.c/TestNormCoverage()
1022 // equivalent if case-insensitive
1023 // 32..33
1024 "\\uAD8B\\uAD8B\\uAD8B\\uAD8B"
1025 "\\U0001d15e\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1026 "\\U0001d15e\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1027 "\\U0001d15e\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1028 "\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1029 "\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1030 "aaaaaaaaaaaaaaaaaazzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz"
1031 "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb"
1032 "ccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc"
1033 "ddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd"
1034 "\\uAD8B\\uAD8B\\uAD8B\\uAD8B"
1035 "d\\u031B\\u0307\\u0323",
1036
1037 "\\u1100\\u116f\\u11aa\\uAD8B\\uAD8B\\u1100\\u116f\\u11aa"
1038 "\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1039 "\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1040 "\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1041 "\\U0001d15e\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1042 "\\U0001d15e\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1043 "aaaaaaaaaaAAAAAAAAZZZZZZZZZZZZZZZZzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz"
1044 "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb"
1045 "ccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc"
1046 "ddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd"
1047 "\\u1100\\u116f\\u11aa\\uAD8B\\uAD8B\\u1100\\u116f\\u11aa"
1048 "\\u1E0C\\u031B\\u0307",
1049
1050 // some strings that may make a difference whether the compare function
1051 // case-folds or decomposes first
1052 // 34..41
1053 "\\u0360\\u0345\\u0334",
1054 "\\u0360\\u03b9\\u0334",
1055
1056 "\\u0360\\u1f80\\u0334",
1057 "\\u0360\\u03b1\\u0313\\u03b9\\u0334",
1058
1059 "\\u0360\\u1ffc\\u0334",
1060 "\\u0360\\u03c9\\u03b9\\u0334",
1061
1062 "a\\u0360\\u0345\\u0360\\u0345b",
1063 "a\\u0345\\u0360\\u0345\\u0360b",
1064
1065 // interesting cases for canonical caseless match with turkic i handling
1066 // 42..43
1067 "\\u00cc",
1068 "\\u0069\\u0300",
1069
1070 // strings with post-Unicode 3.2 normalization or normalization corrections
1071 // 44..45
1072 "\\u00e4\\u193b\\U0002f868",
1073 "\\u0061\\u193b\\u0308\\u36fc",
1074
1075 // empty string
1076 // 46
1077 ""
1078 };
1079
1080 UnicodeString s[100]; // at least as many items as in strings[] !
1081
1082 // all combinations of options
1083 // UNORM_INPUT_IS_FCD is set automatically if both input strings fulfill FCD conditions
1084 // set UNORM_UNICODE_3_2 in one additional combination
1085 static const struct {
1086 uint32_t options;
1087 const char *name;
1088 } opt[]={
1089 { 0, "default" },
1090 { U_COMPARE_CODE_POINT_ORDER, "c.p. order" },
1091 { U_COMPARE_IGNORE_CASE, "ignore case" },
1092 { U_COMPARE_CODE_POINT_ORDER|U_COMPARE_IGNORE_CASE, "c.p. order & ignore case" },
1093 { U_COMPARE_IGNORE_CASE|U_FOLD_CASE_EXCLUDE_SPECIAL_I, "ignore case & special i" },
1094 { U_COMPARE_CODE_POINT_ORDER|U_COMPARE_IGNORE_CASE|U_FOLD_CASE_EXCLUDE_SPECIAL_I, "c.p. order & ignore case & special i" },
1095 { UNORM_UNICODE_3_2<<UNORM_COMPARE_NORM_OPTIONS_SHIFT, "Unicode 3.2" }
1096 };
1097
1098 int32_t i, j, k, count=UPRV_LENGTHOF(strings);
1099 int32_t result, refResult;
1100
1101 UErrorCode errorCode;
1102
1103 // create the UnicodeStrings
1104 for(i=0; i<count; ++i) {
1105 s[i]=UnicodeString(strings[i], "").unescape();
1106 }
1107
1108 // test them each with each other
1109 for(i=0; i<count; ++i) {
1110 for(j=i; j<count; ++j) {
1111 for(k=0; k<UPRV_LENGTHOF(opt); ++k) {
1112 // test Normalizer::compare
1113 errorCode=U_ZERO_ERROR;
1114 result=_norm_compare(s[i], s[j], opt[k].options, errorCode);
1115 refResult=ref_norm_compare(s[i], s[j], opt[k].options, errorCode);
1116 if(_sign(result)!=_sign(refResult)) {
1117 errln("Normalizer::compare(%d, %d, %s)%s should be %s %s",
1118 i, j, opt[k].name, _signString(result), _signString(refResult),
1119 U_SUCCESS(errorCode) ? "" : u_errorName(errorCode));
1120 }
1121
1122 // test UnicodeString::caseCompare - same internal implementation function
1123 if(opt[k].options&U_COMPARE_IGNORE_CASE) {
1124 errorCode=U_ZERO_ERROR;
1125 result=s[i].caseCompare(s[j], opt[k].options);
1126 refResult=ref_case_compare(s[i], s[j], opt[k].options);
1127 if(_sign(result)!=_sign(refResult)) {
1128 errln("UniStr::caseCompare(%d, %d, %s)%s should be %s %s",
1129 i, j, opt[k].name, _signString(result), _signString(refResult),
1130 U_SUCCESS(errorCode) ? "" : u_errorName(errorCode));
1131 }
1132 }
1133 }
1134 }
1135 }
1136
1137 // test cases with i and I to make sure Turkic works
1138 static const char16_t iI[]={ 0x49, 0x69, 0x130, 0x131 };
1139 UnicodeSet iSet, set;
1140
1141 UnicodeString s1, s2;
1142
1143 const Normalizer2Impl *nfcImpl=Normalizer2Factory::getNFCImpl(errorCode);
1144 if(U_FAILURE(errorCode) || !nfcImpl->ensureCanonIterData(errorCode)) {
1145 dataerrln("Normalizer2Factory::getNFCImpl().ensureCanonIterData() failed: %s",
1146 u_errorName(errorCode));
1147 return;
1148 }
1149
1150 // collect all sets into one for contiguous output
1151 for(i=0; i<UPRV_LENGTHOF(iI); ++i) {
1152 if(nfcImpl->getCanonStartSet(iI[i], iSet)) {
1153 set.addAll(iSet);
1154 }
1155 }
1156
1157 // test all of these precomposed characters
1158 const Normalizer2 *nfcNorm2=Normalizer2::getNFCInstance(errorCode);
1159 UnicodeSetIterator it(set);
1160 while(it.next() && !it.isString()) {
1161 UChar32 c=it.getCodepoint();
1162 if(!nfcNorm2->getDecomposition(c, s2)) {
1163 dataerrln("NFC.getDecomposition(i-composite U+%04lx) failed", static_cast<long>(c));
1164 return;
1165 }
1166
1167 s1.setTo(c);
1168 for(k=0; k<UPRV_LENGTHOF(opt); ++k) {
1169 // test Normalizer::compare
1170 errorCode=U_ZERO_ERROR;
1171 result=_norm_compare(s1, s2, opt[k].options, errorCode);
1172 refResult=ref_norm_compare(s1, s2, opt[k].options, errorCode);
1173 if(_sign(result)!=_sign(refResult)) {
1174 errln("Normalizer::compare(U+%04x with its NFD, %s)%s should be %s %s",
1175 c, opt[k].name, _signString(result), _signString(refResult),
1176 U_SUCCESS(errorCode) ? "" : u_errorName(errorCode));
1177 }
1178
1179 // test UnicodeString::caseCompare - same internal implementation function
1180 if(opt[k].options&U_COMPARE_IGNORE_CASE) {
1181 errorCode=U_ZERO_ERROR;
1182 result=s1.caseCompare(s2, opt[k].options);
1183 refResult=ref_case_compare(s1, s2, opt[k].options);
1184 if(_sign(result)!=_sign(refResult)) {
1185 errln("UniStr::caseCompare(U+%04x with its NFD, %s)%s should be %s %s",
1186 c, opt[k].name, _signString(result), _signString(refResult),
1187 U_SUCCESS(errorCode) ? "" : u_errorName(errorCode));
1188 }
1189 }
1190 }
1191 }
1192
1193 // test getDecomposition() for some characters that do not decompose
1194 if( nfcNorm2->getDecomposition(0x20, s2) ||
1195 nfcNorm2->getDecomposition(0x4e00, s2) ||
1196 nfcNorm2->getDecomposition(0x20002, s2)
1197 ) {
1198 errln("NFC.getDecomposition() returns true for characters which do not have decompositions");
1199 }
1200
1201 // test getRawDecomposition() for some characters that do not decompose
1202 if( nfcNorm2->getRawDecomposition(0x20, s2) ||
1203 nfcNorm2->getRawDecomposition(0x4e00, s2) ||
1204 nfcNorm2->getRawDecomposition(0x20002, s2)
1205 ) {
1206 errln("NFC.getRawDecomposition() returns true for characters which do not have decompositions");
1207 }
1208
1209 // test composePair() for some pairs of characters that do not compose
1210 if( nfcNorm2->composePair(0x20, 0x301)>=0 ||
1211 nfcNorm2->composePair(0x61, 0x305)>=0 ||
1212 nfcNorm2->composePair(0x1100, 0x1160)>=0 ||
1213 nfcNorm2->composePair(0xac00, 0x11a7)>=0 ||
1214 nfcNorm2->composePair(0x1100, 0x80000020)>= 0 || // ICU-22635
1215 nfcNorm2->composePair(0xac00, 0x80000020)>= 0 // ICU-22635
1216 ) {
1217 errln("NFC.composePair() incorrectly composes some pairs of characters");
1218 }
1219
1220 // test FilteredNormalizer2::getDecomposition()
1221 UnicodeSet filter(UNICODE_STRING_SIMPLE("[^\\u00a0-\\u00ff]"), errorCode);
1222 FilteredNormalizer2 fn2(*nfcNorm2, filter);
1223 if( fn2.getDecomposition(0xe4, s1) || !fn2.getDecomposition(0x100, s2) ||
1224 s2.length()!=2 || s2[0]!=0x41 || s2[1]!=0x304
1225 ) {
1226 errln("FilteredNormalizer2(NFC, ^A0-FF).getDecomposition() failed");
1227 }
1228
1229 // test FilteredNormalizer2::getRawDecomposition()
1230 if( fn2.getRawDecomposition(0xe4, s1) || !fn2.getRawDecomposition(0x100, s2) ||
1231 s2.length()!=2 || s2[0]!=0x41 || s2[1]!=0x304
1232 ) {
1233 errln("FilteredNormalizer2(NFC, ^A0-FF).getRawDecomposition() failed");
1234 }
1235
1236 // test FilteredNormalizer2::composePair()
1237 if( 0x100!=fn2.composePair(0x41, 0x304) ||
1238 fn2.composePair(0xc7, 0x301)>=0 // unfiltered result: U+1E08
1239 ) {
1240 errln("FilteredNormalizer2(NFC, ^A0-FF).composePair() failed");
1241 }
1242 }
1243
1244 // verify that case-folding does not un-FCD strings
1245 int32_t
countFoldFCDExceptions(uint32_t foldingOptions)1246 BasicNormalizerTest::countFoldFCDExceptions(uint32_t foldingOptions) {
1247 UnicodeString s, fold, d;
1248 UChar32 c;
1249 int32_t count;
1250 uint8_t cc, trailCC, foldCC, foldTrailCC;
1251 UNormalizationCheckResult qcResult;
1252 int8_t category;
1253 UBool isNFD;
1254 UErrorCode errorCode;
1255
1256 logln("Test if case folding may un-FCD a string (folding options %04lx)", foldingOptions);
1257
1258 count=0;
1259 for(c=0; c<=0x10ffff; ++c) {
1260 errorCode = U_ZERO_ERROR;
1261 category=u_charType(c);
1262 if(category==U_UNASSIGNED) {
1263 continue; // skip unassigned code points
1264 }
1265 if(c==0xac00) {
1266 c=0xd7a3; // skip Hangul - no case folding there
1267 continue;
1268 }
1269 // skip Han blocks - no case folding there either
1270 if(c==0x3400) {
1271 c=0x4db5;
1272 continue;
1273 }
1274 if(c==0x4e00) {
1275 c=0x9fa5;
1276 continue;
1277 }
1278 if(c==0x20000) {
1279 c=0x2a6d6;
1280 continue;
1281 }
1282
1283 s.setTo(c);
1284
1285 // get leading and trailing cc for c
1286 Normalizer::decompose(s, false, 0, d, errorCode);
1287 isNFD= s==d;
1288 cc=u_getCombiningClass(d.char32At(0));
1289 trailCC=u_getCombiningClass(d.char32At(d.length()-1));
1290
1291 // get leading and trailing cc for the case-folding of c
1292 s.foldCase(foldingOptions);
1293 Normalizer::decompose(s, false, 0, d, errorCode);
1294 foldCC=u_getCombiningClass(d.char32At(0));
1295 foldTrailCC=u_getCombiningClass(d.char32At(d.length()-1));
1296
1297 qcResult=Normalizer::quickCheck(s, UNORM_FCD, errorCode);
1298
1299 if (U_FAILURE(errorCode)) {
1300 ++count;
1301 dataerrln("U+%04lx: Failed with error %s", u_errorName(errorCode));
1302 }
1303
1304 // bad:
1305 // - character maps to empty string: adjacent characters may then need reordering
1306 // - folding has different leading/trailing cc's, and they don't become just 0
1307 // - folding itself is not FCD
1308 if( qcResult!=UNORM_YES ||
1309 s.isEmpty() ||
1310 (cc!=foldCC && foldCC!=0) || (trailCC!=foldTrailCC && foldTrailCC!=0)
1311 ) {
1312 ++count;
1313 dataerrln("U+%04lx: case-folding may un-FCD a string (folding options %04lx)", c, foldingOptions);
1314 dataerrln(" cc %02x trailCC %02x foldCC(U+%04lx) %02x foldTrailCC(U+%04lx) %02x quickCheck(folded)=%d", cc, trailCC, d.char32At(0), foldCC, d.char32At(d.length()-1), foldTrailCC, qcResult);
1315 continue;
1316 }
1317
1318 // also bad:
1319 // if a code point is in NFD but its case folding is not, then
1320 // unorm_compare will also fail
1321 if(isNFD && UNORM_YES!=Normalizer::quickCheck(s, UNORM_NFD, errorCode)) {
1322 ++count;
1323 errln("U+%04lx: case-folding un-NFDs this character (folding options %04lx)", c, foldingOptions);
1324 }
1325 }
1326
1327 logln("There are %ld code points for which case-folding may un-FCD a string (folding options %04lx)", count, foldingOptions);
1328 return count;
1329 }
1330
1331 void
FindFoldFCDExceptions()1332 BasicNormalizerTest::FindFoldFCDExceptions() {
1333 int32_t count;
1334
1335 count=countFoldFCDExceptions(0);
1336 count+=countFoldFCDExceptions(U_FOLD_CASE_EXCLUDE_SPECIAL_I);
1337 if(count>0) {
1338 /*
1339 * If case-folding un-FCDs any strings, then unorm_compare() must be
1340 * re-implemented.
1341 * It currently assumes that one can check for FCD then case-fold
1342 * and then still have FCD strings for raw decomposition without reordering.
1343 */
1344 dataerrln("error: There are %ld code points for which case-folding may un-FCD a string for all folding options.\n"
1345 "See comment in BasicNormalizerTest::FindFoldFCDExceptions()!", count);
1346 }
1347 }
1348
1349 static void
initExpectedSkippables(UnicodeSet skipSets[UNORM_MODE_COUNT],UErrorCode & errorCode)1350 initExpectedSkippables(UnicodeSet skipSets[UNORM_MODE_COUNT], UErrorCode &errorCode) {
1351 skipSets[UNORM_NFD].applyPattern(
1352 UNICODE_STRING_SIMPLE("[[:NFD_QC=Yes:]&[:ccc=0:]]"), errorCode);
1353 skipSets[UNORM_NFC].applyPattern(
1354 UNICODE_STRING_SIMPLE("[[:NFC_QC=Yes:]&[:ccc=0:]-[:HST=LV:]]"), errorCode);
1355 skipSets[UNORM_NFKD].applyPattern(
1356 UNICODE_STRING_SIMPLE("[[:NFKD_QC=Yes:]&[:ccc=0:]]"), errorCode);
1357 skipSets[UNORM_NFKC].applyPattern(
1358 UNICODE_STRING_SIMPLE("[[:NFKC_QC=Yes:]&[:ccc=0:]-[:HST=LV:]]"), errorCode);
1359
1360 // Remove from the NFC and NFKC sets all those characters that change
1361 // when a back-combining character is added.
1362 // First, get all of the back-combining characters and their combining classes.
1363 UnicodeSet combineBack("[:NFC_QC=Maybe:]", errorCode);
1364 int32_t numCombineBack=combineBack.size();
1365 int32_t *combineBackCharsAndCc=new int32_t[numCombineBack*2];
1366 UnicodeSetIterator iter(combineBack);
1367 for(int32_t i=0; i<numCombineBack; ++i) {
1368 iter.next();
1369 UChar32 c=iter.getCodepoint();
1370 combineBackCharsAndCc[2*i]=c;
1371 combineBackCharsAndCc[2*i+1]=u_getCombiningClass(c);
1372 }
1373
1374 // We need not look at control codes, Han characters nor Hangul LVT syllables because they
1375 // do not combine forward. LV syllables are already removed.
1376 UnicodeSet notInteresting("[[:C:][:Unified_Ideograph:][:HST=LVT:]]", errorCode);
1377 LocalPointer<UnicodeSet> unsure(&(skipSets[UNORM_NFC].clone())->removeAll(notInteresting));
1378 // System.out.format("unsure.size()=%d\n", unsure.size());
1379
1380 // For each character about which we are unsure, see if it changes when we add
1381 // one of the back-combining characters.
1382 const Normalizer2 *norm2=Normalizer2::getNFCInstance(errorCode);
1383 UnicodeString s;
1384 iter.reset(*unsure);
1385 while(iter.next()) {
1386 UChar32 c=iter.getCodepoint();
1387 s.setTo(c);
1388 int32_t cLength=s.length();
1389 int32_t tccc=u_getIntPropertyValue(c, UCHAR_TRAIL_CANONICAL_COMBINING_CLASS);
1390 for(int32_t i=0; i<numCombineBack; ++i) {
1391 // If c's decomposition ends with a character with non-zero combining class, then
1392 // c can only change if it combines with a character with a non-zero combining class.
1393 int32_t cc2=combineBackCharsAndCc[2*i+1];
1394 if(tccc==0 || cc2!=0) {
1395 UChar32 c2=combineBackCharsAndCc[2*i];
1396 s.append(c2);
1397 if(!norm2->isNormalized(s, errorCode)) {
1398 // System.out.format("remove U+%04x (tccc=%d) + U+%04x (cc=%d)\n", c, tccc, c2, cc2);
1399 skipSets[UNORM_NFC].remove(c);
1400 skipSets[UNORM_NFKC].remove(c);
1401 break;
1402 }
1403 s.truncate(cLength);
1404 }
1405 }
1406 }
1407 delete [] combineBackCharsAndCc;
1408 }
1409
1410 static const char *const kModeStrings[UNORM_MODE_COUNT] = {
1411 "?", "none", "D", "KD", "C", "KC", "FCD"
1412 };
1413
1414 void
TestSkippable()1415 BasicNormalizerTest::TestSkippable() {
1416 UnicodeSet diff, skipSets[UNORM_MODE_COUNT], expectSets[UNORM_MODE_COUNT];
1417 UnicodeString s, pattern;
1418
1419 /* build NF*Skippable sets from runtime data */
1420 IcuTestErrorCode errorCode(*this, "TestSkippable");
1421 skipSets[UNORM_NFD].applyPattern(UNICODE_STRING_SIMPLE("[:NFD_Inert:]"), errorCode);
1422 skipSets[UNORM_NFKD].applyPattern(UNICODE_STRING_SIMPLE("[:NFKD_Inert:]"), errorCode);
1423 skipSets[UNORM_NFC].applyPattern(UNICODE_STRING_SIMPLE("[:NFC_Inert:]"), errorCode);
1424 skipSets[UNORM_NFKC].applyPattern(UNICODE_STRING_SIMPLE("[:NFKC_Inert:]"), errorCode);
1425 if(errorCode.errDataIfFailureAndReset("UnicodeSet(NF..._Inert) failed")) {
1426 return;
1427 }
1428
1429 /* get expected sets from hardcoded patterns */
1430 initExpectedSkippables(expectSets, errorCode);
1431 errorCode.assertSuccess();
1432
1433 for(int32_t i=UNORM_NONE; i<UNORM_MODE_COUNT; ++i) {
1434 if(skipSets[i]!=expectSets[i]) {
1435 const char *ms=kModeStrings[i];
1436 errln("error: TestSkippable skipSets[%s]!=expectedSets[%s]\n", ms, ms);
1437 // Note: This used to depend on hardcoded UnicodeSet patterns generated by
1438 // Mark's unicodetools.com.ibm.text.UCD.NFSkippable, by
1439 // running com.ibm.text.UCD.Main with the option NFSkippable.
1440 // Since ICU 4.6/Unicode 6, we are generating the
1441 // expectSets ourselves in initSkippables().
1442
1443 s=UNICODE_STRING_SIMPLE("skip-expect=");
1444 (diff=skipSets[i]).removeAll(expectSets[i]).toPattern(pattern, true);
1445 s.append(pattern);
1446
1447 pattern.remove();
1448 s.append(UNICODE_STRING_SIMPLE("\n\nexpect-skip="));
1449 (diff=expectSets[i]).removeAll(skipSets[i]).toPattern(pattern, true);
1450 s.append(pattern);
1451 s.append(UNICODE_STRING_SIMPLE("\n\n"));
1452
1453 errln(s);
1454 }
1455 }
1456 }
1457
1458 struct StringPair { const char *input, *expected; };
1459
1460 void
TestCustomComp()1461 BasicNormalizerTest::TestCustomComp() {
1462 static const StringPair pairs[]={
1463 // ICU 63 normalization with UCPTrie requires inert surrogate code points.
1464 // { "\\uD801\\uE000\\uDFFE", "" },
1465 // { "\\uD800\\uD801\\uE000\\uDFFE\\uDFFF", "\\uD7FF\\uFFFF" },
1466 // { "\\uD800\\uD801\\uDFFE\\uDFFF", "\\uD7FF\\U000107FE\\uFFFF" },
1467 { "\\uD801\\uE000\\uDFFE", "\\uD801\\uDFFE" },
1468 { "\\uD800\\uD801\\uE000\\uDFFE\\uDFFF", "\\uD800\\uD801\\uDFFE\\uDFFF" },
1469 { "\\uD800\\uD801\\uDFFE\\uDFFF", "\\uD800\\U000107FE\\uDFFF" },
1470
1471 { "\\uE001\\U000110B9\\u0345\\u0308\\u0327", "\\uE002\\U000110B9\\u0327\\u0345" },
1472 { "\\uE010\\U000F0011\\uE012", "\\uE011\\uE012" },
1473 { "\\uE010\\U000F0011\\U000F0011\\uE012", "\\uE011\\U000F0010" },
1474 { "\\uE111\\u1161\\uE112\\u1162", "\\uAE4C\\u1102\\u0062\\u1162" },
1475 { "\\uFFF3\\uFFF7\\U00010036\\U00010077", "\\U00010037\\U00010037\\uFFF6\\U00010037" }
1476 };
1477 IcuTestErrorCode errorCode(*this, "BasicNormalizerTest/TestCustomComp");
1478 const Normalizer2 *customNorm2=
1479 Normalizer2::getInstance(loadTestData(errorCode), "testnorm",
1480 UNORM2_COMPOSE, errorCode);
1481 if(errorCode.errDataIfFailureAndReset("unable to load testdata/testnorm.nrm")) {
1482 return;
1483 }
1484 for(int32_t i=0; i<UPRV_LENGTHOF(pairs); ++i) {
1485 const StringPair &pair=pairs[i];
1486 UnicodeString input=UnicodeString(pair.input, -1, US_INV).unescape();
1487 UnicodeString expected=UnicodeString(pair.expected, -1, US_INV).unescape();
1488 UnicodeString result=customNorm2->normalize(input, errorCode);
1489 if(result!=expected) {
1490 errln("custom compose Normalizer2 did not normalize input %d as expected", i);
1491 }
1492 }
1493 }
1494
1495 void
TestCustomFCC()1496 BasicNormalizerTest::TestCustomFCC() {
1497 static const StringPair pairs[]={
1498 // ICU 63 normalization with UCPTrie requires inert surrogate code points.
1499 // { "\\uD801\\uE000\\uDFFE", "" },
1500 // { "\\uD800\\uD801\\uE000\\uDFFE\\uDFFF", "\\uD7FF\\uFFFF" },
1501 // { "\\uD800\\uD801\\uDFFE\\uDFFF", "\\uD7FF\\U000107FE\\uFFFF" },
1502 { "\\uD801\\uE000\\uDFFE", "\\uD801\\uDFFE" },
1503 { "\\uD800\\uD801\\uE000\\uDFFE\\uDFFF", "\\uD800\\uD801\\uDFFE\\uDFFF" },
1504 { "\\uD800\\uD801\\uDFFE\\uDFFF", "\\uD800\\U000107FE\\uDFFF" },
1505
1506 // The following expected result is different from CustomComp
1507 // because of only-contiguous composition.
1508 { "\\uE001\\U000110B9\\u0345\\u0308\\u0327", "\\uE001\\U000110B9\\u0327\\u0308\\u0345" },
1509 { "\\uE010\\U000F0011\\uE012", "\\uE011\\uE012" },
1510 { "\\uE010\\U000F0011\\U000F0011\\uE012", "\\uE011\\U000F0010" },
1511 { "\\uE111\\u1161\\uE112\\u1162", "\\uAE4C\\u1102\\u0062\\u1162" },
1512 { "\\uFFF3\\uFFF7\\U00010036\\U00010077", "\\U00010037\\U00010037\\uFFF6\\U00010037" }
1513 };
1514 IcuTestErrorCode errorCode(*this, "BasicNormalizerTest/TestCustomFCC");
1515 const Normalizer2 *customNorm2=
1516 Normalizer2::getInstance(loadTestData(errorCode), "testnorm",
1517 UNORM2_COMPOSE_CONTIGUOUS, errorCode);
1518 if(errorCode.errDataIfFailureAndReset("unable to load testdata/testnorm.nrm")) {
1519 return;
1520 }
1521 for(int32_t i=0; i<UPRV_LENGTHOF(pairs); ++i) {
1522 const StringPair &pair=pairs[i];
1523 UnicodeString input=UnicodeString(pair.input, -1, US_INV).unescape();
1524 UnicodeString expected=UnicodeString(pair.expected, -1, US_INV).unescape();
1525 UnicodeString result=customNorm2->normalize(input, errorCode);
1526 if(result!=expected) {
1527 errln("custom FCC Normalizer2 did not normalize input %d as expected", i);
1528 }
1529 }
1530 }
1531
1532 /* Improve code coverage of Normalizer2 */
1533 void
TestFilteredNormalizer2Coverage()1534 BasicNormalizerTest::TestFilteredNormalizer2Coverage() {
1535 UErrorCode errorCode = U_ZERO_ERROR;
1536 const Normalizer2 *nfcNorm2=Normalizer2::getNFCInstance(errorCode);
1537 if (U_FAILURE(errorCode)) {
1538 dataerrln("Normalizer2::getNFCInstance() call failed - %s", u_errorName(errorCode));
1539 return;
1540 }
1541 UnicodeSet filter(UNICODE_STRING_SIMPLE("[^\\u00a0-\\u00ff\\u0310-\\u031f]"), errorCode);
1542 FilteredNormalizer2 fn2(*nfcNorm2, filter);
1543
1544 UChar32 char32 = 0x0054;
1545
1546 if (fn2.isInert(char32)) {
1547 errln("FilteredNormalizer2.isInert() failed.");
1548 }
1549
1550 if (fn2.hasBoundaryAfter(char32)) {
1551 errln("FilteredNormalizer2.hasBoundaryAfter() failed.");
1552 }
1553
1554 UChar32 c;
1555 for(c=0; c<=0x3ff; ++c) {
1556 uint8_t expectedCC= filter.contains(c) ? nfcNorm2->getCombiningClass(c) : 0;
1557 uint8_t cc=fn2.getCombiningClass(c);
1558 if(cc!=expectedCC) {
1559 errln(
1560 UnicodeString("FilteredNormalizer2(NFC, ^A0-FF,310-31F).getCombiningClass(U+")+
1561 hex(c)+
1562 ")==filtered NFC.getCC()");
1563 }
1564 }
1565
1566 UnicodeString newString1 = UNICODE_STRING_SIMPLE("[^\\u0100-\\u01ff]");
1567 UnicodeString newString2 = UNICODE_STRING_SIMPLE("[^\\u0200-\\u02ff]");
1568 fn2.append(newString1, newString2, errorCode);
1569 if (U_FAILURE(errorCode)) {
1570 errln("FilteredNormalizer2.append() failed.");
1571 }
1572 }
1573
1574 void
TestComposeUTF8WithEdits()1575 BasicNormalizerTest::TestComposeUTF8WithEdits() {
1576 IcuTestErrorCode errorCode(*this, "TestComposeUTF8WithEdits");
1577 const Normalizer2 *nfkc_cf=Normalizer2::getNFKCCasefoldInstance(errorCode);
1578 if(errorCode.errDataIfFailureAndReset("Normalizer2::getNFKCCasefoldInstance() call failed")) {
1579 return;
1580 }
1581 static const StringPiece src =
1582 u8" AÄA\u0308A\u0308\u00ad\u0323Ä\u0323,\u00ad\u1100\u1161가\u11A8가\u3133 ";
1583 StringPiece expected = u8" aääạ\u0308ạ\u0308,가각갃 ";
1584 std::string result;
1585 StringByteSink<std::string> sink(&result, static_cast<int32_t>(expected.length()));
1586 Edits edits;
1587 nfkc_cf->normalizeUTF8(0, src, sink, &edits, errorCode);
1588 assertSuccess("normalizeUTF8 with Edits", errorCode.get());
1589 assertEquals("normalizeUTF8 with Edits", expected.data(), result.c_str());
1590 static const EditChange expectedChanges[] = {
1591 { false, 2, 2 }, // 2 spaces
1592 { true, 1, 1 }, // A→a
1593 { true, 2, 2 }, // Ä→ä
1594 { true, 3, 2 }, // A\u0308→ä
1595 { true, 7, 5 }, // A\u0308\u00ad\u0323→ạ\u0308 removes the soft hyphen
1596 { true, 4, 5 }, // Ä\u0323→ạ\u0308
1597 { false, 1, 1 }, // comma
1598 { true, 2, 0 }, // U+00AD soft hyphen maps to empty
1599 { true, 6, 3 }, // \u1100\u1161→가
1600 { true, 6, 3 }, // 가\u11A8→각
1601 { true, 6, 3 }, // 가\u3133→갃
1602 { false, 2, 2 } // 2 spaces
1603 };
1604 assertTrue("normalizeUTF8 with Edits hasChanges", edits.hasChanges());
1605 assertEquals("normalizeUTF8 with Edits numberOfChanges", 9, edits.numberOfChanges());
1606 TestUtility::checkEditsIter(*this, u"normalizeUTF8 with Edits",
1607 edits.getFineIterator(), edits.getFineIterator(),
1608 expectedChanges, UPRV_LENGTHOF(expectedChanges),
1609 true, errorCode);
1610
1611 assertFalse("isNormalizedUTF8(source)", nfkc_cf->isNormalizedUTF8(src, errorCode));
1612 assertTrue("isNormalizedUTF8(normalized)", nfkc_cf->isNormalizedUTF8(result, errorCode));
1613
1614 // Omit unchanged text.
1615 expected = u8"aääạ\u0308ạ\u0308가각갃";
1616 result.clear();
1617 edits.reset();
1618 nfkc_cf->normalizeUTF8(U_OMIT_UNCHANGED_TEXT, src, sink, &edits, errorCode);
1619 assertSuccess("normalizeUTF8 omit unchanged", errorCode.get());
1620 assertEquals("normalizeUTF8 omit unchanged", expected.data(), result.c_str());
1621 assertTrue("normalizeUTF8 omit unchanged hasChanges", edits.hasChanges());
1622 assertEquals("normalizeUTF8 omit unchanged numberOfChanges", 9, edits.numberOfChanges());
1623 TestUtility::checkEditsIter(*this, u"normalizeUTF8 omit unchanged",
1624 edits.getFineIterator(), edits.getFineIterator(),
1625 expectedChanges, UPRV_LENGTHOF(expectedChanges),
1626 true, errorCode);
1627
1628 // With filter: The normalization code does not see the "A" substrings.
1629 UnicodeSet filter(u"[^A]", errorCode);
1630 FilteredNormalizer2 fn2(*nfkc_cf, filter);
1631 expected = u8" AäA\u0308A\u0323\u0308ạ\u0308,가각갃 ";
1632 result.clear();
1633 edits.reset();
1634 fn2.normalizeUTF8(0, src, sink, &edits, errorCode);
1635 assertSuccess("filtered normalizeUTF8", errorCode.get());
1636 assertEquals("filtered normalizeUTF8", expected.data(), result.c_str());
1637 static const EditChange filteredChanges[] = {
1638 { false, 3, 3 }, // 2 spaces + A
1639 { true, 2, 2 }, // Ä→ä
1640 { false, 4, 4 }, // A\u0308A
1641 { true, 6, 4 }, // \u0308\u00ad\u0323→\u0323\u0308 removes the soft hyphen
1642 { true, 4, 5 }, // Ä\u0323→ạ\u0308
1643 { false, 1, 1 }, // comma
1644 { true, 2, 0 }, // U+00AD soft hyphen maps to empty
1645 { true, 6, 3 }, // \u1100\u1161→가
1646 { true, 6, 3 }, // 가\u11A8→각
1647 { true, 6, 3 }, // 가\u3133→갃
1648 { false, 2, 2 } // 2 spaces
1649 };
1650 assertTrue("filtered normalizeUTF8 hasChanges", edits.hasChanges());
1651 assertEquals("filtered normalizeUTF8 numberOfChanges", 7, edits.numberOfChanges());
1652 TestUtility::checkEditsIter(*this, u"filtered normalizeUTF8",
1653 edits.getFineIterator(), edits.getFineIterator(),
1654 filteredChanges, UPRV_LENGTHOF(filteredChanges),
1655 true, errorCode);
1656
1657 assertFalse("filtered isNormalizedUTF8(source)", fn2.isNormalizedUTF8(src, errorCode));
1658 assertTrue("filtered isNormalizedUTF8(normalized)", fn2.isNormalizedUTF8(result, errorCode));
1659
1660 // Omit unchanged text.
1661 // Note that the result is not normalized because the inner normalizer
1662 // does not see text across filter spans.
1663 expected = u8"ä\u0323\u0308ạ\u0308가각갃";
1664 result.clear();
1665 edits.reset();
1666 fn2.normalizeUTF8(U_OMIT_UNCHANGED_TEXT, src, sink, &edits, errorCode);
1667 assertSuccess("filtered normalizeUTF8 omit unchanged", errorCode.get());
1668 assertEquals("filtered normalizeUTF8 omit unchanged", expected.data(), result.c_str());
1669 assertTrue("filtered normalizeUTF8 omit unchanged hasChanges", edits.hasChanges());
1670 assertEquals("filtered normalizeUTF8 omit unchanged numberOfChanges", 7, edits.numberOfChanges());
1671 TestUtility::checkEditsIter(*this, u"filtered normalizeUTF8 omit unchanged",
1672 edits.getFineIterator(), edits.getFineIterator(),
1673 filteredChanges, UPRV_LENGTHOF(filteredChanges),
1674 true, errorCode);
1675 }
1676
1677 void
TestDecomposeUTF8WithEdits()1678 BasicNormalizerTest::TestDecomposeUTF8WithEdits() {
1679 IcuTestErrorCode errorCode(*this, "TestDecomposeUTF8WithEdits");
1680 const Normalizer2 *nfkd_cf =
1681 Normalizer2::getInstance(nullptr, "nfkc_cf", UNORM2_DECOMPOSE, errorCode);
1682 if(errorCode.errDataIfFailureAndReset("Normalizer2::getInstance(nfkc_cf/decompose) call failed")) {
1683 return;
1684 }
1685 static const StringPiece src =
1686 u8" AÄA\u0308A\u0308\u00ad\u0323Ä\u0323,\u00ad\u1100\u1161가\u11A8가\u3133 ";
1687 StringPiece expected =
1688 u8" aa\u0308a\u0308a\u0323\u0308a\u0323\u0308,"
1689 u8"\u1100\u1161\u1100\u1161\u11A8\u1100\u1161\u11AA ";
1690 std::string result;
1691 StringByteSink<std::string> sink(&result, static_cast<int32_t>(expected.length()));
1692 Edits edits;
1693 nfkd_cf->normalizeUTF8(0, src, sink, &edits, errorCode);
1694 assertSuccess("normalizeUTF8 with Edits", errorCode.get());
1695 assertEquals("normalizeUTF8 with Edits", expected.data(), result.c_str());
1696 static const EditChange expectedChanges[] = {
1697 { false, 2, 2 }, // 2 spaces
1698 { true, 1, 1 }, // A→a
1699 { true, 2, 3 }, // Ä→a\u0308
1700 { true, 1, 1 }, // A→a
1701 { false, 2, 2 }, // \u0308→\u0308 unchanged
1702 { true, 1, 1 }, // A→a
1703 { true, 6, 4 }, // \u0308\u00ad\u0323→\u0323\u0308 removes the soft hyphen
1704 { true, 4, 5 }, // Ä\u0323→a\u0323\u0308
1705 { false, 1, 1 }, // comma
1706 { true, 2, 0 }, // U+00AD soft hyphen maps to empty
1707 { false, 6, 6 }, // \u1100\u1161 unchanged
1708 { true, 3, 6 }, // 가→\u1100\u1161
1709 { false, 3, 3 }, // \u11A8 unchanged
1710 { true, 3, 6 }, // 가→\u1100\u1161
1711 { true, 3, 3 }, // \u3133→\u11AA
1712 { false, 2, 2 } // 2 spaces
1713 };
1714 assertTrue("normalizeUTF8 with Edits hasChanges", edits.hasChanges());
1715 assertEquals("normalizeUTF8 with Edits numberOfChanges", 10, edits.numberOfChanges());
1716 TestUtility::checkEditsIter(*this, u"normalizeUTF8 with Edits",
1717 edits.getFineIterator(), edits.getFineIterator(),
1718 expectedChanges, UPRV_LENGTHOF(expectedChanges),
1719 true, errorCode);
1720
1721 assertFalse("isNormalizedUTF8(source)", nfkd_cf->isNormalizedUTF8(src, errorCode));
1722 assertTrue("isNormalizedUTF8(normalized)", nfkd_cf->isNormalizedUTF8(result, errorCode));
1723
1724 // Omit unchanged text.
1725 expected = u8"aa\u0308aa\u0323\u0308a\u0323\u0308\u1100\u1161\u1100\u1161\u11AA";
1726 result.clear();
1727 edits.reset();
1728 nfkd_cf->normalizeUTF8(U_OMIT_UNCHANGED_TEXT, src, sink, &edits, errorCode);
1729 assertSuccess("normalizeUTF8 omit unchanged", errorCode.get());
1730 assertEquals("normalizeUTF8 omit unchanged", expected.data(), result.c_str());
1731 assertTrue("normalizeUTF8 omit unchanged hasChanges", edits.hasChanges());
1732 assertEquals("normalizeUTF8 omit unchanged numberOfChanges", 10, edits.numberOfChanges());
1733 TestUtility::checkEditsIter(*this, u"normalizeUTF8 omit unchanged",
1734 edits.getFineIterator(), edits.getFineIterator(),
1735 expectedChanges, UPRV_LENGTHOF(expectedChanges),
1736 true, errorCode);
1737
1738 // Not testing FilteredNormalizer2:
1739 // The code there is the same for all normalization modes, and
1740 // TestComposeUTF8WithEdits() covers it well.
1741 }
1742
1743 void
TestLowMappingToEmpty_D()1744 BasicNormalizerTest::TestLowMappingToEmpty_D() {
1745 IcuTestErrorCode errorCode(*this, "TestLowMappingToEmpty_D");
1746 const Normalizer2 *n2 = Normalizer2::getInstance(
1747 nullptr, "nfkc_cf", UNORM2_DECOMPOSE, errorCode);
1748 if (errorCode.errDataIfFailureAndReset("Normalizer2::getInstance() call failed")) {
1749 return;
1750 }
1751 checkLowMappingToEmpty(*n2);
1752
1753 UnicodeString sh(u'\u00AD');
1754 assertFalse("soft hyphen is not normalized", n2->isNormalized(sh, errorCode));
1755 UnicodeString result = n2->normalize(sh, errorCode);
1756 assertTrue("soft hyphen normalizes to empty", result.isEmpty());
1757 assertEquals("soft hyphen QC=No", UNORM_NO, n2->quickCheck(sh, errorCode));
1758 assertEquals("soft hyphen spanQuickCheckYes", 0, n2->spanQuickCheckYes(sh, errorCode));
1759
1760 UnicodeString s(u"\u00ADÄ\u00AD\u0323");
1761 result = n2->normalize(s, errorCode);
1762 assertEquals("normalize string with soft hyphens", u"a\u0323\u0308", result);
1763 }
1764
1765 void
TestLowMappingToEmpty_FCD()1766 BasicNormalizerTest::TestLowMappingToEmpty_FCD() {
1767 IcuTestErrorCode errorCode(*this, "TestLowMappingToEmpty_FCD");
1768 const Normalizer2 *n2 = Normalizer2::getInstance(
1769 nullptr, "nfkc_cf", UNORM2_FCD, errorCode);
1770 if (errorCode.errDataIfFailureAndReset("Normalizer2::getInstance() call failed")) {
1771 return;
1772 }
1773 checkLowMappingToEmpty(*n2);
1774
1775 UnicodeString sh(u'\u00AD');
1776 assertTrue("soft hyphen is FCD", n2->isNormalized(sh, errorCode));
1777
1778 UnicodeString s(u"\u00ADÄ\u00AD\u0323");
1779 UnicodeString result = n2->normalize(s, errorCode);
1780 assertEquals("normalize string with soft hyphens", u"\u00ADa\u0323\u0308", result);
1781 }
1782
1783 void
checkLowMappingToEmpty(const Normalizer2 & n2)1784 BasicNormalizerTest::checkLowMappingToEmpty(const Normalizer2 &n2) {
1785 UnicodeString mapping;
1786 assertTrue("getDecomposition(soft hyphen)", n2.getDecomposition(0xad, mapping));
1787 assertTrue("soft hyphen maps to empty", mapping.isEmpty());
1788 assertFalse("soft hyphen has no boundary before", n2.hasBoundaryBefore(0xad));
1789 assertFalse("soft hyphen has no boundary after", n2.hasBoundaryAfter(0xad));
1790 assertFalse("soft hyphen is not inert", n2.isInert(0xad));
1791 }
1792
1793 void
TestNormalizeIllFormedText()1794 BasicNormalizerTest::TestNormalizeIllFormedText() {
1795 IcuTestErrorCode errorCode(*this, "TestNormalizeIllFormedText");
1796 const Normalizer2 *nfkc_cf = Normalizer2::getNFKCCasefoldInstance(errorCode);
1797 if(errorCode.errDataIfFailureAndReset("Normalizer2::getNFKCCasefoldInstance() call failed")) {
1798 return;
1799 }
1800 // Normalization behavior for ill-formed text is not defined.
1801 // ICU currently treats ill-formed sequences as normalization-inert
1802 // and copies them unchanged.
1803 UnicodeString src(u" A");
1804 src.append(static_cast<char16_t>(0xD800)).append(u"ÄA\u0308").append(static_cast<char16_t>(0xD900)).
1805 append(u"A\u0308\u00ad\u0323").append(static_cast<char16_t>(0xDBFF)).
1806 append(u"Ä\u0323,\u00ad").append(static_cast<char16_t>(0xDC00)).
1807 append(u"\u1100\u1161가\u11A8가\u3133 ").append(static_cast<char16_t>(0xDFFF));
1808 UnicodeString expected(u" a");
1809 expected.append(static_cast<char16_t>(0xD800)).append(u"ää").append(static_cast<char16_t>(0xD900)).
1810 append(u"ạ\u0308").append(static_cast<char16_t>(0xDBFF)).
1811 append(u"ạ\u0308,").append(static_cast<char16_t>(0xDC00)).
1812 append(u"가각갃 ").append(static_cast<char16_t>(0xDFFF));
1813 UnicodeString result = nfkc_cf->normalize(src, errorCode);
1814 assertSuccess("normalize", errorCode.get());
1815 assertEquals("normalize", expected, result);
1816
1817 std::string src8(reinterpret_cast<const char*>(u8" A"));
1818 src8.append("\x80").append(reinterpret_cast<const char*>(u8"ÄA\u0308")).append("\xC0\x80").
1819 append(reinterpret_cast<const char*>(u8"A\u0308\u00ad\u0323")).append("\xED\xA0\x80").
1820 append(reinterpret_cast<const char*>(u8"Ä\u0323,\u00ad")).append("\xF4\x90\x80\x80").
1821 append(reinterpret_cast<const char*>(u8"\u1100\u1161가\u11A8가\u3133 ")).append("\xF0");
1822 std::string expected8(reinterpret_cast<const char*>(u8" a"));
1823 expected8.append("\x80").append(reinterpret_cast<const char*>(u8"ää")).append("\xC0\x80").
1824 append(reinterpret_cast<const char*>(u8"ạ\u0308")).append("\xED\xA0\x80").
1825 append(reinterpret_cast<const char*>(u8"ạ\u0308,")).append("\xF4\x90\x80\x80").
1826 append(reinterpret_cast<const char*>(u8"가각갃 ")).append("\xF0");
1827 std::string result8;
1828 StringByteSink<std::string> sink(&result8);
1829 nfkc_cf->normalizeUTF8(0, src8, sink, nullptr, errorCode);
1830 assertSuccess("normalizeUTF8", errorCode.get());
1831 assertEquals("normalizeUTF8", expected8.c_str(), result8.c_str());
1832 }
1833
1834 void
TestComposeJamoTBase()1835 BasicNormalizerTest::TestComposeJamoTBase() {
1836 // Algorithmic composition of Hangul syllables must not combine with JAMO_T_BASE = U+11A7
1837 // which is not a conjoining Jamo Trailing consonant.
1838 IcuTestErrorCode errorCode(*this, "TestComposeJamoTBase");
1839 const Normalizer2 *nfkc = Normalizer2::getNFKCInstance(errorCode);
1840 if(errorCode.errDataIfFailureAndReset("Normalizer2::getNFKCInstance() call failed")) {
1841 return;
1842 }
1843 UnicodeString s(u"\u1100\u1161\u11A7\u1100\u314F\u11A7가\u11A7");
1844 UnicodeString expected(u"가\u11A7가\u11A7가\u11A7");
1845 UnicodeString result = nfkc->normalize(s, errorCode);
1846 assertSuccess("normalize(LV+11A7)", errorCode.get());
1847 assertEquals("normalize(LV+11A7)", expected, result);
1848 assertFalse("isNormalized(LV+11A7)", nfkc->isNormalized(s, errorCode));
1849 assertTrue("isNormalized(normalized)", nfkc->isNormalized(result, errorCode));
1850
1851 StringPiece s8(u8"\u1100\u1161\u11A7\u1100\u314F\u11A7가\u11A7");
1852 StringPiece expected8(u8"가\u11A7가\u11A7가\u11A7");
1853 std::string result8;
1854 StringByteSink<std::string> sink(&result8, expected8.length());
1855 nfkc->normalizeUTF8(0, s8, sink, nullptr, errorCode);
1856 assertSuccess("normalizeUTF8(LV+11A7)", errorCode.get());
1857 assertEquals("normalizeUTF8(LV+11A7)", expected8.data(), result8.c_str());
1858 assertFalse("isNormalizedUTF8(LV+11A7)", nfkc->isNormalizedUTF8(s8, errorCode));
1859 assertTrue("isNormalizedUTF8(normalized)", nfkc->isNormalizedUTF8(result8, errorCode));
1860 }
1861
1862 void
TestComposeBoundaryAfter()1863 BasicNormalizerTest::TestComposeBoundaryAfter() {
1864 IcuTestErrorCode errorCode(*this, "TestComposeBoundaryAfter");
1865 const Normalizer2 *nfkc = Normalizer2::getNFKCInstance(errorCode);
1866 if(errorCode.errDataIfFailureAndReset("Normalizer2::getNFKCInstance() call failed")) {
1867 return;
1868 }
1869 // U+02DA and U+FB2C do not have compose-boundaries-after.
1870 UnicodeString s(u"\u02DA\u0339 \uFB2C\u05B6");
1871 UnicodeString expected(u" \u0339\u030A \u05E9\u05B6\u05BC\u05C1");
1872 UnicodeString result = nfkc->normalize(s, errorCode);
1873 assertSuccess("nfkc", errorCode.get());
1874 assertEquals("nfkc", expected, result);
1875 assertFalse("U+02DA boundary-after", nfkc->hasBoundaryAfter(0x2DA));
1876 assertFalse("U+FB2C boundary-after", nfkc->hasBoundaryAfter(0xFB2C));
1877 }
1878
1879 void
TestNFKC_SCF()1880 BasicNormalizerTest::TestNFKC_SCF() {
1881 IcuTestErrorCode errorCode(*this, "TestNFKC_SCF");
1882 const Normalizer2 *nfkc_scf = Normalizer2::getNFKCSimpleCasefoldInstance(errorCode);
1883 if(errorCode.errDataIfFailureAndReset(
1884 "Normalizer2::getNFKCSimpleCasefoldInstance() call failed")) {
1885 return;
1886 }
1887 // Uses only Simple_Casefolding mappings.
1888 UnicodeString s(u"aA\u0308 ßẞ \u1F80\u1F88");
1889 UnicodeString expected(u"aä ßß \u1F80\u1F80");
1890 UnicodeString result = nfkc_scf->normalize(s, errorCode);
1891 assertSuccess("nfkc_scf", errorCode.get());
1892 assertEquals("nfkc_scf", expected, result);
1893 }
1894
1895 #endif /* #if !UCONFIG_NO_NORMALIZATION */
1896