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/uniset.h"
18 #include "unicode/usetiter.h"
19 #include "unicode/schriter.h"
20 #include "unicode/utf16.h"
21 #include "cmemory.h"
22 #include "cstring.h"
23 #include "normalizer2impl.h"
24 #include "testutil.h"
25 #include "tstnorm.h"
26
27 #define ARRAY_LENGTH(array) UPRV_LENGTHOF(array)
28
runIndexedTest(int32_t index,UBool exec,const char * & name,char *)29 void BasicNormalizerTest::runIndexedTest(int32_t index, UBool exec,
30 const char* &name, char* /*par*/) {
31 if(exec) {
32 logln("TestSuite BasicNormalizerTest: ");
33 }
34 TESTCASE_AUTO_BEGIN;
35 TESTCASE_AUTO(TestDecomp);
36 TESTCASE_AUTO(TestCompatDecomp);
37 TESTCASE_AUTO(TestCanonCompose);
38 TESTCASE_AUTO(TestCompatCompose);
39 TESTCASE_AUTO(TestPrevious);
40 TESTCASE_AUTO(TestHangulDecomp);
41 TESTCASE_AUTO(TestHangulCompose);
42 TESTCASE_AUTO(TestTibetan);
43 TESTCASE_AUTO(TestCompositionExclusion);
44 TESTCASE_AUTO(TestZeroIndex);
45 TESTCASE_AUTO(TestVerisign);
46 TESTCASE_AUTO(TestPreviousNext);
47 TESTCASE_AUTO(TestNormalizerAPI);
48 TESTCASE_AUTO(TestConcatenate);
49 TESTCASE_AUTO(FindFoldFCDExceptions);
50 TESTCASE_AUTO(TestCompare);
51 TESTCASE_AUTO(TestSkippable);
52 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
53 TESTCASE_AUTO(TestCustomComp);
54 TESTCASE_AUTO(TestCustomFCC);
55 #endif
56 TESTCASE_AUTO(TestFilteredNormalizer2Coverage);
57 TESTCASE_AUTO(TestNormalizeUTF8WithEdits);
58 TESTCASE_AUTO(TestLowMappingToEmpty_D);
59 TESTCASE_AUTO(TestLowMappingToEmpty_FCD);
60 TESTCASE_AUTO(TestNormalizeIllFormedText);
61 TESTCASE_AUTO(TestComposeJamoTBase);
62 TESTCASE_AUTO(TestComposeBoundaryAfter);
63 TESTCASE_AUTO_END;
64 }
65
66 /**
67 * Convert Java-style strings with \u Unicode escapes into UnicodeString objects
68 */
str(const char * input)69 static UnicodeString str(const char *input)
70 {
71 UnicodeString str(input, ""); // Invariant conversion
72 return str.unescape();
73 }
74
75
BasicNormalizerTest()76 BasicNormalizerTest::BasicNormalizerTest()
77 {
78 // canonTest
79 // Input Decomposed Composed
80
81 canonTests[0][0] = str("cat"); canonTests[0][1] = str("cat"); canonTests[0][2] = str("cat");
82
83 canonTests[1][0] = str("\\u00e0ardvark"); canonTests[1][1] = str("a\\u0300ardvark"); canonTests[1][2] = str("\\u00e0ardvark");
84
85 canonTests[2][0] = str("\\u1e0a"); canonTests[2][1] = str("D\\u0307"); canonTests[2][2] = str("\\u1e0a"); // D-dot_above
86
87 canonTests[3][0] = str("D\\u0307"); canonTests[3][1] = str("D\\u0307"); canonTests[3][2] = str("\\u1e0a"); // D dot_above
88
89 canonTests[4][0] = str("\\u1e0c\\u0307"); canonTests[4][1] = str("D\\u0323\\u0307"); canonTests[4][2] = str("\\u1e0c\\u0307"); // D-dot_below dot_above
90
91 canonTests[5][0] = str("\\u1e0a\\u0323"); canonTests[5][1] = str("D\\u0323\\u0307"); canonTests[5][2] = str("\\u1e0c\\u0307"); // D-dot_above dot_below
92
93 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
94
95 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
96
97 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
98
99 canonTests[9][0] = str("\\u1E14"); canonTests[9][1] = str("E\\u0304\\u0300"); canonTests[9][2] = str("\\u1E14"); // E-macron-grave
100
101 canonTests[10][0] = str("\\u0112\\u0300"); canonTests[10][1] = str("E\\u0304\\u0300"); canonTests[10][2] = str("\\u1E14"); // E-macron + grave
102
103 canonTests[11][0] = str("\\u00c8\\u0304"); canonTests[11][1] = str("E\\u0300\\u0304"); canonTests[11][2] = str("\\u00c8\\u0304"); // E-grave + macron
104
105 canonTests[12][0] = str("\\u212b"); canonTests[12][1] = str("A\\u030a"); canonTests[12][2] = str("\\u00c5"); // angstrom_sign
106
107 canonTests[13][0] = str("\\u00c5"); canonTests[13][1] = str("A\\u030a"); canonTests[13][2] = str("\\u00c5"); // A-ring
108
109 canonTests[14][0] = str("\\u00C4ffin"); canonTests[14][1] = str("A\\u0308ffin"); canonTests[14][2] = str("\\u00C4ffin");
110
111 canonTests[15][0] = str("\\u00C4\\uFB03n"); canonTests[15][1] = str("A\\u0308\\uFB03n"); canonTests[15][2] = str("\\u00C4\\uFB03n");
112
113 canonTests[16][0] = str("Henry IV"); canonTests[16][1] = str("Henry IV"); canonTests[16][2] = str("Henry IV");
114
115 canonTests[17][0] = str("Henry \\u2163"); canonTests[17][1] = str("Henry \\u2163"); canonTests[17][2] = str("Henry \\u2163");
116
117 canonTests[18][0] = str("\\u30AC"); canonTests[18][1] = str("\\u30AB\\u3099"); canonTests[18][2] = str("\\u30AC"); // ga (Katakana)
118
119 canonTests[19][0] = str("\\u30AB\\u3099"); canonTests[19][1] = str("\\u30AB\\u3099"); canonTests[19][2] = str("\\u30AC"); // ka + ten
120
121 canonTests[20][0] = str("\\uFF76\\uFF9E"); canonTests[20][1] = str("\\uFF76\\uFF9E"); canonTests[20][2] = str("\\uFF76\\uFF9E"); // hw_ka + hw_ten
122
123 canonTests[21][0] = str("\\u30AB\\uFF9E"); canonTests[21][1] = str("\\u30AB\\uFF9E"); canonTests[21][2] = str("\\u30AB\\uFF9E"); // ka + hw_ten
124
125 canonTests[22][0] = str("\\uFF76\\u3099"); canonTests[22][1] = str("\\uFF76\\u3099"); canonTests[22][2] = str("\\uFF76\\u3099"); // hw_ka + ten
126
127 canonTests[23][0] = str("A\\u0300\\u0316"); canonTests[23][1] = str("A\\u0316\\u0300"); canonTests[23][2] = str("\\u00C0\\u0316");
128
129 /* compatTest */
130 // Input Decomposed Composed
131 compatTests[0][0] = str("cat"); compatTests[0][1] = str("cat"); compatTests[0][2] = str("cat") ;
132
133 compatTests[1][0] = str("\\uFB4f"); compatTests[1][1] = str("\\u05D0\\u05DC"); compatTests[1][2] = str("\\u05D0\\u05DC"); // Alef-Lamed vs. Alef, Lamed
134
135 compatTests[2][0] = str("\\u00C4ffin"); compatTests[2][1] = str("A\\u0308ffin"); compatTests[2][2] = str("\\u00C4ffin") ;
136
137 compatTests[3][0] = str("\\u00C4\\uFB03n"); compatTests[3][1] = str("A\\u0308ffin"); compatTests[3][2] = str("\\u00C4ffin") ; // ffi ligature -> f + f + i
138
139 compatTests[4][0] = str("Henry IV"); compatTests[4][1] = str("Henry IV"); compatTests[4][2] = str("Henry IV") ;
140
141 compatTests[5][0] = str("Henry \\u2163"); compatTests[5][1] = str("Henry IV"); compatTests[5][2] = str("Henry IV") ;
142
143 compatTests[6][0] = str("\\u30AC"); compatTests[6][1] = str("\\u30AB\\u3099"); compatTests[6][2] = str("\\u30AC") ; // ga (Katakana)
144
145 compatTests[7][0] = str("\\u30AB\\u3099"); compatTests[7][1] = str("\\u30AB\\u3099"); compatTests[7][2] = str("\\u30AC") ; // ka + ten
146
147 compatTests[8][0] = str("\\uFF76\\u3099"); compatTests[8][1] = str("\\u30AB\\u3099"); compatTests[8][2] = str("\\u30AC") ; // hw_ka + ten
148
149 /* These two are broken in Unicode 2.1.2 but fixed in 2.1.5 and later */
150 compatTests[9][0] = str("\\uFF76\\uFF9E"); compatTests[9][1] = str("\\u30AB\\u3099"); compatTests[9][2] = str("\\u30AC") ; // hw_ka + hw_ten
151
152 compatTests[10][0] = str("\\u30AB\\uFF9E"); compatTests[10][1] = str("\\u30AB\\u3099"); compatTests[10][2] = str("\\u30AC") ; // ka + hw_ten
153
154 /* Hangul Canonical */
155 // Input Decomposed Composed
156 hangulCanon[0][0] = str("\\ud4db"); hangulCanon[0][1] = str("\\u1111\\u1171\\u11b6"); hangulCanon[0][2] = str("\\ud4db") ;
157
158 hangulCanon[1][0] = str("\\u1111\\u1171\\u11b6"), hangulCanon[1][1] = str("\\u1111\\u1171\\u11b6"), hangulCanon[1][2] = str("\\ud4db");
159 }
160
~BasicNormalizerTest()161 BasicNormalizerTest::~BasicNormalizerTest()
162 {
163 }
164
TestPrevious()165 void BasicNormalizerTest::TestPrevious()
166 {
167 Normalizer* norm = new Normalizer("", UNORM_NFD);
168
169 logln("testing decomp...");
170 uint32_t i;
171 for (i = 0; i < ARRAY_LENGTH(canonTests); i++) {
172 backAndForth(norm, canonTests[i][0]);
173 }
174
175 logln("testing compose...");
176 norm->setMode(UNORM_NFC);
177 for (i = 0; i < ARRAY_LENGTH(canonTests); i++) {
178 backAndForth(norm, canonTests[i][0]);
179 }
180
181 delete norm;
182 }
183
TestDecomp()184 void BasicNormalizerTest::TestDecomp()
185 {
186 Normalizer* norm = new Normalizer("", UNORM_NFD);
187 iterateTest(norm, canonTests, ARRAY_LENGTH(canonTests), 1);
188 staticTest(UNORM_NFD, 0, canonTests, ARRAY_LENGTH(canonTests), 1);
189 delete norm;
190 }
191
TestCompatDecomp()192 void BasicNormalizerTest::TestCompatDecomp()
193 {
194 Normalizer* norm = new Normalizer("", UNORM_NFKD);
195 iterateTest(norm, compatTests, ARRAY_LENGTH(compatTests), 1);
196
197 staticTest(UNORM_NFKD, 0,
198 compatTests, ARRAY_LENGTH(compatTests), 1);
199 delete norm;
200 }
201
TestCanonCompose()202 void BasicNormalizerTest::TestCanonCompose()
203 {
204 Normalizer* norm = new Normalizer("", UNORM_NFC);
205 iterateTest(norm, canonTests, ARRAY_LENGTH(canonTests), 2);
206
207 staticTest(UNORM_NFC, 0, canonTests,
208 ARRAY_LENGTH(canonTests), 2);
209 delete norm;
210 }
211
TestCompatCompose()212 void BasicNormalizerTest::TestCompatCompose()
213 {
214 Normalizer* norm = new Normalizer("", UNORM_NFKC);
215 iterateTest(norm, compatTests, ARRAY_LENGTH(compatTests), 2);
216
217 staticTest(UNORM_NFKC, 0,
218 compatTests, ARRAY_LENGTH(compatTests), 2);
219 delete norm;
220 }
221
222
223 //-------------------------------------------------------------------------------
224
TestHangulCompose()225 void BasicNormalizerTest::TestHangulCompose()
226 {
227 // Make sure that the static composition methods work
228 logln("Canonical composition...");
229 staticTest(UNORM_NFC, 0, hangulCanon, ARRAY_LENGTH(hangulCanon), 2);
230 logln("Compatibility composition...");
231
232 // Now try iterative composition....
233 logln("Static composition...");
234 Normalizer* norm = new Normalizer("", UNORM_NFC);
235 iterateTest(norm, hangulCanon, ARRAY_LENGTH(hangulCanon), 2);
236 norm->setMode(UNORM_NFKC);
237
238 // And finally, make sure you can do it in reverse too
239 logln("Reverse iteration...");
240 norm->setMode(UNORM_NFC);
241 for (uint32_t i = 0; i < ARRAY_LENGTH(hangulCanon); i++) {
242 backAndForth(norm, hangulCanon[i][0]);
243 }
244 delete norm;
245 }
246
TestHangulDecomp()247 void BasicNormalizerTest::TestHangulDecomp()
248 {
249 // Make sure that the static decomposition methods work
250 logln("Canonical decomposition...");
251 staticTest(UNORM_NFD, 0, hangulCanon, ARRAY_LENGTH(hangulCanon), 1);
252 logln("Compatibility decomposition...");
253
254 // Now the iterative decomposition methods...
255 logln("Iterative decomposition...");
256 Normalizer* norm = new Normalizer("", UNORM_NFD);
257 iterateTest(norm, hangulCanon, ARRAY_LENGTH(hangulCanon), 1);
258 norm->setMode(UNORM_NFKD);
259
260 // And finally, make sure you can do it in reverse too
261 logln("Reverse iteration...");
262 norm->setMode(UNORM_NFD);
263 for (uint32_t i = 0; i < ARRAY_LENGTH(hangulCanon); i++) {
264 backAndForth(norm, hangulCanon[i][0]);
265 }
266 delete norm;
267 }
268
269 /**
270 * The Tibetan vowel sign AA, 0f71, was messed up prior to Unicode version 2.1.9.
271 */
TestTibetan(void)272 void BasicNormalizerTest::TestTibetan(void) {
273 UnicodeString decomp[1][3];
274 decomp[0][0] = str("\\u0f77");
275 decomp[0][1] = str("\\u0f77");
276 decomp[0][2] = str("\\u0fb2\\u0f71\\u0f80");
277
278 UnicodeString compose[1][3];
279 compose[0][0] = str("\\u0fb2\\u0f71\\u0f80");
280 compose[0][1] = str("\\u0fb2\\u0f71\\u0f80");
281 compose[0][2] = str("\\u0fb2\\u0f71\\u0f80");
282
283 staticTest(UNORM_NFD, 0, decomp, ARRAY_LENGTH(decomp), 1);
284 staticTest(UNORM_NFKD, 0, decomp, ARRAY_LENGTH(decomp), 2);
285 staticTest(UNORM_NFC, 0, compose, ARRAY_LENGTH(compose), 1);
286 staticTest(UNORM_NFKC, 0, compose, ARRAY_LENGTH(compose), 2);
287 }
288
289 /**
290 * Make sure characters in the CompositionExclusion.txt list do not get
291 * composed to.
292 */
TestCompositionExclusion(void)293 void BasicNormalizerTest::TestCompositionExclusion(void) {
294 // This list is generated from CompositionExclusion.txt.
295 // Update whenever the normalizer tables are updated. Note
296 // that we test all characters listed, even those that can be
297 // derived from the Unicode DB and are therefore commented
298 // out.
299 // ### TODO read composition exclusion from source/data/unidata file
300 // and test against that
301 UnicodeString EXCLUDED = str(
302 "\\u0340\\u0341\\u0343\\u0344\\u0374\\u037E\\u0387\\u0958"
303 "\\u0959\\u095A\\u095B\\u095C\\u095D\\u095E\\u095F\\u09DC"
304 "\\u09DD\\u09DF\\u0A33\\u0A36\\u0A59\\u0A5A\\u0A5B\\u0A5E"
305 "\\u0B5C\\u0B5D\\u0F43\\u0F4D\\u0F52\\u0F57\\u0F5C\\u0F69"
306 "\\u0F73\\u0F75\\u0F76\\u0F78\\u0F81\\u0F93\\u0F9D\\u0FA2"
307 "\\u0FA7\\u0FAC\\u0FB9\\u1F71\\u1F73\\u1F75\\u1F77\\u1F79"
308 "\\u1F7B\\u1F7D\\u1FBB\\u1FBE\\u1FC9\\u1FCB\\u1FD3\\u1FDB"
309 "\\u1FE3\\u1FEB\\u1FEE\\u1FEF\\u1FF9\\u1FFB\\u1FFD\\u2000"
310 "\\u2001\\u2126\\u212A\\u212B\\u2329\\u232A\\uF900\\uFA10"
311 "\\uFA12\\uFA15\\uFA20\\uFA22\\uFA25\\uFA26\\uFA2A\\uFB1F"
312 "\\uFB2A\\uFB2B\\uFB2C\\uFB2D\\uFB2E\\uFB2F\\uFB30\\uFB31"
313 "\\uFB32\\uFB33\\uFB34\\uFB35\\uFB36\\uFB38\\uFB39\\uFB3A"
314 "\\uFB3B\\uFB3C\\uFB3E\\uFB40\\uFB41\\uFB43\\uFB44\\uFB46"
315 "\\uFB47\\uFB48\\uFB49\\uFB4A\\uFB4B\\uFB4C\\uFB4D\\uFB4E"
316 );
317 UErrorCode status = U_ZERO_ERROR;
318 for (int32_t i=0; i<EXCLUDED.length(); ++i) {
319 UnicodeString a(EXCLUDED.charAt(i));
320 UnicodeString b;
321 UnicodeString c;
322 Normalizer::normalize(a, UNORM_NFKD, 0, b, status);
323 Normalizer::normalize(b, UNORM_NFC, 0, c, status);
324 if (c == a) {
325 errln("FAIL: " + hex(a) + " x DECOMP_COMPAT => " +
326 hex(b) + " x COMPOSE => " +
327 hex(c));
328 } else if (verbose) {
329 logln("Ok: " + hex(a) + " x DECOMP_COMPAT => " +
330 hex(b) + " x COMPOSE => " +
331 hex(c));
332 }
333 }
334 }
335
336 /**
337 * Test for a problem that showed up just before ICU 1.6 release
338 * having to do with combining characters with an index of zero.
339 * Such characters do not participate in any canonical
340 * decompositions. However, having an index of zero means that
341 * they all share one typeMask[] entry, that is, they all have to
342 * map to the same canonical class, which is not the case, in
343 * reality.
344 */
TestZeroIndex(void)345 void BasicNormalizerTest::TestZeroIndex(void) {
346 const char* DATA[] = {
347 // Expect col1 x COMPOSE_COMPAT => col2
348 // Expect col2 x DECOMP => col3
349 "A\\u0316\\u0300", "\\u00C0\\u0316", "A\\u0316\\u0300",
350 "A\\u0300\\u0316", "\\u00C0\\u0316", "A\\u0316\\u0300",
351 "A\\u0327\\u0300", "\\u00C0\\u0327", "A\\u0327\\u0300",
352 "c\\u0321\\u0327", "c\\u0321\\u0327", "c\\u0321\\u0327",
353 "c\\u0327\\u0321", "\\u00E7\\u0321", "c\\u0327\\u0321",
354 };
355 int32_t DATA_length = UPRV_LENGTHOF(DATA);
356
357 for (int32_t i=0; i<DATA_length; i+=3) {
358 UErrorCode status = U_ZERO_ERROR;
359 UnicodeString a(DATA[i], "");
360 a = a.unescape();
361 UnicodeString b;
362 Normalizer::normalize(a, UNORM_NFKC, 0, b, status);
363 if (U_FAILURE(status)) {
364 dataerrln("Error calling normalize UNORM_NFKC: %s", u_errorName(status));
365 } else {
366 UnicodeString exp(DATA[i+1], "");
367 exp = exp.unescape();
368 if (b == exp) {
369 logln((UnicodeString)"Ok: " + hex(a) + " x COMPOSE_COMPAT => " + hex(b));
370 } else {
371 errln((UnicodeString)"FAIL: " + hex(a) + " x COMPOSE_COMPAT => " + hex(b) +
372 ", expect " + hex(exp));
373 }
374 }
375 Normalizer::normalize(b, UNORM_NFD, 0, a, status);
376 if (U_FAILURE(status)) {
377 dataerrln("Error calling normalize UNORM_NFD: %s", u_errorName(status));
378 } else {
379 UnicodeString exp = UnicodeString(DATA[i+2], "").unescape();
380 if (a == exp) {
381 logln((UnicodeString)"Ok: " + hex(b) + " x DECOMP => " + hex(a));
382 } else {
383 errln((UnicodeString)"FAIL: " + hex(b) + " x DECOMP => " + hex(a) +
384 ", expect " + hex(exp));
385 }
386 }
387 }
388 }
389
390 /**
391 * Run a few specific cases that are failing for Verisign.
392 */
TestVerisign(void)393 void BasicNormalizerTest::TestVerisign(void) {
394 /*
395 > Their input:
396 > 05B8 05B9 05B1 0591 05C3 05B0 05AC 059F
397 > Their output (supposedly from ICU):
398 > 05B8 05B1 05B9 0591 05C3 05B0 05AC 059F
399 > My output from charlint:
400 > 05B1 05B8 05B9 0591 05C3 05B0 05AC 059F
401
402 05B8 05B9 05B1 0591 05C3 05B0 05AC 059F => 05B1 05B8 05B9 0591 05C3 05B0
403 05AC 059F
404
405 U+05B8 18 E HEBREW POINT QAMATS
406 U+05B9 19 F HEBREW POINT HOLAM
407 U+05B1 11 HEBREW POINT HATAF SEGOL
408 U+0591 220 HEBREW ACCENT ETNAHTA
409 U+05C3 0 HEBREW PUNCTUATION SOF PASUQ
410 U+05B0 10 HEBREW POINT SHEVA
411 U+05AC 230 HEBREW ACCENT ILUY
412 U+059F 230 HEBREW ACCENT QARNEY PARA
413
414 U+05B1 11 HEBREW POINT HATAF SEGOL
415 U+05B8 18 HEBREW POINT QAMATS
416 U+05B9 19 HEBREW POINT HOLAM
417 U+0591 220 HEBREW ACCENT ETNAHTA
418 U+05C3 0 HEBREW PUNCTUATION SOF PASUQ
419 U+05B0 10 HEBREW POINT SHEVA
420 U+05AC 230 HEBREW ACCENT ILUY
421 U+059F 230 HEBREW ACCENT QARNEY PARA
422
423 Wrong result:
424 U+05B8 18 HEBREW POINT QAMATS
425 U+05B1 11 HEBREW POINT HATAF SEGOL
426 U+05B9 19 HEBREW POINT HOLAM
427 U+0591 220 HEBREW ACCENT ETNAHTA
428 U+05C3 0 HEBREW PUNCTUATION SOF PASUQ
429 U+05B0 10 HEBREW POINT SHEVA
430 U+05AC 230 HEBREW ACCENT ILUY
431 U+059F 230 HEBREW ACCENT QARNEY PARA
432
433
434 > Their input:
435 >0592 05B7 05BC 05A5 05B0 05C0 05C4 05AD
436 >Their output (supposedly from ICU):
437 >0592 05B0 05B7 05BC 05A5 05C0 05AD 05C4
438 >My output from charlint:
439 >05B0 05B7 05BC 05A5 0592 05C0 05AD 05C4
440
441 0592 05B7 05BC 05A5 05B0 05C0 05C4 05AD => 05B0 05B7 05BC 05A5 0592 05C0
442 05AD 05C4
443
444 U+0592 230 HEBREW ACCENT SEGOL
445 U+05B7 17 HEBREW POINT PATAH
446 U+05BC 21 HEBREW POINT DAGESH OR MAPIQ
447 U+05A5 220 HEBREW ACCENT MERKHA
448 U+05B0 10 HEBREW POINT SHEVA
449 U+05C0 0 HEBREW PUNCTUATION PASEQ
450 U+05C4 230 HEBREW MARK UPPER DOT
451 U+05AD 222 HEBREW ACCENT DEHI
452
453 U+05B0 10 HEBREW POINT SHEVA
454 U+05B7 17 HEBREW POINT PATAH
455 U+05BC 21 HEBREW POINT DAGESH OR MAPIQ
456 U+05A5 220 HEBREW ACCENT MERKHA
457 U+0592 230 HEBREW ACCENT SEGOL
458 U+05C0 0 HEBREW PUNCTUATION PASEQ
459 U+05AD 222 HEBREW ACCENT DEHI
460 U+05C4 230 HEBREW MARK UPPER DOT
461
462 Wrong result:
463 U+0592 230 HEBREW ACCENT SEGOL
464 U+05B0 10 HEBREW POINT SHEVA
465 U+05B7 17 HEBREW POINT PATAH
466 U+05BC 21 HEBREW POINT DAGESH OR MAPIQ
467 U+05A5 220 HEBREW ACCENT MERKHA
468 U+05C0 0 HEBREW PUNCTUATION PASEQ
469 U+05AD 222 HEBREW ACCENT DEHI
470 U+05C4 230 HEBREW MARK UPPER DOT
471 */
472 UnicodeString data[2][3];
473 data[0][0] = str("\\u05B8\\u05B9\\u05B1\\u0591\\u05C3\\u05B0\\u05AC\\u059F");
474 data[0][1] = str("\\u05B1\\u05B8\\u05B9\\u0591\\u05C3\\u05B0\\u05AC\\u059F");
475 data[0][2] = str("");
476 data[1][0] = str("\\u0592\\u05B7\\u05BC\\u05A5\\u05B0\\u05C0\\u05C4\\u05AD");
477 data[1][1] = str("\\u05B0\\u05B7\\u05BC\\u05A5\\u0592\\u05C0\\u05AD\\u05C4");
478 data[1][2] = str("");
479
480 staticTest(UNORM_NFD, 0, data, ARRAY_LENGTH(data), 1);
481 staticTest(UNORM_NFC, 0, data, ARRAY_LENGTH(data), 1);
482 }
483
484 //------------------------------------------------------------------------
485 // Internal utilities
486 //
487
hex(UChar ch)488 UnicodeString BasicNormalizerTest::hex(UChar ch) {
489 UnicodeString result;
490 return appendHex(ch, 4, result);
491 }
492
hex(const UnicodeString & s)493 UnicodeString BasicNormalizerTest::hex(const UnicodeString& s) {
494 UnicodeString result;
495 for (int i = 0; i < s.length(); ++i) {
496 if (i != 0) result += (UChar)0x2c/*,*/;
497 appendHex(s[i], 4, result);
498 }
499 return result;
500 }
501
502
insert(UnicodeString & dest,int pos,UChar32 ch)503 inline static void insert(UnicodeString& dest, int pos, UChar32 ch)
504 {
505 dest.replace(pos, 0, ch);
506 }
507
backAndForth(Normalizer * iter,const UnicodeString & input)508 void BasicNormalizerTest::backAndForth(Normalizer* iter, const UnicodeString& input)
509 {
510 UChar32 ch;
511 UErrorCode status = U_ZERO_ERROR;
512 iter->setText(input, status);
513
514 // Run through the iterator forwards and stick it into a StringBuffer
515 UnicodeString forward;
516 for (ch = iter->first(); ch != iter->DONE; ch = iter->next()) {
517 forward += ch;
518 }
519
520 // Now do it backwards
521 UnicodeString reverse;
522 for (ch = iter->last(); ch != iter->DONE; ch = iter->previous()) {
523 insert(reverse, 0, ch);
524 }
525
526 if (forward != reverse) {
527 errln("Forward/reverse mismatch for input " + hex(input)
528 + ", forward: " + hex(forward) + ", backward: " + hex(reverse));
529 }
530 }
531
staticTest(UNormalizationMode mode,int options,UnicodeString tests[][3],int length,int outCol)532 void BasicNormalizerTest::staticTest(UNormalizationMode mode, int options,
533 UnicodeString tests[][3], int length,
534 int outCol)
535 {
536 UErrorCode status = U_ZERO_ERROR;
537 for (int i = 0; i < length; i++)
538 {
539 UnicodeString& input = tests[i][0];
540 UnicodeString& expect = tests[i][outCol];
541
542 logln("Normalizing '" + input + "' (" + hex(input) + ")" );
543
544 UnicodeString output;
545 Normalizer::normalize(input, mode, options, output, status);
546
547 if (output != expect) {
548 dataerrln(UnicodeString("ERROR: case ") + i + " normalized " + hex(input) + "\n"
549 + " expected " + hex(expect) + "\n"
550 + " static got " + hex(output) );
551 }
552 }
553 }
554
iterateTest(Normalizer * iter,UnicodeString tests[][3],int length,int outCol)555 void BasicNormalizerTest::iterateTest(Normalizer* iter,
556 UnicodeString tests[][3], int length,
557 int outCol)
558 {
559 UErrorCode status = U_ZERO_ERROR;
560 for (int i = 0; i < length; i++)
561 {
562 UnicodeString& input = tests[i][0];
563 UnicodeString& expect = tests[i][outCol];
564
565 logln("Normalizing '" + input + "' (" + hex(input) + ")" );
566
567 iter->setText(input, status);
568 assertEqual(input, expect, iter, UnicodeString("ERROR: case ") + i + " ");
569 }
570 }
571
assertEqual(const UnicodeString & input,const UnicodeString & expected,Normalizer * iter,const UnicodeString & errPrefix)572 void BasicNormalizerTest::assertEqual(const UnicodeString& input,
573 const UnicodeString& expected,
574 Normalizer* iter,
575 const UnicodeString& errPrefix)
576 {
577 UnicodeString result;
578
579 for (UChar32 ch = iter->first(); ch != iter->DONE; ch = iter->next()) {
580 result += ch;
581 }
582 if (result != expected) {
583 dataerrln(errPrefix + "normalized " + hex(input) + "\n"
584 + " expected " + hex(expected) + "\n"
585 + " iterate got " + hex(result) );
586 }
587 }
588
589 // helper class for TestPreviousNext()
590 // simple UTF-32 character iterator
591 class UChar32Iterator {
592 public:
UChar32Iterator(const UChar32 * text,int32_t len,int32_t index)593 UChar32Iterator(const UChar32 *text, int32_t len, int32_t index) :
594 s(text), length(len), i(index) {}
595
current()596 UChar32 current() {
597 if(i<length) {
598 return s[i];
599 } else {
600 return 0xffff;
601 }
602 }
603
next()604 UChar32 next() {
605 if(i<length) {
606 return s[i++];
607 } else {
608 return 0xffff;
609 }
610 }
611
previous()612 UChar32 previous() {
613 if(i>0) {
614 return s[--i];
615 } else {
616 return 0xffff;
617 }
618 }
619
getIndex()620 int32_t getIndex() {
621 return i;
622 }
623 private:
624 const UChar32 *s;
625 int32_t length, i;
626 };
627
628 void
TestPreviousNext(const UChar * 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)629 BasicNormalizerTest::TestPreviousNext(const UChar *src, int32_t srcLength,
630 const UChar32 *expect, int32_t expectLength,
631 const int32_t *expectIndex, // its length=expectLength+1
632 int32_t srcMiddle, int32_t expectMiddle,
633 const char *moves,
634 UNormalizationMode mode,
635 const char *name) {
636 // Sanity check non-iterative normalization.
637 {
638 IcuTestErrorCode errorCode(*this, "TestPreviousNext");
639 UnicodeString result;
640 Normalizer::normalize(UnicodeString(src, srcLength), mode, 0, result, errorCode);
641 if (errorCode.isFailure()) {
642 dataerrln("error: non-iterative normalization of %s failed: %s",
643 name, errorCode.errorName());
644 errorCode.reset();
645 return;
646 }
647 // UnicodeString::fromUTF32(expect, expectLength)
648 // would turn unpaired surrogates into U+FFFD.
649 for (int32_t i = 0, j = 0; i < result.length(); ++j) {
650 UChar32 c = result.char32At(i);
651 if (c != expect[j]) {
652 errln("error: non-iterative normalization of %s did not yield the expected result",
653 name);
654 }
655 i += U16_LENGTH(c);
656 }
657 }
658
659 // iterators
660 Normalizer iter(src, srcLength, mode);
661
662 // test getStaticClassID and getDynamicClassID
663 if(iter.getDynamicClassID() != Normalizer::getStaticClassID()) {
664 errln("getStaticClassID != getDynamicClassID for Normalizer.");
665 }
666
667 UChar32Iterator iter32(expect, expectLength, expectMiddle);
668
669 UChar32 c1, c2;
670 char m;
671
672 // initially set the indexes into the middle of the strings
673 iter.setIndexOnly(srcMiddle);
674
675 // move around and compare the iteration code points with
676 // the expected ones
677 const char *move=moves;
678 while((m=*move++)!=0) {
679 if(m=='-') {
680 c1=iter.previous();
681 c2=iter32.previous();
682 } else if(m=='0') {
683 c1=iter.current();
684 c2=iter32.current();
685 } else /* m=='+' */ {
686 c1=iter.next();
687 c2=iter32.next();
688 }
689
690 // compare results
691 if(c1!=c2) {
692 // copy the moves until the current (m) move, and terminate
693 char history[64];
694 uprv_strcpy(history, moves);
695 history[move-moves]=0;
696 dataerrln("error: mismatch in Normalizer iteration (%s) at %s: "
697 "got c1=U+%04lx != expected c2=U+%04lx",
698 name, history, c1, c2);
699 break;
700 }
701
702 // compare indexes
703 if(iter.getIndex()!=expectIndex[iter32.getIndex()]) {
704 // copy the moves until the current (m) move, and terminate
705 char history[64];
706 uprv_strcpy(history, moves);
707 history[move-moves]=0;
708 errln("error: index mismatch in Normalizer iteration (%s) at %s: "
709 "Normalizer index %ld expected %ld\n",
710 name, history, iter.getIndex(), expectIndex[iter32.getIndex()]);
711 break;
712 }
713 }
714 }
715
716 void
TestPreviousNext()717 BasicNormalizerTest::TestPreviousNext() {
718 // src and expect strings
719 static const UChar src[]={
720 U16_LEAD(0x2f999), U16_TRAIL(0x2f999),
721 U16_LEAD(0x1d15f), U16_TRAIL(0x1d15f),
722 0xc4,
723 0x1ed0
724 };
725 static const UChar32 expect[]={
726 0x831d,
727 0x1d158, 0x1d165,
728 0x41, 0x308,
729 0x4f, 0x302, 0x301
730 };
731
732 // expected src indexes corresponding to expect indexes
733 static const int32_t expectIndex[]={
734 0,
735 2, 2,
736 4, 4,
737 5, 5, 5,
738 6 // behind last character
739 };
740
741 // src and expect strings for regression test for j2911
742 static const UChar src_j2911[]={
743 U16_LEAD(0x2f999), U16_TRAIL(0x2f999),
744 0xdd00, 0xd900, // unpaired surrogates - regression test for j2911
745 0xc4,
746 0x4f, 0x302, 0x301
747 };
748 static const UChar32 expect_j2911[]={
749 0x831d,
750 0xdd00, 0xd900, // unpaired surrogates - regression test for j2911
751 0xc4,
752 0x1ed0
753 };
754
755 // expected src indexes corresponding to expect indexes
756 static const int32_t expectIndex_j2911[]={
757 0,
758 2, 3,
759 4,
760 5,
761 8 // behind last character
762 };
763
764 // initial indexes into the src and expect strings
765 // for both sets of test data
766 enum {
767 SRC_MIDDLE=4,
768 EXPECT_MIDDLE=3,
769 SRC_MIDDLE_2=2,
770 EXPECT_MIDDLE_2=1
771 };
772
773 // movement vector
774 // - for previous(), 0 for current(), + for next()
775 // for both sets of test data
776 static const char *const moves="0+0+0--0-0-+++0--+++++++0--------";
777
778 TestPreviousNext(src, UPRV_LENGTHOF(src),
779 expect, UPRV_LENGTHOF(expect),
780 expectIndex,
781 SRC_MIDDLE, EXPECT_MIDDLE,
782 moves, UNORM_NFD, "basic");
783
784 TestPreviousNext(src_j2911, UPRV_LENGTHOF(src_j2911),
785 expect_j2911, UPRV_LENGTHOF(expect_j2911),
786 expectIndex_j2911,
787 SRC_MIDDLE, EXPECT_MIDDLE,
788 moves, UNORM_NFKC, "j2911");
789
790 // try again from different "middle" indexes
791 TestPreviousNext(src, UPRV_LENGTHOF(src),
792 expect, UPRV_LENGTHOF(expect),
793 expectIndex,
794 SRC_MIDDLE_2, EXPECT_MIDDLE_2,
795 moves, UNORM_NFD, "basic_2");
796
797 TestPreviousNext(src_j2911, UPRV_LENGTHOF(src_j2911),
798 expect_j2911, UPRV_LENGTHOF(expect_j2911),
799 expectIndex_j2911,
800 SRC_MIDDLE_2, EXPECT_MIDDLE_2,
801 moves, UNORM_NFKC, "j2911_2");
802 }
803
TestConcatenate()804 void BasicNormalizerTest::TestConcatenate() {
805 static const char *const
806 cases[][4]={
807 /* mode, left, right, result */
808 {
809 "C",
810 "re",
811 "\\u0301sum\\u00e9",
812 "r\\u00e9sum\\u00e9"
813 },
814 {
815 "C",
816 "a\\u1100",
817 "\\u1161bcdefghijk",
818 "a\\uac00bcdefghijk"
819 },
820 /* ### TODO: add more interesting cases */
821 {
822 "D",
823 "\\u03B1\\u0345",
824 "\\u0C4D\\U000110BA\\U0001D169",
825 "\\u03B1\\U0001D169\\U000110BA\\u0C4D\\u0345"
826 }
827 };
828
829 UnicodeString left, right, expect, result, r;
830 UErrorCode errorCode;
831 UNormalizationMode mode;
832 int32_t i;
833
834 /* test concatenation */
835 for(i=0; i<UPRV_LENGTHOF(cases); ++i) {
836 switch(*cases[i][0]) {
837 case 'C': mode=UNORM_NFC; break;
838 case 'D': mode=UNORM_NFD; break;
839 case 'c': mode=UNORM_NFKC; break;
840 case 'd': mode=UNORM_NFKD; break;
841 default: mode=UNORM_NONE; break;
842 }
843
844 left=UnicodeString(cases[i][1], "").unescape();
845 right=UnicodeString(cases[i][2], "").unescape();
846 expect=UnicodeString(cases[i][3], "").unescape();
847
848 //result=r=UnicodeString();
849 errorCode=U_ZERO_ERROR;
850
851 r=Normalizer::concatenate(left, right, result, mode, 0, errorCode);
852 if(U_FAILURE(errorCode) || /*result!=r ||*/ result!=expect) {
853 dataerrln("error in Normalizer::concatenate(), cases[] fails with "+
854 UnicodeString(u_errorName(errorCode))+", result==expect: expected: "+
855 hex(expect)+" =========> got: " + hex(result));
856 }
857 }
858
859 /* test error cases */
860
861 /* left.getBuffer()==result.getBuffer() */
862 result=r=expect=UnicodeString("zz", "");
863 errorCode=U_UNEXPECTED_TOKEN;
864 r=Normalizer::concatenate(left, right, result, mode, 0, errorCode);
865 if(errorCode!=U_UNEXPECTED_TOKEN || result!=r || !result.isBogus()) {
866 errln("error in Normalizer::concatenate(), violates UErrorCode protocol");
867 }
868
869 left.setToBogus();
870 errorCode=U_ZERO_ERROR;
871 r=Normalizer::concatenate(left, right, result, mode, 0, errorCode);
872 if(errorCode!=U_ILLEGAL_ARGUMENT_ERROR || result!=r || !result.isBogus()) {
873 errln("error in Normalizer::concatenate(), does not detect left.isBogus()");
874 }
875 }
876
877 // reference implementation of Normalizer::compare
878 static int32_t
ref_norm_compare(const UnicodeString & s1,const UnicodeString & s2,uint32_t options,UErrorCode & errorCode)879 ref_norm_compare(const UnicodeString &s1, const UnicodeString &s2, uint32_t options, UErrorCode &errorCode) {
880 UnicodeString r1, r2, t1, t2;
881 int32_t normOptions=(int32_t)(options>>UNORM_COMPARE_NORM_OPTIONS_SHIFT);
882
883 if(options&U_COMPARE_IGNORE_CASE) {
884 Normalizer::decompose(s1, FALSE, normOptions, r1, errorCode);
885 Normalizer::decompose(s2, FALSE, normOptions, r2, errorCode);
886
887 r1.foldCase(options);
888 r2.foldCase(options);
889 } else {
890 r1=s1;
891 r2=s2;
892 }
893
894 Normalizer::decompose(r1, FALSE, normOptions, t1, errorCode);
895 Normalizer::decompose(r2, FALSE, normOptions, t2, errorCode);
896
897 if(options&U_COMPARE_CODE_POINT_ORDER) {
898 return t1.compareCodePointOrder(t2);
899 } else {
900 return t1.compare(t2);
901 }
902 }
903
904 // test wrapper for Normalizer::compare, sets UNORM_INPUT_IS_FCD appropriately
905 static int32_t
_norm_compare(const UnicodeString & s1,const UnicodeString & s2,uint32_t options,UErrorCode & errorCode)906 _norm_compare(const UnicodeString &s1, const UnicodeString &s2, uint32_t options, UErrorCode &errorCode) {
907 int32_t normOptions=(int32_t)(options>>UNORM_COMPARE_NORM_OPTIONS_SHIFT);
908
909 if( UNORM_YES==Normalizer::quickCheck(s1, UNORM_FCD, normOptions, errorCode) &&
910 UNORM_YES==Normalizer::quickCheck(s2, UNORM_FCD, normOptions, errorCode)) {
911 options|=UNORM_INPUT_IS_FCD;
912 }
913
914 return Normalizer::compare(s1, s2, options, errorCode);
915 }
916
917 // reference implementation of UnicodeString::caseCompare
918 static int32_t
ref_case_compare(const UnicodeString & s1,const UnicodeString & s2,uint32_t options)919 ref_case_compare(const UnicodeString &s1, const UnicodeString &s2, uint32_t options) {
920 UnicodeString t1, t2;
921
922 t1=s1;
923 t2=s2;
924
925 t1.foldCase(options);
926 t2.foldCase(options);
927
928 if(options&U_COMPARE_CODE_POINT_ORDER) {
929 return t1.compareCodePointOrder(t2);
930 } else {
931 return t1.compare(t2);
932 }
933 }
934
935 // reduce an integer to -1/0/1
936 static inline int32_t
_sign(int32_t value)937 _sign(int32_t value) {
938 if(value==0) {
939 return 0;
940 } else {
941 return (value>>31)|1;
942 }
943 }
944
945 static const char *
_signString(int32_t value)946 _signString(int32_t value) {
947 if(value<0) {
948 return "<0";
949 } else if(value==0) {
950 return "=0";
951 } else /* value>0 */ {
952 return ">0";
953 }
954 }
955
956 void
TestCompare()957 BasicNormalizerTest::TestCompare() {
958 // test Normalizer::compare and unorm_compare (thinly wrapped by the former)
959 // by comparing it with its semantic equivalent
960 // since we trust the pieces, this is sufficient
961
962 // test each string with itself and each other
963 // each time with all options
964 static const char *const
965 strings[]={
966 // some cases from NormalizationTest.txt
967 // 0..3
968 "D\\u031B\\u0307\\u0323",
969 "\\u1E0C\\u031B\\u0307",
970 "D\\u031B\\u0323\\u0307",
971 "d\\u031B\\u0323\\u0307",
972
973 // 4..6
974 "\\u00E4",
975 "a\\u0308",
976 "A\\u0308",
977
978 // Angstrom sign = A ring
979 // 7..10
980 "\\u212B",
981 "\\u00C5",
982 "A\\u030A",
983 "a\\u030A",
984
985 // 11.14
986 "a\\u059A\\u0316\\u302A\\u032Fb",
987 "a\\u302A\\u0316\\u032F\\u059Ab",
988 "a\\u302A\\u0316\\u032F\\u059Ab",
989 "A\\u059A\\u0316\\u302A\\u032Fb",
990
991 // from ICU case folding tests
992 // 15..20
993 "A\\u00df\\u00b5\\ufb03\\U0001040c\\u0131",
994 "ass\\u03bcffi\\U00010434i",
995 "\\u0061\\u0042\\u0131\\u03a3\\u00df\\ufb03\\ud93f\\udfff",
996 "\\u0041\\u0062\\u0069\\u03c3\\u0073\\u0053\\u0046\\u0066\\u0049\\ud93f\\udfff",
997 "\\u0041\\u0062\\u0131\\u03c3\\u0053\\u0073\\u0066\\u0046\\u0069\\ud93f\\udfff",
998 "\\u0041\\u0062\\u0069\\u03c3\\u0073\\u0053\\u0046\\u0066\\u0049\\ud93f\\udffd",
999
1000 // U+d800 U+10001 see implementation comment in unorm_cmpEquivFold
1001 // vs. U+10000 at bottom - code point order
1002 // 21..22
1003 "\\ud800\\ud800\\udc01",
1004 "\\ud800\\udc00",
1005
1006 // other code point order tests from ustrtest.cpp
1007 // 23..31
1008 "\\u20ac\\ud801",
1009 "\\u20ac\\ud800\\udc00",
1010 "\\ud800",
1011 "\\ud800\\uff61",
1012 "\\udfff",
1013 "\\uff61\\udfff",
1014 "\\uff61\\ud800\\udc02",
1015 "\\ud800\\udc02",
1016 "\\ud84d\\udc56",
1017
1018 // long strings, see cnormtst.c/TestNormCoverage()
1019 // equivalent if case-insensitive
1020 // 32..33
1021 "\\uAD8B\\uAD8B\\uAD8B\\uAD8B"
1022 "\\U0001d15e\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1023 "\\U0001d15e\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1024 "\\U0001d15e\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1025 "\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1026 "\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1027 "aaaaaaaaaaaaaaaaaazzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz"
1028 "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb"
1029 "ccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc"
1030 "ddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd"
1031 "\\uAD8B\\uAD8B\\uAD8B\\uAD8B"
1032 "d\\u031B\\u0307\\u0323",
1033
1034 "\\u1100\\u116f\\u11aa\\uAD8B\\uAD8B\\u1100\\u116f\\u11aa"
1035 "\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1036 "\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1037 "\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1038 "\\U0001d15e\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1039 "\\U0001d15e\\U0001d157\\U0001d165\\U0001d15e\\U0001d15e\\U0001d15e\\U0001d15e"
1040 "aaaaaaaaaaAAAAAAAAZZZZZZZZZZZZZZZZzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz"
1041 "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb"
1042 "ccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc"
1043 "ddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd"
1044 "\\u1100\\u116f\\u11aa\\uAD8B\\uAD8B\\u1100\\u116f\\u11aa"
1045 "\\u1E0C\\u031B\\u0307",
1046
1047 // some strings that may make a difference whether the compare function
1048 // case-folds or decomposes first
1049 // 34..41
1050 "\\u0360\\u0345\\u0334",
1051 "\\u0360\\u03b9\\u0334",
1052
1053 "\\u0360\\u1f80\\u0334",
1054 "\\u0360\\u03b1\\u0313\\u03b9\\u0334",
1055
1056 "\\u0360\\u1ffc\\u0334",
1057 "\\u0360\\u03c9\\u03b9\\u0334",
1058
1059 "a\\u0360\\u0345\\u0360\\u0345b",
1060 "a\\u0345\\u0360\\u0345\\u0360b",
1061
1062 // interesting cases for canonical caseless match with turkic i handling
1063 // 42..43
1064 "\\u00cc",
1065 "\\u0069\\u0300",
1066
1067 // strings with post-Unicode 3.2 normalization or normalization corrections
1068 // 44..45
1069 "\\u00e4\\u193b\\U0002f868",
1070 "\\u0061\\u193b\\u0308\\u36fc",
1071
1072 // empty string
1073 // 46
1074 ""
1075 };
1076
1077 UnicodeString s[100]; // at least as many items as in strings[] !
1078
1079 // all combinations of options
1080 // UNORM_INPUT_IS_FCD is set automatically if both input strings fulfill FCD conditions
1081 // set UNORM_UNICODE_3_2 in one additional combination
1082 static const struct {
1083 uint32_t options;
1084 const char *name;
1085 } opt[]={
1086 { 0, "default" },
1087 { U_COMPARE_CODE_POINT_ORDER, "c.p. order" },
1088 { U_COMPARE_IGNORE_CASE, "ignore case" },
1089 { U_COMPARE_CODE_POINT_ORDER|U_COMPARE_IGNORE_CASE, "c.p. order & ignore case" },
1090 { U_COMPARE_IGNORE_CASE|U_FOLD_CASE_EXCLUDE_SPECIAL_I, "ignore case & special i" },
1091 { U_COMPARE_CODE_POINT_ORDER|U_COMPARE_IGNORE_CASE|U_FOLD_CASE_EXCLUDE_SPECIAL_I, "c.p. order & ignore case & special i" },
1092 { UNORM_UNICODE_3_2<<UNORM_COMPARE_NORM_OPTIONS_SHIFT, "Unicode 3.2" }
1093 };
1094
1095 int32_t i, j, k, count=UPRV_LENGTHOF(strings);
1096 int32_t result, refResult;
1097
1098 UErrorCode errorCode;
1099
1100 // create the UnicodeStrings
1101 for(i=0; i<count; ++i) {
1102 s[i]=UnicodeString(strings[i], "").unescape();
1103 }
1104
1105 // test them each with each other
1106 for(i=0; i<count; ++i) {
1107 for(j=i; j<count; ++j) {
1108 for(k=0; k<UPRV_LENGTHOF(opt); ++k) {
1109 // test Normalizer::compare
1110 errorCode=U_ZERO_ERROR;
1111 result=_norm_compare(s[i], s[j], opt[k].options, errorCode);
1112 refResult=ref_norm_compare(s[i], s[j], opt[k].options, errorCode);
1113 if(_sign(result)!=_sign(refResult)) {
1114 errln("Normalizer::compare(%d, %d, %s)%s should be %s %s",
1115 i, j, opt[k].name, _signString(result), _signString(refResult),
1116 U_SUCCESS(errorCode) ? "" : u_errorName(errorCode));
1117 }
1118
1119 // test UnicodeString::caseCompare - same internal implementation function
1120 if(opt[k].options&U_COMPARE_IGNORE_CASE) {
1121 errorCode=U_ZERO_ERROR;
1122 result=s[i].caseCompare(s[j], opt[k].options);
1123 refResult=ref_case_compare(s[i], s[j], opt[k].options);
1124 if(_sign(result)!=_sign(refResult)) {
1125 errln("UniStr::caseCompare(%d, %d, %s)%s should be %s %s",
1126 i, j, opt[k].name, _signString(result), _signString(refResult),
1127 U_SUCCESS(errorCode) ? "" : u_errorName(errorCode));
1128 }
1129 }
1130 }
1131 }
1132 }
1133
1134 // test cases with i and I to make sure Turkic works
1135 static const UChar iI[]={ 0x49, 0x69, 0x130, 0x131 };
1136 UnicodeSet iSet, set;
1137
1138 UnicodeString s1, s2;
1139
1140 const Normalizer2Impl *nfcImpl=Normalizer2Factory::getNFCImpl(errorCode);
1141 if(U_FAILURE(errorCode) || !nfcImpl->ensureCanonIterData(errorCode)) {
1142 dataerrln("Normalizer2Factory::getNFCImpl().ensureCanonIterData() failed: %s",
1143 u_errorName(errorCode));
1144 return;
1145 }
1146
1147 // collect all sets into one for contiguous output
1148 for(i=0; i<UPRV_LENGTHOF(iI); ++i) {
1149 if(nfcImpl->getCanonStartSet(iI[i], iSet)) {
1150 set.addAll(iSet);
1151 }
1152 }
1153
1154 // test all of these precomposed characters
1155 const Normalizer2 *nfcNorm2=Normalizer2::getNFCInstance(errorCode);
1156 UnicodeSetIterator it(set);
1157 while(it.next() && !it.isString()) {
1158 UChar32 c=it.getCodepoint();
1159 if(!nfcNorm2->getDecomposition(c, s2)) {
1160 dataerrln("NFC.getDecomposition(i-composite U+%04lx) failed", (long)c);
1161 return;
1162 }
1163
1164 s1.setTo(c);
1165 for(k=0; k<UPRV_LENGTHOF(opt); ++k) {
1166 // test Normalizer::compare
1167 errorCode=U_ZERO_ERROR;
1168 result=_norm_compare(s1, s2, opt[k].options, errorCode);
1169 refResult=ref_norm_compare(s1, s2, opt[k].options, errorCode);
1170 if(_sign(result)!=_sign(refResult)) {
1171 errln("Normalizer::compare(U+%04x with its NFD, %s)%s should be %s %s",
1172 c, opt[k].name, _signString(result), _signString(refResult),
1173 U_SUCCESS(errorCode) ? "" : u_errorName(errorCode));
1174 }
1175
1176 // test UnicodeString::caseCompare - same internal implementation function
1177 if(opt[k].options&U_COMPARE_IGNORE_CASE) {
1178 errorCode=U_ZERO_ERROR;
1179 result=s1.caseCompare(s2, opt[k].options);
1180 refResult=ref_case_compare(s1, s2, opt[k].options);
1181 if(_sign(result)!=_sign(refResult)) {
1182 errln("UniStr::caseCompare(U+%04x with its NFD, %s)%s should be %s %s",
1183 c, opt[k].name, _signString(result), _signString(refResult),
1184 U_SUCCESS(errorCode) ? "" : u_errorName(errorCode));
1185 }
1186 }
1187 }
1188 }
1189
1190 // test getDecomposition() for some characters that do not decompose
1191 if( nfcNorm2->getDecomposition(0x20, s2) ||
1192 nfcNorm2->getDecomposition(0x4e00, s2) ||
1193 nfcNorm2->getDecomposition(0x20002, s2)
1194 ) {
1195 errln("NFC.getDecomposition() returns TRUE for characters which do not have decompositions");
1196 }
1197
1198 // test getRawDecomposition() for some characters that do not decompose
1199 if( nfcNorm2->getRawDecomposition(0x20, s2) ||
1200 nfcNorm2->getRawDecomposition(0x4e00, s2) ||
1201 nfcNorm2->getRawDecomposition(0x20002, s2)
1202 ) {
1203 errln("NFC.getRawDecomposition() returns TRUE for characters which do not have decompositions");
1204 }
1205
1206 // test composePair() for some pairs of characters that do not compose
1207 if( nfcNorm2->composePair(0x20, 0x301)>=0 ||
1208 nfcNorm2->composePair(0x61, 0x305)>=0 ||
1209 nfcNorm2->composePair(0x1100, 0x1160)>=0 ||
1210 nfcNorm2->composePair(0xac00, 0x11a7)>=0
1211 ) {
1212 errln("NFC.composePair() incorrectly composes some pairs of characters");
1213 }
1214
1215 // test FilteredNormalizer2::getDecomposition()
1216 UnicodeSet filter(UNICODE_STRING_SIMPLE("[^\\u00a0-\\u00ff]"), errorCode);
1217 FilteredNormalizer2 fn2(*nfcNorm2, filter);
1218 if( fn2.getDecomposition(0xe4, s1) || !fn2.getDecomposition(0x100, s2) ||
1219 s2.length()!=2 || s2[0]!=0x41 || s2[1]!=0x304
1220 ) {
1221 errln("FilteredNormalizer2(NFC, ^A0-FF).getDecomposition() failed");
1222 }
1223
1224 // test FilteredNormalizer2::getRawDecomposition()
1225 if( fn2.getRawDecomposition(0xe4, s1) || !fn2.getRawDecomposition(0x100, s2) ||
1226 s2.length()!=2 || s2[0]!=0x41 || s2[1]!=0x304
1227 ) {
1228 errln("FilteredNormalizer2(NFC, ^A0-FF).getRawDecomposition() failed");
1229 }
1230
1231 // test FilteredNormalizer2::composePair()
1232 if( 0x100!=fn2.composePair(0x41, 0x304) ||
1233 fn2.composePair(0xc7, 0x301)>=0 // unfiltered result: U+1E08
1234 ) {
1235 errln("FilteredNormalizer2(NFC, ^A0-FF).composePair() failed");
1236 }
1237 }
1238
1239 // verify that case-folding does not un-FCD strings
1240 int32_t
countFoldFCDExceptions(uint32_t foldingOptions)1241 BasicNormalizerTest::countFoldFCDExceptions(uint32_t foldingOptions) {
1242 UnicodeString s, fold, d;
1243 UChar32 c;
1244 int32_t count;
1245 uint8_t cc, trailCC, foldCC, foldTrailCC;
1246 UNormalizationCheckResult qcResult;
1247 int8_t category;
1248 UBool isNFD;
1249 UErrorCode errorCode;
1250
1251 logln("Test if case folding may un-FCD a string (folding options %04lx)", foldingOptions);
1252
1253 count=0;
1254 for(c=0; c<=0x10ffff; ++c) {
1255 errorCode = U_ZERO_ERROR;
1256 category=u_charType(c);
1257 if(category==U_UNASSIGNED) {
1258 continue; // skip unassigned code points
1259 }
1260 if(c==0xac00) {
1261 c=0xd7a3; // skip Hangul - no case folding there
1262 continue;
1263 }
1264 // skip Han blocks - no case folding there either
1265 if(c==0x3400) {
1266 c=0x4db5;
1267 continue;
1268 }
1269 if(c==0x4e00) {
1270 c=0x9fa5;
1271 continue;
1272 }
1273 if(c==0x20000) {
1274 c=0x2a6d6;
1275 continue;
1276 }
1277
1278 s.setTo(c);
1279
1280 // get leading and trailing cc for c
1281 Normalizer::decompose(s, FALSE, 0, d, errorCode);
1282 isNFD= s==d;
1283 cc=u_getCombiningClass(d.char32At(0));
1284 trailCC=u_getCombiningClass(d.char32At(d.length()-1));
1285
1286 // get leading and trailing cc for the case-folding of c
1287 s.foldCase(foldingOptions);
1288 Normalizer::decompose(s, FALSE, 0, d, errorCode);
1289 foldCC=u_getCombiningClass(d.char32At(0));
1290 foldTrailCC=u_getCombiningClass(d.char32At(d.length()-1));
1291
1292 qcResult=Normalizer::quickCheck(s, UNORM_FCD, errorCode);
1293
1294 if (U_FAILURE(errorCode)) {
1295 ++count;
1296 dataerrln("U+%04lx: Failed with error %s", u_errorName(errorCode));
1297 }
1298
1299 // bad:
1300 // - character maps to empty string: adjacent characters may then need reordering
1301 // - folding has different leading/trailing cc's, and they don't become just 0
1302 // - folding itself is not FCD
1303 if( qcResult!=UNORM_YES ||
1304 s.isEmpty() ||
1305 (cc!=foldCC && foldCC!=0) || (trailCC!=foldTrailCC && foldTrailCC!=0)
1306 ) {
1307 ++count;
1308 dataerrln("U+%04lx: case-folding may un-FCD a string (folding options %04lx)", c, foldingOptions);
1309 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);
1310 continue;
1311 }
1312
1313 // also bad:
1314 // if a code point is in NFD but its case folding is not, then
1315 // unorm_compare will also fail
1316 if(isNFD && UNORM_YES!=Normalizer::quickCheck(s, UNORM_NFD, errorCode)) {
1317 ++count;
1318 errln("U+%04lx: case-folding un-NFDs this character (folding options %04lx)", c, foldingOptions);
1319 }
1320 }
1321
1322 logln("There are %ld code points for which case-folding may un-FCD a string (folding options %04lx)", count, foldingOptions);
1323 return count;
1324 }
1325
1326 void
FindFoldFCDExceptions()1327 BasicNormalizerTest::FindFoldFCDExceptions() {
1328 int32_t count;
1329
1330 count=countFoldFCDExceptions(0);
1331 count+=countFoldFCDExceptions(U_FOLD_CASE_EXCLUDE_SPECIAL_I);
1332 if(count>0) {
1333 /*
1334 * If case-folding un-FCDs any strings, then unorm_compare() must be
1335 * re-implemented.
1336 * It currently assumes that one can check for FCD then case-fold
1337 * and then still have FCD strings for raw decomposition without reordering.
1338 */
1339 dataerrln("error: There are %ld code points for which case-folding may un-FCD a string for all folding options.\n"
1340 "See comment in BasicNormalizerTest::FindFoldFCDExceptions()!", count);
1341 }
1342 }
1343
1344 static void
initExpectedSkippables(UnicodeSet skipSets[UNORM_MODE_COUNT],UErrorCode & errorCode)1345 initExpectedSkippables(UnicodeSet skipSets[UNORM_MODE_COUNT], UErrorCode &errorCode) {
1346 skipSets[UNORM_NFD].applyPattern(
1347 UNICODE_STRING_SIMPLE("[[:NFD_QC=Yes:]&[:ccc=0:]]"), errorCode);
1348 skipSets[UNORM_NFC].applyPattern(
1349 UNICODE_STRING_SIMPLE("[[:NFC_QC=Yes:]&[:ccc=0:]-[:HST=LV:]]"), errorCode);
1350 skipSets[UNORM_NFKD].applyPattern(
1351 UNICODE_STRING_SIMPLE("[[:NFKD_QC=Yes:]&[:ccc=0:]]"), errorCode);
1352 skipSets[UNORM_NFKC].applyPattern(
1353 UNICODE_STRING_SIMPLE("[[:NFKC_QC=Yes:]&[:ccc=0:]-[:HST=LV:]]"), errorCode);
1354
1355 // Remove from the NFC and NFKC sets all those characters that change
1356 // when a back-combining character is added.
1357 // First, get all of the back-combining characters and their combining classes.
1358 UnicodeSet combineBack("[:NFC_QC=Maybe:]", errorCode);
1359 int32_t numCombineBack=combineBack.size();
1360 int32_t *combineBackCharsAndCc=new int32_t[numCombineBack*2];
1361 UnicodeSetIterator iter(combineBack);
1362 for(int32_t i=0; i<numCombineBack; ++i) {
1363 iter.next();
1364 UChar32 c=iter.getCodepoint();
1365 combineBackCharsAndCc[2*i]=c;
1366 combineBackCharsAndCc[2*i+1]=u_getCombiningClass(c);
1367 }
1368
1369 // We need not look at control codes, Han characters nor Hangul LVT syllables because they
1370 // do not combine forward. LV syllables are already removed.
1371 UnicodeSet notInteresting("[[:C:][:Unified_Ideograph:][:HST=LVT:]]", errorCode);
1372 LocalPointer<UnicodeSet> unsure(&((UnicodeSet *)(skipSets[UNORM_NFC].clone()))->removeAll(notInteresting));
1373 // System.out.format("unsure.size()=%d\n", unsure.size());
1374
1375 // For each character about which we are unsure, see if it changes when we add
1376 // one of the back-combining characters.
1377 const Normalizer2 *norm2=Normalizer2::getNFCInstance(errorCode);
1378 UnicodeString s;
1379 iter.reset(*unsure);
1380 while(iter.next()) {
1381 UChar32 c=iter.getCodepoint();
1382 s.setTo(c);
1383 int32_t cLength=s.length();
1384 int32_t tccc=u_getIntPropertyValue(c, UCHAR_TRAIL_CANONICAL_COMBINING_CLASS);
1385 for(int32_t i=0; i<numCombineBack; ++i) {
1386 // If c's decomposition ends with a character with non-zero combining class, then
1387 // c can only change if it combines with a character with a non-zero combining class.
1388 int32_t cc2=combineBackCharsAndCc[2*i+1];
1389 if(tccc==0 || cc2!=0) {
1390 UChar32 c2=combineBackCharsAndCc[2*i];
1391 s.append(c2);
1392 if(!norm2->isNormalized(s, errorCode)) {
1393 // System.out.format("remove U+%04x (tccc=%d) + U+%04x (cc=%d)\n", c, tccc, c2, cc2);
1394 skipSets[UNORM_NFC].remove(c);
1395 skipSets[UNORM_NFKC].remove(c);
1396 break;
1397 }
1398 s.truncate(cLength);
1399 }
1400 }
1401 }
1402 delete [] combineBackCharsAndCc;
1403 }
1404
1405 static const char *const kModeStrings[UNORM_MODE_COUNT] = {
1406 "?", "none", "D", "KD", "C", "KC", "FCD"
1407 };
1408
1409 void
TestSkippable()1410 BasicNormalizerTest::TestSkippable() {
1411 UnicodeSet diff, skipSets[UNORM_MODE_COUNT], expectSets[UNORM_MODE_COUNT];
1412 UnicodeString s, pattern;
1413
1414 /* build NF*Skippable sets from runtime data */
1415 IcuTestErrorCode errorCode(*this, "TestSkippable");
1416 skipSets[UNORM_NFD].applyPattern(UNICODE_STRING_SIMPLE("[:NFD_Inert:]"), errorCode);
1417 skipSets[UNORM_NFKD].applyPattern(UNICODE_STRING_SIMPLE("[:NFKD_Inert:]"), errorCode);
1418 skipSets[UNORM_NFC].applyPattern(UNICODE_STRING_SIMPLE("[:NFC_Inert:]"), errorCode);
1419 skipSets[UNORM_NFKC].applyPattern(UNICODE_STRING_SIMPLE("[:NFKC_Inert:]"), errorCode);
1420 if(errorCode.errDataIfFailureAndReset("UnicodeSet(NF..._Inert) failed")) {
1421 return;
1422 }
1423
1424 /* get expected sets from hardcoded patterns */
1425 initExpectedSkippables(expectSets, errorCode);
1426 errorCode.assertSuccess();
1427
1428 for(int32_t i=UNORM_NONE; i<UNORM_MODE_COUNT; ++i) {
1429 if(skipSets[i]!=expectSets[i]) {
1430 const char *ms=kModeStrings[i];
1431 errln("error: TestSkippable skipSets[%s]!=expectedSets[%s]\n", ms, ms);
1432 // Note: This used to depend on hardcoded UnicodeSet patterns generated by
1433 // Mark's unicodetools.com.ibm.text.UCD.NFSkippable, by
1434 // running com.ibm.text.UCD.Main with the option NFSkippable.
1435 // Since ICU 4.6/Unicode 6, we are generating the
1436 // expectSets ourselves in initSkippables().
1437
1438 s=UNICODE_STRING_SIMPLE("skip-expect=");
1439 (diff=skipSets[i]).removeAll(expectSets[i]).toPattern(pattern, TRUE);
1440 s.append(pattern);
1441
1442 pattern.remove();
1443 s.append(UNICODE_STRING_SIMPLE("\n\nexpect-skip="));
1444 (diff=expectSets[i]).removeAll(skipSets[i]).toPattern(pattern, TRUE);
1445 s.append(pattern);
1446 s.append(UNICODE_STRING_SIMPLE("\n\n"));
1447
1448 errln(s);
1449 }
1450 }
1451 }
1452
1453 struct StringPair { const char *input, *expected; };
1454
1455 void
TestCustomComp()1456 BasicNormalizerTest::TestCustomComp() {
1457 static const StringPair pairs[]={
1458 // ICU 63 normalization with UCPTrie requires inert surrogate code points.
1459 // { "\\uD801\\uE000\\uDFFE", "" },
1460 // { "\\uD800\\uD801\\uE000\\uDFFE\\uDFFF", "\\uD7FF\\uFFFF" },
1461 // { "\\uD800\\uD801\\uDFFE\\uDFFF", "\\uD7FF\\U000107FE\\uFFFF" },
1462 { "\\uD801\\uE000\\uDFFE", "\\uD801\\uDFFE" },
1463 { "\\uD800\\uD801\\uE000\\uDFFE\\uDFFF", "\\uD800\\uD801\\uDFFE\\uDFFF" },
1464 { "\\uD800\\uD801\\uDFFE\\uDFFF", "\\uD800\\U000107FE\\uDFFF" },
1465
1466 { "\\uE001\\U000110B9\\u0345\\u0308\\u0327", "\\uE002\\U000110B9\\u0327\\u0345" },
1467 { "\\uE010\\U000F0011\\uE012", "\\uE011\\uE012" },
1468 { "\\uE010\\U000F0011\\U000F0011\\uE012", "\\uE011\\U000F0010" },
1469 { "\\uE111\\u1161\\uE112\\u1162", "\\uAE4C\\u1102\\u0062\\u1162" },
1470 { "\\uFFF3\\uFFF7\\U00010036\\U00010077", "\\U00010037\\U00010037\\uFFF6\\U00010037" }
1471 };
1472 IcuTestErrorCode errorCode(*this, "BasicNormalizerTest/TestCustomComp");
1473 const Normalizer2 *customNorm2=
1474 Normalizer2::getInstance(loadTestData(errorCode), "testnorm",
1475 UNORM2_COMPOSE, errorCode);
1476 if(errorCode.errDataIfFailureAndReset("unable to load testdata/testnorm.nrm")) {
1477 return;
1478 }
1479 for(int32_t i=0; i<UPRV_LENGTHOF(pairs); ++i) {
1480 const StringPair &pair=pairs[i];
1481 UnicodeString input=UnicodeString(pair.input, -1, US_INV).unescape();
1482 UnicodeString expected=UnicodeString(pair.expected, -1, US_INV).unescape();
1483 UnicodeString result=customNorm2->normalize(input, errorCode);
1484 if(result!=expected) {
1485 errln("custom compose Normalizer2 did not normalize input %d as expected", i);
1486 }
1487 }
1488 }
1489
1490 void
TestCustomFCC()1491 BasicNormalizerTest::TestCustomFCC() {
1492 static const StringPair pairs[]={
1493 // ICU 63 normalization with UCPTrie requires inert surrogate code points.
1494 // { "\\uD801\\uE000\\uDFFE", "" },
1495 // { "\\uD800\\uD801\\uE000\\uDFFE\\uDFFF", "\\uD7FF\\uFFFF" },
1496 // { "\\uD800\\uD801\\uDFFE\\uDFFF", "\\uD7FF\\U000107FE\\uFFFF" },
1497 { "\\uD801\\uE000\\uDFFE", "\\uD801\\uDFFE" },
1498 { "\\uD800\\uD801\\uE000\\uDFFE\\uDFFF", "\\uD800\\uD801\\uDFFE\\uDFFF" },
1499 { "\\uD800\\uD801\\uDFFE\\uDFFF", "\\uD800\\U000107FE\\uDFFF" },
1500
1501 // The following expected result is different from CustomComp
1502 // because of only-contiguous composition.
1503 { "\\uE001\\U000110B9\\u0345\\u0308\\u0327", "\\uE001\\U000110B9\\u0327\\u0308\\u0345" },
1504 { "\\uE010\\U000F0011\\uE012", "\\uE011\\uE012" },
1505 { "\\uE010\\U000F0011\\U000F0011\\uE012", "\\uE011\\U000F0010" },
1506 { "\\uE111\\u1161\\uE112\\u1162", "\\uAE4C\\u1102\\u0062\\u1162" },
1507 { "\\uFFF3\\uFFF7\\U00010036\\U00010077", "\\U00010037\\U00010037\\uFFF6\\U00010037" }
1508 };
1509 IcuTestErrorCode errorCode(*this, "BasicNormalizerTest/TestCustomFCC");
1510 const Normalizer2 *customNorm2=
1511 Normalizer2::getInstance(loadTestData(errorCode), "testnorm",
1512 UNORM2_COMPOSE_CONTIGUOUS, errorCode);
1513 if(errorCode.errDataIfFailureAndReset("unable to load testdata/testnorm.nrm")) {
1514 return;
1515 }
1516 for(int32_t i=0; i<UPRV_LENGTHOF(pairs); ++i) {
1517 const StringPair &pair=pairs[i];
1518 UnicodeString input=UnicodeString(pair.input, -1, US_INV).unescape();
1519 UnicodeString expected=UnicodeString(pair.expected, -1, US_INV).unescape();
1520 UnicodeString result=customNorm2->normalize(input, errorCode);
1521 if(result!=expected) {
1522 errln("custom FCC Normalizer2 did not normalize input %d as expected", i);
1523 }
1524 }
1525 }
1526
1527 /* Improve code coverage of Normalizer2 */
1528 void
TestFilteredNormalizer2Coverage()1529 BasicNormalizerTest::TestFilteredNormalizer2Coverage() {
1530 UErrorCode errorCode = U_ZERO_ERROR;
1531 const Normalizer2 *nfcNorm2=Normalizer2::getNFCInstance(errorCode);
1532 if (U_FAILURE(errorCode)) {
1533 dataerrln("Normalizer2::getNFCInstance() call failed - %s", u_errorName(errorCode));
1534 return;
1535 }
1536 UnicodeSet filter(UNICODE_STRING_SIMPLE("[^\\u00a0-\\u00ff\\u0310-\\u031f]"), errorCode);
1537 FilteredNormalizer2 fn2(*nfcNorm2, filter);
1538
1539 UChar32 char32 = 0x0054;
1540
1541 if (fn2.isInert(char32)) {
1542 errln("FilteredNormalizer2.isInert() failed.");
1543 }
1544
1545 if (fn2.hasBoundaryAfter(char32)) {
1546 errln("FilteredNormalizer2.hasBoundaryAfter() failed.");
1547 }
1548
1549 UChar32 c;
1550 for(c=0; c<=0x3ff; ++c) {
1551 uint8_t expectedCC= filter.contains(c) ? nfcNorm2->getCombiningClass(c) : 0;
1552 uint8_t cc=fn2.getCombiningClass(c);
1553 if(cc!=expectedCC) {
1554 errln(
1555 UnicodeString("FilteredNormalizer2(NFC, ^A0-FF,310-31F).getCombiningClass(U+")+
1556 hex(c)+
1557 ")==filtered NFC.getCC()");
1558 }
1559 }
1560
1561 UnicodeString newString1 = UNICODE_STRING_SIMPLE("[^\\u0100-\\u01ff]");
1562 UnicodeString newString2 = UNICODE_STRING_SIMPLE("[^\\u0200-\\u02ff]");
1563 fn2.append(newString1, newString2, errorCode);
1564 if (U_FAILURE(errorCode)) {
1565 errln("FilteredNormalizer2.append() failed.");
1566 }
1567 }
1568
1569 void
TestNormalizeUTF8WithEdits()1570 BasicNormalizerTest::TestNormalizeUTF8WithEdits() {
1571 IcuTestErrorCode errorCode(*this, "TestNormalizeUTF8WithEdits");
1572 const Normalizer2 *nfkc_cf=Normalizer2::getNFKCCasefoldInstance(errorCode);
1573 if(errorCode.errDataIfFailureAndReset("Normalizer2::getNFKCCasefoldInstance() call failed")) {
1574 return;
1575 }
1576 static const char *const src =
1577 u8" AÄA\u0308A\u0308\u00ad\u0323Ä\u0323,\u00ad\u1100\u1161가\u11A8가\u3133 ";
1578 std::string expected = u8" aääạ\u0308ạ\u0308,가각갃 ";
1579 std::string result;
1580 StringByteSink<std::string> sink(&result, expected.length());
1581 Edits edits;
1582 nfkc_cf->normalizeUTF8(0, src, sink, &edits, errorCode);
1583 assertSuccess("normalizeUTF8 with Edits", errorCode.get());
1584 assertEquals("normalizeUTF8 with Edits", expected.c_str(), result.c_str());
1585 static const EditChange expectedChanges[] = {
1586 { FALSE, 2, 2 }, // 2 spaces
1587 { TRUE, 1, 1 }, // A→a
1588 { TRUE, 2, 2 }, // Ä→ä
1589 { TRUE, 3, 2 }, // A\u0308→ä
1590 { TRUE, 7, 5 }, // A\u0308\u00ad\u0323→ạ\u0308 removes the soft hyphen
1591 { TRUE, 4, 5 }, // Ä\u0323→ ạ\u0308
1592 { FALSE, 1, 1 }, // comma
1593 { TRUE, 2, 0 }, // U+00AD soft hyphen maps to empty
1594 { TRUE, 6, 3 }, // \u1100\u1161→ 가
1595 { TRUE, 6, 3 }, // 가\u11A8→ 각
1596 { TRUE, 6, 3 }, // 가\u3133→ 갃
1597 { FALSE, 2, 2 } // 2 spaces
1598 };
1599 assertTrue("normalizeUTF8 with Edits hasChanges", edits.hasChanges());
1600 assertEquals("normalizeUTF8 with Edits numberOfChanges", 9, edits.numberOfChanges());
1601 TestUtility::checkEditsIter(*this, u"normalizeUTF8 with Edits",
1602 edits.getFineIterator(), edits.getFineIterator(),
1603 expectedChanges, UPRV_LENGTHOF(expectedChanges),
1604 TRUE, errorCode);
1605
1606 assertFalse("isNormalizedUTF8(source)", nfkc_cf->isNormalizedUTF8(src, errorCode));
1607 assertTrue("isNormalizedUTF8(normalized)", nfkc_cf->isNormalizedUTF8(result, errorCode));
1608
1609 // Omit unchanged text.
1610 expected = u8"aääạ\u0308ạ\u0308가각갃";
1611 result.clear();
1612 edits.reset();
1613 nfkc_cf->normalizeUTF8(U_OMIT_UNCHANGED_TEXT, src, sink, &edits, errorCode);
1614 assertSuccess("normalizeUTF8 omit unchanged", errorCode.get());
1615 assertEquals("normalizeUTF8 omit unchanged", expected.c_str(), result.c_str());
1616 assertTrue("normalizeUTF8 omit unchanged hasChanges", edits.hasChanges());
1617 assertEquals("normalizeUTF8 omit unchanged numberOfChanges", 9, edits.numberOfChanges());
1618 TestUtility::checkEditsIter(*this, u"normalizeUTF8 omit unchanged",
1619 edits.getFineIterator(), edits.getFineIterator(),
1620 expectedChanges, UPRV_LENGTHOF(expectedChanges),
1621 TRUE, errorCode);
1622
1623 // With filter: The normalization code does not see the "A" substrings.
1624 UnicodeSet filter(u"[^A]", errorCode);
1625 FilteredNormalizer2 fn2(*nfkc_cf, filter);
1626 expected = u8" AäA\u0308A\u0323\u0308ạ\u0308,가각갃 ";
1627 result.clear();
1628 edits.reset();
1629 fn2.normalizeUTF8(0, src, sink, &edits, errorCode);
1630 assertSuccess("filtered normalizeUTF8", errorCode.get());
1631 assertEquals("filtered normalizeUTF8", expected.c_str(), result.c_str());
1632 static const EditChange filteredChanges[] = {
1633 { FALSE, 3, 3 }, // 2 spaces + A
1634 { TRUE, 2, 2 }, // Ä→ä
1635 { FALSE, 4, 4 }, // A\u0308A
1636 { TRUE, 6, 4 }, // \u0308\u00ad\u0323→\u0323\u0308 removes the soft hyphen
1637 { TRUE, 4, 5 }, // Ä\u0323→ ạ\u0308
1638 { FALSE, 1, 1 }, // comma
1639 { TRUE, 2, 0 }, // U+00AD soft hyphen maps to empty
1640 { TRUE, 6, 3 }, // \u1100\u1161→ 가
1641 { TRUE, 6, 3 }, // 가\u11A8→ 각
1642 { TRUE, 6, 3 }, // 가\u3133→ 갃
1643 { FALSE, 2, 2 } // 2 spaces
1644 };
1645 assertTrue("filtered normalizeUTF8 hasChanges", edits.hasChanges());
1646 assertEquals("filtered normalizeUTF8 numberOfChanges", 7, edits.numberOfChanges());
1647 TestUtility::checkEditsIter(*this, u"filtered normalizeUTF8",
1648 edits.getFineIterator(), edits.getFineIterator(),
1649 filteredChanges, UPRV_LENGTHOF(filteredChanges),
1650 TRUE, errorCode);
1651
1652 assertFalse("filtered isNormalizedUTF8(source)", fn2.isNormalizedUTF8(src, errorCode));
1653 assertTrue("filtered isNormalizedUTF8(normalized)", fn2.isNormalizedUTF8(result, errorCode));
1654
1655 // Omit unchanged text.
1656 // Note that the result is not normalized because the inner normalizer
1657 // does not see text across filter spans.
1658 expected = u8"ä\u0323\u0308ạ\u0308가각갃";
1659 result.clear();
1660 edits.reset();
1661 fn2.normalizeUTF8(U_OMIT_UNCHANGED_TEXT, src, sink, &edits, errorCode);
1662 assertSuccess("filtered normalizeUTF8 omit unchanged", errorCode.get());
1663 assertEquals("filtered normalizeUTF8 omit unchanged", expected.c_str(), result.c_str());
1664 assertTrue("filtered normalizeUTF8 omit unchanged hasChanges", edits.hasChanges());
1665 assertEquals("filtered normalizeUTF8 omit unchanged numberOfChanges", 7, edits.numberOfChanges());
1666 TestUtility::checkEditsIter(*this, u"filtered normalizeUTF8 omit unchanged",
1667 edits.getFineIterator(), edits.getFineIterator(),
1668 filteredChanges, UPRV_LENGTHOF(filteredChanges),
1669 TRUE, errorCode);
1670 }
1671
1672 void
TestLowMappingToEmpty_D()1673 BasicNormalizerTest::TestLowMappingToEmpty_D() {
1674 IcuTestErrorCode errorCode(*this, "TestLowMappingToEmpty_D");
1675 const Normalizer2 *n2 = Normalizer2::getInstance(
1676 nullptr, "nfkc_cf", UNORM2_DECOMPOSE, errorCode);
1677 if (errorCode.errDataIfFailureAndReset("Normalizer2::getInstance() call failed")) {
1678 return;
1679 }
1680 checkLowMappingToEmpty(*n2);
1681
1682 UnicodeString sh(u'\u00AD');
1683 assertFalse("soft hyphen is not normalized", n2->isNormalized(sh, errorCode));
1684 UnicodeString result = n2->normalize(sh, errorCode);
1685 assertTrue("soft hyphen normalizes to empty", result.isEmpty());
1686 assertEquals("soft hyphen QC=No", UNORM_NO, n2->quickCheck(sh, errorCode));
1687 assertEquals("soft hyphen spanQuickCheckYes", 0, n2->spanQuickCheckYes(sh, errorCode));
1688
1689 UnicodeString s(u"\u00ADÄ\u00AD\u0323");
1690 result = n2->normalize(s, errorCode);
1691 assertEquals("normalize string with soft hyphens", u"a\u0323\u0308", result);
1692 }
1693
1694 void
TestLowMappingToEmpty_FCD()1695 BasicNormalizerTest::TestLowMappingToEmpty_FCD() {
1696 IcuTestErrorCode errorCode(*this, "TestLowMappingToEmpty_FCD");
1697 const Normalizer2 *n2 = Normalizer2::getInstance(
1698 nullptr, "nfkc_cf", UNORM2_FCD, errorCode);
1699 if (errorCode.errDataIfFailureAndReset("Normalizer2::getInstance() call failed")) {
1700 return;
1701 }
1702 checkLowMappingToEmpty(*n2);
1703
1704 UnicodeString sh(u'\u00AD');
1705 assertTrue("soft hyphen is FCD", n2->isNormalized(sh, errorCode));
1706
1707 UnicodeString s(u"\u00ADÄ\u00AD\u0323");
1708 UnicodeString result = n2->normalize(s, errorCode);
1709 assertEquals("normalize string with soft hyphens", u"\u00ADa\u0323\u0308", result);
1710 }
1711
1712 void
checkLowMappingToEmpty(const Normalizer2 & n2)1713 BasicNormalizerTest::checkLowMappingToEmpty(const Normalizer2 &n2) {
1714 UnicodeString mapping;
1715 assertTrue("getDecomposition(soft hyphen)", n2.getDecomposition(0xad, mapping));
1716 assertTrue("soft hyphen maps to empty", mapping.isEmpty());
1717 assertFalse("soft hyphen has no boundary before", n2.hasBoundaryBefore(0xad));
1718 assertFalse("soft hyphen has no boundary after", n2.hasBoundaryAfter(0xad));
1719 assertFalse("soft hyphen is not inert", n2.isInert(0xad));
1720 }
1721
1722 void
TestNormalizeIllFormedText()1723 BasicNormalizerTest::TestNormalizeIllFormedText() {
1724 IcuTestErrorCode errorCode(*this, "TestNormalizeIllFormedText");
1725 const Normalizer2 *nfkc_cf = Normalizer2::getNFKCCasefoldInstance(errorCode);
1726 if(errorCode.errDataIfFailureAndReset("Normalizer2::getNFKCCasefoldInstance() call failed")) {
1727 return;
1728 }
1729 // Normalization behavior for ill-formed text is not defined.
1730 // ICU currently treats ill-formed sequences as normalization-inert
1731 // and copies them unchanged.
1732 UnicodeString src(u" A");
1733 src.append((char16_t)0xD800).append(u"ÄA\u0308").append((char16_t)0xD900).
1734 append(u"A\u0308\u00ad\u0323").append((char16_t)0xDBFF).
1735 append(u"Ä\u0323,\u00ad").append((char16_t)0xDC00).
1736 append(u"\u1100\u1161가\u11A8가\u3133 ").append((char16_t)0xDFFF);
1737 UnicodeString expected(u" a");
1738 expected.append((char16_t)0xD800).append(u"ää").append((char16_t)0xD900).
1739 append(u"ạ\u0308").append((char16_t)0xDBFF).
1740 append(u"ạ\u0308,").append((char16_t)0xDC00).
1741 append(u"가각갃 ").append((char16_t)0xDFFF);
1742 UnicodeString result = nfkc_cf->normalize(src, errorCode);
1743 assertSuccess("normalize", errorCode.get());
1744 assertEquals("normalize", expected, result);
1745
1746 std::string src8(u8" A");
1747 src8.append("\x80").append(u8"ÄA\u0308").append("\xC0\x80").
1748 append(u8"A\u0308\u00ad\u0323").append("\xED\xA0\x80").
1749 append(u8"Ä\u0323,\u00ad").append("\xF4\x90\x80\x80").
1750 append(u8"\u1100\u1161가\u11A8가\u3133 ").append("\xF0");
1751 std::string expected8(u8" a");
1752 expected8.append("\x80").append(u8"ää").append("\xC0\x80").
1753 append(u8"ạ\u0308").append("\xED\xA0\x80").
1754 append(u8"ạ\u0308,").append("\xF4\x90\x80\x80").
1755 append(u8"가각갃 ").append("\xF0");
1756 std::string result8;
1757 StringByteSink<std::string> sink(&result8);
1758 nfkc_cf->normalizeUTF8(0, src8, sink, nullptr, errorCode);
1759 assertSuccess("normalizeUTF8", errorCode.get());
1760 assertEquals("normalizeUTF8", expected8.c_str(), result8.c_str());
1761 }
1762
1763 void
TestComposeJamoTBase()1764 BasicNormalizerTest::TestComposeJamoTBase() {
1765 // Algorithmic composition of Hangul syllables must not combine with JAMO_T_BASE = U+11A7
1766 // which is not a conjoining Jamo Trailing consonant.
1767 IcuTestErrorCode errorCode(*this, "TestComposeJamoTBase");
1768 const Normalizer2 *nfkc = Normalizer2::getNFKCInstance(errorCode);
1769 if(errorCode.errDataIfFailureAndReset("Normalizer2::getNFKCInstance() call failed")) {
1770 return;
1771 }
1772 UnicodeString s(u"\u1100\u1161\u11A7\u1100\u314F\u11A7가\u11A7");
1773 UnicodeString expected(u"가\u11A7가\u11A7가\u11A7");
1774 UnicodeString result = nfkc->normalize(s, errorCode);
1775 assertSuccess("normalize(LV+11A7)", errorCode.get());
1776 assertEquals("normalize(LV+11A7)", expected, result);
1777 assertFalse("isNormalized(LV+11A7)", nfkc->isNormalized(s, errorCode));
1778 assertTrue("isNormalized(normalized)", nfkc->isNormalized(result, errorCode));
1779
1780 std::string s8(u8"\u1100\u1161\u11A7\u1100\u314F\u11A7가\u11A7");
1781 std::string expected8(u8"가\u11A7가\u11A7가\u11A7");
1782 std::string result8;
1783 StringByteSink<std::string> sink(&result8, expected8.length());
1784 nfkc->normalizeUTF8(0, s8, sink, nullptr, errorCode);
1785 assertSuccess("normalizeUTF8(LV+11A7)", errorCode.get());
1786 assertEquals("normalizeUTF8(LV+11A7)", expected8.c_str(), result8.c_str());
1787 assertFalse("isNormalizedUTF8(LV+11A7)", nfkc->isNormalizedUTF8(s8, errorCode));
1788 assertTrue("isNormalizedUTF8(normalized)", nfkc->isNormalizedUTF8(result8, errorCode));
1789 }
1790
1791 void
TestComposeBoundaryAfter()1792 BasicNormalizerTest::TestComposeBoundaryAfter() {
1793 IcuTestErrorCode errorCode(*this, "TestComposeBoundaryAfter");
1794 const Normalizer2 *nfkc = Normalizer2::getNFKCInstance(errorCode);
1795 if(errorCode.errDataIfFailureAndReset("Normalizer2::getNFKCInstance() call failed")) {
1796 return;
1797 }
1798 // U+02DA and U+FB2C do not have compose-boundaries-after.
1799 UnicodeString s(u"\u02DA\u0339 \uFB2C\u05B6");
1800 UnicodeString expected(u" \u0339\u030A \u05E9\u05B6\u05BC\u05C1");
1801 UnicodeString result = nfkc->normalize(s, errorCode);
1802 assertSuccess("nfkc", errorCode.get());
1803 assertEquals("nfkc", expected, result);
1804 assertFalse("U+02DA boundary-after", nfkc->hasBoundaryAfter(0x2DA));
1805 assertFalse("U+FB2C boundary-after", nfkc->hasBoundaryAfter(0xFB2C));
1806 }
1807
1808 #endif /* #if !UCONFIG_NO_NORMALIZATION */
1809