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