1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 *******************************************************************************
5 *
6 * Copyright (C) 2003-2014, International Business Machines
7 * Corporation and others. All Rights Reserved.
8 *
9 *******************************************************************************
10 * file name: convtest.cpp
11 * encoding: UTF-8
12 * tab size: 8 (not used)
13 * indentation:4
14 *
15 * created on: 2003jul15
16 * created by: Markus W. Scherer
17 *
18 * Test file for data-driven conversion tests.
19 */
20
21 #include "unicode/utypes.h"
22
23 #if !UCONFIG_NO_LEGACY_CONVERSION
24 /*
25 * Note: Turning off all of convtest.cpp if !UCONFIG_NO_LEGACY_CONVERSION
26 * is slightly unnecessary - it removes tests for Unicode charsets
27 * like UTF-8 that should work.
28 * However, there is no easy way for the test to detect whether a test case
29 * is for a Unicode charset, so it would be difficult to only exclude those.
30 * Also, regular testing of ICU is done with all modules on, therefore
31 * not testing conversion for a custom configuration like this should be ok.
32 */
33
34 #include "unicode/ucnv.h"
35 #include "unicode/unistr.h"
36 #include "unicode/parsepos.h"
37 #include "unicode/uniset.h"
38 #include "unicode/ustring.h"
39 #include "unicode/ures.h"
40 #include "unicode/utf16.h"
41 #include "convtest.h"
42 #include "cmemory.h"
43 #include "unicode/tstdtmod.h"
44 #include <string.h>
45 #include <stdlib.h>
46
47 enum {
48 // characters used in test data for callbacks
49 SUB_CB='?',
50 SKIP_CB='0',
51 STOP_CB='.',
52 ESC_CB='&'
53 };
54
ConversionTest()55 ConversionTest::ConversionTest() {
56 UErrorCode errorCode=U_ZERO_ERROR;
57 utf8Cnv=ucnv_open("UTF-8", &errorCode);
58 ucnv_setToUCallBack(utf8Cnv, UCNV_TO_U_CALLBACK_STOP, NULL, NULL, NULL, &errorCode);
59 if(U_FAILURE(errorCode)) {
60 errln("unable to open UTF-8 converter");
61 }
62 }
63
~ConversionTest()64 ConversionTest::~ConversionTest() {
65 ucnv_close(utf8Cnv);
66 }
67
68 void
runIndexedTest(int32_t index,UBool exec,const char * & name,char *)69 ConversionTest::runIndexedTest(int32_t index, UBool exec, const char *&name, char * /*par*/) {
70 if (exec) logln("TestSuite ConversionTest: ");
71 TESTCASE_AUTO_BEGIN;
72 #if !UCONFIG_NO_FILE_IO
73 TESTCASE_AUTO(TestToUnicode);
74 TESTCASE_AUTO(TestFromUnicode);
75 TESTCASE_AUTO(TestGetUnicodeSet);
76 #endif
77 TESTCASE_AUTO(TestGetUnicodeSet2);
78 TESTCASE_AUTO(TestDefaultIgnorableCallback);
79 TESTCASE_AUTO(TestUTF8ToUTF8Overflow);
80 TESTCASE_AUTO_END;
81 }
82
83 // test data interface ----------------------------------------------------- ***
84
85 void
TestToUnicode()86 ConversionTest::TestToUnicode() {
87 ConversionCase cc;
88 char charset[100], cbopt[4];
89 const char *option;
90 UnicodeString s, unicode;
91 int32_t offsetsLength;
92 UConverterToUCallback callback;
93
94 TestDataModule *dataModule;
95 TestData *testData;
96 const DataMap *testCase;
97 UErrorCode errorCode;
98 int32_t i;
99
100 errorCode=U_ZERO_ERROR;
101 dataModule=TestDataModule::getTestDataModule("conversion", *this, errorCode);
102 if(U_SUCCESS(errorCode)) {
103 testData=dataModule->createTestData("toUnicode", errorCode);
104 if(U_SUCCESS(errorCode)) {
105 for(i=0; testData->nextCase(testCase, errorCode); ++i) {
106 if(U_FAILURE(errorCode)) {
107 errln("error retrieving conversion/toUnicode test case %d - %s",
108 i, u_errorName(errorCode));
109 errorCode=U_ZERO_ERROR;
110 continue;
111 }
112
113 cc.caseNr=i;
114
115 s=testCase->getString("charset", errorCode);
116 s.extract(0, 0x7fffffff, charset, sizeof(charset), "");
117 cc.charset=charset;
118
119 // BEGIN android-added
120 // To save space, Android does not build full ISO-2022-CN tables.
121 // We skip the TestGetKeywordValuesForLocale for counting available collations.
122 if (strlen(charset) >= 8 &&
123 strncmp(charset+4, "2022-CN", 4) == 0) {
124 continue;
125 }
126 // END android-added
127
128 cc.bytes=testCase->getBinary(cc.bytesLength, "bytes", errorCode);
129 unicode=testCase->getString("unicode", errorCode);
130 cc.unicode=unicode.getBuffer();
131 cc.unicodeLength=unicode.length();
132
133 offsetsLength=0;
134 cc.offsets=testCase->getIntVector(offsetsLength, "offsets", errorCode);
135 if(offsetsLength==0) {
136 cc.offsets=NULL;
137 } else if(offsetsLength!=unicode.length()) {
138 errln("toUnicode[%d] unicode[%d] and offsets[%d] must have the same length",
139 i, unicode.length(), offsetsLength);
140 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
141 }
142
143 cc.finalFlush= 0!=testCase->getInt28("flush", errorCode);
144 cc.fallbacks= 0!=testCase->getInt28("fallbacks", errorCode);
145
146 s=testCase->getString("errorCode", errorCode);
147 if(s==UNICODE_STRING("invalid", 7)) {
148 cc.outErrorCode=U_INVALID_CHAR_FOUND;
149 } else if(s==UNICODE_STRING("illegal", 7)) {
150 cc.outErrorCode=U_ILLEGAL_CHAR_FOUND;
151 } else if(s==UNICODE_STRING("truncated", 9)) {
152 cc.outErrorCode=U_TRUNCATED_CHAR_FOUND;
153 } else if(s==UNICODE_STRING("illesc", 6)) {
154 cc.outErrorCode=U_ILLEGAL_ESCAPE_SEQUENCE;
155 } else if(s==UNICODE_STRING("unsuppesc", 9)) {
156 cc.outErrorCode=U_UNSUPPORTED_ESCAPE_SEQUENCE;
157 } else {
158 cc.outErrorCode=U_ZERO_ERROR;
159 }
160
161 s=testCase->getString("callback", errorCode);
162 s.extract(0, 0x7fffffff, cbopt, sizeof(cbopt), "");
163 cc.cbopt=cbopt;
164 switch(cbopt[0]) {
165 case SUB_CB:
166 callback=UCNV_TO_U_CALLBACK_SUBSTITUTE;
167 break;
168 case SKIP_CB:
169 callback=UCNV_TO_U_CALLBACK_SKIP;
170 break;
171 case STOP_CB:
172 callback=UCNV_TO_U_CALLBACK_STOP;
173 break;
174 case ESC_CB:
175 callback=UCNV_TO_U_CALLBACK_ESCAPE;
176 break;
177 default:
178 callback=NULL;
179 break;
180 }
181 option=callback==NULL ? cbopt : cbopt+1;
182 if(*option==0) {
183 option=NULL;
184 }
185
186 cc.invalidChars=testCase->getBinary(cc.invalidLength, "invalidChars", errorCode);
187
188 if(U_FAILURE(errorCode)) {
189 errln("error parsing conversion/toUnicode test case %d - %s",
190 i, u_errorName(errorCode));
191 errorCode=U_ZERO_ERROR;
192 } else {
193 logln("TestToUnicode[%d] %s", i, charset);
194 ToUnicodeCase(cc, callback, option);
195 }
196 }
197 delete testData;
198 }
199 delete dataModule;
200 }
201 else {
202 dataerrln("Could not load test conversion data");
203 }
204 }
205
206 void
TestFromUnicode()207 ConversionTest::TestFromUnicode() {
208 ConversionCase cc;
209 char charset[100], cbopt[4];
210 const char *option;
211 UnicodeString s, unicode, invalidUChars;
212 int32_t offsetsLength, index;
213 UConverterFromUCallback callback;
214
215 TestDataModule *dataModule;
216 TestData *testData;
217 const DataMap *testCase;
218 const UChar *p;
219 UErrorCode errorCode;
220 int32_t i, length;
221
222 errorCode=U_ZERO_ERROR;
223 dataModule=TestDataModule::getTestDataModule("conversion", *this, errorCode);
224 if(U_SUCCESS(errorCode)) {
225 testData=dataModule->createTestData("fromUnicode", errorCode);
226 if(U_SUCCESS(errorCode)) {
227 for(i=0; testData->nextCase(testCase, errorCode); ++i) {
228 if(U_FAILURE(errorCode)) {
229 errln("error retrieving conversion/fromUnicode test case %d - %s",
230 i, u_errorName(errorCode));
231 errorCode=U_ZERO_ERROR;
232 continue;
233 }
234
235 cc.caseNr=i;
236
237 s=testCase->getString("charset", errorCode);
238 s.extract(0, 0x7fffffff, charset, sizeof(charset), "");
239 cc.charset=charset;
240
241 // BEGIN android-added
242 // To save space, Android does not build full ISO-2022-CN tables.
243 // We skip the TestGetKeywordValuesForLocale for counting available collations.
244 if (strlen(charset) >= 8 &&
245 strncmp(charset+4, "2022-CN", 4) == 0) {
246 continue;
247 }
248 // END android-added
249
250 unicode=testCase->getString("unicode", errorCode);
251 cc.unicode=unicode.getBuffer();
252 cc.unicodeLength=unicode.length();
253 cc.bytes=testCase->getBinary(cc.bytesLength, "bytes", errorCode);
254
255 offsetsLength=0;
256 cc.offsets=testCase->getIntVector(offsetsLength, "offsets", errorCode);
257 if(offsetsLength==0) {
258 cc.offsets=NULL;
259 } else if(offsetsLength!=cc.bytesLength) {
260 errln("fromUnicode[%d] bytes[%d] and offsets[%d] must have the same length",
261 i, cc.bytesLength, offsetsLength);
262 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
263 }
264
265 cc.finalFlush= 0!=testCase->getInt28("flush", errorCode);
266 cc.fallbacks= 0!=testCase->getInt28("fallbacks", errorCode);
267
268 s=testCase->getString("errorCode", errorCode);
269 if(s==UNICODE_STRING("invalid", 7)) {
270 cc.outErrorCode=U_INVALID_CHAR_FOUND;
271 } else if(s==UNICODE_STRING("illegal", 7)) {
272 cc.outErrorCode=U_ILLEGAL_CHAR_FOUND;
273 } else if(s==UNICODE_STRING("truncated", 9)) {
274 cc.outErrorCode=U_TRUNCATED_CHAR_FOUND;
275 } else {
276 cc.outErrorCode=U_ZERO_ERROR;
277 }
278
279 s=testCase->getString("callback", errorCode);
280 cc.setSub=0; // default: no subchar
281
282 if((index=s.indexOf((UChar)0))>0) {
283 // read NUL-separated subchar first, if any
284 // copy the subchar from Latin-1 characters
285 // start after the NUL
286 p=s.getTerminatedBuffer();
287 length=index+1;
288 p+=length;
289 length=s.length()-length;
290 if(length<=0 || length>=(int32_t)sizeof(cc.subchar)) {
291 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
292 } else {
293 int32_t j;
294
295 for(j=0; j<length; ++j) {
296 cc.subchar[j]=(char)p[j];
297 }
298 // NUL-terminate the subchar
299 cc.subchar[j]=0;
300 cc.setSub=1;
301 }
302
303 // remove the NUL and subchar from s
304 s.truncate(index);
305 } else if((index=s.indexOf((UChar)0x3d))>0) /* '=' */ {
306 // read a substitution string, separated by an equal sign
307 p=s.getBuffer()+index+1;
308 length=s.length()-(index+1);
309 if(length<0 || length>=UPRV_LENGTHOF(cc.subString)) {
310 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
311 } else {
312 u_memcpy(cc.subString, p, length);
313 // NUL-terminate the subString
314 cc.subString[length]=0;
315 cc.setSub=-1;
316 }
317
318 // remove the equal sign and subString from s
319 s.truncate(index);
320 }
321
322 s.extract(0, 0x7fffffff, cbopt, sizeof(cbopt), "");
323 cc.cbopt=cbopt;
324 switch(cbopt[0]) {
325 case SUB_CB:
326 callback=UCNV_FROM_U_CALLBACK_SUBSTITUTE;
327 break;
328 case SKIP_CB:
329 callback=UCNV_FROM_U_CALLBACK_SKIP;
330 break;
331 case STOP_CB:
332 callback=UCNV_FROM_U_CALLBACK_STOP;
333 break;
334 case ESC_CB:
335 callback=UCNV_FROM_U_CALLBACK_ESCAPE;
336 break;
337 default:
338 callback=NULL;
339 break;
340 }
341 option=callback==NULL ? cbopt : cbopt+1;
342 if(*option==0) {
343 option=NULL;
344 }
345
346 invalidUChars=testCase->getString("invalidUChars", errorCode);
347 cc.invalidUChars=invalidUChars.getBuffer();
348 cc.invalidLength=invalidUChars.length();
349
350 if(U_FAILURE(errorCode)) {
351 errln("error parsing conversion/fromUnicode test case %d - %s",
352 i, u_errorName(errorCode));
353 errorCode=U_ZERO_ERROR;
354 } else {
355 logln("TestFromUnicode[%d] %s", i, charset);
356 FromUnicodeCase(cc, callback, option);
357 }
358 }
359 delete testData;
360 }
361 delete dataModule;
362 }
363 else {
364 dataerrln("Could not load test conversion data");
365 }
366 }
367
368 static const UChar ellipsis[]={ 0x2e, 0x2e, 0x2e };
369
370 void
TestGetUnicodeSet()371 ConversionTest::TestGetUnicodeSet() {
372 char charset[100];
373 UnicodeString s, map, mapnot;
374 int32_t which;
375
376 ParsePosition pos;
377 UnicodeSet cnvSet, mapSet, mapnotSet, diffSet;
378 UnicodeSet *cnvSetPtr = &cnvSet;
379 LocalUConverterPointer cnv;
380
381 TestDataModule *dataModule;
382 TestData *testData;
383 const DataMap *testCase;
384 UErrorCode errorCode;
385 int32_t i;
386
387 errorCode=U_ZERO_ERROR;
388 dataModule=TestDataModule::getTestDataModule("conversion", *this, errorCode);
389 if(U_SUCCESS(errorCode)) {
390 testData=dataModule->createTestData("getUnicodeSet", errorCode);
391 if(U_SUCCESS(errorCode)) {
392 for(i=0; testData->nextCase(testCase, errorCode); ++i) {
393 if(U_FAILURE(errorCode)) {
394 errln("error retrieving conversion/getUnicodeSet test case %d - %s",
395 i, u_errorName(errorCode));
396 errorCode=U_ZERO_ERROR;
397 continue;
398 }
399
400 s=testCase->getString("charset", errorCode);
401 s.extract(0, 0x7fffffff, charset, sizeof(charset), "");
402
403 // BEGIN android-added
404 // To save space, Android does not build full ISO-2022-CN tables.
405 // We skip the TestGetKeywordValuesForLocale for counting available collations.
406 if (strlen(charset) >= 8 &&
407 strncmp(charset+4, "2022-CN", 4) == 0) {
408 continue;
409 }
410 // END android-added
411
412 map=testCase->getString("map", errorCode);
413 mapnot=testCase->getString("mapnot", errorCode);
414
415 which=testCase->getInt28("which", errorCode);
416
417 if(U_FAILURE(errorCode)) {
418 errln("error parsing conversion/getUnicodeSet test case %d - %s",
419 i, u_errorName(errorCode));
420 errorCode=U_ZERO_ERROR;
421 continue;
422 }
423
424 // test this test case
425 mapSet.clear();
426 mapnotSet.clear();
427
428 pos.setIndex(0);
429 mapSet.applyPattern(map, pos, 0, NULL, errorCode);
430 if(U_FAILURE(errorCode) || pos.getIndex()!=map.length()) {
431 errln("error creating the map set for conversion/getUnicodeSet test case %d - %s\n"
432 " error index %d index %d U+%04x",
433 i, u_errorName(errorCode), pos.getErrorIndex(), pos.getIndex(), map.char32At(pos.getIndex()));
434 errorCode=U_ZERO_ERROR;
435 continue;
436 }
437
438 pos.setIndex(0);
439 mapnotSet.applyPattern(mapnot, pos, 0, NULL, errorCode);
440 if(U_FAILURE(errorCode) || pos.getIndex()!=mapnot.length()) {
441 errln("error creating the mapnot set for conversion/getUnicodeSet test case %d - %s\n"
442 " error index %d index %d U+%04x",
443 i, u_errorName(errorCode), pos.getErrorIndex(), pos.getIndex(), mapnot.char32At(pos.getIndex()));
444 errorCode=U_ZERO_ERROR;
445 continue;
446 }
447
448 logln("TestGetUnicodeSet[%d] %s", i, charset);
449
450 cnv.adoptInstead(cnv_open(charset, errorCode));
451 if(U_FAILURE(errorCode)) {
452 errcheckln(errorCode, "error opening \"%s\" for conversion/getUnicodeSet test case %d - %s",
453 charset, i, u_errorName(errorCode));
454 errorCode=U_ZERO_ERROR;
455 continue;
456 }
457
458 ucnv_getUnicodeSet(cnv.getAlias(), cnvSetPtr->toUSet(), (UConverterUnicodeSet)which, &errorCode);
459
460 if(U_FAILURE(errorCode)) {
461 errln("error in ucnv_getUnicodeSet(\"%s\") for conversion/getUnicodeSet test case %d - %s",
462 charset, i, u_errorName(errorCode));
463 errorCode=U_ZERO_ERROR;
464 continue;
465 }
466
467 // are there items that must be in cnvSet but are not?
468 (diffSet=mapSet).removeAll(cnvSet);
469 if(!diffSet.isEmpty()) {
470 diffSet.toPattern(s, TRUE);
471 if(s.length()>100) {
472 s.replace(100, 0x7fffffff, ellipsis, UPRV_LENGTHOF(ellipsis));
473 }
474 errln("error: ucnv_getUnicodeSet(\"%s\") is missing items - conversion/getUnicodeSet test case %d",
475 charset, i);
476 errln(s);
477 }
478
479 // are there items that must not be in cnvSet but are?
480 (diffSet=mapnotSet).retainAll(cnvSet);
481 if(!diffSet.isEmpty()) {
482 diffSet.toPattern(s, TRUE);
483 if(s.length()>100) {
484 s.replace(100, 0x7fffffff, ellipsis, UPRV_LENGTHOF(ellipsis));
485 }
486 errln("error: ucnv_getUnicodeSet(\"%s\") contains unexpected items - conversion/getUnicodeSet test case %d",
487 charset, i);
488 errln(s);
489 }
490 }
491 delete testData;
492 }
493 delete dataModule;
494 }
495 else {
496 dataerrln("Could not load test conversion data");
497 }
498 }
499
500 U_CDECL_BEGIN
501 static void U_CALLCONV
getUnicodeSetCallback(const void * context,UConverterFromUnicodeArgs *,const UChar *,int32_t,UChar32 codePoint,UConverterCallbackReason reason,UErrorCode * pErrorCode)502 getUnicodeSetCallback(const void *context,
503 UConverterFromUnicodeArgs * /*fromUArgs*/,
504 const UChar* /*codeUnits*/,
505 int32_t /*length*/,
506 UChar32 codePoint,
507 UConverterCallbackReason reason,
508 UErrorCode *pErrorCode) {
509 if(reason<=UCNV_IRREGULAR) {
510 ((UnicodeSet *)context)->remove(codePoint); // the converter cannot convert this code point
511 *pErrorCode=U_ZERO_ERROR; // skip
512 } // else ignore the reset, close and clone calls.
513 }
514 U_CDECL_END
515
516 // Compare ucnv_getUnicodeSet() with the set of characters that can be converted.
517 void
TestGetUnicodeSet2()518 ConversionTest::TestGetUnicodeSet2() {
519 // Build a string with all code points.
520 UChar32 cpLimit;
521 int32_t s0Length;
522 if(quick) {
523 cpLimit=s0Length=0x10000; // BMP only
524 } else {
525 cpLimit=0x110000;
526 s0Length=0x10000+0x200000; // BMP + surrogate pairs
527 }
528 UChar *s0=new UChar[s0Length];
529 if(s0==NULL) {
530 return;
531 }
532 UChar *s=s0;
533 UChar32 c;
534 UChar c2;
535 // low BMP
536 for(c=0; c<=0xd7ff; ++c) {
537 *s++=(UChar)c;
538 }
539 // trail surrogates
540 for(c=0xdc00; c<=0xdfff; ++c) {
541 *s++=(UChar)c;
542 }
543 // lead surrogates
544 // (after trails so that there is not even one surrogate pair in between)
545 for(c=0xd800; c<=0xdbff; ++c) {
546 *s++=(UChar)c;
547 }
548 // high BMP
549 for(c=0xe000; c<=0xffff; ++c) {
550 *s++=(UChar)c;
551 }
552 // supplementary code points = surrogate pairs
553 if(cpLimit==0x110000) {
554 for(c=0xd800; c<=0xdbff; ++c) {
555 for(c2=0xdc00; c2<=0xdfff; ++c2) {
556 *s++=(UChar)c;
557 *s++=c2;
558 }
559 }
560 }
561
562 static const char *const cnvNames[]={
563 "UTF-8",
564 "UTF-7",
565 "UTF-16",
566 "US-ASCII",
567 "ISO-8859-1",
568 "windows-1252",
569 "Shift-JIS",
570 "ibm-1390", // EBCDIC_STATEFUL table
571 "ibm-16684", // DBCS-only extension table based on EBCDIC_STATEFUL table
572 "HZ",
573 "ISO-2022-JP",
574 "JIS7",
575 "ISO-2022-CN",
576 "ISO-2022-CN-EXT",
577 "LMBCS"
578 };
579 LocalUConverterPointer cnv;
580 char buffer[1024];
581 int32_t i;
582 for(i=0; i<UPRV_LENGTHOF(cnvNames); ++i) {
583 UErrorCode errorCode=U_ZERO_ERROR;
584 cnv.adoptInstead(cnv_open(cnvNames[i], errorCode));
585 if(U_FAILURE(errorCode)) {
586 errcheckln(errorCode, "failed to open converter %s - %s", cnvNames[i], u_errorName(errorCode));
587 continue;
588 }
589 UnicodeSet expected;
590 ucnv_setFromUCallBack(cnv.getAlias(), getUnicodeSetCallback, &expected, NULL, NULL, &errorCode);
591 if(U_FAILURE(errorCode)) {
592 errln("failed to set the callback on converter %s - %s", cnvNames[i], u_errorName(errorCode));
593 continue;
594 }
595 UConverterUnicodeSet which;
596 for(which=UCNV_ROUNDTRIP_SET; which<UCNV_SET_COUNT; which=(UConverterUnicodeSet)((int)which+1)) {
597 if(which==UCNV_ROUNDTRIP_AND_FALLBACK_SET) {
598 ucnv_setFallback(cnv.getAlias(), TRUE);
599 }
600 expected.add(0, cpLimit-1);
601 s=s0;
602 UBool flush;
603 do {
604 char *t=buffer;
605 flush=(UBool)(s==s0+s0Length);
606 ucnv_fromUnicode(cnv.getAlias(), &t, buffer+sizeof(buffer), (const UChar **)&s, s0+s0Length, NULL, flush, &errorCode);
607 if(U_FAILURE(errorCode)) {
608 if(errorCode==U_BUFFER_OVERFLOW_ERROR) {
609 errorCode=U_ZERO_ERROR;
610 continue;
611 } else {
612 break; // unexpected error, should not occur
613 }
614 }
615 } while(!flush);
616 UnicodeSet set;
617 ucnv_getUnicodeSet(cnv.getAlias(), set.toUSet(), which, &errorCode);
618 if(cpLimit<0x110000) {
619 set.remove(cpLimit, 0x10ffff);
620 }
621 if(which==UCNV_ROUNDTRIP_SET) {
622 // ignore PUA code points because they will be converted even if they
623 // are fallbacks and when other fallbacks are turned off,
624 // but ucnv_getUnicodeSet(UCNV_ROUNDTRIP_SET) delivers true roundtrips
625 expected.remove(0xe000, 0xf8ff);
626 expected.remove(0xf0000, 0xffffd);
627 expected.remove(0x100000, 0x10fffd);
628 set.remove(0xe000, 0xf8ff);
629 set.remove(0xf0000, 0xffffd);
630 set.remove(0x100000, 0x10fffd);
631 }
632 if(set!=expected) {
633 // First try to see if we have different sets because ucnv_getUnicodeSet()
634 // added strings: The above conversion method does not tell us what strings might be convertible.
635 // Remove strings from the set and compare again.
636 // Unfortunately, there are no good, direct set methods for finding out whether there are strings
637 // in the set, nor for enumerating or removing just them.
638 // Intersect all code points with the set. The intersection will not contain strings.
639 UnicodeSet temp(0, 0x10ffff);
640 temp.retainAll(set);
641 set=temp;
642 }
643 if(set!=expected) {
644 UnicodeSet diffSet;
645 UnicodeString out;
646
647 // are there items that must be in the set but are not?
648 (diffSet=expected).removeAll(set);
649 if(!diffSet.isEmpty()) {
650 diffSet.toPattern(out, TRUE);
651 if(out.length()>100) {
652 out.replace(100, 0x7fffffff, ellipsis, UPRV_LENGTHOF(ellipsis));
653 }
654 errln("error: ucnv_getUnicodeSet(\"%s\") is missing items - which set: %d",
655 cnvNames[i], which);
656 errln(out);
657 }
658
659 // are there items that must not be in the set but are?
660 (diffSet=set).removeAll(expected);
661 if(!diffSet.isEmpty()) {
662 diffSet.toPattern(out, TRUE);
663 if(out.length()>100) {
664 out.replace(100, 0x7fffffff, ellipsis, UPRV_LENGTHOF(ellipsis));
665 }
666 errln("error: ucnv_getUnicodeSet(\"%s\") contains unexpected items - which set: %d",
667 cnvNames[i], which);
668 errln(out);
669 }
670 }
671 }
672 }
673
674 delete [] s0;
675 }
676
677 // Test all codepoints which has the default ignorable Unicode property are ignored if they have no mapping
678 // If there are any failures, the hard coded list (IS_DEFAULT_IGNORABLE_CODE_POINT) in ucnv_err.c should be updated
679 void
TestDefaultIgnorableCallback()680 ConversionTest::TestDefaultIgnorableCallback() {
681 UErrorCode status = U_ZERO_ERROR;
682 const char *cnv_name = "euc-jp-2007";
683 const char *pattern_ignorable = "[:Default_Ignorable_Code_Point:]";
684 const char *pattern_not_ignorable = "[:^Default_Ignorable_Code_Point:]";
685
686 UnicodeSet *set_ignorable = new UnicodeSet(pattern_ignorable, status);
687 if (U_FAILURE(status)) {
688 dataerrln("Unable to create Unicodeset: %s - %s\n", pattern_ignorable, u_errorName(status));
689 return;
690 }
691
692 UnicodeSet *set_not_ignorable = new UnicodeSet(pattern_not_ignorable, status);
693 if (U_FAILURE(status)) {
694 dataerrln("Unable to create Unicodeset: %s - %s\n", pattern_not_ignorable, u_errorName(status));
695 return;
696 }
697
698 UConverter *cnv = cnv_open(cnv_name, status);
699 if (U_FAILURE(status)) {
700 dataerrln("Unable to open converter: %s - %s\n", cnv_name, u_errorName(status));
701 return;
702 }
703
704 // set callback for the converter
705 ucnv_setFromUCallBack(cnv, UCNV_FROM_U_CALLBACK_SUBSTITUTE, NULL, NULL, NULL, &status);
706
707 UChar32 input[1];
708 char output[10];
709 int32_t outputLength;
710
711 // test default ignorables are ignored
712 int size = set_ignorable->size();
713 for (int i = 0; i < size; i++) {
714 status = U_ZERO_ERROR;
715 outputLength= 0;
716
717 input[0] = set_ignorable->charAt(i);
718
719 outputLength = ucnv_fromUChars(cnv, output, 10, UnicodeString::fromUTF32(input, 1).getTerminatedBuffer(), -1, &status);
720 if (U_FAILURE(status) || outputLength != 0) {
721 errln("Ignorable code point: U+%04X not skipped as expected - %s", input[0], u_errorName(status));
722 }
723 }
724
725 // test non-ignorables are not ignored
726 size = set_not_ignorable->size();
727 for (int i = 0; i < size; i++) {
728 status = U_ZERO_ERROR;
729 outputLength= 0;
730
731 input[0] = set_not_ignorable->charAt(i);
732
733 if (input[0] == 0) {
734 continue;
735 }
736
737 outputLength = ucnv_fromUChars(cnv, output, 10, UnicodeString::fromUTF32(input, 1).getTerminatedBuffer(), -1, &status);
738 if (U_FAILURE(status) || outputLength <= 0) {
739 errln("Non-ignorable code point: U+%04X skipped unexpectedly - %s", input[0], u_errorName(status));
740 }
741 }
742
743 ucnv_close(cnv);
744 delete set_not_ignorable;
745 delete set_ignorable;
746 }
747
748 void
TestUTF8ToUTF8Overflow()749 ConversionTest::TestUTF8ToUTF8Overflow() {
750 IcuTestErrorCode errorCode(*this, "TestUTF8ToUTF8Overflow");
751 LocalUConverterPointer cnv1(ucnv_open("UTF-8", errorCode));
752 LocalUConverterPointer cnv2(ucnv_open("UTF-8", errorCode));
753 static const char *text = "aä"; // ä: 2 bytes
754 const char *source = text;
755 const char *sourceLimit = text + strlen(text);
756 char result[20];
757 char *target = result;
758 const char *targetLimit = result + sizeof(result);
759 UChar buffer16[20];
760 UChar *pivotSource = buffer16;
761 UChar *pivotTarget = buffer16;
762 const UChar *pivotLimit = buffer16 + UPRV_LENGTHOF(buffer16);
763
764 // Convert with insufficient target capacity.
765 result[2] = 5;
766 ucnv_convertEx(cnv2.getAlias(), cnv1.getAlias(),
767 &target, result + 2, &source, sourceLimit,
768 buffer16, &pivotSource, &pivotTarget, pivotLimit,
769 FALSE, FALSE, errorCode);
770 assertEquals("overflow", U_BUFFER_OVERFLOW_ERROR, errorCode.reset());
771 int32_t length = (int32_t)(target - result);
772 assertEquals("number of bytes written", 2, length);
773 assertEquals("next byte not clobbered", 5, result[2]);
774
775 // Convert the rest and flush.
776 ucnv_convertEx(cnv2.getAlias(), cnv1.getAlias(),
777 &target, targetLimit, &source, sourceLimit,
778 buffer16, &pivotSource, &pivotTarget, pivotLimit,
779 FALSE, TRUE, errorCode);
780
781 assertSuccess("UTF-8->UTF-8", errorCode);
782 length = (int32_t)(target - result);
783 assertEquals("3 bytes", 3, length);
784 if (length == 3) {
785 assertTrue("result same as input", memcmp(text, result, length) == 0);
786 }
787
788 ucnv_reset(cnv1.getAlias());
789 ucnv_reset(cnv2.getAlias());
790 memset(result, 0, sizeof(result));
791 static const char *text2 = "a��"; // U+1F6B2 bicycle: 4 bytes
792 source = text2;
793 sourceLimit = text2 + strlen(text2);
794 target = result;
795 pivotSource = pivotTarget = buffer16;
796
797 // Convert with insufficient target capacity.
798 result[3] = 5;
799 ucnv_convertEx(cnv2.getAlias(), cnv1.getAlias(),
800 &target, result + 3, &source, sourceLimit,
801 buffer16, &pivotSource, &pivotTarget, pivotLimit,
802 FALSE, FALSE, errorCode);
803 assertEquals("text2 overflow", U_BUFFER_OVERFLOW_ERROR, errorCode.reset());
804 length = (int32_t)(target - result);
805 assertEquals("text2 number of bytes written", 3, length);
806 assertEquals("text2 next byte not clobbered", 5, result[3]);
807
808 // Convert the rest and flush.
809 ucnv_convertEx(cnv2.getAlias(), cnv1.getAlias(),
810 &target, targetLimit, &source, sourceLimit,
811 buffer16, &pivotSource, &pivotTarget, pivotLimit,
812 FALSE, TRUE, errorCode);
813
814 assertSuccess("text2 UTF-8->UTF-8", errorCode);
815 length = (int32_t)(target - result);
816 assertEquals("text2 5 bytes", 5, length);
817 if (length == 5) {
818 assertTrue("text2 result same as input", memcmp(text2, result, length) == 0);
819 }
820 }
821
822 // open testdata or ICU data converter ------------------------------------- ***
823
824 UConverter *
cnv_open(const char * name,UErrorCode & errorCode)825 ConversionTest::cnv_open(const char *name, UErrorCode &errorCode) {
826 if(name!=NULL && *name=='+') {
827 // Converter names that start with '+' are ignored in ICU4J tests.
828 ++name;
829 }
830 if(name!=NULL && *name=='*') {
831 /* loadTestData(): set the data directory */
832 return ucnv_openPackage(loadTestData(errorCode), name+1, &errorCode);
833 } else {
834 return ucnv_open(name, &errorCode);
835 }
836 }
837
838 // output helpers ---------------------------------------------------------- ***
839
840 static inline char
hexDigit(uint8_t digit)841 hexDigit(uint8_t digit) {
842 return digit<=9 ? (char)('0'+digit) : (char)('a'-10+digit);
843 }
844
845 static char *
printBytes(const uint8_t * bytes,int32_t length,char * out)846 printBytes(const uint8_t *bytes, int32_t length, char *out) {
847 uint8_t b;
848
849 if(length>0) {
850 b=*bytes++;
851 --length;
852 *out++=hexDigit((uint8_t)(b>>4));
853 *out++=hexDigit((uint8_t)(b&0xf));
854 }
855
856 while(length>0) {
857 b=*bytes++;
858 --length;
859 *out++=' ';
860 *out++=hexDigit((uint8_t)(b>>4));
861 *out++=hexDigit((uint8_t)(b&0xf));
862 }
863 *out++=0;
864 return out;
865 }
866
867 static char *
printUnicode(const UChar * unicode,int32_t length,char * out)868 printUnicode(const UChar *unicode, int32_t length, char *out) {
869 UChar32 c;
870 int32_t i;
871
872 for(i=0; i<length;) {
873 if(i>0) {
874 *out++=' ';
875 }
876 U16_NEXT(unicode, i, length, c);
877 // write 4..6 digits
878 if(c>=0x100000) {
879 *out++='1';
880 }
881 if(c>=0x10000) {
882 *out++=hexDigit((uint8_t)((c>>16)&0xf));
883 }
884 *out++=hexDigit((uint8_t)((c>>12)&0xf));
885 *out++=hexDigit((uint8_t)((c>>8)&0xf));
886 *out++=hexDigit((uint8_t)((c>>4)&0xf));
887 *out++=hexDigit((uint8_t)(c&0xf));
888 }
889 *out++=0;
890 return out;
891 }
892
893 static char *
printOffsets(const int32_t * offsets,int32_t length,char * out)894 printOffsets(const int32_t *offsets, int32_t length, char *out) {
895 int32_t i, o, d;
896
897 if(offsets==NULL) {
898 length=0;
899 }
900
901 for(i=0; i<length; ++i) {
902 if(i>0) {
903 *out++=' ';
904 }
905 o=offsets[i];
906
907 // print all offsets with 2 characters each (-x, -9..99, xx)
908 if(o<-9) {
909 *out++='-';
910 *out++='x';
911 } else if(o<0) {
912 *out++='-';
913 *out++=(char)('0'-o);
914 } else if(o<=99) {
915 *out++=(d=o/10)==0 ? ' ' : (char)('0'+d);
916 *out++=(char)('0'+o%10);
917 } else /* o>99 */ {
918 *out++='x';
919 *out++='x';
920 }
921 }
922 *out++=0;
923 return out;
924 }
925
926 // toUnicode test worker functions ----------------------------------------- ***
927
928 static int32_t
stepToUnicode(ConversionCase & cc,UConverter * cnv,UChar * result,int32_t resultCapacity,int32_t * resultOffsets,int32_t step,UErrorCode * pErrorCode)929 stepToUnicode(ConversionCase &cc, UConverter *cnv,
930 UChar *result, int32_t resultCapacity,
931 int32_t *resultOffsets, /* also resultCapacity */
932 int32_t step,
933 UErrorCode *pErrorCode) {
934 const char *source, *sourceLimit, *bytesLimit;
935 UChar *target, *targetLimit, *resultLimit;
936 UBool flush;
937
938 source=(const char *)cc.bytes;
939 target=result;
940 bytesLimit=source+cc.bytesLength;
941 resultLimit=result+resultCapacity;
942
943 if(step>=0) {
944 // call ucnv_toUnicode() with in/out buffers no larger than (step) at a time
945 // move only one buffer (in vs. out) at a time to be extra mean
946 // step==0 performs bulk conversion and generates offsets
947
948 // initialize the partial limits for the loop
949 if(step==0) {
950 // use the entire buffers
951 sourceLimit=bytesLimit;
952 targetLimit=resultLimit;
953 flush=cc.finalFlush;
954 } else {
955 // start with empty partial buffers
956 sourceLimit=source;
957 targetLimit=target;
958 flush=FALSE;
959
960 // output offsets only for bulk conversion
961 resultOffsets=NULL;
962 }
963
964 for(;;) {
965 // resetting the opposite conversion direction must not affect this one
966 ucnv_resetFromUnicode(cnv);
967
968 // convert
969 ucnv_toUnicode(cnv,
970 &target, targetLimit,
971 &source, sourceLimit,
972 resultOffsets,
973 flush, pErrorCode);
974
975 // check pointers and errors
976 if(source>sourceLimit || target>targetLimit) {
977 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
978 break;
979 } else if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
980 if(target!=targetLimit) {
981 // buffer overflow must only be set when the target is filled
982 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
983 break;
984 } else if(targetLimit==resultLimit) {
985 // not just a partial overflow
986 break;
987 }
988
989 // the partial target is filled, set a new limit, reset the error and continue
990 targetLimit=(resultLimit-target)>=step ? target+step : resultLimit;
991 *pErrorCode=U_ZERO_ERROR;
992 } else if(U_FAILURE(*pErrorCode)) {
993 // some other error occurred, done
994 break;
995 } else {
996 if(source!=sourceLimit) {
997 // when no error occurs, then the input must be consumed
998 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
999 break;
1000 }
1001
1002 if(sourceLimit==bytesLimit) {
1003 // we are done
1004 break;
1005 }
1006
1007 // the partial conversion succeeded, set a new limit and continue
1008 sourceLimit=(bytesLimit-source)>=step ? source+step : bytesLimit;
1009 flush=(UBool)(cc.finalFlush && sourceLimit==bytesLimit);
1010 }
1011 }
1012 } else /* step<0 */ {
1013 /*
1014 * step==-1: call only ucnv_getNextUChar()
1015 * otherwise alternate between ucnv_toUnicode() and ucnv_getNextUChar()
1016 * if step==-2 or -3, then give ucnv_toUnicode() the whole remaining input,
1017 * else give it at most (-step-2)/2 bytes
1018 */
1019 UChar32 c;
1020
1021 // end the loop by getting an index out of bounds error
1022 for(;;) {
1023 // resetting the opposite conversion direction must not affect this one
1024 ucnv_resetFromUnicode(cnv);
1025
1026 // convert
1027 if((step&1)!=0 /* odd: -1, -3, -5, ... */) {
1028 sourceLimit=source; // use sourceLimit not as a real limit
1029 // but to remember the pre-getNextUChar source pointer
1030 c=ucnv_getNextUChar(cnv, &source, bytesLimit, pErrorCode);
1031
1032 // check pointers and errors
1033 if(*pErrorCode==U_INDEX_OUTOFBOUNDS_ERROR) {
1034 if(source!=bytesLimit) {
1035 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1036 } else {
1037 *pErrorCode=U_ZERO_ERROR;
1038 }
1039 break;
1040 } else if(U_FAILURE(*pErrorCode)) {
1041 break;
1042 }
1043 // source may not move if c is from previous overflow
1044
1045 if(target==resultLimit) {
1046 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
1047 break;
1048 }
1049 if(c<=0xffff) {
1050 *target++=(UChar)c;
1051 } else {
1052 *target++=U16_LEAD(c);
1053 if(target==resultLimit) {
1054 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
1055 break;
1056 }
1057 *target++=U16_TRAIL(c);
1058 }
1059
1060 // alternate between -n-1 and -n but leave -1 alone
1061 if(step<-1) {
1062 ++step;
1063 }
1064 } else /* step is even */ {
1065 // allow only one UChar output
1066 targetLimit=target<resultLimit ? target+1 : resultLimit;
1067
1068 // as with ucnv_getNextUChar(), we always flush (if we go to bytesLimit)
1069 // and never output offsets
1070 if(step==-2) {
1071 sourceLimit=bytesLimit;
1072 } else {
1073 sourceLimit=source+(-step-2)/2;
1074 if(sourceLimit>bytesLimit) {
1075 sourceLimit=bytesLimit;
1076 }
1077 }
1078
1079 ucnv_toUnicode(cnv,
1080 &target, targetLimit,
1081 &source, sourceLimit,
1082 NULL, (UBool)(sourceLimit==bytesLimit), pErrorCode);
1083
1084 // check pointers and errors
1085 if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
1086 if(target!=targetLimit) {
1087 // buffer overflow must only be set when the target is filled
1088 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1089 break;
1090 } else if(targetLimit==resultLimit) {
1091 // not just a partial overflow
1092 break;
1093 }
1094
1095 // the partial target is filled, set a new limit and continue
1096 *pErrorCode=U_ZERO_ERROR;
1097 } else if(U_FAILURE(*pErrorCode)) {
1098 // some other error occurred, done
1099 break;
1100 } else {
1101 if(source!=sourceLimit) {
1102 // when no error occurs, then the input must be consumed
1103 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1104 break;
1105 }
1106
1107 // we are done (flush==TRUE) but we continue, to get the index out of bounds error above
1108 }
1109
1110 --step;
1111 }
1112 }
1113 }
1114
1115 return (int32_t)(target-result);
1116 }
1117
1118 UBool
ToUnicodeCase(ConversionCase & cc,UConverterToUCallback callback,const char * option)1119 ConversionTest::ToUnicodeCase(ConversionCase &cc, UConverterToUCallback callback, const char *option) {
1120 // open the converter
1121 IcuTestErrorCode errorCode(*this, "ToUnicodeCase");
1122 LocalUConverterPointer cnv(cnv_open(cc.charset, errorCode));
1123 // with no data, the above crashes with "pointer being freed was not allocated" for charset "x11-compound-text", see #13078
1124 if(errorCode.isFailure()) {
1125 errcheckln(errorCode, "toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_open() failed - %s",
1126 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, errorCode.errorName());
1127 errorCode.reset();
1128 return FALSE;
1129 }
1130
1131 // set the callback
1132 if(callback!=NULL) {
1133 ucnv_setToUCallBack(cnv.getAlias(), callback, option, NULL, NULL, errorCode);
1134 if(U_FAILURE(errorCode)) {
1135 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setToUCallBack() failed - %s",
1136 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1137 return FALSE;
1138 }
1139 }
1140
1141 int32_t resultOffsets[256];
1142 UChar result[256];
1143 int32_t resultLength;
1144 UBool ok;
1145
1146 static const struct {
1147 int32_t step;
1148 const char *name;
1149 } steps[]={
1150 { 0, "bulk" }, // must be first for offsets to be checked
1151 { 1, "step=1" },
1152 { 3, "step=3" },
1153 { 7, "step=7" },
1154 { -1, "getNext" },
1155 { -2, "toU(bulk)+getNext" },
1156 { -3, "getNext+toU(bulk)" },
1157 { -4, "toU(1)+getNext" },
1158 { -5, "getNext+toU(1)" },
1159 { -12, "toU(5)+getNext" },
1160 { -13, "getNext+toU(5)" },
1161 };
1162 int32_t i, step;
1163
1164 ok=TRUE;
1165 for(i=0; i<UPRV_LENGTHOF(steps) && ok; ++i) {
1166 step=steps[i].step;
1167 if(step<0 && !cc.finalFlush) {
1168 // skip ucnv_getNextUChar() if !finalFlush because
1169 // ucnv_getNextUChar() always implies flush
1170 continue;
1171 }
1172 if(step!=0) {
1173 // bulk test is first, then offsets are not checked any more
1174 cc.offsets=NULL;
1175 }
1176 else {
1177 memset(resultOffsets, -1, UPRV_LENGTHOF(resultOffsets));
1178 }
1179 memset(result, -1, UPRV_LENGTHOF(result));
1180 errorCode.reset();
1181 resultLength=stepToUnicode(cc, cnv.getAlias(),
1182 result, UPRV_LENGTHOF(result),
1183 step==0 ? resultOffsets : NULL,
1184 step, errorCode);
1185 ok=checkToUnicode(
1186 cc, cnv.getAlias(), steps[i].name,
1187 result, resultLength,
1188 cc.offsets!=NULL ? resultOffsets : NULL,
1189 errorCode);
1190 if(errorCode.isFailure() || !cc.finalFlush) {
1191 // reset if an error occurred or we did not flush
1192 // otherwise do nothing to make sure that flushing resets
1193 ucnv_resetToUnicode(cnv.getAlias());
1194 }
1195 if (cc.offsets != NULL && resultOffsets[resultLength] != -1) {
1196 errln("toUnicode[%d](%s) Conversion wrote too much to offsets at index %d",
1197 cc.caseNr, cc.charset, resultLength);
1198 }
1199 if (result[resultLength] != (UChar)-1) {
1200 errln("toUnicode[%d](%s) Conversion wrote too much to result at index %d",
1201 cc.caseNr, cc.charset, resultLength);
1202 }
1203 }
1204
1205 // not a real loop, just a convenience for breaking out of the block
1206 while(ok && cc.finalFlush) {
1207 // test ucnv_toUChars()
1208 memset(result, 0, sizeof(result));
1209
1210 errorCode.reset();
1211 resultLength=ucnv_toUChars(cnv.getAlias(),
1212 result, UPRV_LENGTHOF(result),
1213 (const char *)cc.bytes, cc.bytesLength,
1214 errorCode);
1215 ok=checkToUnicode(
1216 cc, cnv.getAlias(), "toUChars",
1217 result, resultLength,
1218 NULL,
1219 errorCode);
1220 if(!ok) {
1221 break;
1222 }
1223
1224 // test preflighting
1225 // keep the correct result for simple checking
1226 errorCode.reset();
1227 resultLength=ucnv_toUChars(cnv.getAlias(),
1228 NULL, 0,
1229 (const char *)cc.bytes, cc.bytesLength,
1230 errorCode);
1231 if(errorCode.get()==U_STRING_NOT_TERMINATED_WARNING || errorCode.get()==U_BUFFER_OVERFLOW_ERROR) {
1232 errorCode.reset();
1233 }
1234 ok=checkToUnicode(
1235 cc, cnv.getAlias(), "preflight toUChars",
1236 result, resultLength,
1237 NULL,
1238 errorCode);
1239 break;
1240 }
1241
1242 errorCode.reset(); // all errors have already been reported
1243 return ok;
1244 }
1245
1246 UBool
checkToUnicode(ConversionCase & cc,UConverter * cnv,const char * name,const UChar * result,int32_t resultLength,const int32_t * resultOffsets,UErrorCode resultErrorCode)1247 ConversionTest::checkToUnicode(ConversionCase &cc, UConverter *cnv, const char *name,
1248 const UChar *result, int32_t resultLength,
1249 const int32_t *resultOffsets,
1250 UErrorCode resultErrorCode) {
1251 char resultInvalidChars[8];
1252 int8_t resultInvalidLength;
1253 UErrorCode errorCode;
1254
1255 const char *msg;
1256
1257 // reset the message; NULL will mean "ok"
1258 msg=NULL;
1259
1260 errorCode=U_ZERO_ERROR;
1261 resultInvalidLength=sizeof(resultInvalidChars);
1262 ucnv_getInvalidChars(cnv, resultInvalidChars, &resultInvalidLength, &errorCode);
1263 if(U_FAILURE(errorCode)) {
1264 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) ucnv_getInvalidChars() failed - %s",
1265 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, u_errorName(errorCode));
1266 return FALSE;
1267 }
1268
1269 // check everything that might have gone wrong
1270 if(cc.unicodeLength!=resultLength) {
1271 msg="wrong result length";
1272 } else if(0!=u_memcmp(cc.unicode, result, cc.unicodeLength)) {
1273 msg="wrong result string";
1274 } else if(cc.offsets!=NULL && 0!=memcmp(cc.offsets, resultOffsets, cc.unicodeLength*sizeof(*cc.offsets))) {
1275 msg="wrong offsets";
1276 } else if(cc.outErrorCode!=resultErrorCode) {
1277 msg="wrong error code";
1278 } else if(cc.invalidLength!=resultInvalidLength) {
1279 msg="wrong length of last invalid input";
1280 } else if(0!=memcmp(cc.invalidChars, resultInvalidChars, cc.invalidLength)) {
1281 msg="wrong last invalid input";
1282 }
1283
1284 if(msg==NULL) {
1285 return TRUE;
1286 } else {
1287 char buffer[2000]; // one buffer for all strings
1288 char *s, *bytesString, *unicodeString, *resultString,
1289 *offsetsString, *resultOffsetsString,
1290 *invalidCharsString, *resultInvalidCharsString;
1291
1292 bytesString=s=buffer;
1293 s=printBytes(cc.bytes, cc.bytesLength, bytesString);
1294 s=printUnicode(cc.unicode, cc.unicodeLength, unicodeString=s);
1295 s=printUnicode(result, resultLength, resultString=s);
1296 s=printOffsets(cc.offsets, cc.unicodeLength, offsetsString=s);
1297 s=printOffsets(resultOffsets, resultLength, resultOffsetsString=s);
1298 s=printBytes(cc.invalidChars, cc.invalidLength, invalidCharsString=s);
1299 s=printBytes((uint8_t *)resultInvalidChars, resultInvalidLength, resultInvalidCharsString=s);
1300
1301 if((s-buffer)>(int32_t)sizeof(buffer)) {
1302 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) fatal error: checkToUnicode() test output buffer overflow writing %d chars\n",
1303 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, (int)(s-buffer));
1304 exit(1);
1305 }
1306
1307 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) failed: %s\n"
1308 " bytes <%s>[%d]\n"
1309 " expected <%s>[%d]\n"
1310 " result <%s>[%d]\n"
1311 " offsets <%s>\n"
1312 " result offsets <%s>\n"
1313 " error code expected %s got %s\n"
1314 " invalidChars expected <%s> got <%s>\n",
1315 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, msg,
1316 bytesString, cc.bytesLength,
1317 unicodeString, cc.unicodeLength,
1318 resultString, resultLength,
1319 offsetsString,
1320 resultOffsetsString,
1321 u_errorName(cc.outErrorCode), u_errorName(resultErrorCode),
1322 invalidCharsString, resultInvalidCharsString);
1323
1324 return FALSE;
1325 }
1326 }
1327
1328 // fromUnicode test worker functions --------------------------------------- ***
1329
1330 static int32_t
stepFromUTF8(ConversionCase & cc,UConverter * utf8Cnv,UConverter * cnv,char * result,int32_t resultCapacity,int32_t step,UErrorCode * pErrorCode)1331 stepFromUTF8(ConversionCase &cc,
1332 UConverter *utf8Cnv, UConverter *cnv,
1333 char *result, int32_t resultCapacity,
1334 int32_t step,
1335 UErrorCode *pErrorCode) {
1336 const char *source, *sourceLimit, *utf8Limit;
1337 UChar pivotBuffer[32];
1338 UChar *pivotSource, *pivotTarget, *pivotLimit;
1339 char *target, *targetLimit, *resultLimit;
1340 UBool flush;
1341
1342 source=cc.utf8;
1343 pivotSource=pivotTarget=pivotBuffer;
1344 target=result;
1345 utf8Limit=source+cc.utf8Length;
1346 resultLimit=result+resultCapacity;
1347
1348 // call ucnv_convertEx() with in/out buffers no larger than (step) at a time
1349 // move only one buffer (in vs. out) at a time to be extra mean
1350 // step==0 performs bulk conversion
1351
1352 // initialize the partial limits for the loop
1353 if(step==0) {
1354 // use the entire buffers
1355 sourceLimit=utf8Limit;
1356 targetLimit=resultLimit;
1357 flush=cc.finalFlush;
1358
1359 pivotLimit=pivotBuffer+UPRV_LENGTHOF(pivotBuffer);
1360 } else {
1361 // start with empty partial buffers
1362 sourceLimit=source;
1363 targetLimit=target;
1364 flush=FALSE;
1365
1366 // empty pivot is not allowed, make it of length step
1367 pivotLimit=pivotBuffer+step;
1368 }
1369
1370 for(;;) {
1371 // resetting the opposite conversion direction must not affect this one
1372 ucnv_resetFromUnicode(utf8Cnv);
1373 ucnv_resetToUnicode(cnv);
1374
1375 // convert
1376 ucnv_convertEx(cnv, utf8Cnv,
1377 &target, targetLimit,
1378 &source, sourceLimit,
1379 pivotBuffer, &pivotSource, &pivotTarget, pivotLimit,
1380 FALSE, flush, pErrorCode);
1381
1382 // check pointers and errors
1383 if(source>sourceLimit || target>targetLimit) {
1384 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1385 break;
1386 } else if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
1387 if(target!=targetLimit) {
1388 // buffer overflow must only be set when the target is filled
1389 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1390 break;
1391 } else if(targetLimit==resultLimit) {
1392 // not just a partial overflow
1393 break;
1394 }
1395
1396 // the partial target is filled, set a new limit, reset the error and continue
1397 targetLimit=(resultLimit-target)>=step ? target+step : resultLimit;
1398 *pErrorCode=U_ZERO_ERROR;
1399 } else if(U_FAILURE(*pErrorCode)) {
1400 if(pivotSource==pivotBuffer) {
1401 // toUnicode error, should not occur
1402 // toUnicode errors are tested in cintltst TestConvertExFromUTF8()
1403 break;
1404 } else {
1405 // fromUnicode error
1406 // some other error occurred, done
1407 break;
1408 }
1409 } else {
1410 if(source!=sourceLimit) {
1411 // when no error occurs, then the input must be consumed
1412 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1413 break;
1414 }
1415
1416 if(sourceLimit==utf8Limit) {
1417 // we are done
1418 if(*pErrorCode==U_STRING_NOT_TERMINATED_WARNING) {
1419 // ucnv_convertEx() warns about not terminating the output
1420 // but ucnv_fromUnicode() does not and so
1421 // checkFromUnicode() does not expect it
1422 *pErrorCode=U_ZERO_ERROR;
1423 }
1424 break;
1425 }
1426
1427 // the partial conversion succeeded, set a new limit and continue
1428 sourceLimit=(utf8Limit-source)>=step ? source+step : utf8Limit;
1429 flush=(UBool)(cc.finalFlush && sourceLimit==utf8Limit);
1430 }
1431 }
1432
1433 return (int32_t)(target-result);
1434 }
1435
1436 static int32_t
stepFromUnicode(ConversionCase & cc,UConverter * cnv,char * result,int32_t resultCapacity,int32_t * resultOffsets,int32_t step,UErrorCode * pErrorCode)1437 stepFromUnicode(ConversionCase &cc, UConverter *cnv,
1438 char *result, int32_t resultCapacity,
1439 int32_t *resultOffsets, /* also resultCapacity */
1440 int32_t step,
1441 UErrorCode *pErrorCode) {
1442 const UChar *source, *sourceLimit, *unicodeLimit;
1443 char *target, *targetLimit, *resultLimit;
1444 UBool flush;
1445
1446 source=cc.unicode;
1447 target=result;
1448 unicodeLimit=source+cc.unicodeLength;
1449 resultLimit=result+resultCapacity;
1450
1451 // call ucnv_fromUnicode() with in/out buffers no larger than (step) at a time
1452 // move only one buffer (in vs. out) at a time to be extra mean
1453 // step==0 performs bulk conversion and generates offsets
1454
1455 // initialize the partial limits for the loop
1456 if(step==0) {
1457 // use the entire buffers
1458 sourceLimit=unicodeLimit;
1459 targetLimit=resultLimit;
1460 flush=cc.finalFlush;
1461 } else {
1462 // start with empty partial buffers
1463 sourceLimit=source;
1464 targetLimit=target;
1465 flush=FALSE;
1466
1467 // output offsets only for bulk conversion
1468 resultOffsets=NULL;
1469 }
1470
1471 for(;;) {
1472 // resetting the opposite conversion direction must not affect this one
1473 ucnv_resetToUnicode(cnv);
1474
1475 // convert
1476 ucnv_fromUnicode(cnv,
1477 &target, targetLimit,
1478 &source, sourceLimit,
1479 resultOffsets,
1480 flush, pErrorCode);
1481
1482 // check pointers and errors
1483 if(source>sourceLimit || target>targetLimit) {
1484 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1485 break;
1486 } else if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
1487 if(target!=targetLimit) {
1488 // buffer overflow must only be set when the target is filled
1489 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1490 break;
1491 } else if(targetLimit==resultLimit) {
1492 // not just a partial overflow
1493 break;
1494 }
1495
1496 // the partial target is filled, set a new limit, reset the error and continue
1497 targetLimit=(resultLimit-target)>=step ? target+step : resultLimit;
1498 *pErrorCode=U_ZERO_ERROR;
1499 } else if(U_FAILURE(*pErrorCode)) {
1500 // some other error occurred, done
1501 break;
1502 } else {
1503 if(source!=sourceLimit) {
1504 // when no error occurs, then the input must be consumed
1505 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1506 break;
1507 }
1508
1509 if(sourceLimit==unicodeLimit) {
1510 // we are done
1511 break;
1512 }
1513
1514 // the partial conversion succeeded, set a new limit and continue
1515 sourceLimit=(unicodeLimit-source)>=step ? source+step : unicodeLimit;
1516 flush=(UBool)(cc.finalFlush && sourceLimit==unicodeLimit);
1517 }
1518 }
1519
1520 return (int32_t)(target-result);
1521 }
1522
1523 UBool
FromUnicodeCase(ConversionCase & cc,UConverterFromUCallback callback,const char * option)1524 ConversionTest::FromUnicodeCase(ConversionCase &cc, UConverterFromUCallback callback, const char *option) {
1525 UConverter *cnv;
1526 UErrorCode errorCode;
1527
1528 // open the converter
1529 errorCode=U_ZERO_ERROR;
1530 cnv=cnv_open(cc.charset, errorCode);
1531 if(U_FAILURE(errorCode)) {
1532 errcheckln(errorCode, "fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_open() failed - %s",
1533 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1534 return FALSE;
1535 }
1536 ucnv_resetToUnicode(utf8Cnv);
1537
1538 // set the callback
1539 if(callback!=NULL) {
1540 ucnv_setFromUCallBack(cnv, callback, option, NULL, NULL, &errorCode);
1541 if(U_FAILURE(errorCode)) {
1542 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setFromUCallBack() failed - %s",
1543 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1544 ucnv_close(cnv);
1545 return FALSE;
1546 }
1547 }
1548
1549 // set the fallbacks flag
1550 // TODO change with Jitterbug 2401, then add a similar call for toUnicode too
1551 ucnv_setFallback(cnv, cc.fallbacks);
1552
1553 // set the subchar
1554 int32_t length;
1555
1556 if(cc.setSub>0) {
1557 length=(int32_t)strlen(cc.subchar);
1558 ucnv_setSubstChars(cnv, cc.subchar, (int8_t)length, &errorCode);
1559 if(U_FAILURE(errorCode)) {
1560 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setSubstChars() failed - %s",
1561 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1562 ucnv_close(cnv);
1563 return FALSE;
1564 }
1565 } else if(cc.setSub<0) {
1566 ucnv_setSubstString(cnv, cc.subString, -1, &errorCode);
1567 if(U_FAILURE(errorCode)) {
1568 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setSubstString() failed - %s",
1569 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1570 ucnv_close(cnv);
1571 return FALSE;
1572 }
1573 }
1574
1575 // convert unicode to utf8
1576 char utf8[256];
1577 cc.utf8=utf8;
1578 u_strToUTF8(utf8, UPRV_LENGTHOF(utf8), &cc.utf8Length,
1579 cc.unicode, cc.unicodeLength,
1580 &errorCode);
1581 if(U_FAILURE(errorCode)) {
1582 // skip UTF-8 testing of a string with an unpaired surrogate,
1583 // or of one that's too long
1584 // toUnicode errors are tested in cintltst TestConvertExFromUTF8()
1585 cc.utf8Length=-1;
1586 }
1587
1588 int32_t resultOffsets[256];
1589 char result[256];
1590 int32_t resultLength;
1591 UBool ok;
1592
1593 static const struct {
1594 int32_t step;
1595 const char *name, *utf8Name;
1596 } steps[]={
1597 { 0, "bulk", "utf8" }, // must be first for offsets to be checked
1598 { 1, "step=1", "utf8 step=1" },
1599 { 3, "step=3", "utf8 step=3" },
1600 { 7, "step=7", "utf8 step=7" }
1601 };
1602 int32_t i, step;
1603
1604 ok=TRUE;
1605 for(i=0; i<UPRV_LENGTHOF(steps) && ok; ++i) {
1606 step=steps[i].step;
1607 memset(resultOffsets, -1, UPRV_LENGTHOF(resultOffsets));
1608 memset(result, -1, UPRV_LENGTHOF(result));
1609 errorCode=U_ZERO_ERROR;
1610 resultLength=stepFromUnicode(cc, cnv,
1611 result, UPRV_LENGTHOF(result),
1612 step==0 ? resultOffsets : NULL,
1613 step, &errorCode);
1614 ok=checkFromUnicode(
1615 cc, cnv, steps[i].name,
1616 (uint8_t *)result, resultLength,
1617 cc.offsets!=NULL ? resultOffsets : NULL,
1618 errorCode);
1619 if(U_FAILURE(errorCode) || !cc.finalFlush) {
1620 // reset if an error occurred or we did not flush
1621 // otherwise do nothing to make sure that flushing resets
1622 ucnv_resetFromUnicode(cnv);
1623 }
1624 if (resultOffsets[resultLength] != -1) {
1625 errln("fromUnicode[%d](%s) Conversion wrote too much to offsets at index %d",
1626 cc.caseNr, cc.charset, resultLength);
1627 }
1628 if (result[resultLength] != (char)-1) {
1629 errln("fromUnicode[%d](%s) Conversion wrote too much to result at index %d",
1630 cc.caseNr, cc.charset, resultLength);
1631 }
1632
1633 // bulk test is first, then offsets are not checked any more
1634 cc.offsets=NULL;
1635
1636 // test direct conversion from UTF-8
1637 if(cc.utf8Length>=0) {
1638 errorCode=U_ZERO_ERROR;
1639 resultLength=stepFromUTF8(cc, utf8Cnv, cnv,
1640 result, UPRV_LENGTHOF(result),
1641 step, &errorCode);
1642 ok=checkFromUnicode(
1643 cc, cnv, steps[i].utf8Name,
1644 (uint8_t *)result, resultLength,
1645 NULL,
1646 errorCode);
1647 if(U_FAILURE(errorCode) || !cc.finalFlush) {
1648 // reset if an error occurred or we did not flush
1649 // otherwise do nothing to make sure that flushing resets
1650 ucnv_resetToUnicode(utf8Cnv);
1651 ucnv_resetFromUnicode(cnv);
1652 }
1653 }
1654 }
1655
1656 // not a real loop, just a convenience for breaking out of the block
1657 while(ok && cc.finalFlush) {
1658 // test ucnv_fromUChars()
1659 memset(result, 0, sizeof(result));
1660
1661 errorCode=U_ZERO_ERROR;
1662 resultLength=ucnv_fromUChars(cnv,
1663 result, UPRV_LENGTHOF(result),
1664 cc.unicode, cc.unicodeLength,
1665 &errorCode);
1666 ok=checkFromUnicode(
1667 cc, cnv, "fromUChars",
1668 (uint8_t *)result, resultLength,
1669 NULL,
1670 errorCode);
1671 if(!ok) {
1672 break;
1673 }
1674
1675 // test preflighting
1676 // keep the correct result for simple checking
1677 errorCode=U_ZERO_ERROR;
1678 resultLength=ucnv_fromUChars(cnv,
1679 NULL, 0,
1680 cc.unicode, cc.unicodeLength,
1681 &errorCode);
1682 if(errorCode==U_STRING_NOT_TERMINATED_WARNING || errorCode==U_BUFFER_OVERFLOW_ERROR) {
1683 errorCode=U_ZERO_ERROR;
1684 }
1685 ok=checkFromUnicode(
1686 cc, cnv, "preflight fromUChars",
1687 (uint8_t *)result, resultLength,
1688 NULL,
1689 errorCode);
1690 break;
1691 }
1692
1693 ucnv_close(cnv);
1694 return ok;
1695 }
1696
1697 UBool
checkFromUnicode(ConversionCase & cc,UConverter * cnv,const char * name,const uint8_t * result,int32_t resultLength,const int32_t * resultOffsets,UErrorCode resultErrorCode)1698 ConversionTest::checkFromUnicode(ConversionCase &cc, UConverter *cnv, const char *name,
1699 const uint8_t *result, int32_t resultLength,
1700 const int32_t *resultOffsets,
1701 UErrorCode resultErrorCode) {
1702 UChar resultInvalidUChars[8];
1703 int8_t resultInvalidLength;
1704 UErrorCode errorCode;
1705
1706 const char *msg;
1707
1708 // reset the message; NULL will mean "ok"
1709 msg=NULL;
1710
1711 errorCode=U_ZERO_ERROR;
1712 resultInvalidLength=UPRV_LENGTHOF(resultInvalidUChars);
1713 ucnv_getInvalidUChars(cnv, resultInvalidUChars, &resultInvalidLength, &errorCode);
1714 if(U_FAILURE(errorCode)) {
1715 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) ucnv_getInvalidUChars() failed - %s",
1716 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, u_errorName(errorCode));
1717 return FALSE;
1718 }
1719
1720 // check everything that might have gone wrong
1721 if(cc.bytesLength!=resultLength) {
1722 msg="wrong result length";
1723 } else if(0!=memcmp(cc.bytes, result, cc.bytesLength)) {
1724 msg="wrong result string";
1725 } else if(cc.offsets!=NULL && 0!=memcmp(cc.offsets, resultOffsets, cc.bytesLength*sizeof(*cc.offsets))) {
1726 msg="wrong offsets";
1727 } else if(cc.outErrorCode!=resultErrorCode) {
1728 msg="wrong error code";
1729 } else if(cc.invalidLength!=resultInvalidLength) {
1730 msg="wrong length of last invalid input";
1731 } else if(0!=u_memcmp(cc.invalidUChars, resultInvalidUChars, cc.invalidLength)) {
1732 msg="wrong last invalid input";
1733 }
1734
1735 if(msg==NULL) {
1736 return TRUE;
1737 } else {
1738 char buffer[2000]; // one buffer for all strings
1739 char *s, *unicodeString, *bytesString, *resultString,
1740 *offsetsString, *resultOffsetsString,
1741 *invalidCharsString, *resultInvalidUCharsString;
1742
1743 unicodeString=s=buffer;
1744 s=printUnicode(cc.unicode, cc.unicodeLength, unicodeString);
1745 s=printBytes(cc.bytes, cc.bytesLength, bytesString=s);
1746 s=printBytes(result, resultLength, resultString=s);
1747 s=printOffsets(cc.offsets, cc.bytesLength, offsetsString=s);
1748 s=printOffsets(resultOffsets, resultLength, resultOffsetsString=s);
1749 s=printUnicode(cc.invalidUChars, cc.invalidLength, invalidCharsString=s);
1750 s=printUnicode(resultInvalidUChars, resultInvalidLength, resultInvalidUCharsString=s);
1751
1752 if((s-buffer)>(int32_t)sizeof(buffer)) {
1753 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) fatal error: checkFromUnicode() test output buffer overflow writing %d chars\n",
1754 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, (int)(s-buffer));
1755 exit(1);
1756 }
1757
1758 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) failed: %s\n"
1759 " unicode <%s>[%d]\n"
1760 " expected <%s>[%d]\n"
1761 " result <%s>[%d]\n"
1762 " offsets <%s>\n"
1763 " result offsets <%s>\n"
1764 " error code expected %s got %s\n"
1765 " invalidChars expected <%s> got <%s>\n",
1766 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, msg,
1767 unicodeString, cc.unicodeLength,
1768 bytesString, cc.bytesLength,
1769 resultString, resultLength,
1770 offsetsString,
1771 resultOffsetsString,
1772 u_errorName(cc.outErrorCode), u_errorName(resultErrorCode),
1773 invalidCharsString, resultInvalidUCharsString);
1774
1775 return FALSE;
1776 }
1777 }
1778
1779 #endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */
1780