]> git.saurik.com Git - apple/icu.git/blob - icuSources/test/intltest/convtest.cpp
ICU-6.2.6.tar.gz
[apple/icu.git] / icuSources / test / intltest / convtest.cpp
1 /*
2 *******************************************************************************
3 *
4 * Copyright (C) 2003-2004, International Business Machines
5 * Corporation and others. All Rights Reserved.
6 *
7 *******************************************************************************
8 * file name: convtest.cpp
9 * encoding: US-ASCII
10 * tab size: 8 (not used)
11 * indentation:4
12 *
13 * created on: 2003jul15
14 * created by: Markus W. Scherer
15 *
16 * Test file for data-driven conversion tests.
17 */
18
19 #include "unicode/utypes.h"
20
21 #if !UCONFIG_NO_LEGACY_CONVERSION
22 /*
23 * Note: Turning off all of convtest.cpp if !UCONFIG_NO_LEGACY_CONVERSION
24 * is slightly unnecessary - it removes tests for Unicode charsets
25 * like UTF-8 that should work.
26 * However, there is no easy way for the test to detect whether a test case
27 * is for a Unicode charset, so it would be difficult to only exclude those.
28 * Also, regular testing of ICU is done with all modules on, therefore
29 * not testing conversion for a custom configuration like this should be ok.
30 */
31
32 #include "unicode/ucnv.h"
33 #include "unicode/unistr.h"
34 #include "unicode/parsepos.h"
35 #include "unicode/uniset.h"
36 #include "unicode/ustring.h"
37 #include "unicode/ures.h"
38 #include "convtest.h"
39 #include "unicode/tstdtmod.h"
40 #include <string.h>
41 #include <stdlib.h>
42
43 #define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
44
45 enum {
46 // characters used in test data for callbacks
47 SUB_CB='?',
48 SKIP_CB='0',
49 STOP_CB='.',
50 ESC_CB='&'
51 };
52
53 ConversionTest::~ConversionTest() {}
54
55 void
56 ConversionTest::runIndexedTest(int32_t index, UBool exec, const char *&name, char * /*par*/) {
57 if (exec) logln("TestSuite ConversionTest: ");
58 switch (index) {
59 case 0: name="TestToUnicode"; if (exec) TestToUnicode(); break;
60 case 1: name="TestFromUnicode"; if (exec) TestFromUnicode(); break;
61 case 2: name="TestGetUnicodeSet"; if (exec) TestGetUnicodeSet(); break;
62 default: name=""; break; //needed to end loop
63 }
64 }
65
66 // test data interface ----------------------------------------------------- ***
67
68 void
69 ConversionTest::TestToUnicode() {
70 ConversionCase cc;
71 char charset[100], cbopt[4];
72 const char *option;
73 UnicodeString s, unicode;
74 int32_t offsetsLength;
75 UConverterToUCallback callback;
76
77 TestDataModule *dataModule;
78 TestData *testData;
79 const DataMap *testCase;
80 UErrorCode errorCode;
81 int32_t i;
82
83 errorCode=U_ZERO_ERROR;
84 dataModule=TestDataModule::getTestDataModule("conversion", *this, errorCode);
85 if(U_SUCCESS(errorCode)) {
86 testData=dataModule->createTestData("toUnicode", errorCode);
87 if(U_SUCCESS(errorCode)) {
88 for(i=0; testData->nextCase(testCase, errorCode); ++i) {
89 if(U_FAILURE(errorCode)) {
90 errln("error retrieving conversion/toUnicode test case %d - %s",
91 i, u_errorName(errorCode));
92 errorCode=U_ZERO_ERROR;
93 continue;
94 }
95
96 cc.caseNr=i;
97
98 s=testCase->getString("charset", errorCode);
99 s.extract(0, 0x7fffffff, charset, sizeof(charset), "");
100 cc.charset=charset;
101
102 cc.bytes=testCase->getBinary(cc.bytesLength, "bytes", errorCode);
103 unicode=testCase->getString("unicode", errorCode);
104 cc.unicode=unicode.getBuffer();
105 cc.unicodeLength=unicode.length();
106
107 offsetsLength=0;
108 cc.offsets=testCase->getIntVector(offsetsLength, "offsets", errorCode);
109 if(offsetsLength==0) {
110 cc.offsets=NULL;
111 } else if(offsetsLength!=unicode.length()) {
112 errln("toUnicode[%d] unicode[%d] and offsets[%d] must have the same length",
113 i, unicode.length(), offsetsLength);
114 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
115 }
116
117 cc.finalFlush= 0!=testCase->getInt28("flush", errorCode);
118 cc.fallbacks= 0!=testCase->getInt28("fallbacks", errorCode);
119
120 s=testCase->getString("errorCode", errorCode);
121 if(s==UNICODE_STRING("invalid", 7)) {
122 cc.outErrorCode=U_INVALID_CHAR_FOUND;
123 } else if(s==UNICODE_STRING("illegal", 7)) {
124 cc.outErrorCode=U_ILLEGAL_CHAR_FOUND;
125 } else if(s==UNICODE_STRING("truncated", 9)) {
126 cc.outErrorCode=U_TRUNCATED_CHAR_FOUND;
127 } else if(s==UNICODE_STRING("illesc", 6)) {
128 cc.outErrorCode=U_ILLEGAL_ESCAPE_SEQUENCE;
129 } else if(s==UNICODE_STRING("unsuppesc", 9)) {
130 cc.outErrorCode=U_UNSUPPORTED_ESCAPE_SEQUENCE;
131 } else {
132 cc.outErrorCode=U_ZERO_ERROR;
133 }
134
135 s=testCase->getString("callback", errorCode);
136 s.extract(0, 0x7fffffff, cbopt, sizeof(cbopt), "");
137 cc.cbopt=cbopt;
138 switch(cbopt[0]) {
139 case SUB_CB:
140 callback=UCNV_TO_U_CALLBACK_SUBSTITUTE;
141 break;
142 case SKIP_CB:
143 callback=UCNV_TO_U_CALLBACK_SKIP;
144 break;
145 case STOP_CB:
146 callback=UCNV_TO_U_CALLBACK_STOP;
147 break;
148 case ESC_CB:
149 callback=UCNV_TO_U_CALLBACK_ESCAPE;
150 break;
151 default:
152 callback=NULL;
153 break;
154 }
155 option=callback==NULL ? cbopt : cbopt+1;
156 if(*option==0) {
157 option=NULL;
158 }
159
160 cc.invalidChars=testCase->getBinary(cc.invalidLength, "invalidChars", errorCode);
161
162 if(U_FAILURE(errorCode)) {
163 errln("error parsing conversion/toUnicode test case %d - %s",
164 i, u_errorName(errorCode));
165 errorCode=U_ZERO_ERROR;
166 } else {
167 logln("TestToUnicode[%d] %s", i, charset);
168 ToUnicodeCase(cc, callback, option);
169 }
170 }
171 delete testData;
172 }
173 delete dataModule;
174 }
175 else {
176 errln("Failed: could not load test conversion data");
177 }
178 }
179
180 void
181 ConversionTest::TestFromUnicode() {
182 ConversionCase cc;
183 char charset[100], cbopt[4];
184 const char *option;
185 UnicodeString s, unicode, invalidUChars;
186 int32_t offsetsLength;
187 UConverterFromUCallback callback;
188
189 TestDataModule *dataModule;
190 TestData *testData;
191 const DataMap *testCase;
192 const UChar *p;
193 UErrorCode errorCode;
194 int32_t i, length;
195
196 errorCode=U_ZERO_ERROR;
197 dataModule=TestDataModule::getTestDataModule("conversion", *this, errorCode);
198 if(U_SUCCESS(errorCode)) {
199 testData=dataModule->createTestData("fromUnicode", errorCode);
200 if(U_SUCCESS(errorCode)) {
201 for(i=0; testData->nextCase(testCase, errorCode); ++i) {
202 if(U_FAILURE(errorCode)) {
203 errln("error retrieving conversion/fromUnicode test case %d - %s",
204 i, u_errorName(errorCode));
205 errorCode=U_ZERO_ERROR;
206 continue;
207 }
208
209 cc.caseNr=i;
210
211 s=testCase->getString("charset", errorCode);
212 s.extract(0, 0x7fffffff, charset, sizeof(charset), "");
213 cc.charset=charset;
214
215 unicode=testCase->getString("unicode", errorCode);
216 cc.unicode=unicode.getBuffer();
217 cc.unicodeLength=unicode.length();
218 cc.bytes=testCase->getBinary(cc.bytesLength, "bytes", errorCode);
219
220 offsetsLength=0;
221 cc.offsets=testCase->getIntVector(offsetsLength, "offsets", errorCode);
222 if(offsetsLength==0) {
223 cc.offsets=NULL;
224 } else if(offsetsLength!=cc.bytesLength) {
225 errln("fromUnicode[%d] bytes[%d] and offsets[%d] must have the same length",
226 i, cc.bytesLength, offsetsLength);
227 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
228 }
229
230 cc.finalFlush= 0!=testCase->getInt28("flush", errorCode);
231 cc.fallbacks= 0!=testCase->getInt28("fallbacks", errorCode);
232
233 s=testCase->getString("errorCode", errorCode);
234 if(s==UNICODE_STRING("invalid", 7)) {
235 cc.outErrorCode=U_INVALID_CHAR_FOUND;
236 } else if(s==UNICODE_STRING("illegal", 7)) {
237 cc.outErrorCode=U_ILLEGAL_CHAR_FOUND;
238 } else if(s==UNICODE_STRING("truncated", 9)) {
239 cc.outErrorCode=U_TRUNCATED_CHAR_FOUND;
240 } else {
241 cc.outErrorCode=U_ZERO_ERROR;
242 }
243
244 s=testCase->getString("callback", errorCode);
245
246 // read NUL-separated subchar first, if any
247 length=u_strlen(p=s.getTerminatedBuffer());
248 if(++length<s.length()) {
249 // copy the subchar from Latin-1 characters
250 // start after the NUL
251 p+=length;
252 length=s.length()-length;
253 if(length>=(int32_t)sizeof(cc.subchar)) {
254 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
255 } else {
256 int32_t j;
257
258 for(j=0; j<length; ++j) {
259 cc.subchar[j]=(char)p[j];
260 }
261 // NUL-terminate the subchar
262 cc.subchar[j]=0;
263 }
264
265 // remove the NUL and subchar from s
266 s.truncate(u_strlen(s.getBuffer()));
267 } else {
268 // no subchar
269 cc.subchar[0]=0;
270 }
271
272 s.extract(0, 0x7fffffff, cbopt, sizeof(cbopt), "");
273 cc.cbopt=cbopt;
274 switch(cbopt[0]) {
275 case SUB_CB:
276 callback=UCNV_FROM_U_CALLBACK_SUBSTITUTE;
277 break;
278 case SKIP_CB:
279 callback=UCNV_FROM_U_CALLBACK_SKIP;
280 break;
281 case STOP_CB:
282 callback=UCNV_FROM_U_CALLBACK_STOP;
283 break;
284 case ESC_CB:
285 callback=UCNV_FROM_U_CALLBACK_ESCAPE;
286 break;
287 default:
288 callback=NULL;
289 break;
290 }
291 option=callback==NULL ? cbopt : cbopt+1;
292 if(*option==0) {
293 option=NULL;
294 }
295
296 invalidUChars=testCase->getString("invalidUChars", errorCode);
297 cc.invalidUChars=invalidUChars.getBuffer();
298 cc.invalidLength=invalidUChars.length();
299
300 if(U_FAILURE(errorCode)) {
301 errln("error parsing conversion/fromUnicode test case %d - %s",
302 i, u_errorName(errorCode));
303 errorCode=U_ZERO_ERROR;
304 } else {
305 logln("TestFromUnicode[%d] %s", i, charset);
306 FromUnicodeCase(cc, callback, option);
307 }
308 }
309 delete testData;
310 }
311 delete dataModule;
312 }
313 else {
314 errln("Failed: could not load test conversion data");
315 }
316 }
317
318 static const UChar ellipsis[]={ 0x2e, 0x2e, 0x2e };
319
320 void
321 ConversionTest::TestGetUnicodeSet() {
322 char charset[100];
323 UnicodeString s, map, mapnot;
324 int32_t which;
325
326 ParsePosition pos;
327 UnicodeSet cnvSet, mapSet, mapnotSet, diffSet;
328 UConverter *cnv;
329
330 TestDataModule *dataModule;
331 TestData *testData;
332 const DataMap *testCase;
333 UErrorCode errorCode;
334 int32_t i;
335
336 errorCode=U_ZERO_ERROR;
337 dataModule=TestDataModule::getTestDataModule("conversion", *this, errorCode);
338 if(U_SUCCESS(errorCode)) {
339 testData=dataModule->createTestData("getUnicodeSet", errorCode);
340 if(U_SUCCESS(errorCode)) {
341 for(i=0; testData->nextCase(testCase, errorCode); ++i) {
342 if(U_FAILURE(errorCode)) {
343 errln("error retrieving conversion/getUnicodeSet test case %d - %s",
344 i, u_errorName(errorCode));
345 errorCode=U_ZERO_ERROR;
346 continue;
347 }
348
349 s=testCase->getString("charset", errorCode);
350 s.extract(0, 0x7fffffff, charset, sizeof(charset), "");
351
352 map=testCase->getString("map", errorCode);
353 mapnot=testCase->getString("mapnot", errorCode);
354
355 which=testCase->getInt28("which", errorCode);
356
357 if(U_FAILURE(errorCode)) {
358 errln("error parsing conversion/getUnicodeSet test case %d - %s",
359 i, u_errorName(errorCode));
360 errorCode=U_ZERO_ERROR;
361 continue;
362 }
363
364 // test this test case
365 mapSet.clear();
366 mapnotSet.clear();
367
368 pos.setIndex(0);
369 mapSet.applyPattern(map, pos, 0, NULL, errorCode);
370 if(U_FAILURE(errorCode) || pos.getIndex()!=map.length()) {
371 errln("error creating the map set for conversion/getUnicodeSet test case %d - %s\n"
372 " error index %d index %d U+%04x",
373 i, u_errorName(errorCode), pos.getErrorIndex(), pos.getIndex(), map.char32At(pos.getIndex()));
374 errorCode=U_ZERO_ERROR;
375 continue;
376 }
377
378 pos.setIndex(0);
379 mapnotSet.applyPattern(mapnot, pos, 0, NULL, errorCode);
380 if(U_FAILURE(errorCode) || pos.getIndex()!=mapnot.length()) {
381 errln("error creating the mapnot set for conversion/getUnicodeSet test case %d - %s\n"
382 " error index %d index %d U+%04x",
383 i, u_errorName(errorCode), pos.getErrorIndex(), pos.getIndex(), mapnot.char32At(pos.getIndex()));
384 errorCode=U_ZERO_ERROR;
385 continue;
386 }
387
388 logln("TestGetUnicodeSet[%d] %s", i, charset);
389
390 cnv=cnv_open(charset, errorCode);
391 if(U_FAILURE(errorCode)) {
392 errln("error opening \"%s\" for conversion/getUnicodeSet test case %d - %s",
393 charset, i, u_errorName(errorCode));
394 errorCode=U_ZERO_ERROR;
395 continue;
396 }
397
398 ucnv_getUnicodeSet(cnv, (USet *)&cnvSet, (UConverterUnicodeSet)which, &errorCode);
399 ucnv_close(cnv);
400
401 if(U_FAILURE(errorCode)) {
402 errln("error in ucnv_getUnicodeSet(\"%s\") for conversion/getUnicodeSet test case %d - %s",
403 charset, i, u_errorName(errorCode));
404 errorCode=U_ZERO_ERROR;
405 continue;
406 }
407
408 // are there items that must be in cnvSet but are not?
409 (diffSet=mapSet).removeAll(cnvSet);
410 if(!diffSet.isEmpty()) {
411 diffSet.toPattern(s, TRUE);
412 if(s.length()>100) {
413 s.replace(100, 0x7fffffff, ellipsis, LENGTHOF(ellipsis));
414 }
415 errln("error: ucnv_getUnicodeSet(\"%s\") is missing items - conversion/getUnicodeSet test case %d",
416 charset, i);
417 errln(s);
418 }
419
420 // are there items that must not be in cnvSet but are?
421 (diffSet=mapnotSet).retainAll(cnvSet);
422 if(!diffSet.isEmpty()) {
423 diffSet.toPattern(s, TRUE);
424 if(s.length()>100) {
425 s.replace(100, 0x7fffffff, ellipsis, LENGTHOF(ellipsis));
426 }
427 errln("error: ucnv_getUnicodeSet(\"%s\") contains unexpected items - conversion/getUnicodeSet test case %d",
428 charset, i);
429 errln(s);
430 }
431 }
432 delete testData;
433 }
434 delete dataModule;
435 }
436 else {
437 errln("Failed: could not load test conversion data");
438 }
439 }
440
441 // open testdata or ICU data converter ------------------------------------- ***
442
443 UConverter *
444 ConversionTest::cnv_open(const char *name, UErrorCode &errorCode) {
445 if(name!=NULL && *name=='*') {
446 /* loadTestData(): set the data directory */
447 return ucnv_openPackage(loadTestData(errorCode), name+1, &errorCode);
448 } else {
449 return ucnv_open(name, &errorCode);
450 }
451 }
452
453 // output helpers ---------------------------------------------------------- ***
454
455 static inline char
456 hexDigit(uint8_t digit) {
457 return digit<=9 ? (char)('0'+digit) : (char)('a'-10+digit);
458 }
459
460 static char *
461 printBytes(const uint8_t *bytes, int32_t length, char *out) {
462 uint8_t b;
463
464 if(length>0) {
465 b=*bytes++;
466 --length;
467 *out++=hexDigit((uint8_t)(b>>4));
468 *out++=hexDigit((uint8_t)(b&0xf));
469 }
470
471 while(length>0) {
472 b=*bytes++;
473 --length;
474 *out++=' ';
475 *out++=hexDigit((uint8_t)(b>>4));
476 *out++=hexDigit((uint8_t)(b&0xf));
477 }
478 *out++=0;
479 return out;
480 }
481
482 static char *
483 printUnicode(const UChar *unicode, int32_t length, char *out) {
484 UChar32 c;
485 int32_t i;
486
487 for(i=0; i<length;) {
488 if(i>0) {
489 *out++=' ';
490 }
491 U16_NEXT(unicode, i, length, c);
492 // write 4..6 digits
493 if(c>=0x100000) {
494 *out++='1';
495 }
496 if(c>=0x10000) {
497 *out++=hexDigit((uint8_t)((c>>16)&0xf));
498 }
499 *out++=hexDigit((uint8_t)((c>>12)&0xf));
500 *out++=hexDigit((uint8_t)((c>>8)&0xf));
501 *out++=hexDigit((uint8_t)((c>>4)&0xf));
502 *out++=hexDigit((uint8_t)(c&0xf));
503 }
504 *out++=0;
505 return out;
506 }
507
508 static char *
509 printOffsets(const int32_t *offsets, int32_t length, char *out) {
510 int32_t i, o, d;
511
512 if(offsets==NULL) {
513 length=0;
514 }
515
516 for(i=0; i<length; ++i) {
517 if(i>0) {
518 *out++=' ';
519 }
520 o=offsets[i];
521
522 // print all offsets with 2 characters each (-x, -9..99, xx)
523 if(o<-9) {
524 *out++='-';
525 *out++='x';
526 } else if(o<0) {
527 *out++='-';
528 *out++=(char)('0'-o);
529 } else if(o<=99) {
530 *out++=(d=o/10)==0 ? ' ' : (char)('0'+d);
531 *out++=(char)('0'+o%10);
532 } else /* o>99 */ {
533 *out++='x';
534 *out++='x';
535 }
536 }
537 *out++=0;
538 return out;
539 }
540
541 // toUnicode test worker functions ----------------------------------------- ***
542
543 static int32_t
544 stepToUnicode(ConversionCase &cc, UConverter *cnv,
545 UChar *result, int32_t resultCapacity,
546 int32_t *resultOffsets, /* also resultCapacity */
547 int32_t step,
548 UErrorCode *pErrorCode) {
549 const char *source, *sourceLimit, *bytesLimit;
550 UChar *target, *targetLimit, *resultLimit;
551 UBool flush;
552
553 source=(const char *)cc.bytes;
554 target=result;
555 bytesLimit=source+cc.bytesLength;
556 resultLimit=result+resultCapacity;
557
558 if(step>=0) {
559 // call ucnv_toUnicode() with in/out buffers no larger than (step) at a time
560 // move only one buffer (in vs. out) at a time to be extra mean
561 // step==0 performs bulk conversion and generates offsets
562
563 // initialize the partial limits for the loop
564 if(step==0) {
565 // use the entire buffers
566 sourceLimit=bytesLimit;
567 targetLimit=resultLimit;
568 flush=cc.finalFlush;
569 } else {
570 // start with empty partial buffers
571 sourceLimit=source;
572 targetLimit=target;
573 flush=FALSE;
574
575 // output offsets only for bulk conversion
576 resultOffsets=NULL;
577 }
578
579 for(;;) {
580 // resetting the opposite conversion direction must not affect this one
581 ucnv_resetFromUnicode(cnv);
582
583 // convert
584 ucnv_toUnicode(cnv,
585 &target, targetLimit,
586 &source, sourceLimit,
587 resultOffsets,
588 flush, pErrorCode);
589
590 // check pointers and errors
591 if(source>sourceLimit || target>targetLimit) {
592 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
593 break;
594 } else if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
595 if(target!=targetLimit) {
596 // buffer overflow must only be set when the target is filled
597 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
598 break;
599 } else if(targetLimit==resultLimit) {
600 // not just a partial overflow
601 break;
602 }
603
604 // the partial target is filled, set a new limit, reset the error and continue
605 targetLimit=(resultLimit-target)>=step ? target+step : resultLimit;
606 *pErrorCode=U_ZERO_ERROR;
607 } else if(U_FAILURE(*pErrorCode)) {
608 // some other error occurred, done
609 break;
610 } else {
611 if(source!=sourceLimit) {
612 // when no error occurs, then the input must be consumed
613 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
614 break;
615 }
616
617 if(sourceLimit==bytesLimit) {
618 // we are done
619 break;
620 }
621
622 // the partial conversion succeeded, set a new limit and continue
623 sourceLimit=(bytesLimit-source)>=step ? source+step : bytesLimit;
624 flush=(UBool)(cc.finalFlush && sourceLimit==bytesLimit);
625 }
626 }
627 } else /* step<0 */ {
628 /*
629 * step==-1: call only ucnv_getNextUChar()
630 * otherwise alternate between ucnv_toUnicode() and ucnv_getNextUChar()
631 * if step==-2 or -3, then give ucnv_toUnicode() the whole remaining input,
632 * else give it at most (-step-2)/2 bytes
633 */
634 UChar32 c;
635
636 // end the loop by getting an index out of bounds error
637 for(;;) {
638 // resetting the opposite conversion direction must not affect this one
639 ucnv_resetFromUnicode(cnv);
640
641 // convert
642 if((step&1)!=0 /* odd: -1, -3, -5, ... */) {
643 sourceLimit=source; // use sourceLimit not as a real limit
644 // but to remember the pre-getNextUChar source pointer
645 c=ucnv_getNextUChar(cnv, &source, bytesLimit, pErrorCode);
646
647 // check pointers and errors
648 if(*pErrorCode==U_INDEX_OUTOFBOUNDS_ERROR) {
649 if(source!=bytesLimit) {
650 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
651 } else {
652 *pErrorCode=U_ZERO_ERROR;
653 }
654 break;
655 } else if(U_FAILURE(*pErrorCode)) {
656 break;
657 }
658 // source may not move if c is from previous overflow
659
660 if(target==resultLimit) {
661 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
662 break;
663 }
664 if(c<=0xffff) {
665 *target++=(UChar)c;
666 } else {
667 *target++=U16_LEAD(c);
668 if(target==resultLimit) {
669 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
670 break;
671 }
672 *target++=U16_TRAIL(c);
673 }
674
675 // alternate between -n-1 and -n but leave -1 alone
676 if(step<-1) {
677 ++step;
678 }
679 } else /* step is even */ {
680 // allow only one UChar output
681 targetLimit=target<resultLimit ? target+1 : resultLimit;
682
683 // as with ucnv_getNextUChar(), we always flush (if we go to bytesLimit)
684 // and never output offsets
685 if(step==-2) {
686 sourceLimit=bytesLimit;
687 } else {
688 sourceLimit=source+(-step-2)/2;
689 if(sourceLimit>bytesLimit) {
690 sourceLimit=bytesLimit;
691 }
692 }
693
694 ucnv_toUnicode(cnv,
695 &target, targetLimit,
696 &source, sourceLimit,
697 NULL, (UBool)(sourceLimit==bytesLimit), pErrorCode);
698
699 // check pointers and errors
700 if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
701 if(target!=targetLimit) {
702 // buffer overflow must only be set when the target is filled
703 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
704 break;
705 } else if(targetLimit==resultLimit) {
706 // not just a partial overflow
707 break;
708 }
709
710 // the partial target is filled, set a new limit and continue
711 *pErrorCode=U_ZERO_ERROR;
712 } else if(U_FAILURE(*pErrorCode)) {
713 // some other error occurred, done
714 break;
715 } else {
716 if(source!=sourceLimit) {
717 // when no error occurs, then the input must be consumed
718 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
719 break;
720 }
721
722 // we are done (flush==TRUE) but we continue, to get the index out of bounds error above
723 }
724
725 --step;
726 }
727 }
728 }
729
730 return (int32_t)(target-result);
731 }
732
733 UBool
734 ConversionTest::ToUnicodeCase(ConversionCase &cc, UConverterToUCallback callback, const char *option) {
735 UConverter *cnv;
736 UErrorCode errorCode;
737
738 // open the converter
739 errorCode=U_ZERO_ERROR;
740 cnv=cnv_open(cc.charset, errorCode);
741 if(U_FAILURE(errorCode)) {
742 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_open() failed - %s",
743 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
744 return FALSE;
745 }
746
747 // set the callback
748 if(callback!=NULL) {
749 ucnv_setToUCallBack(cnv, callback, option, NULL, NULL, &errorCode);
750 if(U_FAILURE(errorCode)) {
751 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setToUCallBack() failed - %s",
752 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
753 ucnv_close(cnv);
754 return FALSE;
755 }
756 }
757
758 int32_t resultOffsets[200];
759 UChar result[200];
760 int32_t resultLength;
761 UBool ok;
762
763 static const struct {
764 int32_t step;
765 const char *name;
766 } steps[]={
767 { 0, "bulk" }, // must be first for offsets to be checked
768 { 1, "step=1" },
769 { 3, "step=3" },
770 { 7, "step=7" },
771 { -1, "getNext" },
772 { -2, "toU(bulk)+getNext" },
773 { -3, "getNext+toU(bulk)" },
774 { -4, "toU(1)+getNext" },
775 { -5, "getNext+toU(1)" },
776 { -12, "toU(5)+getNext" },
777 { -13, "getNext+toU(5)" },
778 };
779 int32_t i, step;
780
781 ok=TRUE;
782 for(i=0; i<LENGTHOF(steps) && ok; ++i) {
783 step=steps[i].step;
784 if(step<0 && !cc.finalFlush) {
785 // skip ucnv_getNextUChar() if !finalFlush because
786 // ucnv_getNextUChar() always implies flush
787 continue;
788 }
789 if(step!=0) {
790 // bulk test is first, then offsets are not checked any more
791 cc.offsets=NULL;
792 }
793 errorCode=U_ZERO_ERROR;
794 resultLength=stepToUnicode(cc, cnv,
795 result, LENGTHOF(result),
796 step==0 ? resultOffsets : NULL,
797 step, &errorCode);
798 ok=checkToUnicode(
799 cc, cnv, steps[i].name,
800 result, resultLength,
801 cc.offsets!=NULL ? resultOffsets : NULL,
802 errorCode);
803 if(U_FAILURE(errorCode) || !cc.finalFlush) {
804 // reset if an error occurred or we did not flush
805 // otherwise do nothing to make sure that flushing resets
806 ucnv_resetToUnicode(cnv);
807 }
808 }
809
810 // not a real loop, just a convenience for breaking out of the block
811 while(ok && cc.finalFlush) {
812 // test ucnv_toUChars()
813 memset(result, 0, sizeof(result));
814
815 errorCode=U_ZERO_ERROR;
816 resultLength=ucnv_toUChars(cnv,
817 result, LENGTHOF(result),
818 (const char *)cc.bytes, cc.bytesLength,
819 &errorCode);
820 ok=checkToUnicode(
821 cc, cnv, "toUChars",
822 result, resultLength,
823 NULL,
824 errorCode);
825 if(!ok) {
826 break;
827 }
828
829 // test preflighting
830 // keep the correct result for simple checking
831 errorCode=U_ZERO_ERROR;
832 resultLength=ucnv_toUChars(cnv,
833 NULL, 0,
834 (const char *)cc.bytes, cc.bytesLength,
835 &errorCode);
836 if(errorCode==U_STRING_NOT_TERMINATED_WARNING || errorCode==U_BUFFER_OVERFLOW_ERROR) {
837 errorCode=U_ZERO_ERROR;
838 }
839 ok=checkToUnicode(
840 cc, cnv, "preflight toUChars",
841 result, resultLength,
842 NULL,
843 errorCode);
844 break;
845 }
846
847 ucnv_close(cnv);
848 return ok;
849 }
850
851 UBool
852 ConversionTest::checkToUnicode(ConversionCase &cc, UConverter *cnv, const char *name,
853 const UChar *result, int32_t resultLength,
854 const int32_t *resultOffsets,
855 UErrorCode resultErrorCode) {
856 char resultInvalidChars[8];
857 int8_t resultInvalidLength;
858 UErrorCode errorCode;
859
860 const char *msg;
861
862 // reset the message; NULL will mean "ok"
863 msg=NULL;
864
865 errorCode=U_ZERO_ERROR;
866 resultInvalidLength=sizeof(resultInvalidChars);
867 ucnv_getInvalidChars(cnv, resultInvalidChars, &resultInvalidLength, &errorCode);
868 if(U_FAILURE(errorCode)) {
869 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) ucnv_getInvalidChars() failed - %s",
870 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, u_errorName(errorCode));
871 return FALSE;
872 }
873
874 // check everything that might have gone wrong
875 if(cc.unicodeLength!=resultLength) {
876 msg="wrong result length";
877 } else if(0!=u_memcmp(cc.unicode, result, cc.unicodeLength)) {
878 msg="wrong result string";
879 } else if(cc.offsets!=NULL && 0!=memcmp(cc.offsets, resultOffsets, cc.unicodeLength*sizeof(*cc.offsets))) {
880 msg="wrong offsets";
881 } else if(cc.outErrorCode!=resultErrorCode) {
882 msg="wrong error code";
883 } else if(cc.invalidLength!=resultInvalidLength) {
884 msg="wrong length of last invalid input";
885 } else if(0!=memcmp(cc.invalidChars, resultInvalidChars, cc.invalidLength)) {
886 msg="wrong last invalid input";
887 }
888
889 if(msg==NULL) {
890 return TRUE;
891 } else {
892 char buffer[2000]; // one buffer for all strings
893 char *s, *bytesString, *unicodeString, *resultString,
894 *offsetsString, *resultOffsetsString,
895 *invalidCharsString, *resultInvalidCharsString;
896
897 bytesString=s=buffer;
898 s=printBytes(cc.bytes, cc.bytesLength, bytesString);
899 s=printUnicode(cc.unicode, cc.unicodeLength, unicodeString=s);
900 s=printUnicode(result, resultLength, resultString=s);
901 s=printOffsets(cc.offsets, cc.unicodeLength, offsetsString=s);
902 s=printOffsets(resultOffsets, resultLength, resultOffsetsString=s);
903 s=printBytes(cc.invalidChars, cc.invalidLength, invalidCharsString=s);
904 s=printBytes((uint8_t *)resultInvalidChars, resultInvalidLength, resultInvalidCharsString=s);
905
906 if((s-buffer)>(int32_t)sizeof(buffer)) {
907 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) fatal error: checkToUnicode() test output buffer overflow writing %d chars\n",
908 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, (int)(s-buffer));
909 exit(1);
910 }
911
912 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) failed: %s\n"
913 " bytes <%s>[%d]\n"
914 " expected <%s>[%d]\n"
915 " result <%s>[%d]\n"
916 " offsets <%s>\n"
917 " result offsets <%s>\n"
918 " error code expected %s got %s\n"
919 " invalidChars expected <%s> got <%s>\n",
920 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, msg,
921 bytesString, cc.bytesLength,
922 unicodeString, cc.unicodeLength,
923 resultString, resultLength,
924 offsetsString,
925 resultOffsetsString,
926 u_errorName(cc.outErrorCode), u_errorName(resultErrorCode),
927 invalidCharsString, resultInvalidCharsString);
928
929 return FALSE;
930 }
931 }
932
933 // fromUnicode test worker functions --------------------------------------- ***
934
935 static int32_t
936 stepFromUnicode(ConversionCase &cc, UConverter *cnv,
937 char *result, int32_t resultCapacity,
938 int32_t *resultOffsets, /* also resultCapacity */
939 int32_t step,
940 UErrorCode *pErrorCode) {
941 const UChar *source, *sourceLimit, *unicodeLimit;
942 char *target, *targetLimit, *resultLimit;
943 UBool flush;
944
945 source=cc.unicode;
946 target=result;
947 unicodeLimit=source+cc.unicodeLength;
948 resultLimit=result+resultCapacity;
949
950 // call ucnv_fromUnicode() with in/out buffers no larger than (step) at a time
951 // move only one buffer (in vs. out) at a time to be extra mean
952 // step==0 performs bulk conversion and generates offsets
953
954 // initialize the partial limits for the loop
955 if(step==0) {
956 // use the entire buffers
957 sourceLimit=unicodeLimit;
958 targetLimit=resultLimit;
959 flush=cc.finalFlush;
960 } else {
961 // start with empty partial buffers
962 sourceLimit=source;
963 targetLimit=target;
964 flush=FALSE;
965
966 // output offsets only for bulk conversion
967 resultOffsets=NULL;
968 }
969
970 for(;;) {
971 // resetting the opposite conversion direction must not affect this one
972 ucnv_resetToUnicode(cnv);
973
974 // convert
975 ucnv_fromUnicode(cnv,
976 &target, targetLimit,
977 &source, sourceLimit,
978 resultOffsets,
979 flush, pErrorCode);
980
981 // check pointers and errors
982 if(source>sourceLimit || target>targetLimit) {
983 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
984 break;
985 } else if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
986 if(target!=targetLimit) {
987 // buffer overflow must only be set when the target is filled
988 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
989 break;
990 } else if(targetLimit==resultLimit) {
991 // not just a partial overflow
992 break;
993 }
994
995 // the partial target is filled, set a new limit, reset the error and continue
996 targetLimit=(resultLimit-target)>=step ? target+step : resultLimit;
997 *pErrorCode=U_ZERO_ERROR;
998 } else if(U_FAILURE(*pErrorCode)) {
999 // some other error occurred, done
1000 break;
1001 } else {
1002 if(source!=sourceLimit) {
1003 // when no error occurs, then the input must be consumed
1004 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1005 break;
1006 }
1007
1008 if(sourceLimit==unicodeLimit) {
1009 // we are done
1010 break;
1011 }
1012
1013 // the partial conversion succeeded, set a new limit and continue
1014 sourceLimit=(unicodeLimit-source)>=step ? source+step : unicodeLimit;
1015 flush=(UBool)(cc.finalFlush && sourceLimit==unicodeLimit);
1016 }
1017 }
1018
1019 return (int32_t)(target-result);
1020 }
1021
1022 UBool
1023 ConversionTest::FromUnicodeCase(ConversionCase &cc, UConverterFromUCallback callback, const char *option) {
1024 UConverter *cnv;
1025 UErrorCode errorCode;
1026
1027 // open the converter
1028 errorCode=U_ZERO_ERROR;
1029 cnv=cnv_open(cc.charset, errorCode);
1030 if(U_FAILURE(errorCode)) {
1031 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_open() failed - %s",
1032 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1033 return FALSE;
1034 }
1035
1036 // set the callback
1037 if(callback!=NULL) {
1038 ucnv_setFromUCallBack(cnv, callback, option, NULL, NULL, &errorCode);
1039 if(U_FAILURE(errorCode)) {
1040 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setFromUCallBack() failed - %s",
1041 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1042 ucnv_close(cnv);
1043 return FALSE;
1044 }
1045 }
1046
1047 // set the fallbacks flag
1048 // TODO change with Jitterbug 2401, then add a similar call for toUnicode too
1049 ucnv_setFallback(cnv, cc.fallbacks);
1050
1051 // set the subchar
1052 int32_t length;
1053
1054 if((length=(int32_t)strlen(cc.subchar))!=0) {
1055 ucnv_setSubstChars(cnv, cc.subchar, (int8_t)length, &errorCode);
1056 if(U_FAILURE(errorCode)) {
1057 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setSubChars() failed - %s",
1058 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1059 ucnv_close(cnv);
1060 return FALSE;
1061 }
1062 }
1063
1064 int32_t resultOffsets[200];
1065 char result[200];
1066 int32_t resultLength;
1067 UBool ok;
1068
1069 static const struct {
1070 int32_t step;
1071 const char *name;
1072 } steps[]={
1073 { 0, "bulk" }, // must be first for offsets to be checked
1074 { 1, "step=1" },
1075 { 3, "step=3" },
1076 { 7, "step=7" }
1077 };
1078 int32_t i, step;
1079
1080 ok=TRUE;
1081 for(i=0; i<LENGTHOF(steps) && ok; ++i) {
1082 step=steps[i].step;
1083 if(step!=0) {
1084 // bulk test is first, then offsets are not checked any more
1085 cc.offsets=NULL;
1086 }
1087 errorCode=U_ZERO_ERROR;
1088 resultLength=stepFromUnicode(cc, cnv,
1089 result, LENGTHOF(result),
1090 step==0 ? resultOffsets : NULL,
1091 step, &errorCode);
1092 ok=checkFromUnicode(
1093 cc, cnv, steps[i].name,
1094 (uint8_t *)result, resultLength,
1095 cc.offsets!=NULL ? resultOffsets : NULL,
1096 errorCode);
1097 if(U_FAILURE(errorCode) || !cc.finalFlush) {
1098 // reset if an error occurred or we did not flush
1099 // otherwise do nothing to make sure that flushing resets
1100 ucnv_resetFromUnicode(cnv);
1101 }
1102 }
1103
1104 // not a real loop, just a convenience for breaking out of the block
1105 while(ok && cc.finalFlush) {
1106 // test ucnv_fromUChars()
1107 memset(result, 0, sizeof(result));
1108
1109 errorCode=U_ZERO_ERROR;
1110 resultLength=ucnv_fromUChars(cnv,
1111 result, LENGTHOF(result),
1112 cc.unicode, cc.unicodeLength,
1113 &errorCode);
1114 ok=checkFromUnicode(
1115 cc, cnv, "fromUChars",
1116 (uint8_t *)result, resultLength,
1117 NULL,
1118 errorCode);
1119 if(!ok) {
1120 break;
1121 }
1122
1123 // test preflighting
1124 // keep the correct result for simple checking
1125 errorCode=U_ZERO_ERROR;
1126 resultLength=ucnv_fromUChars(cnv,
1127 NULL, 0,
1128 cc.unicode, cc.unicodeLength,
1129 &errorCode);
1130 if(errorCode==U_STRING_NOT_TERMINATED_WARNING || errorCode==U_BUFFER_OVERFLOW_ERROR) {
1131 errorCode=U_ZERO_ERROR;
1132 }
1133 ok=checkFromUnicode(
1134 cc, cnv, "preflight fromUChars",
1135 (uint8_t *)result, resultLength,
1136 NULL,
1137 errorCode);
1138 break;
1139 }
1140
1141 ucnv_close(cnv);
1142 return ok;
1143 }
1144
1145 UBool
1146 ConversionTest::checkFromUnicode(ConversionCase &cc, UConverter *cnv, const char *name,
1147 const uint8_t *result, int32_t resultLength,
1148 const int32_t *resultOffsets,
1149 UErrorCode resultErrorCode) {
1150 UChar resultInvalidUChars[8];
1151 int8_t resultInvalidLength;
1152 UErrorCode errorCode;
1153
1154 const char *msg;
1155
1156 // reset the message; NULL will mean "ok"
1157 msg=NULL;
1158
1159 errorCode=U_ZERO_ERROR;
1160 resultInvalidLength=LENGTHOF(resultInvalidUChars);
1161 ucnv_getInvalidUChars(cnv, resultInvalidUChars, &resultInvalidLength, &errorCode);
1162 if(U_FAILURE(errorCode)) {
1163 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) ucnv_getInvalidUChars() failed - %s",
1164 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, u_errorName(errorCode));
1165 return FALSE;
1166 }
1167
1168 // check everything that might have gone wrong
1169 if(cc.bytesLength!=resultLength) {
1170 msg="wrong result length";
1171 } else if(0!=memcmp(cc.bytes, result, cc.bytesLength)) {
1172 msg="wrong result string";
1173 } else if(cc.offsets!=NULL && 0!=memcmp(cc.offsets, resultOffsets, cc.bytesLength*sizeof(*cc.offsets))) {
1174 msg="wrong offsets";
1175 } else if(cc.outErrorCode!=resultErrorCode) {
1176 msg="wrong error code";
1177 } else if(cc.invalidLength!=resultInvalidLength) {
1178 msg="wrong length of last invalid input";
1179 } else if(0!=u_memcmp(cc.invalidUChars, resultInvalidUChars, cc.invalidLength)) {
1180 msg="wrong last invalid input";
1181 }
1182
1183 if(msg==NULL) {
1184 return TRUE;
1185 } else {
1186 char buffer[2000]; // one buffer for all strings
1187 char *s, *unicodeString, *bytesString, *resultString,
1188 *offsetsString, *resultOffsetsString,
1189 *invalidCharsString, *resultInvalidUCharsString;
1190
1191 unicodeString=s=buffer;
1192 s=printUnicode(cc.unicode, cc.unicodeLength, unicodeString);
1193 s=printBytes(cc.bytes, cc.bytesLength, bytesString=s);
1194 s=printBytes(result, resultLength, resultString=s);
1195 s=printOffsets(cc.offsets, cc.bytesLength, offsetsString=s);
1196 s=printOffsets(resultOffsets, resultLength, resultOffsetsString=s);
1197 s=printUnicode(cc.invalidUChars, cc.invalidLength, invalidCharsString=s);
1198 s=printUnicode(resultInvalidUChars, resultInvalidLength, resultInvalidUCharsString=s);
1199
1200 if((s-buffer)>(int32_t)sizeof(buffer)) {
1201 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) fatal error: checkFromUnicode() test output buffer overflow writing %d chars\n",
1202 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, (int)(s-buffer));
1203 exit(1);
1204 }
1205
1206 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) failed: %s\n"
1207 " unicode <%s>[%d]\n"
1208 " expected <%s>[%d]\n"
1209 " result <%s>[%d]\n"
1210 " offsets <%s>\n"
1211 " result offsets <%s>\n"
1212 " error code expected %s got %s\n"
1213 " invalidChars expected <%s> got <%s>\n",
1214 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, msg,
1215 unicodeString, cc.unicodeLength,
1216 bytesString, cc.bytesLength,
1217 resultString, resultLength,
1218 offsetsString,
1219 resultOffsetsString,
1220 u_errorName(cc.outErrorCode), u_errorName(resultErrorCode),
1221 invalidCharsString, resultInvalidUCharsString);
1222
1223 return FALSE;
1224 }
1225 }
1226
1227 #endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */