]> git.saurik.com Git - apple/icu.git/blame_incremental - icuSources/test/intltest/convtest.cpp
projects / apple / icu.git / blame_incremental
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (non-incremental) | history | HEAD
ICU-66108.tar.gz
[apple/icu.git] / icuSources / test / intltest / convtest.cpp
... / ...
CommitLineData
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
47enum {
48 // characters used in test data for callbacks
49 SUB_CB='?',
50 SKIP_CB='0',
51 STOP_CB='.',
52 ESC_CB='&'
53};
54
55ConversionTest::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
64ConversionTest::~ConversionTest() {
65 ucnv_close(utf8Cnv);
66}
67
68void
69ConversionTest::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
85void
86ConversionTest::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 cc.bytes=testCase->getBinary(cc.bytesLength, "bytes", errorCode);
120 unicode=testCase->getString("unicode", errorCode);
121 cc.unicode=unicode.getBuffer();
122 cc.unicodeLength=unicode.length();
123
124 offsetsLength=0;
125 cc.offsets=testCase->getIntVector(offsetsLength, "offsets", errorCode);
126 if(offsetsLength==0) {
127 cc.offsets=NULL;
128 } else if(offsetsLength!=unicode.length()) {
129 errln("toUnicode[%d] unicode[%d] and offsets[%d] must have the same length",
130 i, unicode.length(), offsetsLength);
131 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
132 }
133
134 cc.finalFlush= 0!=testCase->getInt28("flush", errorCode);
135 cc.fallbacks= 0!=testCase->getInt28("fallbacks", errorCode);
136
137 s=testCase->getString("errorCode", errorCode);
138 if(s==UNICODE_STRING("invalid", 7)) {
139 cc.outErrorCode=U_INVALID_CHAR_FOUND;
140 } else if(s==UNICODE_STRING("illegal", 7)) {
141 cc.outErrorCode=U_ILLEGAL_CHAR_FOUND;
142 } else if(s==UNICODE_STRING("truncated", 9)) {
143 cc.outErrorCode=U_TRUNCATED_CHAR_FOUND;
144 } else if(s==UNICODE_STRING("illesc", 6)) {
145 cc.outErrorCode=U_ILLEGAL_ESCAPE_SEQUENCE;
146 } else if(s==UNICODE_STRING("unsuppesc", 9)) {
147 cc.outErrorCode=U_UNSUPPORTED_ESCAPE_SEQUENCE;
148 } else {
149 cc.outErrorCode=U_ZERO_ERROR;
150 }
151
152 s=testCase->getString("callback", errorCode);
153 s.extract(0, 0x7fffffff, cbopt, sizeof(cbopt), "");
154 cc.cbopt=cbopt;
155 switch(cbopt[0]) {
156 case SUB_CB:
157 callback=UCNV_TO_U_CALLBACK_SUBSTITUTE;
158 break;
159 case SKIP_CB:
160 callback=UCNV_TO_U_CALLBACK_SKIP;
161 break;
162 case STOP_CB:
163 callback=UCNV_TO_U_CALLBACK_STOP;
164 break;
165 case ESC_CB:
166 callback=UCNV_TO_U_CALLBACK_ESCAPE;
167 break;
168 default:
169 callback=NULL;
170 break;
171 }
172 option=callback==NULL ? cbopt : cbopt+1;
173 if(*option==0) {
174 option=NULL;
175 }
176
177 cc.invalidChars=testCase->getBinary(cc.invalidLength, "invalidChars", errorCode);
178
179 if(U_FAILURE(errorCode)) {
180 errln("error parsing conversion/toUnicode test case %d - %s",
181 i, u_errorName(errorCode));
182 errorCode=U_ZERO_ERROR;
183 } else {
184 logln("TestToUnicode[%d] %s", i, charset);
185 ToUnicodeCase(cc, callback, option);
186 }
187 }
188 delete testData;
189 }
190 delete dataModule;
191 }
192 else {
193 dataerrln("Could not load test conversion data");
194 }
195}
196
197void
198ConversionTest::TestFromUnicode() {
199 ConversionCase cc;
200 char charset[100], cbopt[4];
201 const char *option;
202 UnicodeString s, unicode, invalidUChars;
203 int32_t offsetsLength, index;
204 UConverterFromUCallback callback;
205
206 TestDataModule *dataModule;
207 TestData *testData;
208 const DataMap *testCase;
209 const UChar *p;
210 UErrorCode errorCode;
211 int32_t i, length;
212
213 errorCode=U_ZERO_ERROR;
214 dataModule=TestDataModule::getTestDataModule("conversion", *this, errorCode);
215 if(U_SUCCESS(errorCode)) {
216 testData=dataModule->createTestData("fromUnicode", errorCode);
217 if(U_SUCCESS(errorCode)) {
218 for(i=0; testData->nextCase(testCase, errorCode); ++i) {
219 if(U_FAILURE(errorCode)) {
220 errln("error retrieving conversion/fromUnicode test case %d - %s",
221 i, u_errorName(errorCode));
222 errorCode=U_ZERO_ERROR;
223 continue;
224 }
225
226 cc.caseNr=i;
227
228 s=testCase->getString("charset", errorCode);
229 s.extract(0, 0x7fffffff, charset, sizeof(charset), "");
230 cc.charset=charset;
231
232 unicode=testCase->getString("unicode", errorCode);
233 cc.unicode=unicode.getBuffer();
234 cc.unicodeLength=unicode.length();
235 cc.bytes=testCase->getBinary(cc.bytesLength, "bytes", errorCode);
236
237 offsetsLength=0;
238 cc.offsets=testCase->getIntVector(offsetsLength, "offsets", errorCode);
239 if(offsetsLength==0) {
240 cc.offsets=NULL;
241 } else if(offsetsLength!=cc.bytesLength) {
242 errln("fromUnicode[%d] bytes[%d] and offsets[%d] must have the same length",
243 i, cc.bytesLength, offsetsLength);
244 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
245 }
246
247 cc.finalFlush= 0!=testCase->getInt28("flush", errorCode);
248 cc.fallbacks= 0!=testCase->getInt28("fallbacks", errorCode);
249
250 s=testCase->getString("errorCode", errorCode);
251 if(s==UNICODE_STRING("invalid", 7)) {
252 cc.outErrorCode=U_INVALID_CHAR_FOUND;
253 } else if(s==UNICODE_STRING("illegal", 7)) {
254 cc.outErrorCode=U_ILLEGAL_CHAR_FOUND;
255 } else if(s==UNICODE_STRING("truncated", 9)) {
256 cc.outErrorCode=U_TRUNCATED_CHAR_FOUND;
257 } else {
258 cc.outErrorCode=U_ZERO_ERROR;
259 }
260
261 s=testCase->getString("callback", errorCode);
262 cc.setSub=0; // default: no subchar
263
264 if((index=s.indexOf((UChar)0))>0) {
265 // read NUL-separated subchar first, if any
266 // copy the subchar from Latin-1 characters
267 // start after the NUL
268 p=s.getTerminatedBuffer();
269 length=index+1;
270 p+=length;
271 length=s.length()-length;
272 if(length<=0 || length>=(int32_t)sizeof(cc.subchar)) {
273 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
274 } else {
275 int32_t j;
276
277 for(j=0; j<length; ++j) {
278 cc.subchar[j]=(char)p[j];
279 }
280 // NUL-terminate the subchar
281 cc.subchar[j]=0;
282 cc.setSub=1;
283 }
284
285 // remove the NUL and subchar from s
286 s.truncate(index);
287 } else if((index=s.indexOf((UChar)0x3d))>0) /* '=' */ {
288 // read a substitution string, separated by an equal sign
289 p=s.getBuffer()+index+1;
290 length=s.length()-(index+1);
291 if(length<0 || length>=UPRV_LENGTHOF(cc.subString)) {
292 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
293 } else {
294 u_memcpy(cc.subString, p, length);
295 // NUL-terminate the subString
296 cc.subString[length]=0;
297 cc.setSub=-1;
298 }
299
300 // remove the equal sign and subString from s
301 s.truncate(index);
302 }
303
304 s.extract(0, 0x7fffffff, cbopt, sizeof(cbopt), "");
305 cc.cbopt=cbopt;
306 switch(cbopt[0]) {
307 case SUB_CB:
308 callback=UCNV_FROM_U_CALLBACK_SUBSTITUTE;
309 break;
310 case SKIP_CB:
311 callback=UCNV_FROM_U_CALLBACK_SKIP;
312 break;
313 case STOP_CB:
314 callback=UCNV_FROM_U_CALLBACK_STOP;
315 break;
316 case ESC_CB:
317 callback=UCNV_FROM_U_CALLBACK_ESCAPE;
318 break;
319 default:
320 callback=NULL;
321 break;
322 }
323 option=callback==NULL ? cbopt : cbopt+1;
324 if(*option==0) {
325 option=NULL;
326 }
327
328 invalidUChars=testCase->getString("invalidUChars", errorCode);
329 cc.invalidUChars=invalidUChars.getBuffer();
330 cc.invalidLength=invalidUChars.length();
331
332 if(U_FAILURE(errorCode)) {
333 errln("error parsing conversion/fromUnicode test case %d - %s",
334 i, u_errorName(errorCode));
335 errorCode=U_ZERO_ERROR;
336 } else {
337 logln("TestFromUnicode[%d] %s", i, charset);
338 FromUnicodeCase(cc, callback, option);
339 }
340 }
341 delete testData;
342 }
343 delete dataModule;
344 }
345 else {
346 dataerrln("Could not load test conversion data");
347 }
348}
349
350static const UChar ellipsis[]={ 0x2e, 0x2e, 0x2e };
351
352void
353ConversionTest::TestGetUnicodeSet() {
354 char charset[100];
355 UnicodeString s, map, mapnot;
356 int32_t which;
357
358 ParsePosition pos;
359 UnicodeSet cnvSet, mapSet, mapnotSet, diffSet;
360 UnicodeSet *cnvSetPtr = &cnvSet;
361 LocalUConverterPointer cnv;
362
363 TestDataModule *dataModule;
364 TestData *testData;
365 const DataMap *testCase;
366 UErrorCode errorCode;
367 int32_t i;
368
369 errorCode=U_ZERO_ERROR;
370 dataModule=TestDataModule::getTestDataModule("conversion", *this, errorCode);
371 if(U_SUCCESS(errorCode)) {
372 testData=dataModule->createTestData("getUnicodeSet", errorCode);
373 if(U_SUCCESS(errorCode)) {
374 for(i=0; testData->nextCase(testCase, errorCode); ++i) {
375 if(U_FAILURE(errorCode)) {
376 errln("error retrieving conversion/getUnicodeSet test case %d - %s",
377 i, u_errorName(errorCode));
378 errorCode=U_ZERO_ERROR;
379 continue;
380 }
381
382 s=testCase->getString("charset", errorCode);
383 s.extract(0, 0x7fffffff, charset, sizeof(charset), "");
384
385 map=testCase->getString("map", errorCode);
386 mapnot=testCase->getString("mapnot", errorCode);
387
388 which=testCase->getInt28("which", errorCode);
389
390 if(U_FAILURE(errorCode)) {
391 errln("error parsing conversion/getUnicodeSet test case %d - %s",
392 i, u_errorName(errorCode));
393 errorCode=U_ZERO_ERROR;
394 continue;
395 }
396
397 // test this test case
398 mapSet.clear();
399 mapnotSet.clear();
400
401 pos.setIndex(0);
402 mapSet.applyPattern(map, pos, 0, NULL, errorCode);
403 if(U_FAILURE(errorCode) || pos.getIndex()!=map.length()) {
404 errln("error creating the map set for conversion/getUnicodeSet test case %d - %s\n"
405 " error index %d index %d U+%04x",
406 i, u_errorName(errorCode), pos.getErrorIndex(), pos.getIndex(), map.char32At(pos.getIndex()));
407 errorCode=U_ZERO_ERROR;
408 continue;
409 }
410
411 pos.setIndex(0);
412 mapnotSet.applyPattern(mapnot, pos, 0, NULL, errorCode);
413 if(U_FAILURE(errorCode) || pos.getIndex()!=mapnot.length()) {
414 errln("error creating the mapnot set for conversion/getUnicodeSet test case %d - %s\n"
415 " error index %d index %d U+%04x",
416 i, u_errorName(errorCode), pos.getErrorIndex(), pos.getIndex(), mapnot.char32At(pos.getIndex()));
417 errorCode=U_ZERO_ERROR;
418 continue;
419 }
420
421 logln("TestGetUnicodeSet[%d] %s", i, charset);
422
423 cnv.adoptInstead(cnv_open(charset, errorCode));
424 if(U_FAILURE(errorCode)) {
425 errcheckln(errorCode, "error opening \"%s\" for conversion/getUnicodeSet test case %d - %s",
426 charset, i, u_errorName(errorCode));
427 errorCode=U_ZERO_ERROR;
428 continue;
429 }
430
431 ucnv_getUnicodeSet(cnv.getAlias(), cnvSetPtr->toUSet(), (UConverterUnicodeSet)which, &errorCode);
432
433 if(U_FAILURE(errorCode)) {
434 errln("error in ucnv_getUnicodeSet(\"%s\") for conversion/getUnicodeSet test case %d - %s",
435 charset, i, u_errorName(errorCode));
436 errorCode=U_ZERO_ERROR;
437 continue;
438 }
439
440 // are there items that must be in cnvSet but are not?
441 (diffSet=mapSet).removeAll(cnvSet);
442 if(!diffSet.isEmpty()) {
443 diffSet.toPattern(s, TRUE);
444 if(s.length()>100) {
445 s.replace(100, 0x7fffffff, ellipsis, UPRV_LENGTHOF(ellipsis));
446 }
447 errln("error: ucnv_getUnicodeSet(\"%s\") is missing items - conversion/getUnicodeSet test case %d",
448 charset, i);
449 errln(s);
450 }
451
452 // are there items that must not be in cnvSet but are?
453 (diffSet=mapnotSet).retainAll(cnvSet);
454 if(!diffSet.isEmpty()) {
455 diffSet.toPattern(s, TRUE);
456 if(s.length()>100) {
457 s.replace(100, 0x7fffffff, ellipsis, UPRV_LENGTHOF(ellipsis));
458 }
459 errln("error: ucnv_getUnicodeSet(\"%s\") contains unexpected items - conversion/getUnicodeSet test case %d",
460 charset, i);
461 errln(s);
462 }
463 }
464 delete testData;
465 }
466 delete dataModule;
467 }
468 else {
469 dataerrln("Could not load test conversion data");
470 }
471}
472
473U_CDECL_BEGIN
474static void U_CALLCONV
475getUnicodeSetCallback(const void *context,
476 UConverterFromUnicodeArgs * /*fromUArgs*/,
477 const UChar* /*codeUnits*/,
478 int32_t /*length*/,
479 UChar32 codePoint,
480 UConverterCallbackReason reason,
481 UErrorCode *pErrorCode) {
482 if(reason<=UCNV_IRREGULAR) {
483 ((UnicodeSet *)context)->remove(codePoint); // the converter cannot convert this code point
484 *pErrorCode=U_ZERO_ERROR; // skip
485 } // else ignore the reset, close and clone calls.
486}
487U_CDECL_END
488
489// Compare ucnv_getUnicodeSet() with the set of characters that can be converted.
490void
491ConversionTest::TestGetUnicodeSet2() {
492 // Build a string with all code points.
493 UChar32 cpLimit;
494 int32_t s0Length;
495 if(quick) {
496 cpLimit=s0Length=0x10000; // BMP only
497 } else {
498 cpLimit=0x110000;
499 s0Length=0x10000+0x200000; // BMP + surrogate pairs
500 }
501 UChar *s0=new UChar[s0Length];
502 if(s0==NULL) {
503 return;
504 }
505 UChar *s=s0;
506 UChar32 c;
507 UChar c2;
508 // low BMP
509 for(c=0; c<=0xd7ff; ++c) {
510 *s++=(UChar)c;
511 }
512 // trail surrogates
513 for(c=0xdc00; c<=0xdfff; ++c) {
514 *s++=(UChar)c;
515 }
516 // lead surrogates
517 // (after trails so that there is not even one surrogate pair in between)
518 for(c=0xd800; c<=0xdbff; ++c) {
519 *s++=(UChar)c;
520 }
521 // high BMP
522 for(c=0xe000; c<=0xffff; ++c) {
523 *s++=(UChar)c;
524 }
525 // supplementary code points = surrogate pairs
526 if(cpLimit==0x110000) {
527 for(c=0xd800; c<=0xdbff; ++c) {
528 for(c2=0xdc00; c2<=0xdfff; ++c2) {
529 *s++=(UChar)c;
530 *s++=c2;
531 }
532 }
533 }
534
535 static const char *const cnvNames[]={
536 "UTF-8",
537 "UTF-7",
538 "UTF-16",
539 "US-ASCII",
540 "ISO-8859-1",
541 "windows-1252",
542 "Shift-JIS",
543 "ibm-1390", // EBCDIC_STATEFUL table
544 "ibm-16684", // DBCS-only extension table based on EBCDIC_STATEFUL table
545 "HZ",
546 "ISO-2022-JP",
547 "JIS7",
548 "ISO-2022-CN",
549 "ISO-2022-CN-EXT",
550 "LMBCS"
551 };
552 LocalUConverterPointer cnv;
553 char buffer[1024];
554 int32_t i;
555 for(i=0; i<UPRV_LENGTHOF(cnvNames); ++i) {
556 UErrorCode errorCode=U_ZERO_ERROR;
557 cnv.adoptInstead(cnv_open(cnvNames[i], errorCode));
558 if(U_FAILURE(errorCode)) {
559 errcheckln(errorCode, "failed to open converter %s - %s", cnvNames[i], u_errorName(errorCode));
560 continue;
561 }
562 UnicodeSet expected;
563 ucnv_setFromUCallBack(cnv.getAlias(), getUnicodeSetCallback, &expected, NULL, NULL, &errorCode);
564 if(U_FAILURE(errorCode)) {
565 errln("failed to set the callback on converter %s - %s", cnvNames[i], u_errorName(errorCode));
566 continue;
567 }
568 UConverterUnicodeSet which;
569 for(which=UCNV_ROUNDTRIP_SET; which<UCNV_SET_COUNT; which=(UConverterUnicodeSet)((int)which+1)) {
570 if(which==UCNV_ROUNDTRIP_AND_FALLBACK_SET) {
571 ucnv_setFallback(cnv.getAlias(), TRUE);
572 }
573 expected.add(0, cpLimit-1);
574 s=s0;
575 UBool flush;
576 do {
577 char *t=buffer;
578 flush=(UBool)(s==s0+s0Length);
579 ucnv_fromUnicode(cnv.getAlias(), &t, buffer+sizeof(buffer), (const UChar **)&s, s0+s0Length, NULL, flush, &errorCode);
580 if(U_FAILURE(errorCode)) {
581 if(errorCode==U_BUFFER_OVERFLOW_ERROR) {
582 errorCode=U_ZERO_ERROR;
583 continue;
584 } else {
585 break; // unexpected error, should not occur
586 }
587 }
588 } while(!flush);
589 UnicodeSet set;
590 ucnv_getUnicodeSet(cnv.getAlias(), set.toUSet(), which, &errorCode);
591 if(cpLimit<0x110000) {
592 set.remove(cpLimit, 0x10ffff);
593 }
594 if(which==UCNV_ROUNDTRIP_SET) {
595 // ignore PUA code points because they will be converted even if they
596 // are fallbacks and when other fallbacks are turned off,
597 // but ucnv_getUnicodeSet(UCNV_ROUNDTRIP_SET) delivers true roundtrips
598 expected.remove(0xe000, 0xf8ff);
599 expected.remove(0xf0000, 0xffffd);
600 expected.remove(0x100000, 0x10fffd);
601 set.remove(0xe000, 0xf8ff);
602 set.remove(0xf0000, 0xffffd);
603 set.remove(0x100000, 0x10fffd);
604 }
605 if(set!=expected) {
606 // First try to see if we have different sets because ucnv_getUnicodeSet()
607 // added strings: The above conversion method does not tell us what strings might be convertible.
608 // Remove strings from the set and compare again.
609 set.removeAllStrings();
610 }
611 if(set!=expected) {
612 UnicodeSet diffSet;
613 UnicodeString out;
614
615 // are there items that must be in the set but are not?
616 (diffSet=expected).removeAll(set);
617 if(!diffSet.isEmpty()) {
618 diffSet.toPattern(out, TRUE);
619 if(out.length()>100) {
620 out.replace(100, 0x7fffffff, ellipsis, UPRV_LENGTHOF(ellipsis));
621 }
622 errln("error: ucnv_getUnicodeSet(\"%s\") is missing items - which set: %d",
623 cnvNames[i], which);
624 errln(out);
625 }
626
627 // are there items that must not be in the set but are?
628 (diffSet=set).removeAll(expected);
629 if(!diffSet.isEmpty()) {
630 diffSet.toPattern(out, TRUE);
631 if(out.length()>100) {
632 out.replace(100, 0x7fffffff, ellipsis, UPRV_LENGTHOF(ellipsis));
633 }
634 errln("error: ucnv_getUnicodeSet(\"%s\") contains unexpected items - which set: %d",
635 cnvNames[i], which);
636 errln(out);
637 }
638 }
639 }
640 }
641
642 delete [] s0;
643}
644
645// Test all codepoints which has the default ignorable Unicode property are ignored if they have no mapping
646// If there are any failures, the hard coded list (IS_DEFAULT_IGNORABLE_CODE_POINT) in ucnv_err.c should be updated
647void
648ConversionTest::TestDefaultIgnorableCallback() {
649 UErrorCode status = U_ZERO_ERROR;
650 const char *cnv_name = "euc-jp-2007";
651 const char *pattern_ignorable = "[:Default_Ignorable_Code_Point:]";
652 const char *pattern_not_ignorable = "[:^Default_Ignorable_Code_Point:]";
653
654 LocalPointer<UnicodeSet> set_ignorable(new UnicodeSet(pattern_ignorable, status));
655 if (U_FAILURE(status)) {
656 dataerrln("Unable to create Unicodeset: %s - %s\n", pattern_ignorable, u_errorName(status));
657 return;
658 }
659
660 LocalPointer<UnicodeSet> set_not_ignorable(new UnicodeSet(pattern_not_ignorable, status));
661 if (U_FAILURE(status)) {
662 dataerrln("Unable to create Unicodeset: %s - %s\n", pattern_not_ignorable, u_errorName(status));
663 return;
664 }
665
666 LocalUConverterPointer cnv(cnv_open(cnv_name, status));
667 if (U_FAILURE(status)) {
668 dataerrln("Unable to open converter: %s - %s\n", cnv_name, u_errorName(status));
669 return;
670 }
671
672 // set callback for the converter
673 ucnv_setFromUCallBack(cnv.getAlias(), UCNV_FROM_U_CALLBACK_SUBSTITUTE, NULL, NULL, NULL, &status);
674
675 UChar32 input[1];
676 char output[10];
677 int32_t outputLength;
678
679 // test default ignorables are ignored
680 int size = set_ignorable->size();
681 for (int i = 0; i < size; i++) {
682 status = U_ZERO_ERROR;
683 outputLength= 0;
684
685 input[0] = set_ignorable->charAt(i);
686
687 outputLength = ucnv_fromUChars(cnv.getAlias(), output, 10, UnicodeString::fromUTF32(input, 1).getTerminatedBuffer(), -1, &status);
688 if (U_FAILURE(status) || outputLength != 0) {
689 errln("Ignorable code point: U+%04X not skipped as expected - %s", input[0], u_errorName(status));
690 }
691 }
692
693 // test non-ignorables are not ignored
694 size = set_not_ignorable->size();
695 for (int i = 0; i < size; i++) {
696 status = U_ZERO_ERROR;
697 outputLength= 0;
698
699 input[0] = set_not_ignorable->charAt(i);
700
701 if (input[0] == 0) {
702 continue;
703 }
704
705 outputLength = ucnv_fromUChars(cnv.getAlias(), output, 10, UnicodeString::fromUTF32(input, 1).getTerminatedBuffer(), -1, &status);
706 if (U_FAILURE(status) || outputLength <= 0) {
707 errln("Non-ignorable code point: U+%04X skipped unexpectedly - %s", input[0], u_errorName(status));
708 }
709 }
710}
711
712void
713ConversionTest::TestUTF8ToUTF8Overflow() {
714 IcuTestErrorCode errorCode(*this, "TestUTF8ToUTF8Overflow");
715 LocalUConverterPointer cnv1(ucnv_open("UTF-8", errorCode));
716 LocalUConverterPointer cnv2(ucnv_open("UTF-8", errorCode));
717 static const char *text = "aä"; // ä: 2 bytes
718 const char *source = text;
719 const char *sourceLimit = text + strlen(text);
720 char result[20];
721 char *target = result;
722 const char *targetLimit = result + sizeof(result);
723 UChar buffer16[20];
724 UChar *pivotSource = buffer16;
725 UChar *pivotTarget = buffer16;
726 const UChar *pivotLimit = buffer16 + UPRV_LENGTHOF(buffer16);
727 int32_t length;
728
729 // Convert with insufficient target capacity.
730 result[2] = 5;
731 ucnv_convertEx(cnv2.getAlias(), cnv1.getAlias(),
732 &target, result + 2, &source, sourceLimit,
733 buffer16, &pivotSource, &pivotTarget, pivotLimit,
734 FALSE, FALSE, errorCode);
735 assertEquals("overflow", U_BUFFER_OVERFLOW_ERROR, errorCode.reset());
736 length = (int32_t)(target - result);
737 assertEquals("number of bytes written", 2, length);
738 assertEquals("next byte not clobbered", 5, result[2]);
739
740 // Convert the rest and flush.
741 ucnv_convertEx(cnv2.getAlias(), cnv1.getAlias(),
742 &target, targetLimit, &source, sourceLimit,
743 buffer16, &pivotSource, &pivotTarget, pivotLimit,
744 FALSE, TRUE, errorCode);
745
746 assertSuccess("UTF-8->UTF-8", errorCode);
747 length = (int32_t)(target - result);
748 assertEquals("3 bytes", 3, length);
749 if (length == 3) {
750 assertTrue("result same as input", memcmp(text, result, length) == 0);
751 }
752
753 ucnv_reset(cnv1.getAlias());
754 ucnv_reset(cnv2.getAlias());
755 memset(result, 0, sizeof(result));
756 static const char *text2 = "a🚲"; // U+1F6B2 bicycle: 4 bytes
757 source = text2;
758 sourceLimit = text2 + strlen(text2);
759 target = result;
760 pivotSource = pivotTarget = buffer16;
761
762 // Convert with insufficient target capacity.
763 result[3] = 5;
764 ucnv_convertEx(cnv2.getAlias(), cnv1.getAlias(),
765 &target, result + 3, &source, sourceLimit,
766 buffer16, &pivotSource, &pivotTarget, pivotLimit,
767 FALSE, FALSE, errorCode);
768 assertEquals("text2 overflow", U_BUFFER_OVERFLOW_ERROR, errorCode.reset());
769 length = (int32_t)(target - result);
770 assertEquals("text2 number of bytes written", 3, length);
771 assertEquals("text2 next byte not clobbered", 5, result[3]);
772
773 // Convert the rest and flush.
774 ucnv_convertEx(cnv2.getAlias(), cnv1.getAlias(),
775 &target, targetLimit, &source, sourceLimit,
776 buffer16, &pivotSource, &pivotTarget, pivotLimit,
777 FALSE, TRUE, errorCode);
778
779 assertSuccess("text2 UTF-8->UTF-8", errorCode);
780 length = (int32_t)(target - result);
781 assertEquals("text2 5 bytes", 5, length);
782 if (length == 5) {
783 assertTrue("text2 result same as input", memcmp(text2, result, length) == 0);
784 }
785
786 ucnv_reset(cnv1.getAlias());
787 ucnv_reset(cnv2.getAlias());
788 memset(result, 0, sizeof(result));
789 static const char *illFormed = "\xf1\x91\x93\x96\x91\x94"; // U+514D6 + two more trail bytes
790 source = illFormed;
791 sourceLimit = illFormed + strlen(illFormed);
792 target = result;
793 pivotSource = pivotTarget = buffer16;
794
795 ucnv_setToUCallBack(cnv1.getAlias(), UCNV_TO_U_CALLBACK_STOP, nullptr, nullptr, nullptr, errorCode);
796
797 // Convert only two bytes and flush (but expect failure).
798 char errorBytes[10];
799 int8_t errorLength;
800 result[0] = 5;
801 ucnv_convertEx(cnv2.getAlias(), cnv1.getAlias(),
802 &target, targetLimit, &source, source + 2,
803 buffer16, &pivotSource, &pivotTarget, pivotLimit,
804 FALSE, TRUE, errorCode);
805 assertEquals("illFormed truncated", U_TRUNCATED_CHAR_FOUND, errorCode.reset());
806 length = (int32_t)(target - result);
807 assertEquals("illFormed number of bytes written", 0, length);
808 errorLength = UPRV_LENGTHOF(errorBytes);
809 ucnv_getInvalidChars(cnv1.getAlias(), errorBytes, &errorLength, errorCode);
810 assertEquals("illFormed truncated errorLength", 2, (int32_t)errorLength);
811 if (errorLength == 2) {
812 assertEquals("illFormed truncated errorBytes", 0xf191,
813 ((int32_t)(uint8_t)errorBytes[0] << 8) | (uint8_t)errorBytes[1]);
814 }
815
816 // Continue conversion starting with a trail byte.
817 ucnv_convertEx(cnv2.getAlias(), cnv1.getAlias(),
818 &target, targetLimit, &source, sourceLimit,
819 buffer16, &pivotSource, &pivotTarget, pivotLimit,
820 FALSE, TRUE, errorCode);
821
822 assertEquals("illFormed trail byte", U_ILLEGAL_CHAR_FOUND, errorCode.reset());
823 length = (int32_t)(target - result);
824 assertEquals("illFormed trail byte number of bytes written", 0, length);
825 errorLength = UPRV_LENGTHOF(errorBytes);
826 ucnv_getInvalidChars(cnv1.getAlias(), errorBytes, &errorLength, errorCode);
827 assertEquals("illFormed trail byte errorLength", 1, (int32_t)errorLength);
828 if (errorLength == 1) {
829 assertEquals("illFormed trail byte errorBytes", 0x93, (int32_t)(uint8_t)errorBytes[0]);
830 }
831}
832
833// open testdata or ICU data converter ------------------------------------- ***
834
835UConverter *
836ConversionTest::cnv_open(const char *name, UErrorCode &errorCode) {
837 if(name!=NULL && *name=='+') {
838 // Converter names that start with '+' are ignored in ICU4J tests.
839 ++name;
840 }
841 if(name!=NULL && *name=='*') {
842 /* loadTestData(): set the data directory */
843 return ucnv_openPackage(loadTestData(errorCode), name+1, &errorCode);
844 } else {
845 return ucnv_open(name, &errorCode);
846 }
847}
848
849// output helpers ---------------------------------------------------------- ***
850
851static inline char
852hexDigit(uint8_t digit) {
853 return digit<=9 ? (char)('0'+digit) : (char)('a'-10+digit);
854}
855
856static char *
857printBytes(const uint8_t *bytes, int32_t length, char *out) {
858 uint8_t b;
859
860 if(length>0) {
861 b=*bytes++;
862 --length;
863 *out++=hexDigit((uint8_t)(b>>4));
864 *out++=hexDigit((uint8_t)(b&0xf));
865 }
866
867 while(length>0) {
868 b=*bytes++;
869 --length;
870 *out++=' ';
871 *out++=hexDigit((uint8_t)(b>>4));
872 *out++=hexDigit((uint8_t)(b&0xf));
873 }
874 *out++=0;
875 return out;
876}
877
878static char *
879printUnicode(const UChar *unicode, int32_t length, char *out) {
880 UChar32 c;
881 int32_t i;
882
883 for(i=0; i<length;) {
884 if(i>0) {
885 *out++=' ';
886 }
887 U16_NEXT(unicode, i, length, c);
888 // write 4..6 digits
889 if(c>=0x100000) {
890 *out++='1';
891 }
892 if(c>=0x10000) {
893 *out++=hexDigit((uint8_t)((c>>16)&0xf));
894 }
895 *out++=hexDigit((uint8_t)((c>>12)&0xf));
896 *out++=hexDigit((uint8_t)((c>>8)&0xf));
897 *out++=hexDigit((uint8_t)((c>>4)&0xf));
898 *out++=hexDigit((uint8_t)(c&0xf));
899 }
900 *out++=0;
901 return out;
902}
903
904static char *
905printOffsets(const int32_t *offsets, int32_t length, char *out) {
906 int32_t i, o, d;
907
908 if(offsets==NULL) {
909 length=0;
910 }
911
912 for(i=0; i<length; ++i) {
913 if(i>0) {
914 *out++=' ';
915 }
916 o=offsets[i];
917
918 // print all offsets with 2 characters each (-x, -9..99, xx)
919 if(o<-9) {
920 *out++='-';
921 *out++='x';
922 } else if(o<0) {
923 *out++='-';
924 *out++=(char)('0'-o);
925 } else if(o<=99) {
926 *out++=(d=o/10)==0 ? ' ' : (char)('0'+d);
927 *out++=(char)('0'+o%10);
928 } else /* o>99 */ {
929 *out++='x';
930 *out++='x';
931 }
932 }
933 *out++=0;
934 return out;
935}
936
937// toUnicode test worker functions ----------------------------------------- ***
938
939static int32_t
940stepToUnicode(ConversionCase &cc, UConverter *cnv,
941 UChar *result, int32_t resultCapacity,
942 int32_t *resultOffsets, /* also resultCapacity */
943 int32_t step,
944 UErrorCode *pErrorCode) {
945 const char *source, *sourceLimit, *bytesLimit;
946 UChar *target, *targetLimit, *resultLimit;
947 UBool flush;
948
949 source=(const char *)cc.bytes;
950 target=result;
951 bytesLimit=source+cc.bytesLength;
952 resultLimit=result+resultCapacity;
953
954 if(step>=0) {
955 // call ucnv_toUnicode() with in/out buffers no larger than (step) at a time
956 // move only one buffer (in vs. out) at a time to be extra mean
957 // step==0 performs bulk conversion and generates offsets
958
959 // initialize the partial limits for the loop
960 if(step==0) {
961 // use the entire buffers
962 sourceLimit=bytesLimit;
963 targetLimit=resultLimit;
964 flush=cc.finalFlush;
965 } else {
966 // start with empty partial buffers
967 sourceLimit=source;
968 targetLimit=target;
969 flush=FALSE;
970
971 // output offsets only for bulk conversion
972 resultOffsets=NULL;
973 }
974
975 for(;;) {
976 // resetting the opposite conversion direction must not affect this one
977 ucnv_resetFromUnicode(cnv);
978
979 // convert
980 ucnv_toUnicode(cnv,
981 &target, targetLimit,
982 &source, sourceLimit,
983 resultOffsets,
984 flush, pErrorCode);
985
986 // check pointers and errors
987 if(source>sourceLimit || target>targetLimit) {
988 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
989 break;
990 } else if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
991 if(target!=targetLimit) {
992 // buffer overflow must only be set when the target is filled
993 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
994 break;
995 } else if(targetLimit==resultLimit) {
996 // not just a partial overflow
997 break;
998 }
999
1000 // the partial target is filled, set a new limit, reset the error and continue
1001 targetLimit=(resultLimit-target)>=step ? target+step : resultLimit;
1002 *pErrorCode=U_ZERO_ERROR;
1003 } else if(U_FAILURE(*pErrorCode)) {
1004 // some other error occurred, done
1005 break;
1006 } else {
1007 if(source!=sourceLimit) {
1008 // when no error occurs, then the input must be consumed
1009 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1010 break;
1011 }
1012
1013 if(sourceLimit==bytesLimit) {
1014 // we are done
1015 break;
1016 }
1017
1018 // the partial conversion succeeded, set a new limit and continue
1019 sourceLimit=(bytesLimit-source)>=step ? source+step : bytesLimit;
1020 flush=(UBool)(cc.finalFlush && sourceLimit==bytesLimit);
1021 }
1022 }
1023 } else /* step<0 */ {
1024 /*
1025 * step==-1: call only ucnv_getNextUChar()
1026 * otherwise alternate between ucnv_toUnicode() and ucnv_getNextUChar()
1027 * if step==-2 or -3, then give ucnv_toUnicode() the whole remaining input,
1028 * else give it at most (-step-2)/2 bytes
1029 */
1030 UChar32 c;
1031
1032 // end the loop by getting an index out of bounds error
1033 for(;;) {
1034 // resetting the opposite conversion direction must not affect this one
1035 ucnv_resetFromUnicode(cnv);
1036
1037 // convert
1038 if((step&1)!=0 /* odd: -1, -3, -5, ... */) {
1039 sourceLimit=source; // use sourceLimit not as a real limit
1040 // but to remember the pre-getNextUChar source pointer
1041 c=ucnv_getNextUChar(cnv, &source, bytesLimit, pErrorCode);
1042
1043 // check pointers and errors
1044 if(*pErrorCode==U_INDEX_OUTOFBOUNDS_ERROR) {
1045 if(source!=bytesLimit) {
1046 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1047 } else {
1048 *pErrorCode=U_ZERO_ERROR;
1049 }
1050 break;
1051 } else if(U_FAILURE(*pErrorCode)) {
1052 break;
1053 }
1054 // source may not move if c is from previous overflow
1055
1056 if(target==resultLimit) {
1057 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
1058 break;
1059 }
1060 if(c<=0xffff) {
1061 *target++=(UChar)c;
1062 } else {
1063 *target++=U16_LEAD(c);
1064 if(target==resultLimit) {
1065 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
1066 break;
1067 }
1068 *target++=U16_TRAIL(c);
1069 }
1070
1071 // alternate between -n-1 and -n but leave -1 alone
1072 if(step<-1) {
1073 ++step;
1074 }
1075 } else /* step is even */ {
1076 // allow only one UChar output
1077 targetLimit=target<resultLimit ? target+1 : resultLimit;
1078
1079 // as with ucnv_getNextUChar(), we always flush (if we go to bytesLimit)
1080 // and never output offsets
1081 if(step==-2) {
1082 sourceLimit=bytesLimit;
1083 } else {
1084 sourceLimit=source+(-step-2)/2;
1085 if(sourceLimit>bytesLimit) {
1086 sourceLimit=bytesLimit;
1087 }
1088 }
1089
1090 ucnv_toUnicode(cnv,
1091 &target, targetLimit,
1092 &source, sourceLimit,
1093 NULL, (UBool)(sourceLimit==bytesLimit), pErrorCode);
1094
1095 // check pointers and errors
1096 if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
1097 if(target!=targetLimit) {
1098 // buffer overflow must only be set when the target is filled
1099 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1100 break;
1101 } else if(targetLimit==resultLimit) {
1102 // not just a partial overflow
1103 break;
1104 }
1105
1106 // the partial target is filled, set a new limit and continue
1107 *pErrorCode=U_ZERO_ERROR;
1108 } else if(U_FAILURE(*pErrorCode)) {
1109 // some other error occurred, done
1110 break;
1111 } else {
1112 if(source!=sourceLimit) {
1113 // when no error occurs, then the input must be consumed
1114 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1115 break;
1116 }
1117
1118 // we are done (flush==TRUE) but we continue, to get the index out of bounds error above
1119 }
1120
1121 --step;
1122 }
1123 }
1124 }
1125
1126 return (int32_t)(target-result);
1127}
1128
1129UBool
1130ConversionTest::ToUnicodeCase(ConversionCase &cc, UConverterToUCallback callback, const char *option) {
1131 // open the converter
1132 IcuTestErrorCode errorCode(*this, "ToUnicodeCase");
1133 LocalUConverterPointer cnv(cnv_open(cc.charset, errorCode));
1134 // with no data, the above crashes with "pointer being freed was not allocated" for charset "x11-compound-text", see #13078
1135 if(errorCode.isFailure()) {
1136 errcheckln(errorCode, "toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_open() failed - %s",
1137 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, errorCode.errorName());
1138 errorCode.reset();
1139 return FALSE;
1140 }
1141
1142 // set the callback
1143 if(callback!=NULL) {
1144 ucnv_setToUCallBack(cnv.getAlias(), callback, option, NULL, NULL, errorCode);
1145 if(U_FAILURE(errorCode)) {
1146 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setToUCallBack() failed - %s",
1147 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1148 return FALSE;
1149 }
1150 }
1151
1152 int32_t resultOffsets[256];
1153 UChar result[256];
1154 int32_t resultLength;
1155 UBool ok;
1156
1157 static const struct {
1158 int32_t step;
1159 const char *name;
1160 } steps[]={
1161 { 0, "bulk" }, // must be first for offsets to be checked
1162 { 1, "step=1" },
1163 { 3, "step=3" },
1164 { 7, "step=7" },
1165 { -1, "getNext" },
1166 { -2, "toU(bulk)+getNext" },
1167 { -3, "getNext+toU(bulk)" },
1168 { -4, "toU(1)+getNext" },
1169 { -5, "getNext+toU(1)" },
1170 { -12, "toU(5)+getNext" },
1171 { -13, "getNext+toU(5)" },
1172 };
1173 int32_t i, step;
1174
1175 ok=TRUE;
1176 for(i=0; i<UPRV_LENGTHOF(steps) && ok; ++i) {
1177 step=steps[i].step;
1178 if(step<0 && !cc.finalFlush) {
1179 // skip ucnv_getNextUChar() if !finalFlush because
1180 // ucnv_getNextUChar() always implies flush
1181 continue;
1182 }
1183 if(step!=0) {
1184 // bulk test is first, then offsets are not checked any more
1185 cc.offsets=NULL;
1186 }
1187 else {
1188 memset(resultOffsets, -1, UPRV_LENGTHOF(resultOffsets));
1189 }
1190 memset(result, -1, UPRV_LENGTHOF(result));
1191 errorCode.reset();
1192 resultLength=stepToUnicode(cc, cnv.getAlias(),
1193 result, UPRV_LENGTHOF(result),
1194 step==0 ? resultOffsets : NULL,
1195 step, errorCode);
1196 ok=checkToUnicode(
1197 cc, cnv.getAlias(), steps[i].name,
1198 result, resultLength,
1199 cc.offsets!=NULL ? resultOffsets : NULL,
1200 errorCode);
1201 if(errorCode.isFailure() || !cc.finalFlush) {
1202 // reset if an error occurred or we did not flush
1203 // otherwise do nothing to make sure that flushing resets
1204 ucnv_resetToUnicode(cnv.getAlias());
1205 }
1206 if (cc.offsets != NULL && resultOffsets[resultLength] != -1) {
1207 errln("toUnicode[%d](%s) Conversion wrote too much to offsets at index %d",
1208 cc.caseNr, cc.charset, resultLength);
1209 }
1210 if (result[resultLength] != (UChar)-1) {
1211 errln("toUnicode[%d](%s) Conversion wrote too much to result at index %d",
1212 cc.caseNr, cc.charset, resultLength);
1213 }
1214 }
1215
1216 // not a real loop, just a convenience for breaking out of the block
1217 while(ok && cc.finalFlush) {
1218 // test ucnv_toUChars()
1219 memset(result, 0, sizeof(result));
1220
1221 errorCode.reset();
1222 resultLength=ucnv_toUChars(cnv.getAlias(),
1223 result, UPRV_LENGTHOF(result),
1224 (const char *)cc.bytes, cc.bytesLength,
1225 errorCode);
1226 ok=checkToUnicode(
1227 cc, cnv.getAlias(), "toUChars",
1228 result, resultLength,
1229 NULL,
1230 errorCode);
1231 if(!ok) {
1232 break;
1233 }
1234
1235 // test preflighting
1236 // keep the correct result for simple checking
1237 errorCode.reset();
1238 resultLength=ucnv_toUChars(cnv.getAlias(),
1239 NULL, 0,
1240 (const char *)cc.bytes, cc.bytesLength,
1241 errorCode);
1242 if(errorCode.get()==U_STRING_NOT_TERMINATED_WARNING || errorCode.get()==U_BUFFER_OVERFLOW_ERROR) {
1243 errorCode.reset();
1244 }
1245 ok=checkToUnicode(
1246 cc, cnv.getAlias(), "preflight toUChars",
1247 result, resultLength,
1248 NULL,
1249 errorCode);
1250 break;
1251 }
1252
1253 errorCode.reset(); // all errors have already been reported
1254 return ok;
1255}
1256
1257UBool
1258ConversionTest::checkToUnicode(ConversionCase &cc, UConverter *cnv, const char *name,
1259 const UChar *result, int32_t resultLength,
1260 const int32_t *resultOffsets,
1261 UErrorCode resultErrorCode) {
1262 char resultInvalidChars[8];
1263 int8_t resultInvalidLength;
1264 UErrorCode errorCode;
1265
1266 const char *msg;
1267
1268 // reset the message; NULL will mean "ok"
1269 msg=NULL;
1270
1271 errorCode=U_ZERO_ERROR;
1272 resultInvalidLength=sizeof(resultInvalidChars);
1273 ucnv_getInvalidChars(cnv, resultInvalidChars, &resultInvalidLength, &errorCode);
1274 if(U_FAILURE(errorCode)) {
1275 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) ucnv_getInvalidChars() failed - %s",
1276 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, u_errorName(errorCode));
1277 return FALSE;
1278 }
1279
1280 // check everything that might have gone wrong
1281 if(cc.unicodeLength!=resultLength) {
1282 msg="wrong result length";
1283 } else if(0!=u_memcmp(cc.unicode, result, cc.unicodeLength)) {
1284 msg="wrong result string";
1285 } else if(cc.offsets!=NULL && 0!=memcmp(cc.offsets, resultOffsets, cc.unicodeLength*sizeof(*cc.offsets))) {
1286 msg="wrong offsets";
1287 } else if(cc.outErrorCode!=resultErrorCode) {
1288 msg="wrong error code";
1289 } else if(cc.invalidLength!=resultInvalidLength) {
1290 msg="wrong length of last invalid input";
1291 } else if(0!=memcmp(cc.invalidChars, resultInvalidChars, cc.invalidLength)) {
1292 msg="wrong last invalid input";
1293 }
1294
1295 if(msg==NULL) {
1296 return TRUE;
1297 } else {
1298 char buffer[2000]; // one buffer for all strings
1299 char *s, *bytesString, *unicodeString, *resultString,
1300 *offsetsString, *resultOffsetsString,
1301 *invalidCharsString, *resultInvalidCharsString;
1302
1303 bytesString=s=buffer;
1304 s=printBytes(cc.bytes, cc.bytesLength, bytesString);
1305 s=printUnicode(cc.unicode, cc.unicodeLength, unicodeString=s);
1306 s=printUnicode(result, resultLength, resultString=s);
1307 s=printOffsets(cc.offsets, cc.unicodeLength, offsetsString=s);
1308 s=printOffsets(resultOffsets, resultLength, resultOffsetsString=s);
1309 s=printBytes(cc.invalidChars, cc.invalidLength, invalidCharsString=s);
1310 s=printBytes((uint8_t *)resultInvalidChars, resultInvalidLength, resultInvalidCharsString=s);
1311
1312 if((s-buffer)>(int32_t)sizeof(buffer)) {
1313 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) fatal error: checkToUnicode() test output buffer overflow writing %d chars\n",
1314 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, (int)(s-buffer));
1315 exit(1);
1316 }
1317
1318 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) failed: %s\n"
1319 " bytes <%s>[%d]\n"
1320 " expected <%s>[%d]\n"
1321 " result <%s>[%d]\n"
1322 " offsets <%s>\n"
1323 " result offsets <%s>\n"
1324 " error code expected %s got %s\n"
1325 " invalidChars expected <%s> got <%s>\n",
1326 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, msg,
1327 bytesString, cc.bytesLength,
1328 unicodeString, cc.unicodeLength,
1329 resultString, resultLength,
1330 offsetsString,
1331 resultOffsetsString,
1332 u_errorName(cc.outErrorCode), u_errorName(resultErrorCode),
1333 invalidCharsString, resultInvalidCharsString);
1334
1335 return FALSE;
1336 }
1337}
1338
1339// fromUnicode test worker functions --------------------------------------- ***
1340
1341static int32_t
1342stepFromUTF8(ConversionCase &cc,
1343 UConverter *utf8Cnv, UConverter *cnv,
1344 char *result, int32_t resultCapacity,
1345 int32_t step,
1346 UErrorCode *pErrorCode) {
1347 const char *source, *sourceLimit, *utf8Limit;
1348 UChar pivotBuffer[32];
1349 UChar *pivotSource, *pivotTarget, *pivotLimit;
1350 char *target, *targetLimit, *resultLimit;
1351 UBool flush;
1352
1353 source=cc.utf8;
1354 pivotSource=pivotTarget=pivotBuffer;
1355 target=result;
1356 utf8Limit=source+cc.utf8Length;
1357 resultLimit=result+resultCapacity;
1358
1359 // call ucnv_convertEx() with in/out buffers no larger than (step) at a time
1360 // move only one buffer (in vs. out) at a time to be extra mean
1361 // step==0 performs bulk conversion
1362
1363 // initialize the partial limits for the loop
1364 if(step==0) {
1365 // use the entire buffers
1366 sourceLimit=utf8Limit;
1367 targetLimit=resultLimit;
1368 flush=cc.finalFlush;
1369
1370 pivotLimit=pivotBuffer+UPRV_LENGTHOF(pivotBuffer);
1371 } else {
1372 // start with empty partial buffers
1373 sourceLimit=source;
1374 targetLimit=target;
1375 flush=FALSE;
1376
1377 // empty pivot is not allowed, make it of length step
1378 pivotLimit=pivotBuffer+step;
1379 }
1380
1381 for(;;) {
1382 // resetting the opposite conversion direction must not affect this one
1383 ucnv_resetFromUnicode(utf8Cnv);
1384 ucnv_resetToUnicode(cnv);
1385
1386 // convert
1387 ucnv_convertEx(cnv, utf8Cnv,
1388 &target, targetLimit,
1389 &source, sourceLimit,
1390 pivotBuffer, &pivotSource, &pivotTarget, pivotLimit,
1391 FALSE, flush, pErrorCode);
1392
1393 // check pointers and errors
1394 if(source>sourceLimit || target>targetLimit) {
1395 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1396 break;
1397 } else if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
1398 if(target!=targetLimit) {
1399 // buffer overflow must only be set when the target is filled
1400 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1401 break;
1402 } else if(targetLimit==resultLimit) {
1403 // not just a partial overflow
1404 break;
1405 }
1406
1407 // the partial target is filled, set a new limit, reset the error and continue
1408 targetLimit=(resultLimit-target)>=step ? target+step : resultLimit;
1409 *pErrorCode=U_ZERO_ERROR;
1410 } else if(U_FAILURE(*pErrorCode)) {
1411 if(pivotSource==pivotBuffer) {
1412 // toUnicode error, should not occur
1413 // toUnicode errors are tested in cintltst TestConvertExFromUTF8()
1414 break;
1415 } else {
1416 // fromUnicode error
1417 // some other error occurred, done
1418 break;
1419 }
1420 } else {
1421 if(source!=sourceLimit) {
1422 // when no error occurs, then the input must be consumed
1423 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1424 break;
1425 }
1426
1427 if(sourceLimit==utf8Limit) {
1428 // we are done
1429 if(*pErrorCode==U_STRING_NOT_TERMINATED_WARNING) {
1430 // ucnv_convertEx() warns about not terminating the output
1431 // but ucnv_fromUnicode() does not and so
1432 // checkFromUnicode() does not expect it
1433 *pErrorCode=U_ZERO_ERROR;
1434 }
1435 break;
1436 }
1437
1438 // the partial conversion succeeded, set a new limit and continue
1439 sourceLimit=(utf8Limit-source)>=step ? source+step : utf8Limit;
1440 flush=(UBool)(cc.finalFlush && sourceLimit==utf8Limit);
1441 }
1442 }
1443
1444 return (int32_t)(target-result);
1445}
1446
1447static int32_t
1448stepFromUnicode(ConversionCase &cc, UConverter *cnv,
1449 char *result, int32_t resultCapacity,
1450 int32_t *resultOffsets, /* also resultCapacity */
1451 int32_t step,
1452 UErrorCode *pErrorCode) {
1453 const UChar *source, *sourceLimit, *unicodeLimit;
1454 char *target, *targetLimit, *resultLimit;
1455 UBool flush;
1456
1457 source=cc.unicode;
1458 target=result;
1459 unicodeLimit=source+cc.unicodeLength;
1460 resultLimit=result+resultCapacity;
1461
1462 // call ucnv_fromUnicode() with in/out buffers no larger than (step) at a time
1463 // move only one buffer (in vs. out) at a time to be extra mean
1464 // step==0 performs bulk conversion and generates offsets
1465
1466 // initialize the partial limits for the loop
1467 if(step==0) {
1468 // use the entire buffers
1469 sourceLimit=unicodeLimit;
1470 targetLimit=resultLimit;
1471 flush=cc.finalFlush;
1472 } else {
1473 // start with empty partial buffers
1474 sourceLimit=source;
1475 targetLimit=target;
1476 flush=FALSE;
1477
1478 // output offsets only for bulk conversion
1479 resultOffsets=NULL;
1480 }
1481
1482 for(;;) {
1483 // resetting the opposite conversion direction must not affect this one
1484 ucnv_resetToUnicode(cnv);
1485
1486 // convert
1487 ucnv_fromUnicode(cnv,
1488 &target, targetLimit,
1489 &source, sourceLimit,
1490 resultOffsets,
1491 flush, pErrorCode);
1492
1493 // check pointers and errors
1494 if(source>sourceLimit || target>targetLimit) {
1495 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1496 break;
1497 } else if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
1498 if(target!=targetLimit) {
1499 // buffer overflow must only be set when the target is filled
1500 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1501 break;
1502 } else if(targetLimit==resultLimit) {
1503 // not just a partial overflow
1504 break;
1505 }
1506
1507 // the partial target is filled, set a new limit, reset the error and continue
1508 targetLimit=(resultLimit-target)>=step ? target+step : resultLimit;
1509 *pErrorCode=U_ZERO_ERROR;
1510 } else if(U_FAILURE(*pErrorCode)) {
1511 // some other error occurred, done
1512 break;
1513 } else {
1514 if(source!=sourceLimit) {
1515 // when no error occurs, then the input must be consumed
1516 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1517 break;
1518 }
1519
1520 if(sourceLimit==unicodeLimit) {
1521 // we are done
1522 break;
1523 }
1524
1525 // the partial conversion succeeded, set a new limit and continue
1526 sourceLimit=(unicodeLimit-source)>=step ? source+step : unicodeLimit;
1527 flush=(UBool)(cc.finalFlush && sourceLimit==unicodeLimit);
1528 }
1529 }
1530
1531 return (int32_t)(target-result);
1532}
1533
1534UBool
1535ConversionTest::FromUnicodeCase(ConversionCase &cc, UConverterFromUCallback callback, const char *option) {
1536 UConverter *cnv;
1537 UErrorCode errorCode;
1538
1539 // open the converter
1540 errorCode=U_ZERO_ERROR;
1541 cnv=cnv_open(cc.charset, errorCode);
1542 if(U_FAILURE(errorCode)) {
1543 errcheckln(errorCode, "fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_open() failed - %s",
1544 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1545 return FALSE;
1546 }
1547 ucnv_resetToUnicode(utf8Cnv);
1548
1549 // set the callback
1550 if(callback!=NULL) {
1551 ucnv_setFromUCallBack(cnv, callback, option, NULL, NULL, &errorCode);
1552 if(U_FAILURE(errorCode)) {
1553 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setFromUCallBack() failed - %s",
1554 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1555 ucnv_close(cnv);
1556 return FALSE;
1557 }
1558 }
1559
1560 // set the fallbacks flag
1561 // TODO change with Jitterbug 2401, then add a similar call for toUnicode too
1562 ucnv_setFallback(cnv, cc.fallbacks);
1563
1564 // set the subchar
1565 int32_t length;
1566
1567 if(cc.setSub>0) {
1568 length=(int32_t)strlen(cc.subchar);
1569 ucnv_setSubstChars(cnv, cc.subchar, (int8_t)length, &errorCode);
1570 if(U_FAILURE(errorCode)) {
1571 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setSubstChars() failed - %s",
1572 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1573 ucnv_close(cnv);
1574 return FALSE;
1575 }
1576 } else if(cc.setSub<0) {
1577 ucnv_setSubstString(cnv, cc.subString, -1, &errorCode);
1578 if(U_FAILURE(errorCode)) {
1579 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setSubstString() failed - %s",
1580 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1581 ucnv_close(cnv);
1582 return FALSE;
1583 }
1584 }
1585
1586 // convert unicode to utf8
1587 char utf8[256];
1588 cc.utf8=utf8;
1589 u_strToUTF8(utf8, UPRV_LENGTHOF(utf8), &cc.utf8Length,
1590 cc.unicode, cc.unicodeLength,
1591 &errorCode);
1592 if(U_FAILURE(errorCode)) {
1593 // skip UTF-8 testing of a string with an unpaired surrogate,
1594 // or of one that's too long
1595 // toUnicode errors are tested in cintltst TestConvertExFromUTF8()
1596 cc.utf8Length=-1;
1597 }
1598
1599 int32_t resultOffsets[256];
1600 char result[256];
1601 int32_t resultLength;
1602 UBool ok;
1603
1604 static const struct {
1605 int32_t step;
1606 const char *name, *utf8Name;
1607 } steps[]={
1608 { 0, "bulk", "utf8" }, // must be first for offsets to be checked
1609 { 1, "step=1", "utf8 step=1" },
1610 { 3, "step=3", "utf8 step=3" },
1611 { 7, "step=7", "utf8 step=7" }
1612 };
1613 int32_t i, step;
1614
1615 ok=TRUE;
1616 for(i=0; i<UPRV_LENGTHOF(steps) && ok; ++i) {
1617 step=steps[i].step;
1618 memset(resultOffsets, -1, UPRV_LENGTHOF(resultOffsets));
1619 memset(result, -1, UPRV_LENGTHOF(result));
1620 errorCode=U_ZERO_ERROR;
1621 resultLength=stepFromUnicode(cc, cnv,
1622 result, UPRV_LENGTHOF(result),
1623 step==0 ? resultOffsets : NULL,
1624 step, &errorCode);
1625 ok=checkFromUnicode(
1626 cc, cnv, steps[i].name,
1627 (uint8_t *)result, resultLength,
1628 cc.offsets!=NULL ? resultOffsets : NULL,
1629 errorCode);
1630 if(U_FAILURE(errorCode) || !cc.finalFlush) {
1631 // reset if an error occurred or we did not flush
1632 // otherwise do nothing to make sure that flushing resets
1633 ucnv_resetFromUnicode(cnv);
1634 }
1635 if (resultOffsets[resultLength] != -1) {
1636 errln("fromUnicode[%d](%s) Conversion wrote too much to offsets at index %d",
1637 cc.caseNr, cc.charset, resultLength);
1638 }
1639 if (result[resultLength] != (char)-1) {
1640 errln("fromUnicode[%d](%s) Conversion wrote too much to result at index %d",
1641 cc.caseNr, cc.charset, resultLength);
1642 }
1643
1644 // bulk test is first, then offsets are not checked any more
1645 cc.offsets=NULL;
1646
1647 // test direct conversion from UTF-8
1648 if(cc.utf8Length>=0) {
1649 errorCode=U_ZERO_ERROR;
1650 resultLength=stepFromUTF8(cc, utf8Cnv, cnv,
1651 result, UPRV_LENGTHOF(result),
1652 step, &errorCode);
1653 ok=checkFromUnicode(
1654 cc, cnv, steps[i].utf8Name,
1655 (uint8_t *)result, resultLength,
1656 NULL,
1657 errorCode);
1658 if(U_FAILURE(errorCode) || !cc.finalFlush) {
1659 // reset if an error occurred or we did not flush
1660 // otherwise do nothing to make sure that flushing resets
1661 ucnv_resetToUnicode(utf8Cnv);
1662 ucnv_resetFromUnicode(cnv);
1663 }
1664 }
1665 }
1666
1667 // not a real loop, just a convenience for breaking out of the block
1668 while(ok && cc.finalFlush) {
1669 // test ucnv_fromUChars()
1670 memset(result, 0, sizeof(result));
1671
1672 errorCode=U_ZERO_ERROR;
1673 resultLength=ucnv_fromUChars(cnv,
1674 result, UPRV_LENGTHOF(result),
1675 cc.unicode, cc.unicodeLength,
1676 &errorCode);
1677 ok=checkFromUnicode(
1678 cc, cnv, "fromUChars",
1679 (uint8_t *)result, resultLength,
1680 NULL,
1681 errorCode);
1682 if(!ok) {
1683 break;
1684 }
1685
1686 // test preflighting
1687 // keep the correct result for simple checking
1688 errorCode=U_ZERO_ERROR;
1689 resultLength=ucnv_fromUChars(cnv,
1690 NULL, 0,
1691 cc.unicode, cc.unicodeLength,
1692 &errorCode);
1693 if(errorCode==U_STRING_NOT_TERMINATED_WARNING || errorCode==U_BUFFER_OVERFLOW_ERROR) {
1694 errorCode=U_ZERO_ERROR;
1695 }
1696 ok=checkFromUnicode(
1697 cc, cnv, "preflight fromUChars",
1698 (uint8_t *)result, resultLength,
1699 NULL,
1700 errorCode);
1701 break;
1702 }
1703
1704 ucnv_close(cnv);
1705 return ok;
1706}
1707
1708UBool
1709ConversionTest::checkFromUnicode(ConversionCase &cc, UConverter *cnv, const char *name,
1710 const uint8_t *result, int32_t resultLength,
1711 const int32_t *resultOffsets,
1712 UErrorCode resultErrorCode) {
1713 UChar resultInvalidUChars[8];
1714 int8_t resultInvalidLength;
1715 UErrorCode errorCode;
1716
1717 const char *msg;
1718
1719 // reset the message; NULL will mean "ok"
1720 msg=NULL;
1721
1722 errorCode=U_ZERO_ERROR;
1723 resultInvalidLength=UPRV_LENGTHOF(resultInvalidUChars);
1724 ucnv_getInvalidUChars(cnv, resultInvalidUChars, &resultInvalidLength, &errorCode);
1725 if(U_FAILURE(errorCode)) {
1726 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) ucnv_getInvalidUChars() failed - %s",
1727 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, u_errorName(errorCode));
1728 return FALSE;
1729 }
1730
1731 // check everything that might have gone wrong
1732 if(cc.bytesLength!=resultLength) {
1733 msg="wrong result length";
1734 } else if(0!=memcmp(cc.bytes, result, cc.bytesLength)) {
1735 msg="wrong result string";
1736 } else if(cc.offsets!=NULL && 0!=memcmp(cc.offsets, resultOffsets, cc.bytesLength*sizeof(*cc.offsets))) {
1737 msg="wrong offsets";
1738 } else if(cc.outErrorCode!=resultErrorCode) {
1739 msg="wrong error code";
1740 } else if(cc.invalidLength!=resultInvalidLength) {
1741 msg="wrong length of last invalid input";
1742 } else if(0!=u_memcmp(cc.invalidUChars, resultInvalidUChars, cc.invalidLength)) {
1743 msg="wrong last invalid input";
1744 }
1745
1746 if(msg==NULL) {
1747 return TRUE;
1748 } else {
1749 char buffer[2000]; // one buffer for all strings
1750 char *s, *unicodeString, *bytesString, *resultString,
1751 *offsetsString, *resultOffsetsString,
1752 *invalidCharsString, *resultInvalidUCharsString;
1753
1754 unicodeString=s=buffer;
1755 s=printUnicode(cc.unicode, cc.unicodeLength, unicodeString);
1756 s=printBytes(cc.bytes, cc.bytesLength, bytesString=s);
1757 s=printBytes(result, resultLength, resultString=s);
1758 s=printOffsets(cc.offsets, cc.bytesLength, offsetsString=s);
1759 s=printOffsets(resultOffsets, resultLength, resultOffsetsString=s);
1760 s=printUnicode(cc.invalidUChars, cc.invalidLength, invalidCharsString=s);
1761 s=printUnicode(resultInvalidUChars, resultInvalidLength, resultInvalidUCharsString=s);
1762
1763 if((s-buffer)>(int32_t)sizeof(buffer)) {
1764 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) fatal error: checkFromUnicode() test output buffer overflow writing %d chars\n",
1765 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, (int)(s-buffer));
1766 exit(1);
1767 }
1768
1769 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) failed: %s\n"
1770 " unicode <%s>[%d]\n"
1771 " expected <%s>[%d]\n"
1772 " result <%s>[%d]\n"
1773 " offsets <%s>\n"
1774 " result offsets <%s>\n"
1775 " error code expected %s got %s\n"
1776 " invalidChars expected <%s> got <%s>\n",
1777 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, msg,
1778 unicodeString, cc.unicodeLength,
1779 bytesString, cc.bytesLength,
1780 resultString, resultLength,
1781 offsetsString,
1782 resultOffsetsString,
1783 u_errorName(cc.outErrorCode), u_errorName(resultErrorCode),
1784 invalidCharsString, resultInvalidUCharsString);
1785
1786 return FALSE;
1787 }
1788}
1789
1790#endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */
mirror of ICU