1 /*
2 *******************************************************************************
3 *
4 * Copyright (C) 2003-2010, 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
ConversionTest()53 ConversionTest::ConversionTest() {
54 UErrorCode errorCode=U_ZERO_ERROR;
55 utf8Cnv=ucnv_open("UTF-8", &errorCode);
56 ucnv_setToUCallBack(utf8Cnv, UCNV_TO_U_CALLBACK_STOP, NULL, NULL, NULL, &errorCode);
57 if(U_FAILURE(errorCode)) {
58 errln("unable to open UTF-8 converter");
59 }
60 }
61
~ConversionTest()62 ConversionTest::~ConversionTest() {
63 ucnv_close(utf8Cnv);
64 }
65
66 void
runIndexedTest(int32_t index,UBool exec,const char * & name,char *)67 ConversionTest::runIndexedTest(int32_t index, UBool exec, const char *&name, char * /*par*/) {
68 if (exec) logln("TestSuite ConversionTest: ");
69 switch (index) {
70 #if !UCONFIG_NO_FILE_IO
71 case 0: name="TestToUnicode"; if (exec) TestToUnicode(); break;
72 case 1: name="TestFromUnicode"; if (exec) TestFromUnicode(); break;
73 case 2: name="TestGetUnicodeSet"; if (exec) TestGetUnicodeSet(); break;
74 #else
75 case 0:
76 case 1:
77 case 2: name="skip"; break;
78 #endif
79 case 3: name="TestGetUnicodeSet2"; if (exec) TestGetUnicodeSet2(); break;
80 default: name=""; break; //needed to end loop
81 }
82 }
83
84 // test data interface ----------------------------------------------------- ***
85
86 void
TestToUnicode()87 ConversionTest::TestToUnicode() {
88 ConversionCase cc;
89 char charset[100], cbopt[4];
90 const char *option;
91 UnicodeString s, unicode;
92 int32_t offsetsLength;
93 UConverterToUCallback callback;
94
95 TestDataModule *dataModule;
96 TestData *testData;
97 const DataMap *testCase;
98 UErrorCode errorCode;
99 int32_t i;
100
101 errorCode=U_ZERO_ERROR;
102 dataModule=TestDataModule::getTestDataModule("conversion", *this, errorCode);
103 if(U_SUCCESS(errorCode)) {
104 testData=dataModule->createTestData("toUnicode", errorCode);
105 if(U_SUCCESS(errorCode)) {
106 for(i=0; testData->nextCase(testCase, errorCode); ++i) {
107 if(U_FAILURE(errorCode)) {
108 errln("error retrieving conversion/toUnicode test case %d - %s",
109 i, u_errorName(errorCode));
110 errorCode=U_ZERO_ERROR;
111 continue;
112 }
113
114 cc.caseNr=i;
115
116 s=testCase->getString("charset", errorCode);
117 s.extract(0, 0x7fffffff, charset, sizeof(charset), "");
118 cc.charset=charset;
119
120 // BEGIN android-added
121 // To save space, Android does not build full ISO-2022 tables.
122 // We skip the TestGetKeywordValuesForLocale for counting available collations.
123 if (strlen(charset) >= 8 &&
124 strncmp(charset+4, "2022", 4) == 0) {
125 continue;
126 }
127 // END android-added
128
129 cc.bytes=testCase->getBinary(cc.bytesLength, "bytes", errorCode);
130 unicode=testCase->getString("unicode", errorCode);
131 cc.unicode=unicode.getBuffer();
132 cc.unicodeLength=unicode.length();
133
134 offsetsLength=0;
135 cc.offsets=testCase->getIntVector(offsetsLength, "offsets", errorCode);
136 if(offsetsLength==0) {
137 cc.offsets=NULL;
138 } else if(offsetsLength!=unicode.length()) {
139 errln("toUnicode[%d] unicode[%d] and offsets[%d] must have the same length",
140 i, unicode.length(), offsetsLength);
141 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
142 }
143
144 cc.finalFlush= 0!=testCase->getInt28("flush", errorCode);
145 cc.fallbacks= 0!=testCase->getInt28("fallbacks", errorCode);
146
147 s=testCase->getString("errorCode", errorCode);
148 if(s==UNICODE_STRING("invalid", 7)) {
149 cc.outErrorCode=U_INVALID_CHAR_FOUND;
150 } else if(s==UNICODE_STRING("illegal", 7)) {
151 cc.outErrorCode=U_ILLEGAL_CHAR_FOUND;
152 } else if(s==UNICODE_STRING("truncated", 9)) {
153 cc.outErrorCode=U_TRUNCATED_CHAR_FOUND;
154 } else if(s==UNICODE_STRING("illesc", 6)) {
155 cc.outErrorCode=U_ILLEGAL_ESCAPE_SEQUENCE;
156 } else if(s==UNICODE_STRING("unsuppesc", 9)) {
157 cc.outErrorCode=U_UNSUPPORTED_ESCAPE_SEQUENCE;
158 } else {
159 cc.outErrorCode=U_ZERO_ERROR;
160 }
161
162 s=testCase->getString("callback", errorCode);
163 s.extract(0, 0x7fffffff, cbopt, sizeof(cbopt), "");
164 cc.cbopt=cbopt;
165 switch(cbopt[0]) {
166 case SUB_CB:
167 callback=UCNV_TO_U_CALLBACK_SUBSTITUTE;
168 break;
169 case SKIP_CB:
170 callback=UCNV_TO_U_CALLBACK_SKIP;
171 break;
172 case STOP_CB:
173 callback=UCNV_TO_U_CALLBACK_STOP;
174 break;
175 case ESC_CB:
176 callback=UCNV_TO_U_CALLBACK_ESCAPE;
177 break;
178 default:
179 callback=NULL;
180 break;
181 }
182 option=callback==NULL ? cbopt : cbopt+1;
183 if(*option==0) {
184 option=NULL;
185 }
186
187 cc.invalidChars=testCase->getBinary(cc.invalidLength, "invalidChars", errorCode);
188
189 if(U_FAILURE(errorCode)) {
190 errln("error parsing conversion/toUnicode test case %d - %s",
191 i, u_errorName(errorCode));
192 errorCode=U_ZERO_ERROR;
193 } else {
194 logln("TestToUnicode[%d] %s", i, charset);
195 ToUnicodeCase(cc, callback, option);
196 }
197 }
198 delete testData;
199 }
200 delete dataModule;
201 }
202 else {
203 dataerrln("Could not load test conversion data");
204 }
205 }
206
207 void
TestFromUnicode()208 ConversionTest::TestFromUnicode() {
209 ConversionCase cc;
210 char charset[100], cbopt[4];
211 const char *option;
212 UnicodeString s, unicode, invalidUChars;
213 int32_t offsetsLength, index;
214 UConverterFromUCallback callback;
215
216 TestDataModule *dataModule;
217 TestData *testData;
218 const DataMap *testCase;
219 const UChar *p;
220 UErrorCode errorCode;
221 int32_t i, length;
222
223 errorCode=U_ZERO_ERROR;
224 dataModule=TestDataModule::getTestDataModule("conversion", *this, errorCode);
225 if(U_SUCCESS(errorCode)) {
226 testData=dataModule->createTestData("fromUnicode", errorCode);
227 if(U_SUCCESS(errorCode)) {
228 for(i=0; testData->nextCase(testCase, errorCode); ++i) {
229 if(U_FAILURE(errorCode)) {
230 errln("error retrieving conversion/fromUnicode test case %d - %s",
231 i, u_errorName(errorCode));
232 errorCode=U_ZERO_ERROR;
233 continue;
234 }
235
236 cc.caseNr=i;
237
238 s=testCase->getString("charset", errorCode);
239 s.extract(0, 0x7fffffff, charset, sizeof(charset), "");
240 cc.charset=charset;
241
242 // BEGIN android-added
243 // To save space, Android does not build full ISO-2022 tables.
244 // We skip the TestGetKeywordValuesForLocale for counting available collations.
245 if (strlen(charset) >= 8 &&
246 strncmp(charset+4, "2022", 4) == 0) {
247 continue;
248 }
249 // END android-added
250
251 unicode=testCase->getString("unicode", errorCode);
252 cc.unicode=unicode.getBuffer();
253 cc.unicodeLength=unicode.length();
254 cc.bytes=testCase->getBinary(cc.bytesLength, "bytes", errorCode);
255
256 offsetsLength=0;
257 cc.offsets=testCase->getIntVector(offsetsLength, "offsets", errorCode);
258 if(offsetsLength==0) {
259 cc.offsets=NULL;
260 } else if(offsetsLength!=cc.bytesLength) {
261 errln("fromUnicode[%d] bytes[%d] and offsets[%d] must have the same length",
262 i, cc.bytesLength, offsetsLength);
263 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
264 }
265
266 cc.finalFlush= 0!=testCase->getInt28("flush", errorCode);
267 cc.fallbacks= 0!=testCase->getInt28("fallbacks", errorCode);
268
269 s=testCase->getString("errorCode", errorCode);
270 if(s==UNICODE_STRING("invalid", 7)) {
271 cc.outErrorCode=U_INVALID_CHAR_FOUND;
272 } else if(s==UNICODE_STRING("illegal", 7)) {
273 cc.outErrorCode=U_ILLEGAL_CHAR_FOUND;
274 } else if(s==UNICODE_STRING("truncated", 9)) {
275 cc.outErrorCode=U_TRUNCATED_CHAR_FOUND;
276 } else {
277 cc.outErrorCode=U_ZERO_ERROR;
278 }
279
280 s=testCase->getString("callback", errorCode);
281 cc.setSub=0; // default: no subchar
282
283 if((index=s.indexOf((UChar)0))>0) {
284 // read NUL-separated subchar first, if any
285 // copy the subchar from Latin-1 characters
286 // start after the NUL
287 p=s.getTerminatedBuffer();
288 length=index+1;
289 p+=length;
290 length=s.length()-length;
291 if(length<=0 || length>=(int32_t)sizeof(cc.subchar)) {
292 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
293 } else {
294 int32_t j;
295
296 for(j=0; j<length; ++j) {
297 cc.subchar[j]=(char)p[j];
298 }
299 // NUL-terminate the subchar
300 cc.subchar[j]=0;
301 cc.setSub=1;
302 }
303
304 // remove the NUL and subchar from s
305 s.truncate(index);
306 } else if((index=s.indexOf((UChar)0x3d))>0) /* '=' */ {
307 // read a substitution string, separated by an equal sign
308 p=s.getBuffer()+index+1;
309 length=s.length()-(index+1);
310 if(length<0 || length>=LENGTHOF(cc.subString)) {
311 errorCode=U_ILLEGAL_ARGUMENT_ERROR;
312 } else {
313 u_memcpy(cc.subString, p, length);
314 // NUL-terminate the subString
315 cc.subString[length]=0;
316 cc.setSub=-1;
317 }
318
319 // remove the equal sign and subString from s
320 s.truncate(index);
321 }
322
323 s.extract(0, 0x7fffffff, cbopt, sizeof(cbopt), "");
324 cc.cbopt=cbopt;
325 switch(cbopt[0]) {
326 case SUB_CB:
327 callback=UCNV_FROM_U_CALLBACK_SUBSTITUTE;
328 break;
329 case SKIP_CB:
330 callback=UCNV_FROM_U_CALLBACK_SKIP;
331 break;
332 case STOP_CB:
333 callback=UCNV_FROM_U_CALLBACK_STOP;
334 break;
335 case ESC_CB:
336 callback=UCNV_FROM_U_CALLBACK_ESCAPE;
337 break;
338 default:
339 callback=NULL;
340 break;
341 }
342 option=callback==NULL ? cbopt : cbopt+1;
343 if(*option==0) {
344 option=NULL;
345 }
346
347 invalidUChars=testCase->getString("invalidUChars", errorCode);
348 cc.invalidUChars=invalidUChars.getBuffer();
349 cc.invalidLength=invalidUChars.length();
350
351 if(U_FAILURE(errorCode)) {
352 errln("error parsing conversion/fromUnicode test case %d - %s",
353 i, u_errorName(errorCode));
354 errorCode=U_ZERO_ERROR;
355 } else {
356 logln("TestFromUnicode[%d] %s", i, charset);
357 FromUnicodeCase(cc, callback, option);
358 }
359 }
360 delete testData;
361 }
362 delete dataModule;
363 }
364 else {
365 dataerrln("Could not load test conversion data");
366 }
367 }
368
369 static const UChar ellipsis[]={ 0x2e, 0x2e, 0x2e };
370
371 void
TestGetUnicodeSet()372 ConversionTest::TestGetUnicodeSet() {
373 char charset[100];
374 UnicodeString s, map, mapnot;
375 int32_t which;
376
377 ParsePosition pos;
378 UnicodeSet cnvSet, mapSet, mapnotSet, diffSet;
379 UnicodeSet *cnvSetPtr = &cnvSet;
380 LocalUConverterPointer cnv;
381
382 TestDataModule *dataModule;
383 TestData *testData;
384 const DataMap *testCase;
385 UErrorCode errorCode;
386 int32_t i;
387
388 errorCode=U_ZERO_ERROR;
389 dataModule=TestDataModule::getTestDataModule("conversion", *this, errorCode);
390 if(U_SUCCESS(errorCode)) {
391 testData=dataModule->createTestData("getUnicodeSet", errorCode);
392 if(U_SUCCESS(errorCode)) {
393 for(i=0; testData->nextCase(testCase, errorCode); ++i) {
394 if(U_FAILURE(errorCode)) {
395 errln("error retrieving conversion/getUnicodeSet test case %d - %s",
396 i, u_errorName(errorCode));
397 errorCode=U_ZERO_ERROR;
398 continue;
399 }
400
401 s=testCase->getString("charset", errorCode);
402 s.extract(0, 0x7fffffff, charset, sizeof(charset), "");
403
404 // BEGIN android-added
405 // To save space, Android does not build full ISO-2022 tables.
406 // We skip the TestGetKeywordValuesForLocale for counting available collations.
407 if (strlen(charset) >= 8 &&
408 strncmp(charset+4, "2022", 4) == 0) {
409 continue;
410 }
411 // END android-added
412
413 map=testCase->getString("map", errorCode);
414 mapnot=testCase->getString("mapnot", errorCode);
415
416 which=testCase->getInt28("which", errorCode);
417
418 if(U_FAILURE(errorCode)) {
419 errln("error parsing conversion/getUnicodeSet test case %d - %s",
420 i, u_errorName(errorCode));
421 errorCode=U_ZERO_ERROR;
422 continue;
423 }
424
425 // test this test case
426 mapSet.clear();
427 mapnotSet.clear();
428
429 pos.setIndex(0);
430 mapSet.applyPattern(map, pos, 0, NULL, errorCode);
431 if(U_FAILURE(errorCode) || pos.getIndex()!=map.length()) {
432 errln("error creating the map set for conversion/getUnicodeSet test case %d - %s\n"
433 " error index %d index %d U+%04x",
434 i, u_errorName(errorCode), pos.getErrorIndex(), pos.getIndex(), map.char32At(pos.getIndex()));
435 errorCode=U_ZERO_ERROR;
436 continue;
437 }
438
439 pos.setIndex(0);
440 mapnotSet.applyPattern(mapnot, pos, 0, NULL, errorCode);
441 if(U_FAILURE(errorCode) || pos.getIndex()!=mapnot.length()) {
442 errln("error creating the mapnot set for conversion/getUnicodeSet test case %d - %s\n"
443 " error index %d index %d U+%04x",
444 i, u_errorName(errorCode), pos.getErrorIndex(), pos.getIndex(), mapnot.char32At(pos.getIndex()));
445 errorCode=U_ZERO_ERROR;
446 continue;
447 }
448
449 logln("TestGetUnicodeSet[%d] %s", i, charset);
450
451 cnv.adoptInstead(cnv_open(charset, errorCode));
452 if(U_FAILURE(errorCode)) {
453 errcheckln(errorCode, "error opening \"%s\" for conversion/getUnicodeSet test case %d - %s",
454 charset, i, u_errorName(errorCode));
455 errorCode=U_ZERO_ERROR;
456 continue;
457 }
458
459 ucnv_getUnicodeSet(cnv.getAlias(), cnvSetPtr->toUSet(), (UConverterUnicodeSet)which, &errorCode);
460
461 if(U_FAILURE(errorCode)) {
462 errln("error in ucnv_getUnicodeSet(\"%s\") for conversion/getUnicodeSet test case %d - %s",
463 charset, i, u_errorName(errorCode));
464 errorCode=U_ZERO_ERROR;
465 continue;
466 }
467
468 // are there items that must be in cnvSet but are not?
469 (diffSet=mapSet).removeAll(cnvSet);
470 if(!diffSet.isEmpty()) {
471 diffSet.toPattern(s, TRUE);
472 if(s.length()>100) {
473 s.replace(100, 0x7fffffff, ellipsis, LENGTHOF(ellipsis));
474 }
475 errln("error: ucnv_getUnicodeSet(\"%s\") is missing items - conversion/getUnicodeSet test case %d",
476 charset, i);
477 errln(s);
478 }
479
480 // are there items that must not be in cnvSet but are?
481 (diffSet=mapnotSet).retainAll(cnvSet);
482 if(!diffSet.isEmpty()) {
483 diffSet.toPattern(s, TRUE);
484 if(s.length()>100) {
485 s.replace(100, 0x7fffffff, ellipsis, LENGTHOF(ellipsis));
486 }
487 errln("error: ucnv_getUnicodeSet(\"%s\") contains unexpected items - conversion/getUnicodeSet test case %d",
488 charset, i);
489 errln(s);
490 }
491 }
492 delete testData;
493 }
494 delete dataModule;
495 }
496 else {
497 dataerrln("Could not load test conversion data");
498 }
499 }
500
501 U_CDECL_BEGIN
502 static void U_CALLCONV
getUnicodeSetCallback(const void * context,UConverterFromUnicodeArgs *,const UChar *,int32_t,UChar32 codePoint,UConverterCallbackReason reason,UErrorCode * pErrorCode)503 getUnicodeSetCallback(const void *context,
504 UConverterFromUnicodeArgs * /*fromUArgs*/,
505 const UChar* /*codeUnits*/,
506 int32_t /*length*/,
507 UChar32 codePoint,
508 UConverterCallbackReason reason,
509 UErrorCode *pErrorCode) {
510 if(reason<=UCNV_IRREGULAR) {
511 ((UnicodeSet *)context)->remove(codePoint); // the converter cannot convert this code point
512 *pErrorCode=U_ZERO_ERROR; // skip
513 } // else ignore the reset, close and clone calls.
514 }
515 U_CDECL_END
516
517 // Compare ucnv_getUnicodeSet() with the set of characters that can be converted.
518 void
TestGetUnicodeSet2()519 ConversionTest::TestGetUnicodeSet2() {
520 // Build a string with all code points.
521 UChar32 cpLimit;
522 int32_t s0Length;
523 if(quick) {
524 cpLimit=s0Length=0x10000; // BMP only
525 } else {
526 cpLimit=0x110000;
527 s0Length=0x10000+0x200000; // BMP + surrogate pairs
528 }
529 UChar *s0=new UChar[s0Length];
530 if(s0==NULL) {
531 return;
532 }
533 UChar *s=s0;
534 UChar32 c;
535 UChar c2;
536 // low BMP
537 for(c=0; c<=0xd7ff; ++c) {
538 *s++=(UChar)c;
539 }
540 // trail surrogates
541 for(c=0xdc00; c<=0xdfff; ++c) {
542 *s++=(UChar)c;
543 }
544 // lead surrogates
545 // (after trails so that there is not even one surrogate pair in between)
546 for(c=0xd800; c<=0xdbff; ++c) {
547 *s++=(UChar)c;
548 }
549 // high BMP
550 for(c=0xe000; c<=0xffff; ++c) {
551 *s++=(UChar)c;
552 }
553 // supplementary code points = surrogate pairs
554 if(cpLimit==0x110000) {
555 for(c=0xd800; c<=0xdbff; ++c) {
556 for(c2=0xdc00; c2<=0xdfff; ++c2) {
557 *s++=(UChar)c;
558 *s++=c2;
559 }
560 }
561 }
562
563 static const char *const cnvNames[]={
564 "UTF-8",
565 "UTF-7",
566 "UTF-16",
567 "US-ASCII",
568 "ISO-8859-1",
569 "windows-1252",
570 "Shift-JIS",
571 "ibm-1390", // EBCDIC_STATEFUL table
572 "ibm-16684", // DBCS-only extension table based on EBCDIC_STATEFUL table
573 "HZ",
574 "ISO-2022-JP",
575 "JIS7",
576 "ISO-2022-CN",
577 "ISO-2022-CN-EXT",
578 "LMBCS"
579 };
580 LocalUConverterPointer cnv;
581 char buffer[1024];
582 int32_t i;
583 for(i=0; i<LENGTHOF(cnvNames); ++i) {
584 UErrorCode errorCode=U_ZERO_ERROR;
585 cnv.adoptInstead(cnv_open(cnvNames[i], errorCode));
586 if(U_FAILURE(errorCode)) {
587 errcheckln(errorCode, "failed to open converter %s - %s", cnvNames[i], u_errorName(errorCode));
588 continue;
589 }
590 UnicodeSet expected;
591 ucnv_setFromUCallBack(cnv.getAlias(), getUnicodeSetCallback, &expected, NULL, NULL, &errorCode);
592 if(U_FAILURE(errorCode)) {
593 errln("failed to set the callback on converter %s - %s", cnvNames[i], u_errorName(errorCode));
594 continue;
595 }
596 UConverterUnicodeSet which;
597 for(which=UCNV_ROUNDTRIP_SET; which<UCNV_SET_COUNT; which=(UConverterUnicodeSet)((int)which+1)) {
598 if(which==UCNV_ROUNDTRIP_AND_FALLBACK_SET) {
599 ucnv_setFallback(cnv.getAlias(), TRUE);
600 }
601 expected.add(0, cpLimit-1);
602 s=s0;
603 UBool flush;
604 do {
605 char *t=buffer;
606 flush=(UBool)(s==s0+s0Length);
607 ucnv_fromUnicode(cnv.getAlias(), &t, buffer+sizeof(buffer), (const UChar **)&s, s0+s0Length, NULL, flush, &errorCode);
608 if(U_FAILURE(errorCode)) {
609 if(errorCode==U_BUFFER_OVERFLOW_ERROR) {
610 errorCode=U_ZERO_ERROR;
611 continue;
612 } else {
613 break; // unexpected error, should not occur
614 }
615 }
616 } while(!flush);
617 UnicodeSet set;
618 ucnv_getUnicodeSet(cnv.getAlias(), set.toUSet(), which, &errorCode);
619 if(cpLimit<0x110000) {
620 set.remove(cpLimit, 0x10ffff);
621 }
622 if(which==UCNV_ROUNDTRIP_SET) {
623 // ignore PUA code points because they will be converted even if they
624 // are fallbacks and when other fallbacks are turned off,
625 // but ucnv_getUnicodeSet(UCNV_ROUNDTRIP_SET) delivers true roundtrips
626 expected.remove(0xe000, 0xf8ff);
627 expected.remove(0xf0000, 0xffffd);
628 expected.remove(0x100000, 0x10fffd);
629 set.remove(0xe000, 0xf8ff);
630 set.remove(0xf0000, 0xffffd);
631 set.remove(0x100000, 0x10fffd);
632 }
633 if(set!=expected) {
634 // First try to see if we have different sets because ucnv_getUnicodeSet()
635 // added strings: The above conversion method does not tell us what strings might be convertible.
636 // Remove strings from the set and compare again.
637 // Unfortunately, there are no good, direct set methods for finding out whether there are strings
638 // in the set, nor for enumerating or removing just them.
639 // Intersect all code points with the set. The intersection will not contain strings.
640 UnicodeSet temp(0, 0x10ffff);
641 temp.retainAll(set);
642 set=temp;
643 }
644 if(set!=expected) {
645 UnicodeSet diffSet;
646 UnicodeString out;
647
648 // are there items that must be in the set but are not?
649 (diffSet=expected).removeAll(set);
650 if(!diffSet.isEmpty()) {
651 diffSet.toPattern(out, TRUE);
652 if(out.length()>100) {
653 out.replace(100, 0x7fffffff, ellipsis, LENGTHOF(ellipsis));
654 }
655 errln("error: ucnv_getUnicodeSet(\"%s\") is missing items - which set: %d",
656 cnvNames[i], which);
657 errln(out);
658 }
659
660 // are there items that must not be in the set but are?
661 (diffSet=set).removeAll(expected);
662 if(!diffSet.isEmpty()) {
663 diffSet.toPattern(out, TRUE);
664 if(out.length()>100) {
665 out.replace(100, 0x7fffffff, ellipsis, LENGTHOF(ellipsis));
666 }
667 errln("error: ucnv_getUnicodeSet(\"%s\") contains unexpected items - which set: %d",
668 cnvNames[i], which);
669 errln(out);
670 }
671 }
672 }
673 }
674
675 delete [] s0;
676 }
677
678 // open testdata or ICU data converter ------------------------------------- ***
679
680 UConverter *
cnv_open(const char * name,UErrorCode & errorCode)681 ConversionTest::cnv_open(const char *name, UErrorCode &errorCode) {
682 if(name!=NULL && *name=='*') {
683 /* loadTestData(): set the data directory */
684 return ucnv_openPackage(loadTestData(errorCode), name+1, &errorCode);
685 } else if(name!=NULL && *name=='+') {
686 return ucnv_open((name+1), &errorCode);
687 } else {
688 return ucnv_open(name, &errorCode);
689 }
690 }
691
692 // output helpers ---------------------------------------------------------- ***
693
694 static inline char
hexDigit(uint8_t digit)695 hexDigit(uint8_t digit) {
696 return digit<=9 ? (char)('0'+digit) : (char)('a'-10+digit);
697 }
698
699 static char *
printBytes(const uint8_t * bytes,int32_t length,char * out)700 printBytes(const uint8_t *bytes, int32_t length, char *out) {
701 uint8_t b;
702
703 if(length>0) {
704 b=*bytes++;
705 --length;
706 *out++=hexDigit((uint8_t)(b>>4));
707 *out++=hexDigit((uint8_t)(b&0xf));
708 }
709
710 while(length>0) {
711 b=*bytes++;
712 --length;
713 *out++=' ';
714 *out++=hexDigit((uint8_t)(b>>4));
715 *out++=hexDigit((uint8_t)(b&0xf));
716 }
717 *out++=0;
718 return out;
719 }
720
721 static char *
printUnicode(const UChar * unicode,int32_t length,char * out)722 printUnicode(const UChar *unicode, int32_t length, char *out) {
723 UChar32 c;
724 int32_t i;
725
726 for(i=0; i<length;) {
727 if(i>0) {
728 *out++=' ';
729 }
730 U16_NEXT(unicode, i, length, c);
731 // write 4..6 digits
732 if(c>=0x100000) {
733 *out++='1';
734 }
735 if(c>=0x10000) {
736 *out++=hexDigit((uint8_t)((c>>16)&0xf));
737 }
738 *out++=hexDigit((uint8_t)((c>>12)&0xf));
739 *out++=hexDigit((uint8_t)((c>>8)&0xf));
740 *out++=hexDigit((uint8_t)((c>>4)&0xf));
741 *out++=hexDigit((uint8_t)(c&0xf));
742 }
743 *out++=0;
744 return out;
745 }
746
747 static char *
printOffsets(const int32_t * offsets,int32_t length,char * out)748 printOffsets(const int32_t *offsets, int32_t length, char *out) {
749 int32_t i, o, d;
750
751 if(offsets==NULL) {
752 length=0;
753 }
754
755 for(i=0; i<length; ++i) {
756 if(i>0) {
757 *out++=' ';
758 }
759 o=offsets[i];
760
761 // print all offsets with 2 characters each (-x, -9..99, xx)
762 if(o<-9) {
763 *out++='-';
764 *out++='x';
765 } else if(o<0) {
766 *out++='-';
767 *out++=(char)('0'-o);
768 } else if(o<=99) {
769 *out++=(d=o/10)==0 ? ' ' : (char)('0'+d);
770 *out++=(char)('0'+o%10);
771 } else /* o>99 */ {
772 *out++='x';
773 *out++='x';
774 }
775 }
776 *out++=0;
777 return out;
778 }
779
780 // toUnicode test worker functions ----------------------------------------- ***
781
782 static int32_t
stepToUnicode(ConversionCase & cc,UConverter * cnv,UChar * result,int32_t resultCapacity,int32_t * resultOffsets,int32_t step,UErrorCode * pErrorCode)783 stepToUnicode(ConversionCase &cc, UConverter *cnv,
784 UChar *result, int32_t resultCapacity,
785 int32_t *resultOffsets, /* also resultCapacity */
786 int32_t step,
787 UErrorCode *pErrorCode) {
788 const char *source, *sourceLimit, *bytesLimit;
789 UChar *target, *targetLimit, *resultLimit;
790 UBool flush;
791
792 source=(const char *)cc.bytes;
793 target=result;
794 bytesLimit=source+cc.bytesLength;
795 resultLimit=result+resultCapacity;
796
797 if(step>=0) {
798 // call ucnv_toUnicode() with in/out buffers no larger than (step) at a time
799 // move only one buffer (in vs. out) at a time to be extra mean
800 // step==0 performs bulk conversion and generates offsets
801
802 // initialize the partial limits for the loop
803 if(step==0) {
804 // use the entire buffers
805 sourceLimit=bytesLimit;
806 targetLimit=resultLimit;
807 flush=cc.finalFlush;
808 } else {
809 // start with empty partial buffers
810 sourceLimit=source;
811 targetLimit=target;
812 flush=FALSE;
813
814 // output offsets only for bulk conversion
815 resultOffsets=NULL;
816 }
817
818 for(;;) {
819 // resetting the opposite conversion direction must not affect this one
820 ucnv_resetFromUnicode(cnv);
821
822 // convert
823 ucnv_toUnicode(cnv,
824 &target, targetLimit,
825 &source, sourceLimit,
826 resultOffsets,
827 flush, pErrorCode);
828
829 // check pointers and errors
830 if(source>sourceLimit || target>targetLimit) {
831 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
832 break;
833 } else if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
834 if(target!=targetLimit) {
835 // buffer overflow must only be set when the target is filled
836 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
837 break;
838 } else if(targetLimit==resultLimit) {
839 // not just a partial overflow
840 break;
841 }
842
843 // the partial target is filled, set a new limit, reset the error and continue
844 targetLimit=(resultLimit-target)>=step ? target+step : resultLimit;
845 *pErrorCode=U_ZERO_ERROR;
846 } else if(U_FAILURE(*pErrorCode)) {
847 // some other error occurred, done
848 break;
849 } else {
850 if(source!=sourceLimit) {
851 // when no error occurs, then the input must be consumed
852 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
853 break;
854 }
855
856 if(sourceLimit==bytesLimit) {
857 // we are done
858 break;
859 }
860
861 // the partial conversion succeeded, set a new limit and continue
862 sourceLimit=(bytesLimit-source)>=step ? source+step : bytesLimit;
863 flush=(UBool)(cc.finalFlush && sourceLimit==bytesLimit);
864 }
865 }
866 } else /* step<0 */ {
867 /*
868 * step==-1: call only ucnv_getNextUChar()
869 * otherwise alternate between ucnv_toUnicode() and ucnv_getNextUChar()
870 * if step==-2 or -3, then give ucnv_toUnicode() the whole remaining input,
871 * else give it at most (-step-2)/2 bytes
872 */
873 UChar32 c;
874
875 // end the loop by getting an index out of bounds error
876 for(;;) {
877 // resetting the opposite conversion direction must not affect this one
878 ucnv_resetFromUnicode(cnv);
879
880 // convert
881 if((step&1)!=0 /* odd: -1, -3, -5, ... */) {
882 sourceLimit=source; // use sourceLimit not as a real limit
883 // but to remember the pre-getNextUChar source pointer
884 c=ucnv_getNextUChar(cnv, &source, bytesLimit, pErrorCode);
885
886 // check pointers and errors
887 if(*pErrorCode==U_INDEX_OUTOFBOUNDS_ERROR) {
888 if(source!=bytesLimit) {
889 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
890 } else {
891 *pErrorCode=U_ZERO_ERROR;
892 }
893 break;
894 } else if(U_FAILURE(*pErrorCode)) {
895 break;
896 }
897 // source may not move if c is from previous overflow
898
899 if(target==resultLimit) {
900 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
901 break;
902 }
903 if(c<=0xffff) {
904 *target++=(UChar)c;
905 } else {
906 *target++=U16_LEAD(c);
907 if(target==resultLimit) {
908 *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
909 break;
910 }
911 *target++=U16_TRAIL(c);
912 }
913
914 // alternate between -n-1 and -n but leave -1 alone
915 if(step<-1) {
916 ++step;
917 }
918 } else /* step is even */ {
919 // allow only one UChar output
920 targetLimit=target<resultLimit ? target+1 : resultLimit;
921
922 // as with ucnv_getNextUChar(), we always flush (if we go to bytesLimit)
923 // and never output offsets
924 if(step==-2) {
925 sourceLimit=bytesLimit;
926 } else {
927 sourceLimit=source+(-step-2)/2;
928 if(sourceLimit>bytesLimit) {
929 sourceLimit=bytesLimit;
930 }
931 }
932
933 ucnv_toUnicode(cnv,
934 &target, targetLimit,
935 &source, sourceLimit,
936 NULL, (UBool)(sourceLimit==bytesLimit), pErrorCode);
937
938 // check pointers and errors
939 if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
940 if(target!=targetLimit) {
941 // buffer overflow must only be set when the target is filled
942 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
943 break;
944 } else if(targetLimit==resultLimit) {
945 // not just a partial overflow
946 break;
947 }
948
949 // the partial target is filled, set a new limit and continue
950 *pErrorCode=U_ZERO_ERROR;
951 } else if(U_FAILURE(*pErrorCode)) {
952 // some other error occurred, done
953 break;
954 } else {
955 if(source!=sourceLimit) {
956 // when no error occurs, then the input must be consumed
957 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
958 break;
959 }
960
961 // we are done (flush==TRUE) but we continue, to get the index out of bounds error above
962 }
963
964 --step;
965 }
966 }
967 }
968
969 return (int32_t)(target-result);
970 }
971
972 UBool
ToUnicodeCase(ConversionCase & cc,UConverterToUCallback callback,const char * option)973 ConversionTest::ToUnicodeCase(ConversionCase &cc, UConverterToUCallback callback, const char *option) {
974 // open the converter
975 IcuTestErrorCode errorCode(*this, "ToUnicodeCase");
976 LocalUConverterPointer cnv(cnv_open(cc.charset, errorCode));
977 if(errorCode.isFailure()) {
978 errcheckln(errorCode, "toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_open() failed - %s",
979 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, errorCode.errorName());
980 errorCode.reset();
981 return FALSE;
982 }
983
984 // set the callback
985 if(callback!=NULL) {
986 ucnv_setToUCallBack(cnv.getAlias(), callback, option, NULL, NULL, errorCode);
987 if(U_FAILURE(errorCode)) {
988 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setToUCallBack() failed - %s",
989 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
990 return FALSE;
991 }
992 }
993
994 int32_t resultOffsets[256];
995 UChar result[256];
996 int32_t resultLength;
997 UBool ok;
998
999 static const struct {
1000 int32_t step;
1001 const char *name;
1002 } steps[]={
1003 { 0, "bulk" }, // must be first for offsets to be checked
1004 { 1, "step=1" },
1005 { 3, "step=3" },
1006 { 7, "step=7" },
1007 { -1, "getNext" },
1008 { -2, "toU(bulk)+getNext" },
1009 { -3, "getNext+toU(bulk)" },
1010 { -4, "toU(1)+getNext" },
1011 { -5, "getNext+toU(1)" },
1012 { -12, "toU(5)+getNext" },
1013 { -13, "getNext+toU(5)" },
1014 };
1015 int32_t i, step;
1016
1017 ok=TRUE;
1018 for(i=0; i<LENGTHOF(steps) && ok; ++i) {
1019 step=steps[i].step;
1020 if(step<0 && !cc.finalFlush) {
1021 // skip ucnv_getNextUChar() if !finalFlush because
1022 // ucnv_getNextUChar() always implies flush
1023 continue;
1024 }
1025 if(step!=0) {
1026 // bulk test is first, then offsets are not checked any more
1027 cc.offsets=NULL;
1028 }
1029 else {
1030 memset(resultOffsets, -1, LENGTHOF(resultOffsets));
1031 }
1032 memset(result, -1, LENGTHOF(result));
1033 errorCode.reset();
1034 resultLength=stepToUnicode(cc, cnv.getAlias(),
1035 result, LENGTHOF(result),
1036 step==0 ? resultOffsets : NULL,
1037 step, errorCode);
1038 ok=checkToUnicode(
1039 cc, cnv.getAlias(), steps[i].name,
1040 result, resultLength,
1041 cc.offsets!=NULL ? resultOffsets : NULL,
1042 errorCode);
1043 if(errorCode.isFailure() || !cc.finalFlush) {
1044 // reset if an error occurred or we did not flush
1045 // otherwise do nothing to make sure that flushing resets
1046 ucnv_resetToUnicode(cnv.getAlias());
1047 }
1048 if (cc.offsets != NULL && resultOffsets[resultLength] != -1) {
1049 errln("toUnicode[%d](%s) Conversion wrote too much to offsets at index %d",
1050 cc.caseNr, cc.charset, resultLength);
1051 }
1052 if (result[resultLength] != (UChar)-1) {
1053 errln("toUnicode[%d](%s) Conversion wrote too much to result at index %d",
1054 cc.caseNr, cc.charset, resultLength);
1055 }
1056 }
1057
1058 // not a real loop, just a convenience for breaking out of the block
1059 while(ok && cc.finalFlush) {
1060 // test ucnv_toUChars()
1061 memset(result, 0, sizeof(result));
1062
1063 errorCode.reset();
1064 resultLength=ucnv_toUChars(cnv.getAlias(),
1065 result, LENGTHOF(result),
1066 (const char *)cc.bytes, cc.bytesLength,
1067 errorCode);
1068 ok=checkToUnicode(
1069 cc, cnv.getAlias(), "toUChars",
1070 result, resultLength,
1071 NULL,
1072 errorCode);
1073 if(!ok) {
1074 break;
1075 }
1076
1077 // test preflighting
1078 // keep the correct result for simple checking
1079 errorCode.reset();
1080 resultLength=ucnv_toUChars(cnv.getAlias(),
1081 NULL, 0,
1082 (const char *)cc.bytes, cc.bytesLength,
1083 errorCode);
1084 if(errorCode.get()==U_STRING_NOT_TERMINATED_WARNING || errorCode.get()==U_BUFFER_OVERFLOW_ERROR) {
1085 errorCode.reset();
1086 }
1087 ok=checkToUnicode(
1088 cc, cnv.getAlias(), "preflight toUChars",
1089 result, resultLength,
1090 NULL,
1091 errorCode);
1092 break;
1093 }
1094
1095 errorCode.reset(); // all errors have already been reported
1096 return ok;
1097 }
1098
1099 UBool
checkToUnicode(ConversionCase & cc,UConverter * cnv,const char * name,const UChar * result,int32_t resultLength,const int32_t * resultOffsets,UErrorCode resultErrorCode)1100 ConversionTest::checkToUnicode(ConversionCase &cc, UConverter *cnv, const char *name,
1101 const UChar *result, int32_t resultLength,
1102 const int32_t *resultOffsets,
1103 UErrorCode resultErrorCode) {
1104 char resultInvalidChars[8];
1105 int8_t resultInvalidLength;
1106 UErrorCode errorCode;
1107
1108 const char *msg;
1109
1110 // reset the message; NULL will mean "ok"
1111 msg=NULL;
1112
1113 errorCode=U_ZERO_ERROR;
1114 resultInvalidLength=sizeof(resultInvalidChars);
1115 ucnv_getInvalidChars(cnv, resultInvalidChars, &resultInvalidLength, &errorCode);
1116 if(U_FAILURE(errorCode)) {
1117 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) ucnv_getInvalidChars() failed - %s",
1118 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, u_errorName(errorCode));
1119 return FALSE;
1120 }
1121
1122 // check everything that might have gone wrong
1123 if(cc.unicodeLength!=resultLength) {
1124 msg="wrong result length";
1125 } else if(0!=u_memcmp(cc.unicode, result, cc.unicodeLength)) {
1126 msg="wrong result string";
1127 } else if(cc.offsets!=NULL && 0!=memcmp(cc.offsets, resultOffsets, cc.unicodeLength*sizeof(*cc.offsets))) {
1128 msg="wrong offsets";
1129 } else if(cc.outErrorCode!=resultErrorCode) {
1130 msg="wrong error code";
1131 } else if(cc.invalidLength!=resultInvalidLength) {
1132 msg="wrong length of last invalid input";
1133 } else if(0!=memcmp(cc.invalidChars, resultInvalidChars, cc.invalidLength)) {
1134 msg="wrong last invalid input";
1135 }
1136
1137 if(msg==NULL) {
1138 return TRUE;
1139 } else {
1140 char buffer[2000]; // one buffer for all strings
1141 char *s, *bytesString, *unicodeString, *resultString,
1142 *offsetsString, *resultOffsetsString,
1143 *invalidCharsString, *resultInvalidCharsString;
1144
1145 bytesString=s=buffer;
1146 s=printBytes(cc.bytes, cc.bytesLength, bytesString);
1147 s=printUnicode(cc.unicode, cc.unicodeLength, unicodeString=s);
1148 s=printUnicode(result, resultLength, resultString=s);
1149 s=printOffsets(cc.offsets, cc.unicodeLength, offsetsString=s);
1150 s=printOffsets(resultOffsets, resultLength, resultOffsetsString=s);
1151 s=printBytes(cc.invalidChars, cc.invalidLength, invalidCharsString=s);
1152 s=printBytes((uint8_t *)resultInvalidChars, resultInvalidLength, resultInvalidCharsString=s);
1153
1154 if((s-buffer)>(int32_t)sizeof(buffer)) {
1155 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) fatal error: checkToUnicode() test output buffer overflow writing %d chars\n",
1156 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, (int)(s-buffer));
1157 exit(1);
1158 }
1159
1160 errln("toUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) failed: %s\n"
1161 " bytes <%s>[%d]\n"
1162 " expected <%s>[%d]\n"
1163 " result <%s>[%d]\n"
1164 " offsets <%s>\n"
1165 " result offsets <%s>\n"
1166 " error code expected %s got %s\n"
1167 " invalidChars expected <%s> got <%s>\n",
1168 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, msg,
1169 bytesString, cc.bytesLength,
1170 unicodeString, cc.unicodeLength,
1171 resultString, resultLength,
1172 offsetsString,
1173 resultOffsetsString,
1174 u_errorName(cc.outErrorCode), u_errorName(resultErrorCode),
1175 invalidCharsString, resultInvalidCharsString);
1176
1177 return FALSE;
1178 }
1179 }
1180
1181 // fromUnicode test worker functions --------------------------------------- ***
1182
1183 static int32_t
stepFromUTF8(ConversionCase & cc,UConverter * utf8Cnv,UConverter * cnv,char * result,int32_t resultCapacity,int32_t step,UErrorCode * pErrorCode)1184 stepFromUTF8(ConversionCase &cc,
1185 UConverter *utf8Cnv, UConverter *cnv,
1186 char *result, int32_t resultCapacity,
1187 int32_t step,
1188 UErrorCode *pErrorCode) {
1189 const char *source, *sourceLimit, *utf8Limit;
1190 UChar pivotBuffer[32];
1191 UChar *pivotSource, *pivotTarget, *pivotLimit;
1192 char *target, *targetLimit, *resultLimit;
1193 UBool flush;
1194
1195 source=cc.utf8;
1196 pivotSource=pivotTarget=pivotBuffer;
1197 target=result;
1198 utf8Limit=source+cc.utf8Length;
1199 resultLimit=result+resultCapacity;
1200
1201 // call ucnv_convertEx() with in/out buffers no larger than (step) at a time
1202 // move only one buffer (in vs. out) at a time to be extra mean
1203 // step==0 performs bulk conversion
1204
1205 // initialize the partial limits for the loop
1206 if(step==0) {
1207 // use the entire buffers
1208 sourceLimit=utf8Limit;
1209 targetLimit=resultLimit;
1210 flush=cc.finalFlush;
1211
1212 pivotLimit=pivotBuffer+LENGTHOF(pivotBuffer);
1213 } else {
1214 // start with empty partial buffers
1215 sourceLimit=source;
1216 targetLimit=target;
1217 flush=FALSE;
1218
1219 // empty pivot is not allowed, make it of length step
1220 pivotLimit=pivotBuffer+step;
1221 }
1222
1223 for(;;) {
1224 // resetting the opposite conversion direction must not affect this one
1225 ucnv_resetFromUnicode(utf8Cnv);
1226 ucnv_resetToUnicode(cnv);
1227
1228 // convert
1229 ucnv_convertEx(cnv, utf8Cnv,
1230 &target, targetLimit,
1231 &source, sourceLimit,
1232 pivotBuffer, &pivotSource, &pivotTarget, pivotLimit,
1233 FALSE, flush, pErrorCode);
1234
1235 // check pointers and errors
1236 if(source>sourceLimit || target>targetLimit) {
1237 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1238 break;
1239 } else if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
1240 if(target!=targetLimit) {
1241 // buffer overflow must only be set when the target is filled
1242 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1243 break;
1244 } else if(targetLimit==resultLimit) {
1245 // not just a partial overflow
1246 break;
1247 }
1248
1249 // the partial target is filled, set a new limit, reset the error and continue
1250 targetLimit=(resultLimit-target)>=step ? target+step : resultLimit;
1251 *pErrorCode=U_ZERO_ERROR;
1252 } else if(U_FAILURE(*pErrorCode)) {
1253 if(pivotSource==pivotBuffer) {
1254 // toUnicode error, should not occur
1255 // toUnicode errors are tested in cintltst TestConvertExFromUTF8()
1256 break;
1257 } else {
1258 // fromUnicode error
1259 // some other error occurred, done
1260 break;
1261 }
1262 } else {
1263 if(source!=sourceLimit) {
1264 // when no error occurs, then the input must be consumed
1265 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1266 break;
1267 }
1268
1269 if(sourceLimit==utf8Limit) {
1270 // we are done
1271 if(*pErrorCode==U_STRING_NOT_TERMINATED_WARNING) {
1272 // ucnv_convertEx() warns about not terminating the output
1273 // but ucnv_fromUnicode() does not and so
1274 // checkFromUnicode() does not expect it
1275 *pErrorCode=U_ZERO_ERROR;
1276 }
1277 break;
1278 }
1279
1280 // the partial conversion succeeded, set a new limit and continue
1281 sourceLimit=(utf8Limit-source)>=step ? source+step : utf8Limit;
1282 flush=(UBool)(cc.finalFlush && sourceLimit==utf8Limit);
1283 }
1284 }
1285
1286 return (int32_t)(target-result);
1287 }
1288
1289 static int32_t
stepFromUnicode(ConversionCase & cc,UConverter * cnv,char * result,int32_t resultCapacity,int32_t * resultOffsets,int32_t step,UErrorCode * pErrorCode)1290 stepFromUnicode(ConversionCase &cc, UConverter *cnv,
1291 char *result, int32_t resultCapacity,
1292 int32_t *resultOffsets, /* also resultCapacity */
1293 int32_t step,
1294 UErrorCode *pErrorCode) {
1295 const UChar *source, *sourceLimit, *unicodeLimit;
1296 char *target, *targetLimit, *resultLimit;
1297 UBool flush;
1298
1299 source=cc.unicode;
1300 target=result;
1301 unicodeLimit=source+cc.unicodeLength;
1302 resultLimit=result+resultCapacity;
1303
1304 // call ucnv_fromUnicode() with in/out buffers no larger than (step) at a time
1305 // move only one buffer (in vs. out) at a time to be extra mean
1306 // step==0 performs bulk conversion and generates offsets
1307
1308 // initialize the partial limits for the loop
1309 if(step==0) {
1310 // use the entire buffers
1311 sourceLimit=unicodeLimit;
1312 targetLimit=resultLimit;
1313 flush=cc.finalFlush;
1314 } else {
1315 // start with empty partial buffers
1316 sourceLimit=source;
1317 targetLimit=target;
1318 flush=FALSE;
1319
1320 // output offsets only for bulk conversion
1321 resultOffsets=NULL;
1322 }
1323
1324 for(;;) {
1325 // resetting the opposite conversion direction must not affect this one
1326 ucnv_resetToUnicode(cnv);
1327
1328 // convert
1329 ucnv_fromUnicode(cnv,
1330 &target, targetLimit,
1331 &source, sourceLimit,
1332 resultOffsets,
1333 flush, pErrorCode);
1334
1335 // check pointers and errors
1336 if(source>sourceLimit || target>targetLimit) {
1337 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1338 break;
1339 } else if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR) {
1340 if(target!=targetLimit) {
1341 // buffer overflow must only be set when the target is filled
1342 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1343 break;
1344 } else if(targetLimit==resultLimit) {
1345 // not just a partial overflow
1346 break;
1347 }
1348
1349 // the partial target is filled, set a new limit, reset the error and continue
1350 targetLimit=(resultLimit-target)>=step ? target+step : resultLimit;
1351 *pErrorCode=U_ZERO_ERROR;
1352 } else if(U_FAILURE(*pErrorCode)) {
1353 // some other error occurred, done
1354 break;
1355 } else {
1356 if(source!=sourceLimit) {
1357 // when no error occurs, then the input must be consumed
1358 *pErrorCode=U_INTERNAL_PROGRAM_ERROR;
1359 break;
1360 }
1361
1362 if(sourceLimit==unicodeLimit) {
1363 // we are done
1364 break;
1365 }
1366
1367 // the partial conversion succeeded, set a new limit and continue
1368 sourceLimit=(unicodeLimit-source)>=step ? source+step : unicodeLimit;
1369 flush=(UBool)(cc.finalFlush && sourceLimit==unicodeLimit);
1370 }
1371 }
1372
1373 return (int32_t)(target-result);
1374 }
1375
1376 UBool
FromUnicodeCase(ConversionCase & cc,UConverterFromUCallback callback,const char * option)1377 ConversionTest::FromUnicodeCase(ConversionCase &cc, UConverterFromUCallback callback, const char *option) {
1378 UConverter *cnv;
1379 UErrorCode errorCode;
1380
1381 // open the converter
1382 errorCode=U_ZERO_ERROR;
1383 cnv=cnv_open(cc.charset, errorCode);
1384 if(U_FAILURE(errorCode)) {
1385 errcheckln(errorCode, "fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_open() failed - %s",
1386 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1387 return FALSE;
1388 }
1389 ucnv_resetToUnicode(utf8Cnv);
1390
1391 // set the callback
1392 if(callback!=NULL) {
1393 ucnv_setFromUCallBack(cnv, callback, option, NULL, NULL, &errorCode);
1394 if(U_FAILURE(errorCode)) {
1395 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setFromUCallBack() failed - %s",
1396 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1397 ucnv_close(cnv);
1398 return FALSE;
1399 }
1400 }
1401
1402 // set the fallbacks flag
1403 // TODO change with Jitterbug 2401, then add a similar call for toUnicode too
1404 ucnv_setFallback(cnv, cc.fallbacks);
1405
1406 // set the subchar
1407 int32_t length;
1408
1409 if(cc.setSub>0) {
1410 length=(int32_t)strlen(cc.subchar);
1411 ucnv_setSubstChars(cnv, cc.subchar, (int8_t)length, &errorCode);
1412 if(U_FAILURE(errorCode)) {
1413 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setSubstChars() failed - %s",
1414 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1415 ucnv_close(cnv);
1416 return FALSE;
1417 }
1418 } else if(cc.setSub<0) {
1419 ucnv_setSubstString(cnv, cc.subString, -1, &errorCode);
1420 if(U_FAILURE(errorCode)) {
1421 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d) ucnv_setSubstString() failed - %s",
1422 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, u_errorName(errorCode));
1423 ucnv_close(cnv);
1424 return FALSE;
1425 }
1426 }
1427
1428 // convert unicode to utf8
1429 char utf8[256];
1430 cc.utf8=utf8;
1431 u_strToUTF8(utf8, LENGTHOF(utf8), &cc.utf8Length,
1432 cc.unicode, cc.unicodeLength,
1433 &errorCode);
1434 if(U_FAILURE(errorCode)) {
1435 // skip UTF-8 testing of a string with an unpaired surrogate,
1436 // or of one that's too long
1437 // toUnicode errors are tested in cintltst TestConvertExFromUTF8()
1438 cc.utf8Length=-1;
1439 }
1440
1441 int32_t resultOffsets[256];
1442 char result[256];
1443 int32_t resultLength;
1444 UBool ok;
1445
1446 static const struct {
1447 int32_t step;
1448 const char *name, *utf8Name;
1449 } steps[]={
1450 { 0, "bulk", "utf8" }, // must be first for offsets to be checked
1451 { 1, "step=1", "utf8 step=1" },
1452 { 3, "step=3", "utf8 step=3" },
1453 { 7, "step=7", "utf8 step=7" }
1454 };
1455 int32_t i, step;
1456
1457 ok=TRUE;
1458 for(i=0; i<LENGTHOF(steps) && ok; ++i) {
1459 step=steps[i].step;
1460 memset(resultOffsets, -1, LENGTHOF(resultOffsets));
1461 memset(result, -1, LENGTHOF(result));
1462 errorCode=U_ZERO_ERROR;
1463 resultLength=stepFromUnicode(cc, cnv,
1464 result, LENGTHOF(result),
1465 step==0 ? resultOffsets : NULL,
1466 step, &errorCode);
1467 ok=checkFromUnicode(
1468 cc, cnv, steps[i].name,
1469 (uint8_t *)result, resultLength,
1470 cc.offsets!=NULL ? resultOffsets : NULL,
1471 errorCode);
1472 if(U_FAILURE(errorCode) || !cc.finalFlush) {
1473 // reset if an error occurred or we did not flush
1474 // otherwise do nothing to make sure that flushing resets
1475 ucnv_resetFromUnicode(cnv);
1476 }
1477 if (resultOffsets[resultLength] != -1) {
1478 errln("fromUnicode[%d](%s) Conversion wrote too much to offsets at index %d",
1479 cc.caseNr, cc.charset, resultLength);
1480 }
1481 if (result[resultLength] != (char)-1) {
1482 errln("fromUnicode[%d](%s) Conversion wrote too much to result at index %d",
1483 cc.caseNr, cc.charset, resultLength);
1484 }
1485
1486 // bulk test is first, then offsets are not checked any more
1487 cc.offsets=NULL;
1488
1489 // test direct conversion from UTF-8
1490 if(cc.utf8Length>=0) {
1491 errorCode=U_ZERO_ERROR;
1492 resultLength=stepFromUTF8(cc, utf8Cnv, cnv,
1493 result, LENGTHOF(result),
1494 step, &errorCode);
1495 ok=checkFromUnicode(
1496 cc, cnv, steps[i].utf8Name,
1497 (uint8_t *)result, resultLength,
1498 NULL,
1499 errorCode);
1500 if(U_FAILURE(errorCode) || !cc.finalFlush) {
1501 // reset if an error occurred or we did not flush
1502 // otherwise do nothing to make sure that flushing resets
1503 ucnv_resetToUnicode(utf8Cnv);
1504 ucnv_resetFromUnicode(cnv);
1505 }
1506 }
1507 }
1508
1509 // not a real loop, just a convenience for breaking out of the block
1510 while(ok && cc.finalFlush) {
1511 // test ucnv_fromUChars()
1512 memset(result, 0, sizeof(result));
1513
1514 errorCode=U_ZERO_ERROR;
1515 resultLength=ucnv_fromUChars(cnv,
1516 result, LENGTHOF(result),
1517 cc.unicode, cc.unicodeLength,
1518 &errorCode);
1519 ok=checkFromUnicode(
1520 cc, cnv, "fromUChars",
1521 (uint8_t *)result, resultLength,
1522 NULL,
1523 errorCode);
1524 if(!ok) {
1525 break;
1526 }
1527
1528 // test preflighting
1529 // keep the correct result for simple checking
1530 errorCode=U_ZERO_ERROR;
1531 resultLength=ucnv_fromUChars(cnv,
1532 NULL, 0,
1533 cc.unicode, cc.unicodeLength,
1534 &errorCode);
1535 if(errorCode==U_STRING_NOT_TERMINATED_WARNING || errorCode==U_BUFFER_OVERFLOW_ERROR) {
1536 errorCode=U_ZERO_ERROR;
1537 }
1538 ok=checkFromUnicode(
1539 cc, cnv, "preflight fromUChars",
1540 (uint8_t *)result, resultLength,
1541 NULL,
1542 errorCode);
1543 break;
1544 }
1545
1546 ucnv_close(cnv);
1547 return ok;
1548 }
1549
1550 UBool
checkFromUnicode(ConversionCase & cc,UConverter * cnv,const char * name,const uint8_t * result,int32_t resultLength,const int32_t * resultOffsets,UErrorCode resultErrorCode)1551 ConversionTest::checkFromUnicode(ConversionCase &cc, UConverter *cnv, const char *name,
1552 const uint8_t *result, int32_t resultLength,
1553 const int32_t *resultOffsets,
1554 UErrorCode resultErrorCode) {
1555 UChar resultInvalidUChars[8];
1556 int8_t resultInvalidLength;
1557 UErrorCode errorCode;
1558
1559 const char *msg;
1560
1561 // reset the message; NULL will mean "ok"
1562 msg=NULL;
1563
1564 errorCode=U_ZERO_ERROR;
1565 resultInvalidLength=LENGTHOF(resultInvalidUChars);
1566 ucnv_getInvalidUChars(cnv, resultInvalidUChars, &resultInvalidLength, &errorCode);
1567 if(U_FAILURE(errorCode)) {
1568 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) ucnv_getInvalidUChars() failed - %s",
1569 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, u_errorName(errorCode));
1570 return FALSE;
1571 }
1572
1573 // check everything that might have gone wrong
1574 if(cc.bytesLength!=resultLength) {
1575 msg="wrong result length";
1576 } else if(0!=memcmp(cc.bytes, result, cc.bytesLength)) {
1577 msg="wrong result string";
1578 } else if(cc.offsets!=NULL && 0!=memcmp(cc.offsets, resultOffsets, cc.bytesLength*sizeof(*cc.offsets))) {
1579 msg="wrong offsets";
1580 } else if(cc.outErrorCode!=resultErrorCode) {
1581 msg="wrong error code";
1582 } else if(cc.invalidLength!=resultInvalidLength) {
1583 msg="wrong length of last invalid input";
1584 } else if(0!=u_memcmp(cc.invalidUChars, resultInvalidUChars, cc.invalidLength)) {
1585 msg="wrong last invalid input";
1586 }
1587
1588 if(msg==NULL) {
1589 return TRUE;
1590 } else {
1591 char buffer[2000]; // one buffer for all strings
1592 char *s, *unicodeString, *bytesString, *resultString,
1593 *offsetsString, *resultOffsetsString,
1594 *invalidCharsString, *resultInvalidUCharsString;
1595
1596 unicodeString=s=buffer;
1597 s=printUnicode(cc.unicode, cc.unicodeLength, unicodeString);
1598 s=printBytes(cc.bytes, cc.bytesLength, bytesString=s);
1599 s=printBytes(result, resultLength, resultString=s);
1600 s=printOffsets(cc.offsets, cc.bytesLength, offsetsString=s);
1601 s=printOffsets(resultOffsets, resultLength, resultOffsetsString=s);
1602 s=printUnicode(cc.invalidUChars, cc.invalidLength, invalidCharsString=s);
1603 s=printUnicode(resultInvalidUChars, resultInvalidLength, resultInvalidUCharsString=s);
1604
1605 if((s-buffer)>(int32_t)sizeof(buffer)) {
1606 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) fatal error: checkFromUnicode() test output buffer overflow writing %d chars\n",
1607 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, (int)(s-buffer));
1608 exit(1);
1609 }
1610
1611 errln("fromUnicode[%d](%s cb=\"%s\" fb=%d flush=%d %s) failed: %s\n"
1612 " unicode <%s>[%d]\n"
1613 " expected <%s>[%d]\n"
1614 " result <%s>[%d]\n"
1615 " offsets <%s>\n"
1616 " result offsets <%s>\n"
1617 " error code expected %s got %s\n"
1618 " invalidChars expected <%s> got <%s>\n",
1619 cc.caseNr, cc.charset, cc.cbopt, cc.fallbacks, cc.finalFlush, name, msg,
1620 unicodeString, cc.unicodeLength,
1621 bytesString, cc.bytesLength,
1622 resultString, resultLength,
1623 offsetsString,
1624 resultOffsetsString,
1625 u_errorName(cc.outErrorCode), u_errorName(resultErrorCode),
1626 invalidCharsString, resultInvalidUCharsString);
1627
1628 return FALSE;
1629 }
1630 }
1631
1632 #endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */
1633