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