testRegExp.cpp

   1 /*
   2  *  Copyright (C) 2011 Apple Inc. All rights reserved.
   3  *
   4  *  This library is free software; you can redistribute it and/or
   5  *  modify it under the terms of the GNU Library General Public
   6  *  License as published by the Free Software Foundation; either
   7  *  version 2 of the License, or (at your option) any later version.
   8  *
   9  *  This library is distributed in the hope that it will be useful,
  10  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12  *  Library General Public License for more details.
  13  *
  14  *  You should have received a copy of the GNU Library General Public License
  15  *  along with this library; see the file COPYING.LIB.  If not, write to
  16  *  the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  17  *  Boston, MA 02110-1301, USA.
  18  *
  19  */
  20
  21 #include "config.h"
  22 #include "RegExp.h"
  23
  24 #include <wtf/CurrentTime.h>
  25 #include "InitializeThreading.h"
  26 #include "JSGlobalObject.h"
  27 #include "UStringBuilder.h"
  28 #include <errno.h>
  29 #include <stdio.h>
  30 #include <stdlib.h>
  31 #include <string.h>
  32
  33 #if !OS(WINDOWS)
  34 #include <unistd.h>
  35 #endif
  36
  37 #if HAVE(SYS_TIME_H)
  38 #include <sys/time.h>
  39 #endif
  40
  41 #if COMPILER(MSVC) && !OS(WINCE)
  42 #include <crtdbg.h>
  43 #include <mmsystem.h>
  44 #include <windows.h>
  45 #endif
  46
  47 #if PLATFORM(QT)
  48 #include <QCoreApplication>
  49 #include <QDateTime>
  50 #endif
  51
  52 const int MaxLineLength = 100 * 1024;
  53
  54 using namespace JSC;
  55 using namespace WTF;
  56
  57 struct CommandLine {
  58     CommandLine()
  59         : interactive(false)
  60         , verbose(false)
  61     {
  62     }
  63
  64     bool interactive;
  65     bool verbose;
  66     Vector<UString> arguments;
  67     Vector<UString> files;
  68 };
  69
  70 class StopWatch {
  71 public:
  72     void start();
  73     void stop();
  74     long getElapsedMS(); // call stop() first
  75
  76 private:
  77     double m_startTime;
  78     double m_stopTime;
  79 };
  80
  81 void StopWatch::start()
  82 {
  83     m_startTime = currentTime();
  84 }
  85
  86 void StopWatch::stop()
  87 {
  88     m_stopTime = currentTime();
  89 }
  90
  91 long StopWatch::getElapsedMS()
  92 {
  93     return static_cast<long>((m_stopTime - m_startTime) * 1000);
  94 }
  95
  96 struct RegExpTest {
  97     RegExpTest()
  98         : offset(0)
  99         , result(0)
 100     {
 101     }
 102
 103     UString subject;
 104     int offset;
 105     int result;
 106     Vector<int, 32> expectVector;
 107 };
 108
 109 class GlobalObject : public JSGlobalObject {
 110 private:
 111     GlobalObject(JSGlobalData&, Structure*, const Vector<UString>& arguments);
 112
 113 public:
 114     typedef JSGlobalObject Base;
 115
 116     static GlobalObject* create(JSGlobalData& globalData, Structure* structure, const Vector<UString>& arguments)
 117     {
 118         return new (NotNull, allocateCell<GlobalObject>(globalData.heap)) GlobalObject(globalData, structure, arguments);
 119     }
 120
 121     static const ClassInfo s_info;
 122
 123     static Structure* createStructure(JSGlobalData& globalData, JSValue prototype)
 124     {
 125         return Structure::create(globalData, 0, prototype, TypeInfo(GlobalObjectType, StructureFlags), &s_info);
 126     }
 127
 128 protected:
 129     void finishCreation(JSGlobalData& globalData, const Vector<UString>& arguments)
 130     {
 131         Base::finishCreation(globalData);
 132         UNUSED_PARAM(arguments);
 133     }
 134 };
 135
 136 COMPILE_ASSERT(!IsInteger<GlobalObject>::value, WTF_IsInteger_GlobalObject_false);
 137 ASSERT_CLASS_FITS_IN_CELL(GlobalObject);
 138
 139 const ClassInfo GlobalObject::s_info = { "global", &JSGlobalObject::s_info, 0, ExecState::globalObjectTable, CREATE_METHOD_TABLE(GlobalObject) };
 140
 141 GlobalObject::GlobalObject(JSGlobalData& globalData, Structure* structure, const Vector<UString>& arguments)
 142     : JSGlobalObject(globalData, structure)
 143 {
 144     finishCreation(globalData, arguments);
 145 }
 146
 147 // Use SEH for Release builds only to get rid of the crash report dialog
 148 // (luckily the same tests fail in Release and Debug builds so far). Need to
 149 // be in a separate main function because the realMain function requires object
 150 // unwinding.
 151
 152 #if COMPILER(MSVC) && !COMPILER(INTEL) && !defined(_DEBUG) && !OS(WINCE)
 153 #define TRY       __try {
 154 #define EXCEPT(x) } __except (EXCEPTION_EXECUTE_HANDLER) { x; }
 155 #else
 156 #define TRY
 157 #define EXCEPT(x)
 158 #endif
 159
 160 int realMain(int argc, char** argv);
 161
 162 int main(int argc, char** argv)
 163 {
 164 #if OS(WINDOWS)
 165 #if !OS(WINCE)
 166     // Cygwin calls ::SetErrorMode(SEM_FAILCRITICALERRORS), which we will inherit. This is bad for
 167     // testing/debugging, as it causes the post-mortem debugger not to be invoked. We reset the
 168     // error mode here to work around Cygwin's behavior. See <http://webkit.org/b/55222>.
 169     ::SetErrorMode(0);
 170 #endif
 171
 172 #if defined(_DEBUG)
 173     _CrtSetReportFile(_CRT_WARN, _CRTDBG_FILE_STDERR);
 174     _CrtSetReportMode(_CRT_WARN, _CRTDBG_MODE_FILE);
 175     _CrtSetReportFile(_CRT_ERROR, _CRTDBG_FILE_STDERR);
 176     _CrtSetReportMode(_CRT_ERROR, _CRTDBG_MODE_FILE);
 177     _CrtSetReportFile(_CRT_ASSERT, _CRTDBG_FILE_STDERR);
 178     _CrtSetReportMode(_CRT_ASSERT, _CRTDBG_MODE_FILE);
 179 #endif
 180
 181     timeBeginPeriod(1);
 182 #endif
 183
 184 #if PLATFORM(QT)
 185     QCoreApplication app(argc, argv);
 186 #endif
 187
 188     // Initialize JSC before getting JSGlobalData.
 189     JSC::initializeThreading();
 190
 191     // We can't use destructors in the following code because it uses Windows
 192     // Structured Exception Handling
 193     int res = 0;
 194     TRY
 195         res = realMain(argc, argv);
 196     EXCEPT(res = 3)
 197     return res;
 198 }
 199
 200 static bool testOneRegExp(JSGlobalData& globalData, RegExp* regexp, RegExpTest* regExpTest, bool verbose, unsigned int lineNumber)
 201 {
 202     bool result = true;
 203     Vector<int, 32> outVector;
 204     outVector.resize(regExpTest->expectVector.size());
 205     int matchResult = regexp->match(globalData, regExpTest->subject, regExpTest->offset, outVector);
 206
 207     if (matchResult != regExpTest->result) {
 208         result = false;
 209         if (verbose)
 210             printf("Line %d: results mismatch - expected %d got %d\n", lineNumber, regExpTest->result, matchResult);
 211     } else if (matchResult != -1) {
 212         if (outVector.size() != regExpTest->expectVector.size()) {
 213             result = false;
 214             if (verbose)
 215                 printf("Line %d: output vector size mismatch - expected %lu got %lu\n", lineNumber, regExpTest->expectVector.size(), outVector.size());
 216         } else if (outVector.size() % 2) {
 217             result = false;
 218             if (verbose)
 219                 printf("Line %d: output vector size is odd (%lu), should be even\n", lineNumber, outVector.size());
 220         } else {
 221             // Check in pairs since the first value of the pair could be -1 in which case the second doesn't matter.
 222             size_t pairCount = outVector.size() / 2;
 223             for (size_t i = 0; i < pairCount; ++i) {
 224                 size_t startIndex = i*2;
 225                 if (outVector[startIndex] != regExpTest->expectVector[startIndex]) {
 226                     result = false;
 227                     if (verbose)
 228                         printf("Line %d: output vector mismatch at index %lu - expected %d got %d\n", lineNumber, startIndex, regExpTest->expectVector[startIndex], outVector[startIndex]);
 229                 }
 230                 if ((i > 0) && (regExpTest->expectVector[startIndex] != -1) && (outVector[startIndex+1] != regExpTest->expectVector[startIndex+1])) {
 231                     result = false;
 232                     if (verbose)
 233                         printf("Line %d: output vector mismatch at index %lu - expected %d got %d\n", lineNumber, startIndex+1, regExpTest->expectVector[startIndex+1], outVector[startIndex+1]);
 234                 }
 235             }
 236         }
 237     }
 238
 239     return result;
 240 }
 241
 242 static int scanString(char* buffer, int bufferLength, UStringBuilder& builder, char termChar)
 243 {
 244     bool escape = false;
 245
 246     for (int i = 0; i < bufferLength; ++i) {
 247         UChar c = buffer[i];
 248
 249         if (escape) {
 250             switch (c) {
 251             case '0':
 252                 c = '\0';
 253                 break;
 254             case 'a':
 255                 c = '\a';
 256                 break;
 257             case 'b':
 258                 c = '\b';
 259                 break;
 260             case 'f':
 261                 c = '\f';
 262                 break;
 263             case 'n':
 264                 c = '\n';
 265                 break;
 266             case 'r':
 267                 c = '\r';
 268                 break;
 269             case 't':
 270                 c = '\t';
 271                 break;
 272             case 'v':
 273                 c = '\v';
 274                 break;
 275             case '\\':
 276                 c = '\\';
 277                 break;
 278             case '?':
 279                 c = '\?';
 280                 break;
 281             case 'u':
 282                 if ((i + 4) >= bufferLength)
 283                     return -1;
 284                 unsigned int charValue;
 285                 if (sscanf(buffer+i+1, "%04x", &charValue) != 1)
 286                     return -1;
 287                 c = static_cast<UChar>(charValue);
 288                 i += 4;
 289                 break;
 290             }
 291
 292             builder.append(c);
 293             escape = false;
 294         } else {
 295             if (c == termChar)
 296                 return i;
 297
 298             if (c == '\\')
 299                 escape = true;
 300             else
 301                 builder.append(c);
 302         }
 303     }
 304
 305     return -1;
 306 }
 307
 308 static RegExp* parseRegExpLine(JSGlobalData& globalData, char* line, int lineLength)
 309 {
 310     UStringBuilder pattern;
 311
 312     if (line[0] != '/')
 313         return 0;
 314
 315     int i = scanString(line + 1, lineLength - 1, pattern, '/') + 1;
 316
 317     if ((i >= lineLength) || (line[i] != '/'))
 318         return 0;
 319
 320     ++i;
 321
 322     return RegExp::create(globalData, pattern.toUString(), regExpFlags(line + i));
 323 }
 324
 325 static RegExpTest* parseTestLine(char* line, int lineLength)
 326 {
 327     UStringBuilder subjectString;
 328
 329     if ((line[0] != ' ') || (line[1] != '"'))
 330         return 0;
 331
 332     int i = scanString(line + 2, lineLength - 2, subjectString, '"') + 2;
 333
 334     if ((i >= (lineLength - 2)) || (line[i] != '"') || (line[i+1] != ',') || (line[i+2] != ' '))
 335         return 0;
 336
 337     i += 3;
 338
 339     int offset;
 340
 341     if (sscanf(line + i, "%d, ", &offset) != 1)
 342         return 0;
 343
 344     while (line[i] && line[i] != ' ')
 345         ++i;
 346
 347     ++i;
 348
 349     int matchResult;
 350
 351     if (sscanf(line + i, "%d, ", &matchResult) != 1)
 352         return 0;
 353
 354     while (line[i] && line[i] != ' ')
 355         ++i;
 356
 357     ++i;
 358
 359     if (line[i++] != '(')
 360         return 0;
 361
 362     int start, end;
 363
 364     RegExpTest* result = new RegExpTest();
 365
 366     result->subject = subjectString.toUString();
 367     result->offset = offset;
 368     result->result = matchResult;
 369
 370     while (line[i] && line[i] != ')') {
 371         if (sscanf(line + i, "%d, %d", &start, &end) != 2) {
 372             delete result;
 373             return 0;
 374         }
 375
 376         result->expectVector.append(start);
 377         result->expectVector.append(end);
 378
 379         while (line[i] && (line[i] != ',') && (line[i] != ')'))
 380             i++;
 381         i++;
 382         while (line[i] && (line[i] != ',') && (line[i] != ')'))
 383             i++;
 384
 385         if (line[i] == ')')
 386             break;
 387         if (!line[i] || (line[i] != ',')) {
 388             delete result;
 389             return 0;
 390         }
 391         i++;
 392     }
 393
 394     return result;
 395 }
 396
 397 static bool runFromFiles(GlobalObject* globalObject, const Vector<UString>& files, bool verbose)
 398 {
 399     UString script;
 400     UString fileName;
 401     Vector<char> scriptBuffer;
 402     unsigned tests = 0;
 403     unsigned failures = 0;
 404     char* lineBuffer = new char[MaxLineLength + 1];
 405
 406     JSGlobalData& globalData = globalObject->globalData();
 407
 408     bool success = true;
 409     for (size_t i = 0; i < files.size(); i++) {
 410         FILE* testCasesFile = fopen(files[i].utf8().data(), "rb");
 411
 412         if (!testCasesFile) {
 413             printf("Unable to open test data file \"%s\"\n", files[i].utf8().data());
 414             continue;
 415         }
 416
 417         RegExp* regexp = 0;
 418         size_t lineLength = 0;
 419         char* linePtr = 0;
 420         unsigned int lineNumber = 0;
 421
 422         while ((linePtr = fgets(&lineBuffer[0], MaxLineLength, testCasesFile))) {
 423             lineLength = strlen(linePtr);
 424             if (linePtr[lineLength - 1] == '\n') {
 425                 linePtr[lineLength - 1] = '\0';
 426                 --lineLength;
 427             }
 428             ++lineNumber;
 429
 430             if (linePtr[0] == '#')
 431                 continue;
 432
 433             if (linePtr[0] == '/') {
 434                 regexp = parseRegExpLine(globalData, linePtr, lineLength);
 435             } else if (linePtr[0] == ' ') {
 436                 RegExpTest* regExpTest = parseTestLine(linePtr, lineLength);
 437
 438                 if (regexp && regExpTest) {
 439                     ++tests;
 440                     if (!testOneRegExp(globalData, regexp, regExpTest, verbose, lineNumber)) {
 441                         failures++;
 442                         printf("Failure on line %u\n", lineNumber);
 443                     }
 444                 }
 445
 446                 if (regExpTest)
 447                     delete regExpTest;
 448             }
 449         }
 450
 451         fclose(testCasesFile);
 452     }
 453
 454     if (failures)
 455         printf("%u tests run, %u failures\n", tests, failures);
 456     else
 457         printf("%u tests passed\n", tests);
 458
 459     delete[] lineBuffer;
 460
 461     globalData.dumpSampleData(globalObject->globalExec());
 462 #if ENABLE(REGEXP_TRACING)
 463     globalData.dumpRegExpTrace();
 464 #endif
 465     return success;
 466 }
 467
 468 #define RUNNING_FROM_XCODE 0
 469
 470 static NO_RETURN void printUsageStatement(bool help = false)
 471 {
 472     fprintf(stderr, "Usage: regexp_test [options] file\n");
 473     fprintf(stderr, "  -h|--help  Prints this help message\n");
 474     fprintf(stderr, "  -v|--verbose  Verbose output\n");
 475
 476     exit(help ? EXIT_SUCCESS : EXIT_FAILURE);
 477 }
 478
 479 static void parseArguments(int argc, char** argv, CommandLine& options)
 480 {
 481     int i = 1;
 482     for (; i < argc; ++i) {
 483         const char* arg = argv[i];
 484         if (!strcmp(arg, "-h") || !strcmp(arg, "--help"))
 485             printUsageStatement(true);
 486         if (!strcmp(arg, "-v") || !strcmp(arg, "--verbose"))
 487             options.verbose = true;
 488         else
 489             options.files.append(argv[i]);
 490     }
 491
 492     for (; i < argc; ++i)
 493         options.arguments.append(argv[i]);
 494 }
 495
 496 int realMain(int argc, char** argv)
 497 {
 498     RefPtr<JSGlobalData> globalData = JSGlobalData::create(ThreadStackTypeLarge, LargeHeap);
 499     JSLockHolder lock(globalData.get());
 500
 501     CommandLine options;
 502     parseArguments(argc, argv, options);
 503
 504     GlobalObject* globalObject = GlobalObject::create(*globalData, GlobalObject::createStructure(*globalData, jsNull()), options.arguments);
 505     bool success = runFromFiles(globalObject, options.files, options.verbose);
 506
 507     return success ? 0 : 3;
 508 }