1 //===- FileCheck.cpp - Check that File's Contents match what is expected --===// 2 // 3 // The LLVM Compiler Infrastructure 4 // 5 // This file is distributed under the University of Illinois Open Source 6 // License. See LICENSE.TXT for details. 7 // 8 //===----------------------------------------------------------------------===// 9 // 10 // FileCheck does a line-by line check of a file that validates whether it 11 // contains the expected content. This is useful for regression tests etc. 12 // 13 // This program exits with an error status of 2 on error, exit status of 0 if 14 // the file matched the expected contents, and exit status of 1 if it did not 15 // contain the expected contents. 16 // 17 //===----------------------------------------------------------------------===// 18 19 #include "llvm/Support/CommandLine.h" 20 #include "llvm/Support/MemoryBuffer.h" 21 #include "llvm/Support/PrettyStackTrace.h" 22 #include "llvm/Support/Regex.h" 23 #include "llvm/Support/SourceMgr.h" 24 #include "llvm/Support/raw_ostream.h" 25 #include "llvm/System/Signals.h" 26 #include "llvm/ADT/SmallString.h" 27 #include "llvm/ADT/StringMap.h" 28 #include <algorithm> 29 using namespace llvm; 30 31 static cl::opt<std::string> 32 CheckFilename(cl::Positional, cl::desc("<check-file>"), cl::Required); 33 34 static cl::opt<std::string> 35 InputFilename("input-file", cl::desc("File to check (defaults to stdin)"), 36 cl::init("-"), cl::value_desc("filename")); 37 38 static cl::opt<std::string> 39 CheckPrefix("check-prefix", cl::init("CHECK"), 40 cl::desc("Prefix to use from check file (defaults to 'CHECK')")); 41 42 static cl::opt<bool> 43 NoCanonicalizeWhiteSpace("strict-whitespace", 44 cl::desc("Do not treat all horizontal whitespace as equivalent")); 45 46 //===----------------------------------------------------------------------===// 47 // Pattern Handling Code. 48 //===----------------------------------------------------------------------===// 49 50 class Pattern { 51 SMLoc PatternLoc; 52 53 /// MatchEOF - When set, this pattern only matches the end of file. This is 54 /// used for trailing CHECK-NOTs. 55 bool MatchEOF; 56 57 /// FixedStr - If non-empty, this pattern is a fixed string match with the 58 /// specified fixed string. 59 StringRef FixedStr; 60 61 /// RegEx - If non-empty, this is a regex pattern. 62 std::string RegExStr; 63 64 /// VariableUses - Entries in this vector map to uses of a variable in the 65 /// pattern, e.g. "foo[[bar]]baz". In this case, the RegExStr will contain 66 /// "foobaz" and we'll get an entry in this vector that tells us to insert the 67 /// value of bar at offset 3. 68 std::vector<std::pair<StringRef, unsigned> > VariableUses; 69 70 /// VariableDefs - Entries in this vector map to definitions of a variable in 71 /// the pattern, e.g. "foo[[bar:.*]]baz". In this case, the RegExStr will 72 /// contain "foo(.*)baz" and VariableDefs will contain the pair "bar",1. The 73 /// index indicates what parenthesized value captures the variable value. 74 std::vector<std::pair<StringRef, unsigned> > VariableDefs; 75 76 public: 77 78 Pattern(bool matchEOF = false) : MatchEOF(matchEOF) { } 79 80 bool ParsePattern(StringRef PatternStr, SourceMgr &SM); 81 82 /// Match - Match the pattern string against the input buffer Buffer. This 83 /// returns the position that is matched or npos if there is no match. If 84 /// there is a match, the size of the matched string is returned in MatchLen. 85 /// 86 /// The VariableTable StringMap provides the current values of filecheck 87 /// variables and is updated if this match defines new values. 88 size_t Match(StringRef Buffer, size_t &MatchLen, 89 StringMap<StringRef> &VariableTable) const; 90 91 /// PrintFailureInfo - Print additional information about a failure to match 92 /// involving this pattern. 93 void PrintFailureInfo(const SourceMgr &SM, StringRef Buffer, 94 const StringMap<StringRef> &VariableTable) const; 95 96 private: 97 static void AddFixedStringToRegEx(StringRef FixedStr, std::string &TheStr); 98 bool AddRegExToRegEx(StringRef RegExStr, unsigned &CurParen, SourceMgr &SM); 99 100 /// ComputeMatchDistance - Compute an arbitrary estimate for the quality of 101 /// matching this pattern at the start of \arg Buffer; a distance of zero 102 /// should correspond to a perfect match. 103 unsigned ComputeMatchDistance(StringRef Buffer, 104 const StringMap<StringRef> &VariableTable) const; 105 }; 106 107 108 bool Pattern::ParsePattern(StringRef PatternStr, SourceMgr &SM) { 109 PatternLoc = SMLoc::getFromPointer(PatternStr.data()); 110 111 // Ignore trailing whitespace. 112 while (!PatternStr.empty() && 113 (PatternStr.back() == ' ' || PatternStr.back() == '\t')) 114 PatternStr = PatternStr.substr(0, PatternStr.size()-1); 115 116 // Check that there is something on the line. 117 if (PatternStr.empty()) { 118 SM.PrintMessage(PatternLoc, "found empty check string with prefix '" + 119 CheckPrefix+":'", "error"); 120 return true; 121 } 122 123 // Check to see if this is a fixed string, or if it has regex pieces. 124 if (PatternStr.size() < 2 || 125 (PatternStr.find("{{") == StringRef::npos && 126 PatternStr.find("[[") == StringRef::npos)) { 127 FixedStr = PatternStr; 128 return false; 129 } 130 131 // Paren value #0 is for the fully matched string. Any new parenthesized 132 // values add from their. 133 unsigned CurParen = 1; 134 135 // Otherwise, there is at least one regex piece. Build up the regex pattern 136 // by escaping scary characters in fixed strings, building up one big regex. 137 while (!PatternStr.empty()) { 138 // RegEx matches. 139 if (PatternStr.size() >= 2 && 140 PatternStr[0] == '{' && PatternStr[1] == '{') { 141 142 // Otherwise, this is the start of a regex match. Scan for the }}. 143 size_t End = PatternStr.find("}}"); 144 if (End == StringRef::npos) { 145 SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()), 146 "found start of regex string with no end '}}'", "error"); 147 return true; 148 } 149 150 if (AddRegExToRegEx(PatternStr.substr(2, End-2), CurParen, SM)) 151 return true; 152 PatternStr = PatternStr.substr(End+2); 153 continue; 154 } 155 156 // Named RegEx matches. These are of two forms: [[foo:.*]] which matches .* 157 // (or some other regex) and assigns it to the FileCheck variable 'foo'. The 158 // second form is [[foo]] which is a reference to foo. The variable name 159 // itself must be of the form "[a-zA-Z_][0-9a-zA-Z_]*", otherwise we reject 160 // it. This is to catch some common errors. 161 if (PatternStr.size() >= 2 && 162 PatternStr[0] == '[' && PatternStr[1] == '[') { 163 // Verify that it is terminated properly. 164 size_t End = PatternStr.find("]]"); 165 if (End == StringRef::npos) { 166 SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()), 167 "invalid named regex reference, no ]] found", "error"); 168 return true; 169 } 170 171 StringRef MatchStr = PatternStr.substr(2, End-2); 172 PatternStr = PatternStr.substr(End+2); 173 174 // Get the regex name (e.g. "foo"). 175 size_t NameEnd = MatchStr.find(':'); 176 StringRef Name = MatchStr.substr(0, NameEnd); 177 178 if (Name.empty()) { 179 SM.PrintMessage(SMLoc::getFromPointer(Name.data()), 180 "invalid name in named regex: empty name", "error"); 181 return true; 182 } 183 184 // Verify that the name is well formed. 185 for (unsigned i = 0, e = Name.size(); i != e; ++i) 186 if (Name[i] != '_' && 187 (Name[i] < 'a' || Name[i] > 'z') && 188 (Name[i] < 'A' || Name[i] > 'Z') && 189 (Name[i] < '0' || Name[i] > '9')) { 190 SM.PrintMessage(SMLoc::getFromPointer(Name.data()+i), 191 "invalid name in named regex", "error"); 192 return true; 193 } 194 195 // Name can't start with a digit. 196 if (isdigit(Name[0])) { 197 SM.PrintMessage(SMLoc::getFromPointer(Name.data()), 198 "invalid name in named regex", "error"); 199 return true; 200 } 201 202 // Handle [[foo]]. 203 if (NameEnd == StringRef::npos) { 204 VariableUses.push_back(std::make_pair(Name, RegExStr.size())); 205 continue; 206 } 207 208 // Handle [[foo:.*]]. 209 VariableDefs.push_back(std::make_pair(Name, CurParen)); 210 RegExStr += '('; 211 ++CurParen; 212 213 if (AddRegExToRegEx(MatchStr.substr(NameEnd+1), CurParen, SM)) 214 return true; 215 216 RegExStr += ')'; 217 } 218 219 // Handle fixed string matches. 220 // Find the end, which is the start of the next regex. 221 size_t FixedMatchEnd = PatternStr.find("{{"); 222 FixedMatchEnd = std::min(FixedMatchEnd, PatternStr.find("[[")); 223 AddFixedStringToRegEx(PatternStr.substr(0, FixedMatchEnd), RegExStr); 224 PatternStr = PatternStr.substr(FixedMatchEnd); 225 continue; 226 } 227 228 return false; 229 } 230 231 void Pattern::AddFixedStringToRegEx(StringRef FixedStr, std::string &TheStr) { 232 // Add the characters from FixedStr to the regex, escaping as needed. This 233 // avoids "leaning toothpicks" in common patterns. 234 for (unsigned i = 0, e = FixedStr.size(); i != e; ++i) { 235 switch (FixedStr[i]) { 236 // These are the special characters matched in "p_ere_exp". 237 case '(': 238 case ')': 239 case '^': 240 case '$': 241 case '|': 242 case '*': 243 case '+': 244 case '?': 245 case '.': 246 case '[': 247 case '\\': 248 case '{': 249 TheStr += '\\'; 250 // FALL THROUGH. 251 default: 252 TheStr += FixedStr[i]; 253 break; 254 } 255 } 256 } 257 258 bool Pattern::AddRegExToRegEx(StringRef RegexStr, unsigned &CurParen, 259 SourceMgr &SM) { 260 Regex R(RegexStr); 261 std::string Error; 262 if (!R.isValid(Error)) { 263 SM.PrintMessage(SMLoc::getFromPointer(RegexStr.data()), 264 "invalid regex: " + Error, "error"); 265 return true; 266 } 267 268 RegExStr += RegexStr.str(); 269 CurParen += R.getNumMatches(); 270 return false; 271 } 272 273 /// Match - Match the pattern string against the input buffer Buffer. This 274 /// returns the position that is matched or npos if there is no match. If 275 /// there is a match, the size of the matched string is returned in MatchLen. 276 size_t Pattern::Match(StringRef Buffer, size_t &MatchLen, 277 StringMap<StringRef> &VariableTable) const { 278 // If this is the EOF pattern, match it immediately. 279 if (MatchEOF) { 280 MatchLen = 0; 281 return Buffer.size(); 282 } 283 284 // If this is a fixed string pattern, just match it now. 285 if (!FixedStr.empty()) { 286 MatchLen = FixedStr.size(); 287 return Buffer.find(FixedStr); 288 } 289 290 // Regex match. 291 292 // If there are variable uses, we need to create a temporary string with the 293 // actual value. 294 StringRef RegExToMatch = RegExStr; 295 std::string TmpStr; 296 if (!VariableUses.empty()) { 297 TmpStr = RegExStr; 298 299 unsigned InsertOffset = 0; 300 for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) { 301 StringMap<StringRef>::iterator it = 302 VariableTable.find(VariableUses[i].first); 303 // If the variable is undefined, return an error. 304 if (it == VariableTable.end()) 305 return StringRef::npos; 306 307 // Look up the value and escape it so that we can plop it into the regex. 308 std::string Value; 309 AddFixedStringToRegEx(it->second, Value); 310 311 // Plop it into the regex at the adjusted offset. 312 TmpStr.insert(TmpStr.begin()+VariableUses[i].second+InsertOffset, 313 Value.begin(), Value.end()); 314 InsertOffset += Value.size(); 315 } 316 317 // Match the newly constructed regex. 318 RegExToMatch = TmpStr; 319 } 320 321 322 SmallVector<StringRef, 4> MatchInfo; 323 if (!Regex(RegExToMatch, Regex::Newline).match(Buffer, &MatchInfo)) 324 return StringRef::npos; 325 326 // Successful regex match. 327 assert(!MatchInfo.empty() && "Didn't get any match"); 328 StringRef FullMatch = MatchInfo[0]; 329 330 // If this defines any variables, remember their values. 331 for (unsigned i = 0, e = VariableDefs.size(); i != e; ++i) { 332 assert(VariableDefs[i].second < MatchInfo.size() && 333 "Internal paren error"); 334 VariableTable[VariableDefs[i].first] = MatchInfo[VariableDefs[i].second]; 335 } 336 337 MatchLen = FullMatch.size(); 338 return FullMatch.data()-Buffer.data(); 339 } 340 341 unsigned Pattern::ComputeMatchDistance(StringRef Buffer, 342 const StringMap<StringRef> &VariableTable) const { 343 // Just compute the number of matching characters. For regular expressions, we 344 // just compare against the regex itself and hope for the best. 345 // 346 // FIXME: One easy improvement here is have the regex lib generate a single 347 // example regular expression which matches, and use that as the example 348 // string. 349 StringRef ExampleString(FixedStr); 350 if (ExampleString.empty()) 351 ExampleString = RegExStr; 352 353 // Only compare up to the first line in the buffer, or the string size. 354 StringRef BufferPrefix = Buffer.substr(0, ExampleString.size()); 355 BufferPrefix = BufferPrefix.split('\n').first; 356 return BufferPrefix.edit_distance(ExampleString); 357 } 358 359 void Pattern::PrintFailureInfo(const SourceMgr &SM, StringRef Buffer, 360 const StringMap<StringRef> &VariableTable) const{ 361 // If this was a regular expression using variables, print the current 362 // variable values. 363 if (!VariableUses.empty()) { 364 for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) { 365 StringRef Var = VariableUses[i].first; 366 StringMap<StringRef>::const_iterator it = VariableTable.find(Var); 367 SmallString<256> Msg; 368 raw_svector_ostream OS(Msg); 369 370 // Check for undefined variable references. 371 if (it == VariableTable.end()) { 372 OS << "uses undefined variable \""; 373 OS.write_escaped(Var) << "\"";; 374 } else { 375 OS << "with variable \""; 376 OS.write_escaped(Var) << "\" equal to \""; 377 OS.write_escaped(it->second) << "\""; 378 } 379 380 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), OS.str(), "note", 381 /*ShowLine=*/false); 382 } 383 } 384 385 // Attempt to find the closest/best fuzzy match. Usually an error happens 386 // because some string in the output didn't exactly match. In these cases, we 387 // would like to show the user a best guess at what "should have" matched, to 388 // save them having to actually check the input manually. 389 size_t NumLinesForward = 0; 390 size_t Best = StringRef::npos; 391 double BestQuality = 0; 392 393 // Use an arbitrary 4k limit on how far we will search. 394 for (size_t i = 0, e = std::min(size_t(4096), Buffer.size()); i != e; ++i) { 395 if (Buffer[i] == '\n') 396 ++NumLinesForward; 397 398 // Patterns have leading whitespace stripped, so skip whitespace when 399 // looking for something which looks like a pattern. 400 if (Buffer[i] == ' ' || Buffer[i] == '\t') 401 continue; 402 403 // Compute the "quality" of this match as an arbitrary combination of the 404 // match distance and the number of lines skipped to get to this match. 405 unsigned Distance = ComputeMatchDistance(Buffer.substr(i), VariableTable); 406 double Quality = Distance + (NumLinesForward / 100.); 407 408 if (Quality < BestQuality || Best == StringRef::npos) { 409 Best = i; 410 BestQuality = Quality; 411 } 412 } 413 414 // Print the "possible intended match here" line if we found something 415 // reasonable and not equal to what we showed in the "scanning from here" 416 // line. 417 if (Best && Best != StringRef::npos && BestQuality < 50) { 418 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data() + Best), 419 "possible intended match here", "note"); 420 421 // FIXME: If we wanted to be really friendly we would show why the match 422 // failed, as it can be hard to spot simple one character differences. 423 } 424 } 425 426 //===----------------------------------------------------------------------===// 427 // Check Strings. 428 //===----------------------------------------------------------------------===// 429 430 /// CheckString - This is a check that we found in the input file. 431 struct CheckString { 432 /// Pat - The pattern to match. 433 Pattern Pat; 434 435 /// Loc - The location in the match file that the check string was specified. 436 SMLoc Loc; 437 438 /// IsCheckNext - This is true if this is a CHECK-NEXT: directive (as opposed 439 /// to a CHECK: directive. 440 bool IsCheckNext; 441 442 /// NotStrings - These are all of the strings that are disallowed from 443 /// occurring between this match string and the previous one (or start of 444 /// file). 445 std::vector<std::pair<SMLoc, Pattern> > NotStrings; 446 447 CheckString(const Pattern &P, SMLoc L, bool isCheckNext) 448 : Pat(P), Loc(L), IsCheckNext(isCheckNext) {} 449 }; 450 451 /// CanonicalizeInputFile - Remove duplicate horizontal space from the specified 452 /// memory buffer, free it, and return a new one. 453 static MemoryBuffer *CanonicalizeInputFile(MemoryBuffer *MB) { 454 SmallString<128> NewFile; 455 NewFile.reserve(MB->getBufferSize()); 456 457 for (const char *Ptr = MB->getBufferStart(), *End = MB->getBufferEnd(); 458 Ptr != End; ++Ptr) { 459 // If C is not a horizontal whitespace, skip it. 460 if (*Ptr != ' ' && *Ptr != '\t') { 461 NewFile.push_back(*Ptr); 462 continue; 463 } 464 465 // Otherwise, add one space and advance over neighboring space. 466 NewFile.push_back(' '); 467 while (Ptr+1 != End && 468 (Ptr[1] == ' ' || Ptr[1] == '\t')) 469 ++Ptr; 470 } 471 472 // Free the old buffer and return a new one. 473 MemoryBuffer *MB2 = 474 MemoryBuffer::getMemBufferCopy(NewFile.str(), MB->getBufferIdentifier()); 475 476 delete MB; 477 return MB2; 478 } 479 480 481 /// ReadCheckFile - Read the check file, which specifies the sequence of 482 /// expected strings. The strings are added to the CheckStrings vector. 483 static bool ReadCheckFile(SourceMgr &SM, 484 std::vector<CheckString> &CheckStrings) { 485 // Open the check file, and tell SourceMgr about it. 486 std::string ErrorStr; 487 MemoryBuffer *F = 488 MemoryBuffer::getFileOrSTDIN(CheckFilename.c_str(), &ErrorStr); 489 if (F == 0) { 490 errs() << "Could not open check file '" << CheckFilename << "': " 491 << ErrorStr << '\n'; 492 return true; 493 } 494 495 // If we want to canonicalize whitespace, strip excess whitespace from the 496 // buffer containing the CHECK lines. 497 if (!NoCanonicalizeWhiteSpace) 498 F = CanonicalizeInputFile(F); 499 500 SM.AddNewSourceBuffer(F, SMLoc()); 501 502 // Find all instances of CheckPrefix followed by : in the file. 503 StringRef Buffer = F->getBuffer(); 504 505 std::vector<std::pair<SMLoc, Pattern> > NotMatches; 506 507 while (1) { 508 // See if Prefix occurs in the memory buffer. 509 Buffer = Buffer.substr(Buffer.find(CheckPrefix)); 510 511 // If we didn't find a match, we're done. 512 if (Buffer.empty()) 513 break; 514 515 const char *CheckPrefixStart = Buffer.data(); 516 517 // When we find a check prefix, keep track of whether we find CHECK: or 518 // CHECK-NEXT: 519 bool IsCheckNext = false, IsCheckNot = false; 520 521 // Verify that the : is present after the prefix. 522 if (Buffer[CheckPrefix.size()] == ':') { 523 Buffer = Buffer.substr(CheckPrefix.size()+1); 524 } else if (Buffer.size() > CheckPrefix.size()+6 && 525 memcmp(Buffer.data()+CheckPrefix.size(), "-NEXT:", 6) == 0) { 526 Buffer = Buffer.substr(CheckPrefix.size()+7); 527 IsCheckNext = true; 528 } else if (Buffer.size() > CheckPrefix.size()+5 && 529 memcmp(Buffer.data()+CheckPrefix.size(), "-NOT:", 5) == 0) { 530 Buffer = Buffer.substr(CheckPrefix.size()+6); 531 IsCheckNot = true; 532 } else { 533 Buffer = Buffer.substr(1); 534 continue; 535 } 536 537 // Okay, we found the prefix, yay. Remember the rest of the line, but 538 // ignore leading and trailing whitespace. 539 Buffer = Buffer.substr(Buffer.find_first_not_of(" \t")); 540 541 // Scan ahead to the end of line. 542 size_t EOL = Buffer.find_first_of("\n\r"); 543 544 // Remember the location of the start of the pattern, for diagnostics. 545 SMLoc PatternLoc = SMLoc::getFromPointer(Buffer.data()); 546 547 // Parse the pattern. 548 Pattern P; 549 if (P.ParsePattern(Buffer.substr(0, EOL), SM)) 550 return true; 551 552 Buffer = Buffer.substr(EOL); 553 554 555 // Verify that CHECK-NEXT lines have at least one CHECK line before them. 556 if (IsCheckNext && CheckStrings.empty()) { 557 SM.PrintMessage(SMLoc::getFromPointer(CheckPrefixStart), 558 "found '"+CheckPrefix+"-NEXT:' without previous '"+ 559 CheckPrefix+ ": line", "error"); 560 return true; 561 } 562 563 // Handle CHECK-NOT. 564 if (IsCheckNot) { 565 NotMatches.push_back(std::make_pair(SMLoc::getFromPointer(Buffer.data()), 566 P)); 567 continue; 568 } 569 570 571 // Okay, add the string we captured to the output vector and move on. 572 CheckStrings.push_back(CheckString(P, 573 PatternLoc, 574 IsCheckNext)); 575 std::swap(NotMatches, CheckStrings.back().NotStrings); 576 } 577 578 // Add an EOF pattern for any trailing CHECK-NOTs. 579 if (!NotMatches.empty()) { 580 CheckStrings.push_back(CheckString(Pattern(true), 581 SMLoc::getFromPointer(Buffer.data()), 582 false)); 583 std::swap(NotMatches, CheckStrings.back().NotStrings); 584 } 585 586 if (CheckStrings.empty()) { 587 errs() << "error: no check strings found with prefix '" << CheckPrefix 588 << ":'\n"; 589 return true; 590 } 591 592 return false; 593 } 594 595 static void PrintCheckFailed(const SourceMgr &SM, const CheckString &CheckStr, 596 StringRef Buffer, 597 StringMap<StringRef> &VariableTable) { 598 // Otherwise, we have an error, emit an error message. 599 SM.PrintMessage(CheckStr.Loc, "expected string not found in input", 600 "error"); 601 602 // Print the "scanning from here" line. If the current position is at the 603 // end of a line, advance to the start of the next line. 604 Buffer = Buffer.substr(Buffer.find_first_not_of(" \t\n\r")); 605 606 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), "scanning from here", 607 "note"); 608 609 // Allow the pattern to print additional information if desired. 610 CheckStr.Pat.PrintFailureInfo(SM, Buffer, VariableTable); 611 } 612 613 /// CountNumNewlinesBetween - Count the number of newlines in the specified 614 /// range. 615 static unsigned CountNumNewlinesBetween(StringRef Range) { 616 unsigned NumNewLines = 0; 617 while (1) { 618 // Scan for newline. 619 Range = Range.substr(Range.find_first_of("\n\r")); 620 if (Range.empty()) return NumNewLines; 621 622 ++NumNewLines; 623 624 // Handle \n\r and \r\n as a single newline. 625 if (Range.size() > 1 && 626 (Range[1] == '\n' || Range[1] == '\r') && 627 (Range[0] != Range[1])) 628 Range = Range.substr(1); 629 Range = Range.substr(1); 630 } 631 } 632 633 int main(int argc, char **argv) { 634 sys::PrintStackTraceOnErrorSignal(); 635 PrettyStackTraceProgram X(argc, argv); 636 cl::ParseCommandLineOptions(argc, argv); 637 638 SourceMgr SM; 639 640 // Read the expected strings from the check file. 641 std::vector<CheckString> CheckStrings; 642 if (ReadCheckFile(SM, CheckStrings)) 643 return 2; 644 645 // Open the file to check and add it to SourceMgr. 646 std::string ErrorStr; 647 MemoryBuffer *F = 648 MemoryBuffer::getFileOrSTDIN(InputFilename.c_str(), &ErrorStr); 649 if (F == 0) { 650 errs() << "Could not open input file '" << InputFilename << "': " 651 << ErrorStr << '\n'; 652 return true; 653 } 654 655 // Remove duplicate spaces in the input file if requested. 656 if (!NoCanonicalizeWhiteSpace) 657 F = CanonicalizeInputFile(F); 658 659 SM.AddNewSourceBuffer(F, SMLoc()); 660 661 /// VariableTable - This holds all the current filecheck variables. 662 StringMap<StringRef> VariableTable; 663 664 // Check that we have all of the expected strings, in order, in the input 665 // file. 666 StringRef Buffer = F->getBuffer(); 667 668 const char *LastMatch = Buffer.data(); 669 670 for (unsigned StrNo = 0, e = CheckStrings.size(); StrNo != e; ++StrNo) { 671 const CheckString &CheckStr = CheckStrings[StrNo]; 672 673 StringRef SearchFrom = Buffer; 674 675 // Find StrNo in the file. 676 size_t MatchLen = 0; 677 size_t MatchPos = CheckStr.Pat.Match(Buffer, MatchLen, VariableTable); 678 Buffer = Buffer.substr(MatchPos); 679 680 // If we didn't find a match, reject the input. 681 if (MatchPos == StringRef::npos) { 682 PrintCheckFailed(SM, CheckStr, SearchFrom, VariableTable); 683 return 1; 684 } 685 686 StringRef SkippedRegion(LastMatch, Buffer.data()-LastMatch); 687 688 // If this check is a "CHECK-NEXT", verify that the previous match was on 689 // the previous line (i.e. that there is one newline between them). 690 if (CheckStr.IsCheckNext) { 691 // Count the number of newlines between the previous match and this one. 692 assert(LastMatch != F->getBufferStart() && 693 "CHECK-NEXT can't be the first check in a file"); 694 695 unsigned NumNewLines = CountNumNewlinesBetween(SkippedRegion); 696 if (NumNewLines == 0) { 697 SM.PrintMessage(CheckStr.Loc, 698 CheckPrefix+"-NEXT: is on the same line as previous match", 699 "error"); 700 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), 701 "'next' match was here", "note"); 702 SM.PrintMessage(SMLoc::getFromPointer(LastMatch), 703 "previous match was here", "note"); 704 return 1; 705 } 706 707 if (NumNewLines != 1) { 708 SM.PrintMessage(CheckStr.Loc, 709 CheckPrefix+ 710 "-NEXT: is not on the line after the previous match", 711 "error"); 712 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), 713 "'next' match was here", "note"); 714 SM.PrintMessage(SMLoc::getFromPointer(LastMatch), 715 "previous match was here", "note"); 716 return 1; 717 } 718 } 719 720 // If this match had "not strings", verify that they don't exist in the 721 // skipped region. 722 for (unsigned ChunkNo = 0, e = CheckStr.NotStrings.size(); 723 ChunkNo != e; ++ChunkNo) { 724 size_t MatchLen = 0; 725 size_t Pos = CheckStr.NotStrings[ChunkNo].second.Match(SkippedRegion, 726 MatchLen, 727 VariableTable); 728 if (Pos == StringRef::npos) continue; 729 730 SM.PrintMessage(SMLoc::getFromPointer(LastMatch+Pos), 731 CheckPrefix+"-NOT: string occurred!", "error"); 732 SM.PrintMessage(CheckStr.NotStrings[ChunkNo].first, 733 CheckPrefix+"-NOT: pattern specified here", "note"); 734 return 1; 735 } 736 737 738 // Otherwise, everything is good. Step over the matched text and remember 739 // the position after the match as the end of the last match. 740 Buffer = Buffer.substr(MatchLen); 741 LastMatch = Buffer.data(); 742 } 743 744 return 0; 745 } 746