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/ADT/OwningPtr.h" 20 #include "llvm/Support/CommandLine.h" 21 #include "llvm/Support/MemoryBuffer.h" 22 #include "llvm/Support/PrettyStackTrace.h" 23 #include "llvm/Support/Regex.h" 24 #include "llvm/Support/SourceMgr.h" 25 #include "llvm/Support/raw_ostream.h" 26 #include "llvm/Support/Signals.h" 27 #include "llvm/Support/system_error.h" 28 #include "llvm/ADT/SmallString.h" 29 #include "llvm/ADT/StringMap.h" 30 #include <algorithm> 31 using namespace llvm; 32 33 static cl::opt<std::string> 34 CheckFilename(cl::Positional, cl::desc("<check-file>"), cl::Required); 35 36 static cl::opt<std::string> 37 InputFilename("input-file", cl::desc("File to check (defaults to stdin)"), 38 cl::init("-"), cl::value_desc("filename")); 39 40 static cl::opt<std::string> 41 CheckPrefix("check-prefix", cl::init("CHECK"), 42 cl::desc("Prefix to use from check file (defaults to 'CHECK')")); 43 44 static cl::opt<bool> 45 NoCanonicalizeWhiteSpace("strict-whitespace", 46 cl::desc("Do not treat all horizontal whitespace as equivalent")); 47 48 //===----------------------------------------------------------------------===// 49 // Pattern Handling Code. 50 //===----------------------------------------------------------------------===// 51 52 class Pattern { 53 SMLoc PatternLoc; 54 55 /// MatchEOF - When set, this pattern only matches the end of file. This is 56 /// used for trailing CHECK-NOTs. 57 bool MatchEOF; 58 59 /// FixedStr - If non-empty, this pattern is a fixed string match with the 60 /// specified fixed string. 61 StringRef FixedStr; 62 63 /// RegEx - If non-empty, this is a regex pattern. 64 std::string RegExStr; 65 66 /// VariableUses - Entries in this vector map to uses of a variable in the 67 /// pattern, e.g. "foo[[bar]]baz". In this case, the RegExStr will contain 68 /// "foobaz" and we'll get an entry in this vector that tells us to insert the 69 /// value of bar at offset 3. 70 std::vector<std::pair<StringRef, unsigned> > VariableUses; 71 72 /// VariableDefs - Entries in this vector map to definitions of a variable in 73 /// the pattern, e.g. "foo[[bar:.*]]baz". In this case, the RegExStr will 74 /// contain "foo(.*)baz" and VariableDefs will contain the pair "bar",1. The 75 /// index indicates what parenthesized value captures the variable value. 76 std::vector<std::pair<StringRef, unsigned> > VariableDefs; 77 78 public: 79 80 Pattern(bool matchEOF = false) : MatchEOF(matchEOF) { } 81 82 bool ParsePattern(StringRef PatternStr, SourceMgr &SM); 83 84 /// Match - Match the pattern string against the input buffer Buffer. This 85 /// returns the position that is matched or npos if there is no match. If 86 /// there is a match, the size of the matched string is returned in MatchLen. 87 /// 88 /// The VariableTable StringMap provides the current values of filecheck 89 /// variables and is updated if this match defines new values. 90 size_t Match(StringRef Buffer, size_t &MatchLen, 91 StringMap<StringRef> &VariableTable) const; 92 93 /// PrintFailureInfo - Print additional information about a failure to match 94 /// involving this pattern. 95 void PrintFailureInfo(const SourceMgr &SM, StringRef Buffer, 96 const StringMap<StringRef> &VariableTable) const; 97 98 private: 99 static void AddFixedStringToRegEx(StringRef FixedStr, std::string &TheStr); 100 bool AddRegExToRegEx(StringRef RegExStr, unsigned &CurParen, SourceMgr &SM); 101 102 /// ComputeMatchDistance - Compute an arbitrary estimate for the quality of 103 /// matching this pattern at the start of \arg Buffer; a distance of zero 104 /// should correspond to a perfect match. 105 unsigned ComputeMatchDistance(StringRef Buffer, 106 const StringMap<StringRef> &VariableTable) const; 107 }; 108 109 110 bool Pattern::ParsePattern(StringRef PatternStr, SourceMgr &SM) { 111 PatternLoc = SMLoc::getFromPointer(PatternStr.data()); 112 113 // Ignore trailing whitespace. 114 while (!PatternStr.empty() && 115 (PatternStr.back() == ' ' || PatternStr.back() == '\t')) 116 PatternStr = PatternStr.substr(0, PatternStr.size()-1); 117 118 // Check that there is something on the line. 119 if (PatternStr.empty()) { 120 SM.PrintMessage(PatternLoc, "found empty check string with prefix '" + 121 CheckPrefix+":'", "error"); 122 return true; 123 } 124 125 // Check to see if this is a fixed string, or if it has regex pieces. 126 if (PatternStr.size() < 2 || 127 (PatternStr.find("{{") == StringRef::npos && 128 PatternStr.find("[[") == StringRef::npos)) { 129 FixedStr = PatternStr; 130 return false; 131 } 132 133 // Paren value #0 is for the fully matched string. Any new parenthesized 134 // values add from their. 135 unsigned CurParen = 1; 136 137 // Otherwise, there is at least one regex piece. Build up the regex pattern 138 // by escaping scary characters in fixed strings, building up one big regex. 139 while (!PatternStr.empty()) { 140 // RegEx matches. 141 if (PatternStr.size() >= 2 && 142 PatternStr[0] == '{' && PatternStr[1] == '{') { 143 144 // Otherwise, this is the start of a regex match. Scan for the }}. 145 size_t End = PatternStr.find("}}"); 146 if (End == StringRef::npos) { 147 SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()), 148 "found start of regex string with no end '}}'", "error"); 149 return true; 150 } 151 152 if (AddRegExToRegEx(PatternStr.substr(2, End-2), CurParen, SM)) 153 return true; 154 PatternStr = PatternStr.substr(End+2); 155 continue; 156 } 157 158 // Named RegEx matches. These are of two forms: [[foo:.*]] which matches .* 159 // (or some other regex) and assigns it to the FileCheck variable 'foo'. The 160 // second form is [[foo]] which is a reference to foo. The variable name 161 // itself must be of the form "[a-zA-Z_][0-9a-zA-Z_]*", otherwise we reject 162 // it. This is to catch some common errors. 163 if (PatternStr.size() >= 2 && 164 PatternStr[0] == '[' && PatternStr[1] == '[') { 165 // Verify that it is terminated properly. 166 size_t End = PatternStr.find("]]"); 167 if (End == StringRef::npos) { 168 SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()), 169 "invalid named regex reference, no ]] found", "error"); 170 return true; 171 } 172 173 StringRef MatchStr = PatternStr.substr(2, End-2); 174 PatternStr = PatternStr.substr(End+2); 175 176 // Get the regex name (e.g. "foo"). 177 size_t NameEnd = MatchStr.find(':'); 178 StringRef Name = MatchStr.substr(0, NameEnd); 179 180 if (Name.empty()) { 181 SM.PrintMessage(SMLoc::getFromPointer(Name.data()), 182 "invalid name in named regex: empty name", "error"); 183 return true; 184 } 185 186 // Verify that the name is well formed. 187 for (unsigned i = 0, e = Name.size(); i != e; ++i) 188 if (Name[i] != '_' && 189 (Name[i] < 'a' || Name[i] > 'z') && 190 (Name[i] < 'A' || Name[i] > 'Z') && 191 (Name[i] < '0' || Name[i] > '9')) { 192 SM.PrintMessage(SMLoc::getFromPointer(Name.data()+i), 193 "invalid name in named regex", "error"); 194 return true; 195 } 196 197 // Name can't start with a digit. 198 if (isdigit(Name[0])) { 199 SM.PrintMessage(SMLoc::getFromPointer(Name.data()), 200 "invalid name in named regex", "error"); 201 return true; 202 } 203 204 // Handle [[foo]]. 205 if (NameEnd == StringRef::npos) { 206 VariableUses.push_back(std::make_pair(Name, RegExStr.size())); 207 continue; 208 } 209 210 // Handle [[foo:.*]]. 211 VariableDefs.push_back(std::make_pair(Name, CurParen)); 212 RegExStr += '('; 213 ++CurParen; 214 215 if (AddRegExToRegEx(MatchStr.substr(NameEnd+1), CurParen, SM)) 216 return true; 217 218 RegExStr += ')'; 219 } 220 221 // Handle fixed string matches. 222 // Find the end, which is the start of the next regex. 223 size_t FixedMatchEnd = PatternStr.find("{{"); 224 FixedMatchEnd = std::min(FixedMatchEnd, PatternStr.find("[[")); 225 AddFixedStringToRegEx(PatternStr.substr(0, FixedMatchEnd), RegExStr); 226 PatternStr = PatternStr.substr(FixedMatchEnd); 227 continue; 228 } 229 230 return false; 231 } 232 233 void Pattern::AddFixedStringToRegEx(StringRef FixedStr, std::string &TheStr) { 234 // Add the characters from FixedStr to the regex, escaping as needed. This 235 // avoids "leaning toothpicks" in common patterns. 236 for (unsigned i = 0, e = FixedStr.size(); i != e; ++i) { 237 switch (FixedStr[i]) { 238 // These are the special characters matched in "p_ere_exp". 239 case '(': 240 case ')': 241 case '^': 242 case '$': 243 case '|': 244 case '*': 245 case '+': 246 case '?': 247 case '.': 248 case '[': 249 case '\\': 250 case '{': 251 TheStr += '\\'; 252 // FALL THROUGH. 253 default: 254 TheStr += FixedStr[i]; 255 break; 256 } 257 } 258 } 259 260 bool Pattern::AddRegExToRegEx(StringRef RegexStr, unsigned &CurParen, 261 SourceMgr &SM) { 262 Regex R(RegexStr); 263 std::string Error; 264 if (!R.isValid(Error)) { 265 SM.PrintMessage(SMLoc::getFromPointer(RegexStr.data()), 266 "invalid regex: " + Error, "error"); 267 return true; 268 } 269 270 RegExStr += RegexStr.str(); 271 CurParen += R.getNumMatches(); 272 return false; 273 } 274 275 /// Match - Match the pattern string against the input buffer Buffer. This 276 /// returns the position that is matched or npos if there is no match. If 277 /// there is a match, the size of the matched string is returned in MatchLen. 278 size_t Pattern::Match(StringRef Buffer, size_t &MatchLen, 279 StringMap<StringRef> &VariableTable) const { 280 // If this is the EOF pattern, match it immediately. 281 if (MatchEOF) { 282 MatchLen = 0; 283 return Buffer.size(); 284 } 285 286 // If this is a fixed string pattern, just match it now. 287 if (!FixedStr.empty()) { 288 MatchLen = FixedStr.size(); 289 return Buffer.find(FixedStr); 290 } 291 292 // Regex match. 293 294 // If there are variable uses, we need to create a temporary string with the 295 // actual value. 296 StringRef RegExToMatch = RegExStr; 297 std::string TmpStr; 298 if (!VariableUses.empty()) { 299 TmpStr = RegExStr; 300 301 unsigned InsertOffset = 0; 302 for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) { 303 StringMap<StringRef>::iterator it = 304 VariableTable.find(VariableUses[i].first); 305 // If the variable is undefined, return an error. 306 if (it == VariableTable.end()) 307 return StringRef::npos; 308 309 // Look up the value and escape it so that we can plop it into the regex. 310 std::string Value; 311 AddFixedStringToRegEx(it->second, Value); 312 313 // Plop it into the regex at the adjusted offset. 314 TmpStr.insert(TmpStr.begin()+VariableUses[i].second+InsertOffset, 315 Value.begin(), Value.end()); 316 InsertOffset += Value.size(); 317 } 318 319 // Match the newly constructed regex. 320 RegExToMatch = TmpStr; 321 } 322 323 324 SmallVector<StringRef, 4> MatchInfo; 325 if (!Regex(RegExToMatch, Regex::Newline).match(Buffer, &MatchInfo)) 326 return StringRef::npos; 327 328 // Successful regex match. 329 assert(!MatchInfo.empty() && "Didn't get any match"); 330 StringRef FullMatch = MatchInfo[0]; 331 332 // If this defines any variables, remember their values. 333 for (unsigned i = 0, e = VariableDefs.size(); i != e; ++i) { 334 assert(VariableDefs[i].second < MatchInfo.size() && 335 "Internal paren error"); 336 VariableTable[VariableDefs[i].first] = MatchInfo[VariableDefs[i].second]; 337 } 338 339 MatchLen = FullMatch.size(); 340 return FullMatch.data()-Buffer.data(); 341 } 342 343 unsigned Pattern::ComputeMatchDistance(StringRef Buffer, 344 const StringMap<StringRef> &VariableTable) const { 345 // Just compute the number of matching characters. For regular expressions, we 346 // just compare against the regex itself and hope for the best. 347 // 348 // FIXME: One easy improvement here is have the regex lib generate a single 349 // example regular expression which matches, and use that as the example 350 // string. 351 StringRef ExampleString(FixedStr); 352 if (ExampleString.empty()) 353 ExampleString = RegExStr; 354 355 // Only compare up to the first line in the buffer, or the string size. 356 StringRef BufferPrefix = Buffer.substr(0, ExampleString.size()); 357 BufferPrefix = BufferPrefix.split('\n').first; 358 return BufferPrefix.edit_distance(ExampleString); 359 } 360 361 void Pattern::PrintFailureInfo(const SourceMgr &SM, StringRef Buffer, 362 const StringMap<StringRef> &VariableTable) const{ 363 // If this was a regular expression using variables, print the current 364 // variable values. 365 if (!VariableUses.empty()) { 366 for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) { 367 StringRef Var = VariableUses[i].first; 368 StringMap<StringRef>::const_iterator it = VariableTable.find(Var); 369 SmallString<256> Msg; 370 raw_svector_ostream OS(Msg); 371 372 // Check for undefined variable references. 373 if (it == VariableTable.end()) { 374 OS << "uses undefined variable \""; 375 OS.write_escaped(Var) << "\"";; 376 } else { 377 OS << "with variable \""; 378 OS.write_escaped(Var) << "\" equal to \""; 379 OS.write_escaped(it->second) << "\""; 380 } 381 382 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), OS.str(), "note", 383 /*ShowLine=*/false); 384 } 385 } 386 387 // Attempt to find the closest/best fuzzy match. Usually an error happens 388 // because some string in the output didn't exactly match. In these cases, we 389 // would like to show the user a best guess at what "should have" matched, to 390 // save them having to actually check the input manually. 391 size_t NumLinesForward = 0; 392 size_t Best = StringRef::npos; 393 double BestQuality = 0; 394 395 // Use an arbitrary 4k limit on how far we will search. 396 for (size_t i = 0, e = std::min(size_t(4096), Buffer.size()); i != e; ++i) { 397 if (Buffer[i] == '\n') 398 ++NumLinesForward; 399 400 // Patterns have leading whitespace stripped, so skip whitespace when 401 // looking for something which looks like a pattern. 402 if (Buffer[i] == ' ' || Buffer[i] == '\t') 403 continue; 404 405 // Compute the "quality" of this match as an arbitrary combination of the 406 // match distance and the number of lines skipped to get to this match. 407 unsigned Distance = ComputeMatchDistance(Buffer.substr(i), VariableTable); 408 double Quality = Distance + (NumLinesForward / 100.); 409 410 if (Quality < BestQuality || Best == StringRef::npos) { 411 Best = i; 412 BestQuality = Quality; 413 } 414 } 415 416 // Print the "possible intended match here" line if we found something 417 // reasonable and not equal to what we showed in the "scanning from here" 418 // line. 419 if (Best && Best != StringRef::npos && BestQuality < 50) { 420 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data() + Best), 421 "possible intended match here", "note"); 422 423 // FIXME: If we wanted to be really friendly we would show why the match 424 // failed, as it can be hard to spot simple one character differences. 425 } 426 } 427 428 //===----------------------------------------------------------------------===// 429 // Check Strings. 430 //===----------------------------------------------------------------------===// 431 432 /// CheckString - This is a check that we found in the input file. 433 struct CheckString { 434 /// Pat - The pattern to match. 435 Pattern Pat; 436 437 /// Loc - The location in the match file that the check string was specified. 438 SMLoc Loc; 439 440 /// IsCheckNext - This is true if this is a CHECK-NEXT: directive (as opposed 441 /// to a CHECK: directive. 442 bool IsCheckNext; 443 444 /// NotStrings - These are all of the strings that are disallowed from 445 /// occurring between this match string and the previous one (or start of 446 /// file). 447 std::vector<std::pair<SMLoc, Pattern> > NotStrings; 448 449 CheckString(const Pattern &P, SMLoc L, bool isCheckNext) 450 : Pat(P), Loc(L), IsCheckNext(isCheckNext) {} 451 }; 452 453 /// CanonicalizeInputFile - Remove duplicate horizontal space from the specified 454 /// memory buffer, free it, and return a new one. 455 static MemoryBuffer *CanonicalizeInputFile(MemoryBuffer *MB) { 456 SmallString<128> NewFile; 457 NewFile.reserve(MB->getBufferSize()); 458 459 for (const char *Ptr = MB->getBufferStart(), *End = MB->getBufferEnd(); 460 Ptr != End; ++Ptr) { 461 // Eliminate trailing dosish \r. 462 if (Ptr <= End - 2 && Ptr[0] == '\r' && Ptr[1] == '\n') { 463 continue; 464 } 465 466 // If C is not a horizontal whitespace, skip it. 467 if (*Ptr != ' ' && *Ptr != '\t') { 468 NewFile.push_back(*Ptr); 469 continue; 470 } 471 472 // Otherwise, add one space and advance over neighboring space. 473 NewFile.push_back(' '); 474 while (Ptr+1 != End && 475 (Ptr[1] == ' ' || Ptr[1] == '\t')) 476 ++Ptr; 477 } 478 479 // Free the old buffer and return a new one. 480 MemoryBuffer *MB2 = 481 MemoryBuffer::getMemBufferCopy(NewFile.str(), MB->getBufferIdentifier()); 482 483 delete MB; 484 return MB2; 485 } 486 487 488 /// ReadCheckFile - Read the check file, which specifies the sequence of 489 /// expected strings. The strings are added to the CheckStrings vector. 490 static bool ReadCheckFile(SourceMgr &SM, 491 std::vector<CheckString> &CheckStrings) { 492 // Open the check file, and tell SourceMgr about it. 493 OwningPtr<MemoryBuffer> File; 494 if (error_code ec = 495 MemoryBuffer::getFileOrSTDIN(CheckFilename.c_str(), File)) { 496 errs() << "Could not open check file '" << CheckFilename << "': " 497 << ec.message() << '\n'; 498 return true; 499 } 500 MemoryBuffer *F = File.take(); 501 502 // If we want to canonicalize whitespace, strip excess whitespace from the 503 // buffer containing the CHECK lines. 504 if (!NoCanonicalizeWhiteSpace) 505 F = CanonicalizeInputFile(F); 506 507 SM.AddNewSourceBuffer(F, SMLoc()); 508 509 // Find all instances of CheckPrefix followed by : in the file. 510 StringRef Buffer = F->getBuffer(); 511 512 std::vector<std::pair<SMLoc, Pattern> > NotMatches; 513 514 while (1) { 515 // See if Prefix occurs in the memory buffer. 516 Buffer = Buffer.substr(Buffer.find(CheckPrefix)); 517 518 // If we didn't find a match, we're done. 519 if (Buffer.empty()) 520 break; 521 522 const char *CheckPrefixStart = Buffer.data(); 523 524 // When we find a check prefix, keep track of whether we find CHECK: or 525 // CHECK-NEXT: 526 bool IsCheckNext = false, IsCheckNot = false; 527 528 // Verify that the : is present after the prefix. 529 if (Buffer[CheckPrefix.size()] == ':') { 530 Buffer = Buffer.substr(CheckPrefix.size()+1); 531 } else if (Buffer.size() > CheckPrefix.size()+6 && 532 memcmp(Buffer.data()+CheckPrefix.size(), "-NEXT:", 6) == 0) { 533 Buffer = Buffer.substr(CheckPrefix.size()+7); 534 IsCheckNext = true; 535 } else if (Buffer.size() > CheckPrefix.size()+5 && 536 memcmp(Buffer.data()+CheckPrefix.size(), "-NOT:", 5) == 0) { 537 Buffer = Buffer.substr(CheckPrefix.size()+6); 538 IsCheckNot = true; 539 } else { 540 Buffer = Buffer.substr(1); 541 continue; 542 } 543 544 // Okay, we found the prefix, yay. Remember the rest of the line, but 545 // ignore leading and trailing whitespace. 546 Buffer = Buffer.substr(Buffer.find_first_not_of(" \t")); 547 548 // Scan ahead to the end of line. 549 size_t EOL = Buffer.find_first_of("\n\r"); 550 551 // Remember the location of the start of the pattern, for diagnostics. 552 SMLoc PatternLoc = SMLoc::getFromPointer(Buffer.data()); 553 554 // Parse the pattern. 555 Pattern P; 556 if (P.ParsePattern(Buffer.substr(0, EOL), SM)) 557 return true; 558 559 Buffer = Buffer.substr(EOL); 560 561 562 // Verify that CHECK-NEXT lines have at least one CHECK line before them. 563 if (IsCheckNext && CheckStrings.empty()) { 564 SM.PrintMessage(SMLoc::getFromPointer(CheckPrefixStart), 565 "found '"+CheckPrefix+"-NEXT:' without previous '"+ 566 CheckPrefix+ ": line", "error"); 567 return true; 568 } 569 570 // Handle CHECK-NOT. 571 if (IsCheckNot) { 572 NotMatches.push_back(std::make_pair(SMLoc::getFromPointer(Buffer.data()), 573 P)); 574 continue; 575 } 576 577 578 // Okay, add the string we captured to the output vector and move on. 579 CheckStrings.push_back(CheckString(P, 580 PatternLoc, 581 IsCheckNext)); 582 std::swap(NotMatches, CheckStrings.back().NotStrings); 583 } 584 585 // Add an EOF pattern for any trailing CHECK-NOTs. 586 if (!NotMatches.empty()) { 587 CheckStrings.push_back(CheckString(Pattern(true), 588 SMLoc::getFromPointer(Buffer.data()), 589 false)); 590 std::swap(NotMatches, CheckStrings.back().NotStrings); 591 } 592 593 if (CheckStrings.empty()) { 594 errs() << "error: no check strings found with prefix '" << CheckPrefix 595 << ":'\n"; 596 return true; 597 } 598 599 return false; 600 } 601 602 static void PrintCheckFailed(const SourceMgr &SM, const CheckString &CheckStr, 603 StringRef Buffer, 604 StringMap<StringRef> &VariableTable) { 605 // Otherwise, we have an error, emit an error message. 606 SM.PrintMessage(CheckStr.Loc, "expected string not found in input", 607 "error"); 608 609 // Print the "scanning from here" line. If the current position is at the 610 // end of a line, advance to the start of the next line. 611 Buffer = Buffer.substr(Buffer.find_first_not_of(" \t\n\r")); 612 613 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), "scanning from here", 614 "note"); 615 616 // Allow the pattern to print additional information if desired. 617 CheckStr.Pat.PrintFailureInfo(SM, Buffer, VariableTable); 618 } 619 620 /// CountNumNewlinesBetween - Count the number of newlines in the specified 621 /// range. 622 static unsigned CountNumNewlinesBetween(StringRef Range) { 623 unsigned NumNewLines = 0; 624 while (1) { 625 // Scan for newline. 626 Range = Range.substr(Range.find_first_of("\n\r")); 627 if (Range.empty()) return NumNewLines; 628 629 ++NumNewLines; 630 631 // Handle \n\r and \r\n as a single newline. 632 if (Range.size() > 1 && 633 (Range[1] == '\n' || Range[1] == '\r') && 634 (Range[0] != Range[1])) 635 Range = Range.substr(1); 636 Range = Range.substr(1); 637 } 638 } 639 640 int main(int argc, char **argv) { 641 sys::PrintStackTraceOnErrorSignal(); 642 PrettyStackTraceProgram X(argc, argv); 643 cl::ParseCommandLineOptions(argc, argv); 644 645 SourceMgr SM; 646 647 // Read the expected strings from the check file. 648 std::vector<CheckString> CheckStrings; 649 if (ReadCheckFile(SM, CheckStrings)) 650 return 2; 651 652 // Open the file to check and add it to SourceMgr. 653 OwningPtr<MemoryBuffer> File; 654 if (error_code ec = 655 MemoryBuffer::getFileOrSTDIN(InputFilename.c_str(), File)) { 656 errs() << "Could not open input file '" << InputFilename << "': " 657 << ec.message() << '\n'; 658 return true; 659 } 660 MemoryBuffer *F = File.take(); 661 662 // Remove duplicate spaces in the input file if requested. 663 if (!NoCanonicalizeWhiteSpace) 664 F = CanonicalizeInputFile(F); 665 666 SM.AddNewSourceBuffer(F, SMLoc()); 667 668 /// VariableTable - This holds all the current filecheck variables. 669 StringMap<StringRef> VariableTable; 670 671 // Check that we have all of the expected strings, in order, in the input 672 // file. 673 StringRef Buffer = F->getBuffer(); 674 675 const char *LastMatch = Buffer.data(); 676 677 for (unsigned StrNo = 0, e = CheckStrings.size(); StrNo != e; ++StrNo) { 678 const CheckString &CheckStr = CheckStrings[StrNo]; 679 680 StringRef SearchFrom = Buffer; 681 682 // Find StrNo in the file. 683 size_t MatchLen = 0; 684 size_t MatchPos = CheckStr.Pat.Match(Buffer, MatchLen, VariableTable); 685 Buffer = Buffer.substr(MatchPos); 686 687 // If we didn't find a match, reject the input. 688 if (MatchPos == StringRef::npos) { 689 PrintCheckFailed(SM, CheckStr, SearchFrom, VariableTable); 690 return 1; 691 } 692 693 StringRef SkippedRegion(LastMatch, Buffer.data()-LastMatch); 694 695 // If this check is a "CHECK-NEXT", verify that the previous match was on 696 // the previous line (i.e. that there is one newline between them). 697 if (CheckStr.IsCheckNext) { 698 // Count the number of newlines between the previous match and this one. 699 assert(LastMatch != F->getBufferStart() && 700 "CHECK-NEXT can't be the first check in a file"); 701 702 unsigned NumNewLines = CountNumNewlinesBetween(SkippedRegion); 703 if (NumNewLines == 0) { 704 SM.PrintMessage(CheckStr.Loc, 705 CheckPrefix+"-NEXT: is on the same line as previous match", 706 "error"); 707 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), 708 "'next' match was here", "note"); 709 SM.PrintMessage(SMLoc::getFromPointer(LastMatch), 710 "previous match was here", "note"); 711 return 1; 712 } 713 714 if (NumNewLines != 1) { 715 SM.PrintMessage(CheckStr.Loc, 716 CheckPrefix+ 717 "-NEXT: is not on the line after the previous match", 718 "error"); 719 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), 720 "'next' match was here", "note"); 721 SM.PrintMessage(SMLoc::getFromPointer(LastMatch), 722 "previous match was here", "note"); 723 return 1; 724 } 725 } 726 727 // If this match had "not strings", verify that they don't exist in the 728 // skipped region. 729 for (unsigned ChunkNo = 0, e = CheckStr.NotStrings.size(); 730 ChunkNo != e; ++ChunkNo) { 731 size_t MatchLen = 0; 732 size_t Pos = CheckStr.NotStrings[ChunkNo].second.Match(SkippedRegion, 733 MatchLen, 734 VariableTable); 735 if (Pos == StringRef::npos) continue; 736 737 SM.PrintMessage(SMLoc::getFromPointer(LastMatch+Pos), 738 CheckPrefix+"-NOT: string occurred!", "error"); 739 SM.PrintMessage(CheckStr.NotStrings[ChunkNo].first, 740 CheckPrefix+"-NOT: pattern specified here", "note"); 741 return 1; 742 } 743 744 745 // Otherwise, everything is good. Step over the matched text and remember 746 // the position after the match as the end of the last match. 747 Buffer = Buffer.substr(MatchLen); 748 LastMatch = Buffer.data(); 749 } 750 751 return 0; 752 } 753