FileCheck.cpp revision e3a1e506809acbddceb6840914879c498d454f26
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>
29using namespace llvm;
30
31static cl::opt<std::string>
32CheckFilename(cl::Positional, cl::desc("<check-file>"), cl::Required);
33
34static cl::opt<std::string>
35InputFilename("input-file", cl::desc("File to check (defaults to stdin)"),
36              cl::init("-"), cl::value_desc("filename"));
37
38static cl::opt<std::string>
39CheckPrefix("check-prefix", cl::init("CHECK"),
40            cl::desc("Prefix to use from check file (defaults to 'CHECK')"));
41
42static cl::opt<bool>
43NoCanonicalizeWhiteSpace("strict-whitespace",
44              cl::desc("Do not treat all horizontal whitespace as equivalent"));
45
46//===----------------------------------------------------------------------===//
47// Pattern Handling Code.
48//===----------------------------------------------------------------------===//
49
50class Pattern {
51  SMLoc PatternLoc;
52
53  /// FixedStr - If non-empty, this pattern is a fixed string match with the
54  /// specified fixed string.
55  StringRef FixedStr;
56
57  /// RegEx - If non-empty, this is a regex pattern.
58  std::string RegExStr;
59
60  /// VariableUses - Entries in this vector map to uses of a variable in the
61  /// pattern, e.g. "foo[[bar]]baz".  In this case, the RegExStr will contain
62  /// "foobaz" and we'll get an entry in this vector that tells us to insert the
63  /// value of bar at offset 3.
64  std::vector<std::pair<StringRef, unsigned> > VariableUses;
65
66  /// VariableDefs - Entries in this vector map to definitions of a variable in
67  /// the pattern, e.g. "foo[[bar:.*]]baz".  In this case, the RegExStr will
68  /// contain "foo(.*)baz" and VariableDefs will contain the pair "bar",1.  The
69  /// index indicates what parenthesized value captures the variable value.
70  std::vector<std::pair<StringRef, unsigned> > VariableDefs;
71
72public:
73
74  Pattern() { }
75
76  bool ParsePattern(StringRef PatternStr, SourceMgr &SM);
77
78  /// Match - Match the pattern string against the input buffer Buffer.  This
79  /// returns the position that is matched or npos if there is no match.  If
80  /// there is a match, the size of the matched string is returned in MatchLen.
81  ///
82  /// The VariableTable StringMap provides the current values of filecheck
83  /// variables and is updated if this match defines new values.
84  size_t Match(StringRef Buffer, size_t &MatchLen,
85               StringMap<StringRef> &VariableTable) const;
86
87  /// PrintFailureInfo - Print additional information about a failure to match
88  /// involving this pattern.
89  void PrintFailureInfo(const SourceMgr &SM, StringRef Buffer,
90                        const StringMap<StringRef> &VariableTable) const;
91
92private:
93  static void AddFixedStringToRegEx(StringRef FixedStr, std::string &TheStr);
94  bool AddRegExToRegEx(StringRef RegExStr, unsigned &CurParen, SourceMgr &SM);
95
96  /// ComputeMatchDistance - Compute an arbitrary estimate for the quality of
97  /// matching this pattern at the start of \arg Buffer; a distance of zero
98  /// should correspond to a perfect match.
99  unsigned ComputeMatchDistance(StringRef Buffer,
100                               const StringMap<StringRef> &VariableTable) const;
101};
102
103
104bool Pattern::ParsePattern(StringRef PatternStr, SourceMgr &SM) {
105  PatternLoc = SMLoc::getFromPointer(PatternStr.data());
106
107  // Ignore trailing whitespace.
108  while (!PatternStr.empty() &&
109         (PatternStr.back() == ' ' || PatternStr.back() == '\t'))
110    PatternStr = PatternStr.substr(0, PatternStr.size()-1);
111
112  // Check that there is something on the line.
113  if (PatternStr.empty()) {
114    SM.PrintMessage(PatternLoc, "found empty check string with prefix '" +
115                    CheckPrefix+":'", "error");
116    return true;
117  }
118
119  // Check to see if this is a fixed string, or if it has regex pieces.
120  if (PatternStr.size() < 2 ||
121      (PatternStr.find("{{") == StringRef::npos &&
122       PatternStr.find("[[") == StringRef::npos)) {
123    FixedStr = PatternStr;
124    return false;
125  }
126
127  // Paren value #0 is for the fully matched string.  Any new parenthesized
128  // values add from their.
129  unsigned CurParen = 1;
130
131  // Otherwise, there is at least one regex piece.  Build up the regex pattern
132  // by escaping scary characters in fixed strings, building up one big regex.
133  while (!PatternStr.empty()) {
134    // RegEx matches.
135    if (PatternStr.size() >= 2 &&
136        PatternStr[0] == '{' && PatternStr[1] == '{') {
137
138      // Otherwise, this is the start of a regex match.  Scan for the }}.
139      size_t End = PatternStr.find("}}");
140      if (End == StringRef::npos) {
141        SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()),
142                        "found start of regex string with no end '}}'", "error");
143        return true;
144      }
145
146      if (AddRegExToRegEx(PatternStr.substr(2, End-2), CurParen, SM))
147        return true;
148      PatternStr = PatternStr.substr(End+2);
149      continue;
150    }
151
152    // Named RegEx matches.  These are of two forms: [[foo:.*]] which matches .*
153    // (or some other regex) and assigns it to the FileCheck variable 'foo'. The
154    // second form is [[foo]] which is a reference to foo.  The variable name
155    // itself must be of the form "[a-zA-Z_][0-9a-zA-Z_]*", otherwise we reject
156    // it.  This is to catch some common errors.
157    if (PatternStr.size() >= 2 &&
158        PatternStr[0] == '[' && PatternStr[1] == '[') {
159      // Verify that it is terminated properly.
160      size_t End = PatternStr.find("]]");
161      if (End == StringRef::npos) {
162        SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()),
163                        "invalid named regex reference, no ]] found", "error");
164        return true;
165      }
166
167      StringRef MatchStr = PatternStr.substr(2, End-2);
168      PatternStr = PatternStr.substr(End+2);
169
170      // Get the regex name (e.g. "foo").
171      size_t NameEnd = MatchStr.find(':');
172      StringRef Name = MatchStr.substr(0, NameEnd);
173
174      if (Name.empty()) {
175        SM.PrintMessage(SMLoc::getFromPointer(Name.data()),
176                        "invalid name in named regex: empty name", "error");
177        return true;
178      }
179
180      // Verify that the name is well formed.
181      for (unsigned i = 0, e = Name.size(); i != e; ++i)
182        if (Name[i] != '_' &&
183            (Name[i] < 'a' || Name[i] > 'z') &&
184            (Name[i] < 'A' || Name[i] > 'Z') &&
185            (Name[i] < '0' || Name[i] > '9')) {
186          SM.PrintMessage(SMLoc::getFromPointer(Name.data()+i),
187                          "invalid name in named regex", "error");
188          return true;
189        }
190
191      // Name can't start with a digit.
192      if (isdigit(Name[0])) {
193        SM.PrintMessage(SMLoc::getFromPointer(Name.data()),
194                        "invalid name in named regex", "error");
195        return true;
196      }
197
198      // Handle [[foo]].
199      if (NameEnd == StringRef::npos) {
200        VariableUses.push_back(std::make_pair(Name, RegExStr.size()));
201        continue;
202      }
203
204      // Handle [[foo:.*]].
205      VariableDefs.push_back(std::make_pair(Name, CurParen));
206      RegExStr += '(';
207      ++CurParen;
208
209      if (AddRegExToRegEx(MatchStr.substr(NameEnd+1), CurParen, SM))
210        return true;
211
212      RegExStr += ')';
213    }
214
215    // Handle fixed string matches.
216    // Find the end, which is the start of the next regex.
217    size_t FixedMatchEnd = PatternStr.find("{{");
218    FixedMatchEnd = std::min(FixedMatchEnd, PatternStr.find("[["));
219    AddFixedStringToRegEx(PatternStr.substr(0, FixedMatchEnd), RegExStr);
220    PatternStr = PatternStr.substr(FixedMatchEnd);
221    continue;
222  }
223
224  return false;
225}
226
227void Pattern::AddFixedStringToRegEx(StringRef FixedStr, std::string &TheStr) {
228  // Add the characters from FixedStr to the regex, escaping as needed.  This
229  // avoids "leaning toothpicks" in common patterns.
230  for (unsigned i = 0, e = FixedStr.size(); i != e; ++i) {
231    switch (FixedStr[i]) {
232    // These are the special characters matched in "p_ere_exp".
233    case '(':
234    case ')':
235    case '^':
236    case '$':
237    case '|':
238    case '*':
239    case '+':
240    case '?':
241    case '.':
242    case '[':
243    case '\\':
244    case '{':
245      TheStr += '\\';
246      // FALL THROUGH.
247    default:
248      TheStr += FixedStr[i];
249      break;
250    }
251  }
252}
253
254bool Pattern::AddRegExToRegEx(StringRef RegexStr, unsigned &CurParen,
255                              SourceMgr &SM) {
256  Regex R(RegexStr);
257  std::string Error;
258  if (!R.isValid(Error)) {
259    SM.PrintMessage(SMLoc::getFromPointer(RegexStr.data()),
260                    "invalid regex: " + Error, "error");
261    return true;
262  }
263
264  RegExStr += RegexStr.str();
265  CurParen += R.getNumMatches();
266  return false;
267}
268
269/// Match - Match the pattern string against the input buffer Buffer.  This
270/// returns the position that is matched or npos if there is no match.  If
271/// there is a match, the size of the matched string is returned in MatchLen.
272size_t Pattern::Match(StringRef Buffer, size_t &MatchLen,
273                      StringMap<StringRef> &VariableTable) const {
274  // If this is a fixed string pattern, just match it now.
275  if (!FixedStr.empty()) {
276    MatchLen = FixedStr.size();
277    return Buffer.find(FixedStr);
278  }
279
280  // Regex match.
281
282  // If there are variable uses, we need to create a temporary string with the
283  // actual value.
284  StringRef RegExToMatch = RegExStr;
285  std::string TmpStr;
286  if (!VariableUses.empty()) {
287    TmpStr = RegExStr;
288
289    unsigned InsertOffset = 0;
290    for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) {
291      StringMap<StringRef>::iterator it =
292        VariableTable.find(VariableUses[i].first);
293      // If the variable is undefined, return an error.
294      if (it == VariableTable.end())
295        return StringRef::npos;
296
297      // Look up the value and escape it so that we can plop it into the regex.
298      std::string Value;
299      AddFixedStringToRegEx(it->second, Value);
300
301      // Plop it into the regex at the adjusted offset.
302      TmpStr.insert(TmpStr.begin()+VariableUses[i].second+InsertOffset,
303                    Value.begin(), Value.end());
304      InsertOffset += Value.size();
305    }
306
307    // Match the newly constructed regex.
308    RegExToMatch = TmpStr;
309  }
310
311
312  SmallVector<StringRef, 4> MatchInfo;
313  if (!Regex(RegExToMatch, Regex::Newline).match(Buffer, &MatchInfo))
314    return StringRef::npos;
315
316  // Successful regex match.
317  assert(!MatchInfo.empty() && "Didn't get any match");
318  StringRef FullMatch = MatchInfo[0];
319
320  // If this defines any variables, remember their values.
321  for (unsigned i = 0, e = VariableDefs.size(); i != e; ++i) {
322    assert(VariableDefs[i].second < MatchInfo.size() &&
323           "Internal paren error");
324    VariableTable[VariableDefs[i].first] = MatchInfo[VariableDefs[i].second];
325  }
326
327  MatchLen = FullMatch.size();
328  return FullMatch.data()-Buffer.data();
329}
330
331unsigned Pattern::ComputeMatchDistance(StringRef Buffer,
332                              const StringMap<StringRef> &VariableTable) const {
333  // Just compute the number of matching characters. For regular expressions, we
334  // just compare against the regex itself and hope for the best.
335  //
336  // FIXME: One easy improvement here is have the regex lib generate a single
337  // example regular expression which matches, and use that as the example
338  // string.
339  StringRef ExampleString(FixedStr);
340  if (ExampleString.empty())
341    ExampleString = RegExStr;
342
343  return Buffer.substr(0, ExampleString.size()).edit_distance(ExampleString);
344}
345
346void Pattern::PrintFailureInfo(const SourceMgr &SM, StringRef Buffer,
347                               const StringMap<StringRef> &VariableTable) const{
348  // If this was a regular expression using variables, print the current
349  // variable values.
350  if (!VariableUses.empty()) {
351    for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) {
352      StringRef Var = VariableUses[i].first;
353      StringMap<StringRef>::const_iterator it = VariableTable.find(Var);
354      SmallString<256> Msg;
355      raw_svector_ostream OS(Msg);
356
357      // Check for undefined variable references.
358      if (it == VariableTable.end()) {
359        OS << "uses undefined variable \"";
360        OS.write_escaped(Var) << "\"";;
361      } else {
362        OS << "with variable \"";
363        OS.write_escaped(Var) << "\" equal to \"";
364        OS.write_escaped(it->second) << "\"";
365      }
366
367      SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), OS.str(), "note",
368                      /*ShowLine=*/false);
369    }
370  }
371
372  // Attempt to find the closest/best fuzzy match.  Usually an error happens
373  // because some string in the output didn't exactly match. In these cases, we
374  // would like to show the user a best guess at what "should have" matched, to
375  // save them having to actually check the input manually.
376  size_t NumLinesForward = 0;
377  size_t Best = StringRef::npos;
378  double BestQuality = 0;
379
380  // Use an arbitrary 4k limit on how far we will search.
381  for (size_t i = 0, e = std::min(size_t(4096), Buffer.size()); i != e; ++i) {
382    if (Buffer[i] == '\n')
383      ++NumLinesForward;
384
385    // Patterns have leading whitespace stripped, so skip whitespace when
386    // looking for something which looks like a pattern.
387    if (Buffer[i] == ' ' || Buffer[i] == '\t')
388      continue;
389
390    // Compute the "quality" of this match as an arbitrary combination of the
391    // match distance and the number of lines skipped to get to this match.
392    unsigned Distance = ComputeMatchDistance(Buffer.substr(i), VariableTable);
393    double Quality = Distance + (NumLinesForward / 100.);
394
395    if (Quality < BestQuality || Best == StringRef::npos) {
396      Best = i;
397      BestQuality = Quality;
398    }
399  }
400
401  if (Best != StringRef::npos && BestQuality < 50) {
402    // Print the "possible intended match here" line if we found something
403    // reasonable.
404    SM.PrintMessage(SMLoc::getFromPointer(Buffer.data() + Best),
405                    "possible intended match here", "note");
406
407    // FIXME: If we wanted to be really friendly we would show why the match
408    // failed, as it can be hard to spot simple one character differences.
409  }
410}
411
412//===----------------------------------------------------------------------===//
413// Check Strings.
414//===----------------------------------------------------------------------===//
415
416/// CheckString - This is a check that we found in the input file.
417struct CheckString {
418  /// Pat - The pattern to match.
419  Pattern Pat;
420
421  /// Loc - The location in the match file that the check string was specified.
422  SMLoc Loc;
423
424  /// IsCheckNext - This is true if this is a CHECK-NEXT: directive (as opposed
425  /// to a CHECK: directive.
426  bool IsCheckNext;
427
428  /// NotStrings - These are all of the strings that are disallowed from
429  /// occurring between this match string and the previous one (or start of
430  /// file).
431  std::vector<std::pair<SMLoc, Pattern> > NotStrings;
432
433  CheckString(const Pattern &P, SMLoc L, bool isCheckNext)
434    : Pat(P), Loc(L), IsCheckNext(isCheckNext) {}
435};
436
437/// CanonicalizeInputFile - Remove duplicate horizontal space from the specified
438/// memory buffer, free it, and return a new one.
439static MemoryBuffer *CanonicalizeInputFile(MemoryBuffer *MB) {
440  SmallVector<char, 16> NewFile;
441  NewFile.reserve(MB->getBufferSize());
442
443  for (const char *Ptr = MB->getBufferStart(), *End = MB->getBufferEnd();
444       Ptr != End; ++Ptr) {
445    // If C is not a horizontal whitespace, skip it.
446    if (*Ptr != ' ' && *Ptr != '\t') {
447      NewFile.push_back(*Ptr);
448      continue;
449    }
450
451    // Otherwise, add one space and advance over neighboring space.
452    NewFile.push_back(' ');
453    while (Ptr+1 != End &&
454           (Ptr[1] == ' ' || Ptr[1] == '\t'))
455      ++Ptr;
456  }
457
458  // Free the old buffer and return a new one.
459  MemoryBuffer *MB2 =
460    MemoryBuffer::getMemBufferCopy(NewFile.data(),
461                                   NewFile.data() + NewFile.size(),
462                                   MB->getBufferIdentifier());
463
464  delete MB;
465  return MB2;
466}
467
468
469/// ReadCheckFile - Read the check file, which specifies the sequence of
470/// expected strings.  The strings are added to the CheckStrings vector.
471static bool ReadCheckFile(SourceMgr &SM,
472                          std::vector<CheckString> &CheckStrings) {
473  // Open the check file, and tell SourceMgr about it.
474  std::string ErrorStr;
475  MemoryBuffer *F =
476    MemoryBuffer::getFileOrSTDIN(CheckFilename.c_str(), &ErrorStr);
477  if (F == 0) {
478    errs() << "Could not open check file '" << CheckFilename << "': "
479           << ErrorStr << '\n';
480    return true;
481  }
482
483  // If we want to canonicalize whitespace, strip excess whitespace from the
484  // buffer containing the CHECK lines.
485  if (!NoCanonicalizeWhiteSpace)
486    F = CanonicalizeInputFile(F);
487
488  SM.AddNewSourceBuffer(F, SMLoc());
489
490  // Find all instances of CheckPrefix followed by : in the file.
491  StringRef Buffer = F->getBuffer();
492
493  std::vector<std::pair<SMLoc, Pattern> > NotMatches;
494
495  while (1) {
496    // See if Prefix occurs in the memory buffer.
497    Buffer = Buffer.substr(Buffer.find(CheckPrefix));
498
499    // If we didn't find a match, we're done.
500    if (Buffer.empty())
501      break;
502
503    const char *CheckPrefixStart = Buffer.data();
504
505    // When we find a check prefix, keep track of whether we find CHECK: or
506    // CHECK-NEXT:
507    bool IsCheckNext = false, IsCheckNot = false;
508
509    // Verify that the : is present after the prefix.
510    if (Buffer[CheckPrefix.size()] == ':') {
511      Buffer = Buffer.substr(CheckPrefix.size()+1);
512    } else if (Buffer.size() > CheckPrefix.size()+6 &&
513               memcmp(Buffer.data()+CheckPrefix.size(), "-NEXT:", 6) == 0) {
514      Buffer = Buffer.substr(CheckPrefix.size()+7);
515      IsCheckNext = true;
516    } else if (Buffer.size() > CheckPrefix.size()+5 &&
517               memcmp(Buffer.data()+CheckPrefix.size(), "-NOT:", 5) == 0) {
518      Buffer = Buffer.substr(CheckPrefix.size()+6);
519      IsCheckNot = true;
520    } else {
521      Buffer = Buffer.substr(1);
522      continue;
523    }
524
525    // Okay, we found the prefix, yay.  Remember the rest of the line, but
526    // ignore leading and trailing whitespace.
527    Buffer = Buffer.substr(Buffer.find_first_not_of(" \t"));
528
529    // Scan ahead to the end of line.
530    size_t EOL = Buffer.find_first_of("\n\r");
531
532    // Remember the location of the start of the pattern, for diagnostics.
533    SMLoc PatternLoc = SMLoc::getFromPointer(Buffer.data());
534
535    // Parse the pattern.
536    Pattern P;
537    if (P.ParsePattern(Buffer.substr(0, EOL), SM))
538      return true;
539
540    Buffer = Buffer.substr(EOL);
541
542
543    // Verify that CHECK-NEXT lines have at least one CHECK line before them.
544    if (IsCheckNext && CheckStrings.empty()) {
545      SM.PrintMessage(SMLoc::getFromPointer(CheckPrefixStart),
546                      "found '"+CheckPrefix+"-NEXT:' without previous '"+
547                      CheckPrefix+ ": line", "error");
548      return true;
549    }
550
551    // Handle CHECK-NOT.
552    if (IsCheckNot) {
553      NotMatches.push_back(std::make_pair(SMLoc::getFromPointer(Buffer.data()),
554                                          P));
555      continue;
556    }
557
558
559    // Okay, add the string we captured to the output vector and move on.
560    CheckStrings.push_back(CheckString(P,
561                                       PatternLoc,
562                                       IsCheckNext));
563    std::swap(NotMatches, CheckStrings.back().NotStrings);
564  }
565
566  if (CheckStrings.empty()) {
567    errs() << "error: no check strings found with prefix '" << CheckPrefix
568           << ":'\n";
569    return true;
570  }
571
572  if (!NotMatches.empty()) {
573    errs() << "error: '" << CheckPrefix
574           << "-NOT:' not supported after last check line.\n";
575    return true;
576  }
577
578  return false;
579}
580
581static void PrintCheckFailed(const SourceMgr &SM, const CheckString &CheckStr,
582                             StringRef Buffer,
583                             StringMap<StringRef> &VariableTable) {
584  // Otherwise, we have an error, emit an error message.
585  SM.PrintMessage(CheckStr.Loc, "expected string not found in input",
586                  "error");
587
588  // Print the "scanning from here" line.  If the current position is at the
589  // end of a line, advance to the start of the next line.
590  Buffer = Buffer.substr(Buffer.find_first_not_of(" \t\n\r"));
591
592  SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), "scanning from here",
593                  "note");
594
595  // Allow the pattern to print additional information if desired.
596  CheckStr.Pat.PrintFailureInfo(SM, Buffer, VariableTable);
597}
598
599/// CountNumNewlinesBetween - Count the number of newlines in the specified
600/// range.
601static unsigned CountNumNewlinesBetween(StringRef Range) {
602  unsigned NumNewLines = 0;
603  while (1) {
604    // Scan for newline.
605    Range = Range.substr(Range.find_first_of("\n\r"));
606    if (Range.empty()) return NumNewLines;
607
608    ++NumNewLines;
609
610    // Handle \n\r and \r\n as a single newline.
611    if (Range.size() > 1 &&
612        (Range[1] == '\n' || Range[1] == '\r') &&
613        (Range[0] != Range[1]))
614      Range = Range.substr(1);
615    Range = Range.substr(1);
616  }
617}
618
619int main(int argc, char **argv) {
620  sys::PrintStackTraceOnErrorSignal();
621  PrettyStackTraceProgram X(argc, argv);
622  cl::ParseCommandLineOptions(argc, argv);
623
624  SourceMgr SM;
625
626  // Read the expected strings from the check file.
627  std::vector<CheckString> CheckStrings;
628  if (ReadCheckFile(SM, CheckStrings))
629    return 2;
630
631  // Open the file to check and add it to SourceMgr.
632  std::string ErrorStr;
633  MemoryBuffer *F =
634    MemoryBuffer::getFileOrSTDIN(InputFilename.c_str(), &ErrorStr);
635  if (F == 0) {
636    errs() << "Could not open input file '" << InputFilename << "': "
637           << ErrorStr << '\n';
638    return true;
639  }
640
641  // Remove duplicate spaces in the input file if requested.
642  if (!NoCanonicalizeWhiteSpace)
643    F = CanonicalizeInputFile(F);
644
645  SM.AddNewSourceBuffer(F, SMLoc());
646
647  /// VariableTable - This holds all the current filecheck variables.
648  StringMap<StringRef> VariableTable;
649
650  // Check that we have all of the expected strings, in order, in the input
651  // file.
652  StringRef Buffer = F->getBuffer();
653
654  const char *LastMatch = Buffer.data();
655
656  for (unsigned StrNo = 0, e = CheckStrings.size(); StrNo != e; ++StrNo) {
657    const CheckString &CheckStr = CheckStrings[StrNo];
658
659    StringRef SearchFrom = Buffer;
660
661    // Find StrNo in the file.
662    size_t MatchLen = 0;
663    Buffer = Buffer.substr(CheckStr.Pat.Match(Buffer, MatchLen, VariableTable));
664
665    // If we didn't find a match, reject the input.
666    if (Buffer.empty()) {
667      PrintCheckFailed(SM, CheckStr, SearchFrom, VariableTable);
668      return 1;
669    }
670
671    StringRef SkippedRegion(LastMatch, Buffer.data()-LastMatch);
672
673    // If this check is a "CHECK-NEXT", verify that the previous match was on
674    // the previous line (i.e. that there is one newline between them).
675    if (CheckStr.IsCheckNext) {
676      // Count the number of newlines between the previous match and this one.
677      assert(LastMatch != F->getBufferStart() &&
678             "CHECK-NEXT can't be the first check in a file");
679
680      unsigned NumNewLines = CountNumNewlinesBetween(SkippedRegion);
681      if (NumNewLines == 0) {
682        SM.PrintMessage(CheckStr.Loc,
683                    CheckPrefix+"-NEXT: is on the same line as previous match",
684                        "error");
685        SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
686                        "'next' match was here", "note");
687        SM.PrintMessage(SMLoc::getFromPointer(LastMatch),
688                        "previous match was here", "note");
689        return 1;
690      }
691
692      if (NumNewLines != 1) {
693        SM.PrintMessage(CheckStr.Loc,
694                        CheckPrefix+
695                        "-NEXT: is not on the line after the previous match",
696                        "error");
697        SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
698                        "'next' match was here", "note");
699        SM.PrintMessage(SMLoc::getFromPointer(LastMatch),
700                        "previous match was here", "note");
701        return 1;
702      }
703    }
704
705    // If this match had "not strings", verify that they don't exist in the
706    // skipped region.
707    for (unsigned ChunkNo = 0, e = CheckStr.NotStrings.size();
708         ChunkNo != e; ++ChunkNo) {
709      size_t MatchLen = 0;
710      size_t Pos = CheckStr.NotStrings[ChunkNo].second.Match(SkippedRegion,
711                                                             MatchLen,
712                                                             VariableTable);
713      if (Pos == StringRef::npos) continue;
714
715      SM.PrintMessage(SMLoc::getFromPointer(LastMatch+Pos),
716                      CheckPrefix+"-NOT: string occurred!", "error");
717      SM.PrintMessage(CheckStr.NotStrings[ChunkNo].first,
718                      CheckPrefix+"-NOT: pattern specified here", "note");
719      return 1;
720    }
721
722
723    // Otherwise, everything is good.  Step over the matched text and remember
724    // the position after the match as the end of the last match.
725    Buffer = Buffer.substr(MatchLen);
726    LastMatch = Buffer.data();
727  }
728
729  return 0;
730}
731