FileCheck.cpp revision 7112c86fc208145334aaa04812c794ce6feef416
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  // Only compare up to the first line in the buffer, or the string size.
344  StringRef BufferPrefix = Buffer.substr(0, ExampleString.size());
345  BufferPrefix = BufferPrefix.split('\n').first;
346  return BufferPrefix.edit_distance(ExampleString);
347}
348
349void Pattern::PrintFailureInfo(const SourceMgr &SM, StringRef Buffer,
350                               const StringMap<StringRef> &VariableTable) const{
351  // If this was a regular expression using variables, print the current
352  // variable values.
353  if (!VariableUses.empty()) {
354    for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) {
355      StringRef Var = VariableUses[i].first;
356      StringMap<StringRef>::const_iterator it = VariableTable.find(Var);
357      SmallString<256> Msg;
358      raw_svector_ostream OS(Msg);
359
360      // Check for undefined variable references.
361      if (it == VariableTable.end()) {
362        OS << "uses undefined variable \"";
363        OS.write_escaped(Var) << "\"";;
364      } else {
365        OS << "with variable \"";
366        OS.write_escaped(Var) << "\" equal to \"";
367        OS.write_escaped(it->second) << "\"";
368      }
369
370      SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), OS.str(), "note",
371                      /*ShowLine=*/false);
372    }
373  }
374
375  // Attempt to find the closest/best fuzzy match.  Usually an error happens
376  // because some string in the output didn't exactly match. In these cases, we
377  // would like to show the user a best guess at what "should have" matched, to
378  // save them having to actually check the input manually.
379  size_t NumLinesForward = 0;
380  size_t Best = StringRef::npos;
381  double BestQuality = 0;
382
383  // Use an arbitrary 4k limit on how far we will search.
384  for (size_t i = 0, e = std::min(size_t(4096), Buffer.size()); i != e; ++i) {
385    if (Buffer[i] == '\n')
386      ++NumLinesForward;
387
388    // Patterns have leading whitespace stripped, so skip whitespace when
389    // looking for something which looks like a pattern.
390    if (Buffer[i] == ' ' || Buffer[i] == '\t')
391      continue;
392
393    // Compute the "quality" of this match as an arbitrary combination of the
394    // match distance and the number of lines skipped to get to this match.
395    unsigned Distance = ComputeMatchDistance(Buffer.substr(i), VariableTable);
396    double Quality = Distance + (NumLinesForward / 100.);
397
398    if (Quality < BestQuality || Best == StringRef::npos) {
399      Best = i;
400      BestQuality = Quality;
401    }
402  }
403
404  // Print the "possible intended match here" line if we found something
405  // reasonable and not equal to what we showed in the "scanning from here"
406  // line.
407  if (Best && Best != StringRef::npos && BestQuality < 50) {
408      SM.PrintMessage(SMLoc::getFromPointer(Buffer.data() + Best),
409                      "possible intended match here", "note");
410
411    // FIXME: If we wanted to be really friendly we would show why the match
412    // failed, as it can be hard to spot simple one character differences.
413  }
414}
415
416//===----------------------------------------------------------------------===//
417// Check Strings.
418//===----------------------------------------------------------------------===//
419
420/// CheckString - This is a check that we found in the input file.
421struct CheckString {
422  /// Pat - The pattern to match.
423  Pattern Pat;
424
425  /// Loc - The location in the match file that the check string was specified.
426  SMLoc Loc;
427
428  /// IsCheckNext - This is true if this is a CHECK-NEXT: directive (as opposed
429  /// to a CHECK: directive.
430  bool IsCheckNext;
431
432  /// NotStrings - These are all of the strings that are disallowed from
433  /// occurring between this match string and the previous one (or start of
434  /// file).
435  std::vector<std::pair<SMLoc, Pattern> > NotStrings;
436
437  CheckString(const Pattern &P, SMLoc L, bool isCheckNext)
438    : Pat(P), Loc(L), IsCheckNext(isCheckNext) {}
439};
440
441/// CanonicalizeInputFile - Remove duplicate horizontal space from the specified
442/// memory buffer, free it, and return a new one.
443static MemoryBuffer *CanonicalizeInputFile(MemoryBuffer *MB) {
444  SmallString<128> NewFile;
445  NewFile.reserve(MB->getBufferSize());
446
447  for (const char *Ptr = MB->getBufferStart(), *End = MB->getBufferEnd();
448       Ptr != End; ++Ptr) {
449    // If C is not a horizontal whitespace, skip it.
450    if (*Ptr != ' ' && *Ptr != '\t') {
451      NewFile.push_back(*Ptr);
452      continue;
453    }
454
455    // Otherwise, add one space and advance over neighboring space.
456    NewFile.push_back(' ');
457    while (Ptr+1 != End &&
458           (Ptr[1] == ' ' || Ptr[1] == '\t'))
459      ++Ptr;
460  }
461
462  // Free the old buffer and return a new one.
463  MemoryBuffer *MB2 =
464    MemoryBuffer::getMemBufferCopy(NewFile.str(), MB->getBufferIdentifier());
465
466  delete MB;
467  return MB2;
468}
469
470
471/// ReadCheckFile - Read the check file, which specifies the sequence of
472/// expected strings.  The strings are added to the CheckStrings vector.
473static bool ReadCheckFile(SourceMgr &SM,
474                          std::vector<CheckString> &CheckStrings) {
475  // Open the check file, and tell SourceMgr about it.
476  std::string ErrorStr;
477  MemoryBuffer *F =
478    MemoryBuffer::getFileOrSTDIN(CheckFilename.c_str(), &ErrorStr);
479  if (F == 0) {
480    errs() << "Could not open check file '" << CheckFilename << "': "
481           << ErrorStr << '\n';
482    return true;
483  }
484
485  // If we want to canonicalize whitespace, strip excess whitespace from the
486  // buffer containing the CHECK lines.
487  if (!NoCanonicalizeWhiteSpace)
488    F = CanonicalizeInputFile(F);
489
490  SM.AddNewSourceBuffer(F, SMLoc());
491
492  // Find all instances of CheckPrefix followed by : in the file.
493  StringRef Buffer = F->getBuffer();
494
495  std::vector<std::pair<SMLoc, Pattern> > NotMatches;
496
497  while (1) {
498    // See if Prefix occurs in the memory buffer.
499    Buffer = Buffer.substr(Buffer.find(CheckPrefix));
500
501    // If we didn't find a match, we're done.
502    if (Buffer.empty())
503      break;
504
505    const char *CheckPrefixStart = Buffer.data();
506
507    // When we find a check prefix, keep track of whether we find CHECK: or
508    // CHECK-NEXT:
509    bool IsCheckNext = false, IsCheckNot = false;
510
511    // Verify that the : is present after the prefix.
512    if (Buffer[CheckPrefix.size()] == ':') {
513      Buffer = Buffer.substr(CheckPrefix.size()+1);
514    } else if (Buffer.size() > CheckPrefix.size()+6 &&
515               memcmp(Buffer.data()+CheckPrefix.size(), "-NEXT:", 6) == 0) {
516      Buffer = Buffer.substr(CheckPrefix.size()+7);
517      IsCheckNext = true;
518    } else if (Buffer.size() > CheckPrefix.size()+5 &&
519               memcmp(Buffer.data()+CheckPrefix.size(), "-NOT:", 5) == 0) {
520      Buffer = Buffer.substr(CheckPrefix.size()+6);
521      IsCheckNot = true;
522    } else {
523      Buffer = Buffer.substr(1);
524      continue;
525    }
526
527    // Okay, we found the prefix, yay.  Remember the rest of the line, but
528    // ignore leading and trailing whitespace.
529    Buffer = Buffer.substr(Buffer.find_first_not_of(" \t"));
530
531    // Scan ahead to the end of line.
532    size_t EOL = Buffer.find_first_of("\n\r");
533
534    // Remember the location of the start of the pattern, for diagnostics.
535    SMLoc PatternLoc = SMLoc::getFromPointer(Buffer.data());
536
537    // Parse the pattern.
538    Pattern P;
539    if (P.ParsePattern(Buffer.substr(0, EOL), SM))
540      return true;
541
542    Buffer = Buffer.substr(EOL);
543
544
545    // Verify that CHECK-NEXT lines have at least one CHECK line before them.
546    if (IsCheckNext && CheckStrings.empty()) {
547      SM.PrintMessage(SMLoc::getFromPointer(CheckPrefixStart),
548                      "found '"+CheckPrefix+"-NEXT:' without previous '"+
549                      CheckPrefix+ ": line", "error");
550      return true;
551    }
552
553    // Handle CHECK-NOT.
554    if (IsCheckNot) {
555      NotMatches.push_back(std::make_pair(SMLoc::getFromPointer(Buffer.data()),
556                                          P));
557      continue;
558    }
559
560
561    // Okay, add the string we captured to the output vector and move on.
562    CheckStrings.push_back(CheckString(P,
563                                       PatternLoc,
564                                       IsCheckNext));
565    std::swap(NotMatches, CheckStrings.back().NotStrings);
566  }
567
568  if (CheckStrings.empty()) {
569    errs() << "error: no check strings found with prefix '" << CheckPrefix
570           << ":'\n";
571    return true;
572  }
573
574  if (!NotMatches.empty()) {
575    errs() << "error: '" << CheckPrefix
576           << "-NOT:' not supported after last check line.\n";
577    return true;
578  }
579
580  return false;
581}
582
583static void PrintCheckFailed(const SourceMgr &SM, const CheckString &CheckStr,
584                             StringRef Buffer,
585                             StringMap<StringRef> &VariableTable) {
586  // Otherwise, we have an error, emit an error message.
587  SM.PrintMessage(CheckStr.Loc, "expected string not found in input",
588                  "error");
589
590  // Print the "scanning from here" line.  If the current position is at the
591  // end of a line, advance to the start of the next line.
592  Buffer = Buffer.substr(Buffer.find_first_not_of(" \t\n\r"));
593
594  SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), "scanning from here",
595                  "note");
596
597  // Allow the pattern to print additional information if desired.
598  CheckStr.Pat.PrintFailureInfo(SM, Buffer, VariableTable);
599}
600
601/// CountNumNewlinesBetween - Count the number of newlines in the specified
602/// range.
603static unsigned CountNumNewlinesBetween(StringRef Range) {
604  unsigned NumNewLines = 0;
605  while (1) {
606    // Scan for newline.
607    Range = Range.substr(Range.find_first_of("\n\r"));
608    if (Range.empty()) return NumNewLines;
609
610    ++NumNewLines;
611
612    // Handle \n\r and \r\n as a single newline.
613    if (Range.size() > 1 &&
614        (Range[1] == '\n' || Range[1] == '\r') &&
615        (Range[0] != Range[1]))
616      Range = Range.substr(1);
617    Range = Range.substr(1);
618  }
619}
620
621int main(int argc, char **argv) {
622  sys::PrintStackTraceOnErrorSignal();
623  PrettyStackTraceProgram X(argc, argv);
624  cl::ParseCommandLineOptions(argc, argv);
625
626  SourceMgr SM;
627
628  // Read the expected strings from the check file.
629  std::vector<CheckString> CheckStrings;
630  if (ReadCheckFile(SM, CheckStrings))
631    return 2;
632
633  // Open the file to check and add it to SourceMgr.
634  std::string ErrorStr;
635  MemoryBuffer *F =
636    MemoryBuffer::getFileOrSTDIN(InputFilename.c_str(), &ErrorStr);
637  if (F == 0) {
638    errs() << "Could not open input file '" << InputFilename << "': "
639           << ErrorStr << '\n';
640    return true;
641  }
642
643  // Remove duplicate spaces in the input file if requested.
644  if (!NoCanonicalizeWhiteSpace)
645    F = CanonicalizeInputFile(F);
646
647  SM.AddNewSourceBuffer(F, SMLoc());
648
649  /// VariableTable - This holds all the current filecheck variables.
650  StringMap<StringRef> VariableTable;
651
652  // Check that we have all of the expected strings, in order, in the input
653  // file.
654  StringRef Buffer = F->getBuffer();
655
656  const char *LastMatch = Buffer.data();
657
658  for (unsigned StrNo = 0, e = CheckStrings.size(); StrNo != e; ++StrNo) {
659    const CheckString &CheckStr = CheckStrings[StrNo];
660
661    StringRef SearchFrom = Buffer;
662
663    // Find StrNo in the file.
664    size_t MatchLen = 0;
665    Buffer = Buffer.substr(CheckStr.Pat.Match(Buffer, MatchLen, VariableTable));
666
667    // If we didn't find a match, reject the input.
668    if (Buffer.empty()) {
669      PrintCheckFailed(SM, CheckStr, SearchFrom, VariableTable);
670      return 1;
671    }
672
673    StringRef SkippedRegion(LastMatch, Buffer.data()-LastMatch);
674
675    // If this check is a "CHECK-NEXT", verify that the previous match was on
676    // the previous line (i.e. that there is one newline between them).
677    if (CheckStr.IsCheckNext) {
678      // Count the number of newlines between the previous match and this one.
679      assert(LastMatch != F->getBufferStart() &&
680             "CHECK-NEXT can't be the first check in a file");
681
682      unsigned NumNewLines = CountNumNewlinesBetween(SkippedRegion);
683      if (NumNewLines == 0) {
684        SM.PrintMessage(CheckStr.Loc,
685                    CheckPrefix+"-NEXT: is on the same line as previous match",
686                        "error");
687        SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
688                        "'next' match was here", "note");
689        SM.PrintMessage(SMLoc::getFromPointer(LastMatch),
690                        "previous match was here", "note");
691        return 1;
692      }
693
694      if (NumNewLines != 1) {
695        SM.PrintMessage(CheckStr.Loc,
696                        CheckPrefix+
697                        "-NEXT: is not on the line after the previous match",
698                        "error");
699        SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
700                        "'next' match was here", "note");
701        SM.PrintMessage(SMLoc::getFromPointer(LastMatch),
702                        "previous match was here", "note");
703        return 1;
704      }
705    }
706
707    // If this match had "not strings", verify that they don't exist in the
708    // skipped region.
709    for (unsigned ChunkNo = 0, e = CheckStr.NotStrings.size();
710         ChunkNo != e; ++ChunkNo) {
711      size_t MatchLen = 0;
712      size_t Pos = CheckStr.NotStrings[ChunkNo].second.Match(SkippedRegion,
713                                                             MatchLen,
714                                                             VariableTable);
715      if (Pos == StringRef::npos) continue;
716
717      SM.PrintMessage(SMLoc::getFromPointer(LastMatch+Pos),
718                      CheckPrefix+"-NOT: string occurred!", "error");
719      SM.PrintMessage(CheckStr.NotStrings[ChunkNo].first,
720                      CheckPrefix+"-NOT: pattern specified here", "note");
721      return 1;
722    }
723
724
725    // Otherwise, everything is good.  Step over the matched text and remember
726    // the position after the match as the end of the last match.
727    Buffer = Buffer.substr(MatchLen);
728    LastMatch = Buffer.data();
729  }
730
731  return 0;
732}
733