FileCheck.cpp revision 13a38c4cb4c90a6d2b5313a6517d6291cb924730
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>
31using namespace llvm;
32
33static cl::opt<std::string>
34CheckFilename(cl::Positional, cl::desc("<check-file>"), cl::Required);
35
36static cl::opt<std::string>
37InputFilename("input-file", cl::desc("File to check (defaults to stdin)"),
38              cl::init("-"), cl::value_desc("filename"));
39
40static cl::opt<std::string>
41CheckPrefix("check-prefix", cl::init("CHECK"),
42            cl::desc("Prefix to use from check file (defaults to 'CHECK')"));
43
44static cl::opt<bool>
45NoCanonicalizeWhiteSpace("strict-whitespace",
46              cl::desc("Do not treat all horizontal whitespace as equivalent"));
47
48//===----------------------------------------------------------------------===//
49// Pattern Handling Code.
50//===----------------------------------------------------------------------===//
51
52class 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
78public:
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
98private:
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
110bool 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 there.
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.startswith("{{")) {
142
143      // Otherwise, this is the start of a regex match.  Scan for the }}.
144      size_t End = PatternStr.find("}}");
145      if (End == StringRef::npos) {
146        SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()),
147                        "found start of regex string with no end '}}'","error");
148        return true;
149      }
150
151      if (AddRegExToRegEx(PatternStr.substr(2, End-2), CurParen, SM))
152        return true;
153
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.startswith("[[")) {
164      // Verify that it is terminated properly.
165      size_t End = PatternStr.find("]]");
166      if (End == StringRef::npos) {
167        SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()),
168                        "invalid named regex reference, no ]] found", "error");
169        return true;
170      }
171
172      StringRef MatchStr = PatternStr.substr(2, End-2);
173      PatternStr = PatternStr.substr(End+2);
174
175      // Get the regex name (e.g. "foo").
176      size_t NameEnd = MatchStr.find(':');
177      StringRef Name = MatchStr.substr(0, NameEnd);
178
179      if (Name.empty()) {
180        SM.PrintMessage(SMLoc::getFromPointer(Name.data()),
181                        "invalid name in named regex: empty name", "error");
182        return true;
183      }
184
185      // Verify that the name is well formed.
186      for (unsigned i = 0, e = Name.size(); i != e; ++i)
187        if (Name[i] != '_' && !isalnum(Name[i])) {
188          SM.PrintMessage(SMLoc::getFromPointer(Name.data()+i),
189                          "invalid name in named regex", "error");
190          return true;
191        }
192
193      // Name can't start with a digit.
194      if (isdigit(Name[0])) {
195        SM.PrintMessage(SMLoc::getFromPointer(Name.data()),
196                        "invalid name in named regex", "error");
197        return true;
198      }
199
200      // Handle [[foo]].
201      if (NameEnd == StringRef::npos) {
202        VariableUses.push_back(std::make_pair(Name, RegExStr.size()));
203        continue;
204      }
205
206      // Handle [[foo:.*]].
207      VariableDefs.push_back(std::make_pair(Name, CurParen));
208      RegExStr += '(';
209      ++CurParen;
210
211      if (AddRegExToRegEx(MatchStr.substr(NameEnd+1), CurParen, SM))
212        return true;
213
214      RegExStr += ')';
215    }
216
217    // Handle fixed string matches.
218    // Find the end, which is the start of the next regex.
219    size_t FixedMatchEnd = PatternStr.find("{{");
220    FixedMatchEnd = std::min(FixedMatchEnd, PatternStr.find("[["));
221    AddFixedStringToRegEx(PatternStr.substr(0, FixedMatchEnd), RegExStr);
222    PatternStr = PatternStr.substr(FixedMatchEnd);
223    continue;
224  }
225
226  return false;
227}
228
229void Pattern::AddFixedStringToRegEx(StringRef FixedStr, std::string &TheStr) {
230  // Add the characters from FixedStr to the regex, escaping as needed.  This
231  // avoids "leaning toothpicks" in common patterns.
232  for (unsigned i = 0, e = FixedStr.size(); i != e; ++i) {
233    switch (FixedStr[i]) {
234    // These are the special characters matched in "p_ere_exp".
235    case '(':
236    case ')':
237    case '^':
238    case '$':
239    case '|':
240    case '*':
241    case '+':
242    case '?':
243    case '.':
244    case '[':
245    case '\\':
246    case '{':
247      TheStr += '\\';
248      // FALL THROUGH.
249    default:
250      TheStr += FixedStr[i];
251      break;
252    }
253  }
254}
255
256bool Pattern::AddRegExToRegEx(StringRef RegexStr, unsigned &CurParen,
257                              SourceMgr &SM) {
258  Regex R(RegexStr);
259  std::string Error;
260  if (!R.isValid(Error)) {
261    SM.PrintMessage(SMLoc::getFromPointer(RegexStr.data()),
262                    "invalid regex: " + Error, "error");
263    return true;
264  }
265
266  RegExStr += RegexStr.str();
267  CurParen += R.getNumMatches();
268  return false;
269}
270
271/// Match - Match the pattern string against the input buffer Buffer.  This
272/// returns the position that is matched or npos if there is no match.  If
273/// there is a match, the size of the matched string is returned in MatchLen.
274size_t Pattern::Match(StringRef Buffer, size_t &MatchLen,
275                      StringMap<StringRef> &VariableTable) const {
276  // If this is the EOF pattern, match it immediately.
277  if (MatchEOF) {
278    MatchLen = 0;
279    return Buffer.size();
280  }
281
282  // If this is a fixed string pattern, just match it now.
283  if (!FixedStr.empty()) {
284    MatchLen = FixedStr.size();
285    return Buffer.find(FixedStr);
286  }
287
288  // Regex match.
289
290  // If there are variable uses, we need to create a temporary string with the
291  // actual value.
292  StringRef RegExToMatch = RegExStr;
293  std::string TmpStr;
294  if (!VariableUses.empty()) {
295    TmpStr = RegExStr;
296
297    unsigned InsertOffset = 0;
298    for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) {
299      StringMap<StringRef>::iterator it =
300        VariableTable.find(VariableUses[i].first);
301      // If the variable is undefined, return an error.
302      if (it == VariableTable.end())
303        return StringRef::npos;
304
305      // Look up the value and escape it so that we can plop it into the regex.
306      std::string Value;
307      AddFixedStringToRegEx(it->second, Value);
308
309      // Plop it into the regex at the adjusted offset.
310      TmpStr.insert(TmpStr.begin()+VariableUses[i].second+InsertOffset,
311                    Value.begin(), Value.end());
312      InsertOffset += Value.size();
313    }
314
315    // Match the newly constructed regex.
316    RegExToMatch = TmpStr;
317  }
318
319
320  SmallVector<StringRef, 4> MatchInfo;
321  if (!Regex(RegExToMatch, Regex::Newline).match(Buffer, &MatchInfo))
322    return StringRef::npos;
323
324  // Successful regex match.
325  assert(!MatchInfo.empty() && "Didn't get any match");
326  StringRef FullMatch = MatchInfo[0];
327
328  // If this defines any variables, remember their values.
329  for (unsigned i = 0, e = VariableDefs.size(); i != e; ++i) {
330    assert(VariableDefs[i].second < MatchInfo.size() &&
331           "Internal paren error");
332    VariableTable[VariableDefs[i].first] = MatchInfo[VariableDefs[i].second];
333  }
334
335  MatchLen = FullMatch.size();
336  return FullMatch.data()-Buffer.data();
337}
338
339unsigned Pattern::ComputeMatchDistance(StringRef Buffer,
340                              const StringMap<StringRef> &VariableTable) const {
341  // Just compute the number of matching characters. For regular expressions, we
342  // just compare against the regex itself and hope for the best.
343  //
344  // FIXME: One easy improvement here is have the regex lib generate a single
345  // example regular expression which matches, and use that as the example
346  // string.
347  StringRef ExampleString(FixedStr);
348  if (ExampleString.empty())
349    ExampleString = RegExStr;
350
351  // Only compare up to the first line in the buffer, or the string size.
352  StringRef BufferPrefix = Buffer.substr(0, ExampleString.size());
353  BufferPrefix = BufferPrefix.split('\n').first;
354  return BufferPrefix.edit_distance(ExampleString);
355}
356
357void Pattern::PrintFailureInfo(const SourceMgr &SM, StringRef Buffer,
358                               const StringMap<StringRef> &VariableTable) const{
359  // If this was a regular expression using variables, print the current
360  // variable values.
361  if (!VariableUses.empty()) {
362    for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) {
363      StringRef Var = VariableUses[i].first;
364      StringMap<StringRef>::const_iterator it = VariableTable.find(Var);
365      SmallString<256> Msg;
366      raw_svector_ostream OS(Msg);
367
368      // Check for undefined variable references.
369      if (it == VariableTable.end()) {
370        OS << "uses undefined variable \"";
371        OS.write_escaped(Var) << "\"";;
372      } else {
373        OS << "with variable \"";
374        OS.write_escaped(Var) << "\" equal to \"";
375        OS.write_escaped(it->second) << "\"";
376      }
377
378      SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), OS.str(), "note",
379                      /*ShowLine=*/false);
380    }
381  }
382
383  // Attempt to find the closest/best fuzzy match.  Usually an error happens
384  // because some string in the output didn't exactly match. In these cases, we
385  // would like to show the user a best guess at what "should have" matched, to
386  // save them having to actually check the input manually.
387  size_t NumLinesForward = 0;
388  size_t Best = StringRef::npos;
389  double BestQuality = 0;
390
391  // Use an arbitrary 4k limit on how far we will search.
392  for (size_t i = 0, e = std::min(size_t(4096), Buffer.size()); i != e; ++i) {
393    if (Buffer[i] == '\n')
394      ++NumLinesForward;
395
396    // Patterns have leading whitespace stripped, so skip whitespace when
397    // looking for something which looks like a pattern.
398    if (Buffer[i] == ' ' || Buffer[i] == '\t')
399      continue;
400
401    // Compute the "quality" of this match as an arbitrary combination of the
402    // match distance and the number of lines skipped to get to this match.
403    unsigned Distance = ComputeMatchDistance(Buffer.substr(i), VariableTable);
404    double Quality = Distance + (NumLinesForward / 100.);
405
406    if (Quality < BestQuality || Best == StringRef::npos) {
407      Best = i;
408      BestQuality = Quality;
409    }
410  }
411
412  // Print the "possible intended match here" line if we found something
413  // reasonable and not equal to what we showed in the "scanning from here"
414  // line.
415  if (Best && Best != StringRef::npos && BestQuality < 50) {
416      SM.PrintMessage(SMLoc::getFromPointer(Buffer.data() + Best),
417                      "possible intended match here", "note");
418
419    // FIXME: If we wanted to be really friendly we would show why the match
420    // failed, as it can be hard to spot simple one character differences.
421  }
422}
423
424//===----------------------------------------------------------------------===//
425// Check Strings.
426//===----------------------------------------------------------------------===//
427
428/// CheckString - This is a check that we found in the input file.
429struct CheckString {
430  /// Pat - The pattern to match.
431  Pattern Pat;
432
433  /// Loc - The location in the match file that the check string was specified.
434  SMLoc Loc;
435
436  /// IsCheckNext - This is true if this is a CHECK-NEXT: directive (as opposed
437  /// to a CHECK: directive.
438  bool IsCheckNext;
439
440  /// NotStrings - These are all of the strings that are disallowed from
441  /// occurring between this match string and the previous one (or start of
442  /// file).
443  std::vector<std::pair<SMLoc, Pattern> > NotStrings;
444
445  CheckString(const Pattern &P, SMLoc L, bool isCheckNext)
446    : Pat(P), Loc(L), IsCheckNext(isCheckNext) {}
447};
448
449/// CanonicalizeInputFile - Remove duplicate horizontal space from the specified
450/// memory buffer, free it, and return a new one.
451static MemoryBuffer *CanonicalizeInputFile(MemoryBuffer *MB) {
452  SmallString<128> NewFile;
453  NewFile.reserve(MB->getBufferSize());
454
455  for (const char *Ptr = MB->getBufferStart(), *End = MB->getBufferEnd();
456       Ptr != End; ++Ptr) {
457    // Eliminate trailing dosish \r.
458    if (Ptr <= End - 2 && Ptr[0] == '\r' && Ptr[1] == '\n') {
459      continue;
460    }
461
462    // If C is not a horizontal whitespace, skip it.
463    if (*Ptr != ' ' && *Ptr != '\t') {
464      NewFile.push_back(*Ptr);
465      continue;
466    }
467
468    // Otherwise, add one space and advance over neighboring space.
469    NewFile.push_back(' ');
470    while (Ptr+1 != End &&
471           (Ptr[1] == ' ' || Ptr[1] == '\t'))
472      ++Ptr;
473  }
474
475  // Free the old buffer and return a new one.
476  MemoryBuffer *MB2 =
477    MemoryBuffer::getMemBufferCopy(NewFile.str(), MB->getBufferIdentifier());
478
479  delete MB;
480  return MB2;
481}
482
483
484/// ReadCheckFile - Read the check file, which specifies the sequence of
485/// expected strings.  The strings are added to the CheckStrings vector.
486static bool ReadCheckFile(SourceMgr &SM,
487                          std::vector<CheckString> &CheckStrings) {
488  // Open the check file, and tell SourceMgr about it.
489  OwningPtr<MemoryBuffer> File;
490  if (error_code ec =
491        MemoryBuffer::getFileOrSTDIN(CheckFilename.c_str(), File)) {
492    errs() << "Could not open check file '" << CheckFilename << "': "
493           << ec.message() << '\n';
494    return true;
495  }
496  MemoryBuffer *F = File.take();
497
498  // If we want to canonicalize whitespace, strip excess whitespace from the
499  // buffer containing the CHECK lines.
500  if (!NoCanonicalizeWhiteSpace)
501    F = CanonicalizeInputFile(F);
502
503  SM.AddNewSourceBuffer(F, SMLoc());
504
505  // Find all instances of CheckPrefix followed by : in the file.
506  StringRef Buffer = F->getBuffer();
507
508  std::vector<std::pair<SMLoc, Pattern> > NotMatches;
509
510  while (1) {
511    // See if Prefix occurs in the memory buffer.
512    Buffer = Buffer.substr(Buffer.find(CheckPrefix));
513
514    // If we didn't find a match, we're done.
515    if (Buffer.empty())
516      break;
517
518    const char *CheckPrefixStart = Buffer.data();
519
520    // When we find a check prefix, keep track of whether we find CHECK: or
521    // CHECK-NEXT:
522    bool IsCheckNext = false, IsCheckNot = false;
523
524    // Verify that the : is present after the prefix.
525    if (Buffer[CheckPrefix.size()] == ':') {
526      Buffer = Buffer.substr(CheckPrefix.size()+1);
527    } else if (Buffer.size() > CheckPrefix.size()+6 &&
528               memcmp(Buffer.data()+CheckPrefix.size(), "-NEXT:", 6) == 0) {
529      Buffer = Buffer.substr(CheckPrefix.size()+7);
530      IsCheckNext = true;
531    } else if (Buffer.size() > CheckPrefix.size()+5 &&
532               memcmp(Buffer.data()+CheckPrefix.size(), "-NOT:", 5) == 0) {
533      Buffer = Buffer.substr(CheckPrefix.size()+6);
534      IsCheckNot = true;
535    } else {
536      Buffer = Buffer.substr(1);
537      continue;
538    }
539
540    // Okay, we found the prefix, yay.  Remember the rest of the line, but
541    // ignore leading and trailing whitespace.
542    Buffer = Buffer.substr(Buffer.find_first_not_of(" \t"));
543
544    // Scan ahead to the end of line.
545    size_t EOL = Buffer.find_first_of("\n\r");
546
547    // Remember the location of the start of the pattern, for diagnostics.
548    SMLoc PatternLoc = SMLoc::getFromPointer(Buffer.data());
549
550    // Parse the pattern.
551    Pattern P;
552    if (P.ParsePattern(Buffer.substr(0, EOL), SM))
553      return true;
554
555    Buffer = Buffer.substr(EOL);
556
557
558    // Verify that CHECK-NEXT lines have at least one CHECK line before them.
559    if (IsCheckNext && CheckStrings.empty()) {
560      SM.PrintMessage(SMLoc::getFromPointer(CheckPrefixStart),
561                      "found '"+CheckPrefix+"-NEXT:' without previous '"+
562                      CheckPrefix+ ": line", "error");
563      return true;
564    }
565
566    // Handle CHECK-NOT.
567    if (IsCheckNot) {
568      NotMatches.push_back(std::make_pair(SMLoc::getFromPointer(Buffer.data()),
569                                          P));
570      continue;
571    }
572
573
574    // Okay, add the string we captured to the output vector and move on.
575    CheckStrings.push_back(CheckString(P,
576                                       PatternLoc,
577                                       IsCheckNext));
578    std::swap(NotMatches, CheckStrings.back().NotStrings);
579  }
580
581  // Add an EOF pattern for any trailing CHECK-NOTs.
582  if (!NotMatches.empty()) {
583    CheckStrings.push_back(CheckString(Pattern(true),
584                                       SMLoc::getFromPointer(Buffer.data()),
585                                       false));
586    std::swap(NotMatches, CheckStrings.back().NotStrings);
587  }
588
589  if (CheckStrings.empty()) {
590    errs() << "error: no check strings found with prefix '" << CheckPrefix
591           << ":'\n";
592    return true;
593  }
594
595  return false;
596}
597
598static void PrintCheckFailed(const SourceMgr &SM, const CheckString &CheckStr,
599                             StringRef Buffer,
600                             StringMap<StringRef> &VariableTable) {
601  // Otherwise, we have an error, emit an error message.
602  SM.PrintMessage(CheckStr.Loc, "expected string not found in input",
603                  "error");
604
605  // Print the "scanning from here" line.  If the current position is at the
606  // end of a line, advance to the start of the next line.
607  Buffer = Buffer.substr(Buffer.find_first_not_of(" \t\n\r"));
608
609  SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), "scanning from here",
610                  "note");
611
612  // Allow the pattern to print additional information if desired.
613  CheckStr.Pat.PrintFailureInfo(SM, Buffer, VariableTable);
614}
615
616/// CountNumNewlinesBetween - Count the number of newlines in the specified
617/// range.
618static unsigned CountNumNewlinesBetween(StringRef Range) {
619  unsigned NumNewLines = 0;
620  while (1) {
621    // Scan for newline.
622    Range = Range.substr(Range.find_first_of("\n\r"));
623    if (Range.empty()) return NumNewLines;
624
625    ++NumNewLines;
626
627    // Handle \n\r and \r\n as a single newline.
628    if (Range.size() > 1 &&
629        (Range[1] == '\n' || Range[1] == '\r') &&
630        (Range[0] != Range[1]))
631      Range = Range.substr(1);
632    Range = Range.substr(1);
633  }
634}
635
636int main(int argc, char **argv) {
637  sys::PrintStackTraceOnErrorSignal();
638  PrettyStackTraceProgram X(argc, argv);
639  cl::ParseCommandLineOptions(argc, argv);
640
641  SourceMgr SM;
642
643  // Read the expected strings from the check file.
644  std::vector<CheckString> CheckStrings;
645  if (ReadCheckFile(SM, CheckStrings))
646    return 2;
647
648  // Open the file to check and add it to SourceMgr.
649  OwningPtr<MemoryBuffer> File;
650  if (error_code ec =
651        MemoryBuffer::getFileOrSTDIN(InputFilename.c_str(), File)) {
652    errs() << "Could not open input file '" << InputFilename << "': "
653           << ec.message() << '\n';
654    return true;
655  }
656  MemoryBuffer *F = File.take();
657
658  if (F->getBufferSize() == 0) {
659    errs() << "FileCheck error: '" << InputFilename << "' is empty.\n";
660    return 1;
661  }
662
663  // Remove duplicate spaces in the input file if requested.
664  if (!NoCanonicalizeWhiteSpace)
665    F = CanonicalizeInputFile(F);
666
667  SM.AddNewSourceBuffer(F, SMLoc());
668
669  /// VariableTable - This holds all the current filecheck variables.
670  StringMap<StringRef> VariableTable;
671
672  // Check that we have all of the expected strings, in order, in the input
673  // file.
674  StringRef Buffer = F->getBuffer();
675
676  const char *LastMatch = Buffer.data();
677
678  for (unsigned StrNo = 0, e = CheckStrings.size(); StrNo != e; ++StrNo) {
679    const CheckString &CheckStr = CheckStrings[StrNo];
680
681    StringRef SearchFrom = Buffer;
682
683    // Find StrNo in the file.
684    size_t MatchLen = 0;
685    size_t MatchPos = CheckStr.Pat.Match(Buffer, MatchLen, VariableTable);
686    Buffer = Buffer.substr(MatchPos);
687
688    // If we didn't find a match, reject the input.
689    if (MatchPos == StringRef::npos) {
690      PrintCheckFailed(SM, CheckStr, SearchFrom, VariableTable);
691      return 1;
692    }
693
694    StringRef SkippedRegion(LastMatch, Buffer.data()-LastMatch);
695
696    // If this check is a "CHECK-NEXT", verify that the previous match was on
697    // the previous line (i.e. that there is one newline between them).
698    if (CheckStr.IsCheckNext) {
699      // Count the number of newlines between the previous match and this one.
700      assert(LastMatch != F->getBufferStart() &&
701             "CHECK-NEXT can't be the first check in a file");
702
703      unsigned NumNewLines = CountNumNewlinesBetween(SkippedRegion);
704      if (NumNewLines == 0) {
705        SM.PrintMessage(CheckStr.Loc,
706                    CheckPrefix+"-NEXT: is on the same line as previous match",
707                        "error");
708        SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
709                        "'next' match was here", "note");
710        SM.PrintMessage(SMLoc::getFromPointer(LastMatch),
711                        "previous match was here", "note");
712        return 1;
713      }
714
715      if (NumNewLines != 1) {
716        SM.PrintMessage(CheckStr.Loc,
717                        CheckPrefix+
718                        "-NEXT: is not on the line after the previous match",
719                        "error");
720        SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
721                        "'next' match was here", "note");
722        SM.PrintMessage(SMLoc::getFromPointer(LastMatch),
723                        "previous match was here", "note");
724        return 1;
725      }
726    }
727
728    // If this match had "not strings", verify that they don't exist in the
729    // skipped region.
730    for (unsigned ChunkNo = 0, e = CheckStr.NotStrings.size();
731         ChunkNo != e; ++ChunkNo) {
732      size_t MatchLen = 0;
733      size_t Pos = CheckStr.NotStrings[ChunkNo].second.Match(SkippedRegion,
734                                                             MatchLen,
735                                                             VariableTable);
736      if (Pos == StringRef::npos) continue;
737
738      SM.PrintMessage(SMLoc::getFromPointer(LastMatch+Pos),
739                      CheckPrefix+"-NOT: string occurred!", "error");
740      SM.PrintMessage(CheckStr.NotStrings[ChunkNo].first,
741                      CheckPrefix+"-NOT: pattern specified here", "note");
742      return 1;
743    }
744
745
746    // Otherwise, everything is good.  Step over the matched text and remember
747    // the position after the match as the end of the last match.
748    Buffer = Buffer.substr(MatchLen);
749    LastMatch = Buffer.data();
750  }
751
752  return 0;
753}
754