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, SourceMgr::DK_Error,
121                    "found empty check string with prefix '" +
122                    CheckPrefix+":'");
123    return true;
124  }
125
126  // Check to see if this is a fixed string, or if it has regex pieces.
127  if (PatternStr.size() < 2 ||
128      (PatternStr.find("{{") == StringRef::npos &&
129       PatternStr.find("[[") == StringRef::npos)) {
130    FixedStr = PatternStr;
131    return false;
132  }
133
134  // Paren value #0 is for the fully matched string.  Any new parenthesized
135  // values add from there.
136  unsigned CurParen = 1;
137
138  // Otherwise, there is at least one regex piece.  Build up the regex pattern
139  // by escaping scary characters in fixed strings, building up one big regex.
140  while (!PatternStr.empty()) {
141    // RegEx matches.
142    if (PatternStr.startswith("{{")) {
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                        SourceMgr::DK_Error,
149                        "found start of regex string with no end '}}'");
150        return true;
151      }
152
153      // Enclose {{}} patterns in parens just like [[]] even though we're not
154      // capturing the result for any purpose.  This is required in case the
155      // expression contains an alternation like: CHECK:  abc{{x|z}}def.  We
156      // want this to turn into: "abc(x|z)def" not "abcx|zdef".
157      RegExStr += '(';
158      ++CurParen;
159
160      if (AddRegExToRegEx(PatternStr.substr(2, End-2), CurParen, SM))
161        return true;
162      RegExStr += ')';
163
164      PatternStr = PatternStr.substr(End+2);
165      continue;
166    }
167
168    // Named RegEx matches.  These are of two forms: [[foo:.*]] which matches .*
169    // (or some other regex) and assigns it to the FileCheck variable 'foo'. The
170    // second form is [[foo]] which is a reference to foo.  The variable name
171    // itself must be of the form "[a-zA-Z_][0-9a-zA-Z_]*", otherwise we reject
172    // it.  This is to catch some common errors.
173    if (PatternStr.startswith("[[")) {
174      // Verify that it is terminated properly.
175      size_t End = PatternStr.find("]]");
176      if (End == StringRef::npos) {
177        SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()),
178                        SourceMgr::DK_Error,
179                        "invalid named regex reference, no ]] found");
180        return true;
181      }
182
183      StringRef MatchStr = PatternStr.substr(2, End-2);
184      PatternStr = PatternStr.substr(End+2);
185
186      // Get the regex name (e.g. "foo").
187      size_t NameEnd = MatchStr.find(':');
188      StringRef Name = MatchStr.substr(0, NameEnd);
189
190      if (Name.empty()) {
191        SM.PrintMessage(SMLoc::getFromPointer(Name.data()), SourceMgr::DK_Error,
192                        "invalid name in named regex: empty name");
193        return true;
194      }
195
196      // Verify that the name is well formed.
197      for (unsigned i = 0, e = Name.size(); i != e; ++i)
198        if (Name[i] != '_' && !isalnum(Name[i])) {
199          SM.PrintMessage(SMLoc::getFromPointer(Name.data()+i),
200                          SourceMgr::DK_Error, "invalid name in named regex");
201          return true;
202        }
203
204      // Name can't start with a digit.
205      if (isdigit(Name[0])) {
206        SM.PrintMessage(SMLoc::getFromPointer(Name.data()), SourceMgr::DK_Error,
207                        "invalid name in named regex");
208        return true;
209      }
210
211      // Handle [[foo]].
212      if (NameEnd == StringRef::npos) {
213        VariableUses.push_back(std::make_pair(Name, RegExStr.size()));
214        continue;
215      }
216
217      // Handle [[foo:.*]].
218      VariableDefs.push_back(std::make_pair(Name, CurParen));
219      RegExStr += '(';
220      ++CurParen;
221
222      if (AddRegExToRegEx(MatchStr.substr(NameEnd+1), CurParen, SM))
223        return true;
224
225      RegExStr += ')';
226    }
227
228    // Handle fixed string matches.
229    // Find the end, which is the start of the next regex.
230    size_t FixedMatchEnd = PatternStr.find("{{");
231    FixedMatchEnd = std::min(FixedMatchEnd, PatternStr.find("[["));
232    AddFixedStringToRegEx(PatternStr.substr(0, FixedMatchEnd), RegExStr);
233    PatternStr = PatternStr.substr(FixedMatchEnd);
234    continue;
235  }
236
237  return false;
238}
239
240void Pattern::AddFixedStringToRegEx(StringRef FixedStr, std::string &TheStr) {
241  // Add the characters from FixedStr to the regex, escaping as needed.  This
242  // avoids "leaning toothpicks" in common patterns.
243  for (unsigned i = 0, e = FixedStr.size(); i != e; ++i) {
244    switch (FixedStr[i]) {
245    // These are the special characters matched in "p_ere_exp".
246    case '(':
247    case ')':
248    case '^':
249    case '$':
250    case '|':
251    case '*':
252    case '+':
253    case '?':
254    case '.':
255    case '[':
256    case '\\':
257    case '{':
258      TheStr += '\\';
259      // FALL THROUGH.
260    default:
261      TheStr += FixedStr[i];
262      break;
263    }
264  }
265}
266
267bool Pattern::AddRegExToRegEx(StringRef RegexStr, unsigned &CurParen,
268                              SourceMgr &SM) {
269  Regex R(RegexStr);
270  std::string Error;
271  if (!R.isValid(Error)) {
272    SM.PrintMessage(SMLoc::getFromPointer(RegexStr.data()), SourceMgr::DK_Error,
273                    "invalid regex: " + Error);
274    return true;
275  }
276
277  RegExStr += RegexStr.str();
278  CurParen += R.getNumMatches();
279  return false;
280}
281
282/// Match - Match the pattern string against the input buffer Buffer.  This
283/// returns the position that is matched or npos if there is no match.  If
284/// there is a match, the size of the matched string is returned in MatchLen.
285size_t Pattern::Match(StringRef Buffer, size_t &MatchLen,
286                      StringMap<StringRef> &VariableTable) const {
287  // If this is the EOF pattern, match it immediately.
288  if (MatchEOF) {
289    MatchLen = 0;
290    return Buffer.size();
291  }
292
293  // If this is a fixed string pattern, just match it now.
294  if (!FixedStr.empty()) {
295    MatchLen = FixedStr.size();
296    return Buffer.find(FixedStr);
297  }
298
299  // Regex match.
300
301  // If there are variable uses, we need to create a temporary string with the
302  // actual value.
303  StringRef RegExToMatch = RegExStr;
304  std::string TmpStr;
305  if (!VariableUses.empty()) {
306    TmpStr = RegExStr;
307
308    unsigned InsertOffset = 0;
309    for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) {
310      StringMap<StringRef>::iterator it =
311        VariableTable.find(VariableUses[i].first);
312      // If the variable is undefined, return an error.
313      if (it == VariableTable.end())
314        return StringRef::npos;
315
316      // Look up the value and escape it so that we can plop it into the regex.
317      std::string Value;
318      AddFixedStringToRegEx(it->second, Value);
319
320      // Plop it into the regex at the adjusted offset.
321      TmpStr.insert(TmpStr.begin()+VariableUses[i].second+InsertOffset,
322                    Value.begin(), Value.end());
323      InsertOffset += Value.size();
324    }
325
326    // Match the newly constructed regex.
327    RegExToMatch = TmpStr;
328  }
329
330
331  SmallVector<StringRef, 4> MatchInfo;
332  if (!Regex(RegExToMatch, Regex::Newline).match(Buffer, &MatchInfo))
333    return StringRef::npos;
334
335  // Successful regex match.
336  assert(!MatchInfo.empty() && "Didn't get any match");
337  StringRef FullMatch = MatchInfo[0];
338
339  // If this defines any variables, remember their values.
340  for (unsigned i = 0, e = VariableDefs.size(); i != e; ++i) {
341    assert(VariableDefs[i].second < MatchInfo.size() &&
342           "Internal paren error");
343    VariableTable[VariableDefs[i].first] = MatchInfo[VariableDefs[i].second];
344  }
345
346  MatchLen = FullMatch.size();
347  return FullMatch.data()-Buffer.data();
348}
349
350unsigned Pattern::ComputeMatchDistance(StringRef Buffer,
351                              const StringMap<StringRef> &VariableTable) const {
352  // Just compute the number of matching characters. For regular expressions, we
353  // just compare against the regex itself and hope for the best.
354  //
355  // FIXME: One easy improvement here is have the regex lib generate a single
356  // example regular expression which matches, and use that as the example
357  // string.
358  StringRef ExampleString(FixedStr);
359  if (ExampleString.empty())
360    ExampleString = RegExStr;
361
362  // Only compare up to the first line in the buffer, or the string size.
363  StringRef BufferPrefix = Buffer.substr(0, ExampleString.size());
364  BufferPrefix = BufferPrefix.split('\n').first;
365  return BufferPrefix.edit_distance(ExampleString);
366}
367
368void Pattern::PrintFailureInfo(const SourceMgr &SM, StringRef Buffer,
369                               const StringMap<StringRef> &VariableTable) const{
370  // If this was a regular expression using variables, print the current
371  // variable values.
372  if (!VariableUses.empty()) {
373    for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) {
374      StringRef Var = VariableUses[i].first;
375      StringMap<StringRef>::const_iterator it = VariableTable.find(Var);
376      SmallString<256> Msg;
377      raw_svector_ostream OS(Msg);
378
379      // Check for undefined variable references.
380      if (it == VariableTable.end()) {
381        OS << "uses undefined variable \"";
382        OS.write_escaped(Var) << "\"";;
383      } else {
384        OS << "with variable \"";
385        OS.write_escaped(Var) << "\" equal to \"";
386        OS.write_escaped(it->second) << "\"";
387      }
388
389      SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), SourceMgr::DK_Note,
390                      OS.str());
391    }
392  }
393
394  // Attempt to find the closest/best fuzzy match.  Usually an error happens
395  // because some string in the output didn't exactly match. In these cases, we
396  // would like to show the user a best guess at what "should have" matched, to
397  // save them having to actually check the input manually.
398  size_t NumLinesForward = 0;
399  size_t Best = StringRef::npos;
400  double BestQuality = 0;
401
402  // Use an arbitrary 4k limit on how far we will search.
403  for (size_t i = 0, e = std::min(size_t(4096), Buffer.size()); i != e; ++i) {
404    if (Buffer[i] == '\n')
405      ++NumLinesForward;
406
407    // Patterns have leading whitespace stripped, so skip whitespace when
408    // looking for something which looks like a pattern.
409    if (Buffer[i] == ' ' || Buffer[i] == '\t')
410      continue;
411
412    // Compute the "quality" of this match as an arbitrary combination of the
413    // match distance and the number of lines skipped to get to this match.
414    unsigned Distance = ComputeMatchDistance(Buffer.substr(i), VariableTable);
415    double Quality = Distance + (NumLinesForward / 100.);
416
417    if (Quality < BestQuality || Best == StringRef::npos) {
418      Best = i;
419      BestQuality = Quality;
420    }
421  }
422
423  // Print the "possible intended match here" line if we found something
424  // reasonable and not equal to what we showed in the "scanning from here"
425  // line.
426  if (Best && Best != StringRef::npos && BestQuality < 50) {
427      SM.PrintMessage(SMLoc::getFromPointer(Buffer.data() + Best),
428                      SourceMgr::DK_Note, "possible intended match here");
429
430    // FIXME: If we wanted to be really friendly we would show why the match
431    // failed, as it can be hard to spot simple one character differences.
432  }
433}
434
435//===----------------------------------------------------------------------===//
436// Check Strings.
437//===----------------------------------------------------------------------===//
438
439/// CheckString - This is a check that we found in the input file.
440struct CheckString {
441  /// Pat - The pattern to match.
442  Pattern Pat;
443
444  /// Loc - The location in the match file that the check string was specified.
445  SMLoc Loc;
446
447  /// IsCheckNext - This is true if this is a CHECK-NEXT: directive (as opposed
448  /// to a CHECK: directive.
449  bool IsCheckNext;
450
451  /// NotStrings - These are all of the strings that are disallowed from
452  /// occurring between this match string and the previous one (or start of
453  /// file).
454  std::vector<std::pair<SMLoc, Pattern> > NotStrings;
455
456  CheckString(const Pattern &P, SMLoc L, bool isCheckNext)
457    : Pat(P), Loc(L), IsCheckNext(isCheckNext) {}
458};
459
460/// CanonicalizeInputFile - Remove duplicate horizontal space from the specified
461/// memory buffer, free it, and return a new one.
462static MemoryBuffer *CanonicalizeInputFile(MemoryBuffer *MB) {
463  SmallString<128> NewFile;
464  NewFile.reserve(MB->getBufferSize());
465
466  for (const char *Ptr = MB->getBufferStart(), *End = MB->getBufferEnd();
467       Ptr != End; ++Ptr) {
468    // Eliminate trailing dosish \r.
469    if (Ptr <= End - 2 && Ptr[0] == '\r' && Ptr[1] == '\n') {
470      continue;
471    }
472
473    // If C is not a horizontal whitespace, skip it.
474    if (*Ptr != ' ' && *Ptr != '\t') {
475      NewFile.push_back(*Ptr);
476      continue;
477    }
478
479    // Otherwise, add one space and advance over neighboring space.
480    NewFile.push_back(' ');
481    while (Ptr+1 != End &&
482           (Ptr[1] == ' ' || Ptr[1] == '\t'))
483      ++Ptr;
484  }
485
486  // Free the old buffer and return a new one.
487  MemoryBuffer *MB2 =
488    MemoryBuffer::getMemBufferCopy(NewFile.str(), MB->getBufferIdentifier());
489
490  delete MB;
491  return MB2;
492}
493
494
495/// ReadCheckFile - Read the check file, which specifies the sequence of
496/// expected strings.  The strings are added to the CheckStrings vector.
497static bool ReadCheckFile(SourceMgr &SM,
498                          std::vector<CheckString> &CheckStrings) {
499  // Open the check file, and tell SourceMgr about it.
500  OwningPtr<MemoryBuffer> File;
501  if (error_code ec =
502        MemoryBuffer::getFileOrSTDIN(CheckFilename.c_str(), File)) {
503    errs() << "Could not open check file '" << CheckFilename << "': "
504           << ec.message() << '\n';
505    return true;
506  }
507  MemoryBuffer *F = File.take();
508
509  // If we want to canonicalize whitespace, strip excess whitespace from the
510  // buffer containing the CHECK lines.
511  if (!NoCanonicalizeWhiteSpace)
512    F = CanonicalizeInputFile(F);
513
514  SM.AddNewSourceBuffer(F, SMLoc());
515
516  // Find all instances of CheckPrefix followed by : in the file.
517  StringRef Buffer = F->getBuffer();
518
519  std::vector<std::pair<SMLoc, Pattern> > NotMatches;
520
521  while (1) {
522    // See if Prefix occurs in the memory buffer.
523    Buffer = Buffer.substr(Buffer.find(CheckPrefix));
524
525    // If we didn't find a match, we're done.
526    if (Buffer.empty())
527      break;
528
529    const char *CheckPrefixStart = Buffer.data();
530
531    // When we find a check prefix, keep track of whether we find CHECK: or
532    // CHECK-NEXT:
533    bool IsCheckNext = false, IsCheckNot = false;
534
535    // Verify that the : is present after the prefix.
536    if (Buffer[CheckPrefix.size()] == ':') {
537      Buffer = Buffer.substr(CheckPrefix.size()+1);
538    } else if (Buffer.size() > CheckPrefix.size()+6 &&
539               memcmp(Buffer.data()+CheckPrefix.size(), "-NEXT:", 6) == 0) {
540      Buffer = Buffer.substr(CheckPrefix.size()+7);
541      IsCheckNext = true;
542    } else if (Buffer.size() > CheckPrefix.size()+5 &&
543               memcmp(Buffer.data()+CheckPrefix.size(), "-NOT:", 5) == 0) {
544      Buffer = Buffer.substr(CheckPrefix.size()+6);
545      IsCheckNot = true;
546    } else {
547      Buffer = Buffer.substr(1);
548      continue;
549    }
550
551    // Okay, we found the prefix, yay.  Remember the rest of the line, but
552    // ignore leading and trailing whitespace.
553    Buffer = Buffer.substr(Buffer.find_first_not_of(" \t"));
554
555    // Scan ahead to the end of line.
556    size_t EOL = Buffer.find_first_of("\n\r");
557
558    // Remember the location of the start of the pattern, for diagnostics.
559    SMLoc PatternLoc = SMLoc::getFromPointer(Buffer.data());
560
561    // Parse the pattern.
562    Pattern P;
563    if (P.ParsePattern(Buffer.substr(0, EOL), SM))
564      return true;
565
566    Buffer = Buffer.substr(EOL);
567
568
569    // Verify that CHECK-NEXT lines have at least one CHECK line before them.
570    if (IsCheckNext && CheckStrings.empty()) {
571      SM.PrintMessage(SMLoc::getFromPointer(CheckPrefixStart),
572                      SourceMgr::DK_Error,
573                      "found '"+CheckPrefix+"-NEXT:' without previous '"+
574                      CheckPrefix+ ": line");
575      return true;
576    }
577
578    // Handle CHECK-NOT.
579    if (IsCheckNot) {
580      NotMatches.push_back(std::make_pair(SMLoc::getFromPointer(Buffer.data()),
581                                          P));
582      continue;
583    }
584
585
586    // Okay, add the string we captured to the output vector and move on.
587    CheckStrings.push_back(CheckString(P,
588                                       PatternLoc,
589                                       IsCheckNext));
590    std::swap(NotMatches, CheckStrings.back().NotStrings);
591  }
592
593  // Add an EOF pattern for any trailing CHECK-NOTs.
594  if (!NotMatches.empty()) {
595    CheckStrings.push_back(CheckString(Pattern(true),
596                                       SMLoc::getFromPointer(Buffer.data()),
597                                       false));
598    std::swap(NotMatches, CheckStrings.back().NotStrings);
599  }
600
601  if (CheckStrings.empty()) {
602    errs() << "error: no check strings found with prefix '" << CheckPrefix
603           << ":'\n";
604    return true;
605  }
606
607  return false;
608}
609
610static void PrintCheckFailed(const SourceMgr &SM, const CheckString &CheckStr,
611                             StringRef Buffer,
612                             StringMap<StringRef> &VariableTable) {
613  // Otherwise, we have an error, emit an error message.
614  SM.PrintMessage(CheckStr.Loc, SourceMgr::DK_Error,
615                  "expected string not found in input");
616
617  // Print the "scanning from here" line.  If the current position is at the
618  // end of a line, advance to the start of the next line.
619  Buffer = Buffer.substr(Buffer.find_first_not_of(" \t\n\r"));
620
621  SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), SourceMgr::DK_Note,
622                  "scanning from here");
623
624  // Allow the pattern to print additional information if desired.
625  CheckStr.Pat.PrintFailureInfo(SM, Buffer, VariableTable);
626}
627
628/// CountNumNewlinesBetween - Count the number of newlines in the specified
629/// range.
630static unsigned CountNumNewlinesBetween(StringRef Range) {
631  unsigned NumNewLines = 0;
632  while (1) {
633    // Scan for newline.
634    Range = Range.substr(Range.find_first_of("\n\r"));
635    if (Range.empty()) return NumNewLines;
636
637    ++NumNewLines;
638
639    // Handle \n\r and \r\n as a single newline.
640    if (Range.size() > 1 &&
641        (Range[1] == '\n' || Range[1] == '\r') &&
642        (Range[0] != Range[1]))
643      Range = Range.substr(1);
644    Range = Range.substr(1);
645  }
646}
647
648int main(int argc, char **argv) {
649  sys::PrintStackTraceOnErrorSignal();
650  PrettyStackTraceProgram X(argc, argv);
651  cl::ParseCommandLineOptions(argc, argv);
652
653  SourceMgr SM;
654
655  // Read the expected strings from the check file.
656  std::vector<CheckString> CheckStrings;
657  if (ReadCheckFile(SM, CheckStrings))
658    return 2;
659
660  // Open the file to check and add it to SourceMgr.
661  OwningPtr<MemoryBuffer> File;
662  if (error_code ec =
663        MemoryBuffer::getFileOrSTDIN(InputFilename.c_str(), File)) {
664    errs() << "Could not open input file '" << InputFilename << "': "
665           << ec.message() << '\n';
666    return true;
667  }
668  MemoryBuffer *F = File.take();
669
670  if (F->getBufferSize() == 0) {
671    errs() << "FileCheck error: '" << InputFilename << "' is empty.\n";
672    return 1;
673  }
674
675  // Remove duplicate spaces in the input file if requested.
676  if (!NoCanonicalizeWhiteSpace)
677    F = CanonicalizeInputFile(F);
678
679  SM.AddNewSourceBuffer(F, SMLoc());
680
681  /// VariableTable - This holds all the current filecheck variables.
682  StringMap<StringRef> VariableTable;
683
684  // Check that we have all of the expected strings, in order, in the input
685  // file.
686  StringRef Buffer = F->getBuffer();
687
688  const char *LastMatch = Buffer.data();
689
690  for (unsigned StrNo = 0, e = CheckStrings.size(); StrNo != e; ++StrNo) {
691    const CheckString &CheckStr = CheckStrings[StrNo];
692
693    StringRef SearchFrom = Buffer;
694
695    // Find StrNo in the file.
696    size_t MatchLen = 0;
697    size_t MatchPos = CheckStr.Pat.Match(Buffer, MatchLen, VariableTable);
698    Buffer = Buffer.substr(MatchPos);
699
700    // If we didn't find a match, reject the input.
701    if (MatchPos == StringRef::npos) {
702      PrintCheckFailed(SM, CheckStr, SearchFrom, VariableTable);
703      return 1;
704    }
705
706    StringRef SkippedRegion(LastMatch, Buffer.data()-LastMatch);
707
708    // If this check is a "CHECK-NEXT", verify that the previous match was on
709    // the previous line (i.e. that there is one newline between them).
710    if (CheckStr.IsCheckNext) {
711      // Count the number of newlines between the previous match and this one.
712      assert(LastMatch != F->getBufferStart() &&
713             "CHECK-NEXT can't be the first check in a file");
714
715      unsigned NumNewLines = CountNumNewlinesBetween(SkippedRegion);
716      if (NumNewLines == 0) {
717        SM.PrintMessage(CheckStr.Loc, SourceMgr::DK_Error,
718                    CheckPrefix+"-NEXT: is on the same line as previous match");
719        SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
720                        SourceMgr::DK_Note, "'next' match was here");
721        SM.PrintMessage(SMLoc::getFromPointer(LastMatch), SourceMgr::DK_Note,
722                        "previous match was here");
723        return 1;
724      }
725
726      if (NumNewLines != 1) {
727        SM.PrintMessage(CheckStr.Loc, SourceMgr::DK_Error, CheckPrefix+
728                        "-NEXT: is not on the line after the previous match");
729        SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
730                        SourceMgr::DK_Note, "'next' match was here");
731        SM.PrintMessage(SMLoc::getFromPointer(LastMatch), SourceMgr::DK_Note,
732                        "previous match was here");
733        return 1;
734      }
735    }
736
737    // If this match had "not strings", verify that they don't exist in the
738    // skipped region.
739    for (unsigned ChunkNo = 0, e = CheckStr.NotStrings.size();
740         ChunkNo != e; ++ChunkNo) {
741      size_t MatchLen = 0;
742      size_t Pos = CheckStr.NotStrings[ChunkNo].second.Match(SkippedRegion,
743                                                             MatchLen,
744                                                             VariableTable);
745      if (Pos == StringRef::npos) continue;
746
747      SM.PrintMessage(SMLoc::getFromPointer(LastMatch+Pos), SourceMgr::DK_Error,
748                      CheckPrefix+"-NOT: string occurred!");
749      SM.PrintMessage(CheckStr.NotStrings[ChunkNo].first, SourceMgr::DK_Note,
750                      CheckPrefix+"-NOT: pattern specified here");
751      return 1;
752    }
753
754
755    // Otherwise, everything is good.  Step over the matched text and remember
756    // the position after the match as the end of the last match.
757    Buffer = Buffer.substr(MatchLen);
758    LastMatch = Buffer.data();
759  }
760
761  return 0;
762}
763