LLVM  9.0.0svn
FileCheck.cpp
Go to the documentation of this file.
1 //===- FileCheck.cpp - Check that File's Contents match what is expected --===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 // FileCheck does a line-by line check of a file that validates whether it
10 // contains the expected content. This is useful for regression tests etc.
11 //
12 // This file implements most of the API that will be used by the FileCheck utility
13 // as well as various unittests.
14 //===----------------------------------------------------------------------===//
15 
16 #include "llvm/Support/FileCheck.h"
17 #include "llvm/ADT/StringSet.h"
19 #include <cstdint>
20 #include <list>
21 #include <map>
22 #include <tuple>
23 #include <utility>
24 
25 using namespace llvm;
26 
27 /// Parses the given string into the Pattern.
28 ///
29 /// \p Prefix provides which prefix is being matched, \p SM provides the
30 /// SourceMgr used for error reports, and \p LineNumber is the line number in
31 /// the input file from which the pattern string was read. Returns true in
32 /// case of an error, false otherwise.
34  SourceMgr &SM, unsigned LineNumber,
35  const FileCheckRequest &Req) {
36  bool MatchFullLinesHere = Req.MatchFullLines && CheckTy != Check::CheckNot;
37 
38  this->LineNumber = LineNumber;
39  PatternLoc = SMLoc::getFromPointer(PatternStr.data());
40 
41  if (!(Req.NoCanonicalizeWhiteSpace && Req.MatchFullLines))
42  // Ignore trailing whitespace.
43  while (!PatternStr.empty() &&
44  (PatternStr.back() == ' ' || PatternStr.back() == '\t'))
45  PatternStr = PatternStr.substr(0, PatternStr.size() - 1);
46 
47  // Check that there is something on the line.
48  if (PatternStr.empty() && CheckTy != Check::CheckEmpty) {
49  SM.PrintMessage(PatternLoc, SourceMgr::DK_Error,
50  "found empty check string with prefix '" + Prefix + ":'");
51  return true;
52  }
53 
54  if (!PatternStr.empty() && CheckTy == Check::CheckEmpty) {
55  SM.PrintMessage(
56  PatternLoc, SourceMgr::DK_Error,
57  "found non-empty check string for empty check with prefix '" + Prefix +
58  ":'");
59  return true;
60  }
61 
62  if (CheckTy == Check::CheckEmpty) {
63  RegExStr = "(\n$)";
64  return false;
65  }
66 
67  // Check to see if this is a fixed string, or if it has regex pieces.
68  if (!MatchFullLinesHere &&
69  (PatternStr.size() < 2 || (PatternStr.find("{{") == StringRef::npos &&
70  PatternStr.find("[[") == StringRef::npos))) {
71  FixedStr = PatternStr;
72  return false;
73  }
74 
75  if (MatchFullLinesHere) {
76  RegExStr += '^';
77  if (!Req.NoCanonicalizeWhiteSpace)
78  RegExStr += " *";
79  }
80 
81  // Paren value #0 is for the fully matched string. Any new parenthesized
82  // values add from there.
83  unsigned CurParen = 1;
84 
85  // Otherwise, there is at least one regex piece. Build up the regex pattern
86  // by escaping scary characters in fixed strings, building up one big regex.
87  while (!PatternStr.empty()) {
88  // RegEx matches.
89  if (PatternStr.startswith("{{")) {
90  // This is the start of a regex match. Scan for the }}.
91  size_t End = PatternStr.find("}}");
92  if (End == StringRef::npos) {
93  SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()),
95  "found start of regex string with no end '}}'");
96  return true;
97  }
98 
99  // Enclose {{}} patterns in parens just like [[]] even though we're not
100  // capturing the result for any purpose. This is required in case the
101  // expression contains an alternation like: CHECK: abc{{x|z}}def. We
102  // want this to turn into: "abc(x|z)def" not "abcx|zdef".
103  RegExStr += '(';
104  ++CurParen;
105 
106  if (AddRegExToRegEx(PatternStr.substr(2, End - 2), CurParen, SM))
107  return true;
108  RegExStr += ')';
109 
110  PatternStr = PatternStr.substr(End + 2);
111  continue;
112  }
113 
114  // Named RegEx matches. These are of two forms: [[foo:.*]] which matches .*
115  // (or some other regex) and assigns it to the FileCheck variable 'foo'. The
116  // second form is [[foo]] which is a reference to foo. The variable name
117  // itself must be of the form "[a-zA-Z_][0-9a-zA-Z_]*", otherwise we reject
118  // it. This is to catch some common errors.
119  if (PatternStr.startswith("[[")) {
120  // Find the closing bracket pair ending the match. End is going to be an
121  // offset relative to the beginning of the match string.
122  size_t End = FindRegexVarEnd(PatternStr.substr(2), SM);
123 
124  if (End == StringRef::npos) {
125  SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()),
127  "invalid named regex reference, no ]] found");
128  return true;
129  }
130 
131  StringRef MatchStr = PatternStr.substr(2, End);
132  PatternStr = PatternStr.substr(End + 4);
133 
134  // Get the regex name (e.g. "foo").
135  size_t NameEnd = MatchStr.find(':');
136  StringRef Name = MatchStr.substr(0, NameEnd);
137 
138  if (Name.empty()) {
140  "invalid name in named regex: empty name");
141  return true;
142  }
143 
144  // Verify that the name/expression is well formed. FileCheck currently
145  // supports @LINE, @LINE+number, @LINE-number expressions. The check here
146  // is relaxed, more strict check is performed in \c EvaluateExpression.
147  bool IsExpression = false;
148  for (unsigned i = 0, e = Name.size(); i != e; ++i) {
149  if (i == 0) {
150  if (Name[i] == '$') // Global vars start with '$'
151  continue;
152  if (Name[i] == '@') {
153  if (NameEnd != StringRef::npos) {
156  "invalid name in named regex definition");
157  return true;
158  }
159  IsExpression = true;
160  continue;
161  }
162  }
163  if (Name[i] != '_' && !isalnum(Name[i]) &&
164  (!IsExpression || (Name[i] != '+' && Name[i] != '-'))) {
165  SM.PrintMessage(SMLoc::getFromPointer(Name.data() + i),
166  SourceMgr::DK_Error, "invalid name in named regex");
167  return true;
168  }
169  }
170 
171  // Name can't start with a digit.
172  if (isdigit(static_cast<unsigned char>(Name[0]))) {
174  "invalid name in named regex");
175  return true;
176  }
177 
178  // Handle [[foo]].
179  if (NameEnd == StringRef::npos) {
180  // Handle variables that were defined earlier on the same line by
181  // emitting a backreference.
182  if (VariableDefs.find(Name) != VariableDefs.end()) {
183  unsigned VarParenNum = VariableDefs[Name];
184  if (VarParenNum < 1 || VarParenNum > 9) {
187  "Can't back-reference more than 9 variables");
188  return true;
189  }
190  AddBackrefToRegEx(VarParenNum);
191  } else {
192  VariableUses.push_back(std::make_pair(Name, RegExStr.size()));
193  }
194  continue;
195  }
196 
197  // Handle [[foo:.*]].
198  VariableDefs[Name] = CurParen;
199  RegExStr += '(';
200  ++CurParen;
201 
202  if (AddRegExToRegEx(MatchStr.substr(NameEnd + 1), CurParen, SM))
203  return true;
204 
205  RegExStr += ')';
206  }
207 
208  // Handle fixed string matches.
209  // Find the end, which is the start of the next regex.
210  size_t FixedMatchEnd = PatternStr.find("{{");
211  FixedMatchEnd = std::min(FixedMatchEnd, PatternStr.find("[["));
212  RegExStr += Regex::escape(PatternStr.substr(0, FixedMatchEnd));
213  PatternStr = PatternStr.substr(FixedMatchEnd);
214  }
215 
216  if (MatchFullLinesHere) {
217  if (!Req.NoCanonicalizeWhiteSpace)
218  RegExStr += " *";
219  RegExStr += '$';
220  }
221 
222  return false;
223 }
224 
225 bool FileCheckPattern::AddRegExToRegEx(StringRef RS, unsigned &CurParen, SourceMgr &SM) {
226  Regex R(RS);
227  std::string Error;
228  if (!R.isValid(Error)) {
230  "invalid regex: " + Error);
231  return true;
232  }
233 
234  RegExStr += RS.str();
235  CurParen += R.getNumMatches();
236  return false;
237 }
238 
239 void FileCheckPattern::AddBackrefToRegEx(unsigned BackrefNum) {
240  assert(BackrefNum >= 1 && BackrefNum <= 9 && "Invalid backref number");
241  std::string Backref = std::string("\\") + std::string(1, '0' + BackrefNum);
242  RegExStr += Backref;
243 }
244 
245 /// Evaluates expression and stores the result to \p Value.
246 ///
247 /// Returns true on success and false when the expression has invalid syntax.
248 bool FileCheckPattern::EvaluateExpression(StringRef Expr, std::string &Value) const {
249  // The only supported expression is @LINE([\+-]\d+)?
250  if (!Expr.startswith("@LINE"))
251  return false;
252  Expr = Expr.substr(StringRef("@LINE").size());
253  int Offset = 0;
254  if (!Expr.empty()) {
255  if (Expr[0] == '+')
256  Expr = Expr.substr(1);
257  else if (Expr[0] != '-')
258  return false;
259  if (Expr.getAsInteger(10, Offset))
260  return false;
261  }
262  Value = llvm::itostr(LineNumber + Offset);
263  return true;
264 }
265 
266 /// Matches the pattern string against the input buffer \p Buffer
267 ///
268 /// This returns the position that is matched or npos if there is no match. If
269 /// there is a match, the size of the matched string is returned in \p
270 /// MatchLen.
271 ///
272 /// The GlobalVariableTable StringMap in the FileCheckPatternContext class
273 /// instance provides the current values of FileCheck variables and is updated
274 /// if this match defines new values.
275 size_t FileCheckPattern::match(StringRef Buffer, size_t &MatchLen) const {
276  // If this is the EOF pattern, match it immediately.
277  if (CheckTy == Check::CheckEOF) {
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 (const auto &VariableUse : VariableUses) {
299  std::string Value;
300 
301  if (VariableUse.first[0] == '@') {
302  if (!EvaluateExpression(VariableUse.first, Value))
303  return StringRef::npos;
304  } else {
305  llvm::Optional<StringRef> ValueRef =
306  Context->getVarValue(VariableUse.first);
307  // If the variable is undefined, return an error.
308  if (!ValueRef)
309  return StringRef::npos;
310 
311  // Look up the value and escape it so that we can put it into the regex.
312  Value += Regex::escape(*ValueRef);
313  }
314 
315  // Plop it into the regex at the adjusted offset.
316  TmpStr.insert(TmpStr.begin() + VariableUse.second + InsertOffset,
317  Value.begin(), Value.end());
318  InsertOffset += Value.size();
319  }
320 
321  // Match the newly constructed regex.
322  RegExToMatch = TmpStr;
323  }
324 
325  SmallVector<StringRef, 4> MatchInfo;
326  if (!Regex(RegExToMatch, Regex::Newline).match(Buffer, &MatchInfo))
327  return StringRef::npos;
328 
329  // Successful regex match.
330  assert(!MatchInfo.empty() && "Didn't get any match");
331  StringRef FullMatch = MatchInfo[0];
332 
333  // If this defines any variables, remember their values.
334  for (const auto &VariableDef : VariableDefs) {
335  assert(VariableDef.second < MatchInfo.size() && "Internal paren error");
336  Context->GlobalVariableTable[VariableDef.first] =
337  MatchInfo[VariableDef.second];
338  }
339 
340  // Like CHECK-NEXT, CHECK-EMPTY's match range is considered to start after
341  // the required preceding newline, which is consumed by the pattern in the
342  // case of CHECK-EMPTY but not CHECK-NEXT.
343  size_t MatchStartSkip = CheckTy == Check::CheckEmpty;
344  MatchLen = FullMatch.size() - MatchStartSkip;
345  return FullMatch.data() - Buffer.data() + MatchStartSkip;
346 }
347 
348 /// Computes an arbitrary estimate for the quality of matching this pattern at
349 /// the start of \p Buffer; a distance of zero should correspond to a perfect
350 /// match.
351 unsigned FileCheckPattern::computeMatchDistance(StringRef Buffer) 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 
369  SMRange MatchRange) const {
370  // If this was a regular expression using variables, print the current
371  // variable values.
372  if (!VariableUses.empty()) {
373  for (const auto &VariableUse : VariableUses) {
374  SmallString<256> Msg;
375  raw_svector_ostream OS(Msg);
376  StringRef Var = VariableUse.first;
377  if (Var[0] == '@') {
378  std::string Value;
379  if (EvaluateExpression(Var, Value)) {
380  OS << "with expression \"";
381  OS.write_escaped(Var) << "\" equal to \"";
382  OS.write_escaped(Value) << "\"";
383  } else {
384  OS << "uses incorrect expression \"";
385  OS.write_escaped(Var) << "\"";
386  }
387  } else {
388  llvm::Optional<StringRef> VarValue = Context->getVarValue(Var);
389 
390  // Check for undefined variable references.
391  if (!VarValue) {
392  OS << "uses undefined variable \"";
393  OS.write_escaped(Var) << "\"";
394  } else {
395  OS << "with variable \"";
396  OS.write_escaped(Var) << "\" equal to \"";
397  OS.write_escaped(*VarValue) << "\"";
398  }
399  }
400 
401  if (MatchRange.isValid())
402  SM.PrintMessage(MatchRange.Start, SourceMgr::DK_Note, OS.str(),
403  {MatchRange});
404  else
406  SourceMgr::DK_Note, OS.str());
407  }
408  }
409 }
410 
412  const SourceMgr &SM, SMLoc Loc,
413  Check::FileCheckType CheckTy,
414  StringRef Buffer, size_t Pos, size_t Len,
415  std::vector<FileCheckDiag> *Diags,
416  bool AdjustPrevDiag = false) {
417  SMLoc Start = SMLoc::getFromPointer(Buffer.data() + Pos);
418  SMLoc End = SMLoc::getFromPointer(Buffer.data() + Pos + Len);
419  SMRange Range(Start, End);
420  if (Diags) {
421  if (AdjustPrevDiag)
422  Diags->rbegin()->MatchTy = MatchTy;
423  else
424  Diags->emplace_back(SM, CheckTy, Loc, MatchTy, Range);
425  }
426  return Range;
427 }
428 
430  const SourceMgr &SM, StringRef Buffer,
431  std::vector<FileCheckDiag> *Diags) const {
432  // Attempt to find the closest/best fuzzy match. Usually an error happens
433  // because some string in the output didn't exactly match. In these cases, we
434  // would like to show the user a best guess at what "should have" matched, to
435  // save them having to actually check the input manually.
436  size_t NumLinesForward = 0;
437  size_t Best = StringRef::npos;
438  double BestQuality = 0;
439 
440  // Use an arbitrary 4k limit on how far we will search.
441  for (size_t i = 0, e = std::min(size_t(4096), Buffer.size()); i != e; ++i) {
442  if (Buffer[i] == '\n')
443  ++NumLinesForward;
444 
445  // Patterns have leading whitespace stripped, so skip whitespace when
446  // looking for something which looks like a pattern.
447  if (Buffer[i] == ' ' || Buffer[i] == '\t')
448  continue;
449 
450  // Compute the "quality" of this match as an arbitrary combination of the
451  // match distance and the number of lines skipped to get to this match.
452  unsigned Distance = computeMatchDistance(Buffer.substr(i));
453  double Quality = Distance + (NumLinesForward / 100.);
454 
455  if (Quality < BestQuality || Best == StringRef::npos) {
456  Best = i;
457  BestQuality = Quality;
458  }
459  }
460 
461  // Print the "possible intended match here" line if we found something
462  // reasonable and not equal to what we showed in the "scanning from here"
463  // line.
464  if (Best && Best != StringRef::npos && BestQuality < 50) {
465  SMRange MatchRange =
467  getCheckTy(), Buffer, Best, 0, Diags);
468  SM.PrintMessage(MatchRange.Start, SourceMgr::DK_Note,
469  "possible intended match here");
470 
471  // FIXME: If we wanted to be really friendly we would show why the match
472  // failed, as it can be hard to spot simple one character differences.
473  }
474 }
475 
478  auto VarIter = GlobalVariableTable.find(VarName);
479  if (VarIter == GlobalVariableTable.end())
480  return llvm::None;
481 
482  return VarIter->second;
483 }
484 
485 /// Finds the closing sequence of a regex variable usage or definition.
486 ///
487 /// \p Str has to point in the beginning of the definition (right after the
488 /// opening sequence). Returns the offset of the closing sequence within Str,
489 /// or npos if it was not found.
490 size_t FileCheckPattern::FindRegexVarEnd(StringRef Str, SourceMgr &SM) {
491  // Offset keeps track of the current offset within the input Str
492  size_t Offset = 0;
493  // [...] Nesting depth
494  size_t BracketDepth = 0;
495 
496  while (!Str.empty()) {
497  if (Str.startswith("]]") && BracketDepth == 0)
498  return Offset;
499  if (Str[0] == '\\') {
500  // Backslash escapes the next char within regexes, so skip them both.
501  Str = Str.substr(2);
502  Offset += 2;
503  } else {
504  switch (Str[0]) {
505  default:
506  break;
507  case '[':
508  BracketDepth++;
509  break;
510  case ']':
511  if (BracketDepth == 0) {
514  "missing closing \"]\" for regex variable");
515  exit(1);
516  }
517  BracketDepth--;
518  break;
519  }
520  Str = Str.substr(1);
521  Offset++;
522  }
523  }
524 
525  return StringRef::npos;
526 }
527 
528 /// Canonicalize whitespaces in the file. Line endings are replaced with
529 /// UNIX-style '\n'.
530 StringRef
532  SmallVectorImpl<char> &OutputBuffer) {
533  OutputBuffer.reserve(MB.getBufferSize());
534 
535  for (const char *Ptr = MB.getBufferStart(), *End = MB.getBufferEnd();
536  Ptr != End; ++Ptr) {
537  // Eliminate trailing dosish \r.
538  if (Ptr <= End - 2 && Ptr[0] == '\r' && Ptr[1] == '\n') {
539  continue;
540  }
541 
542  // If current char is not a horizontal whitespace or if horizontal
543  // whitespace canonicalization is disabled, dump it to output as is.
544  if (Req.NoCanonicalizeWhiteSpace || (*Ptr != ' ' && *Ptr != '\t')) {
545  OutputBuffer.push_back(*Ptr);
546  continue;
547  }
548 
549  // Otherwise, add one space and advance over neighboring space.
550  OutputBuffer.push_back(' ');
551  while (Ptr + 1 != End && (Ptr[1] == ' ' || Ptr[1] == '\t'))
552  ++Ptr;
553  }
554 
555  // Add a null byte and then return all but that byte.
556  OutputBuffer.push_back('\0');
557  return StringRef(OutputBuffer.data(), OutputBuffer.size() - 1);
558 }
559 
561  const Check::FileCheckType &CheckTy,
562  SMLoc CheckLoc, MatchType MatchTy,
563  SMRange InputRange)
564  : CheckTy(CheckTy), MatchTy(MatchTy) {
565  auto Start = SM.getLineAndColumn(InputRange.Start);
566  auto End = SM.getLineAndColumn(InputRange.End);
567  InputStartLine = Start.first;
568  InputStartCol = Start.second;
569  InputEndLine = End.first;
570  InputEndCol = End.second;
571  Start = SM.getLineAndColumn(CheckLoc);
572  CheckLine = Start.first;
573  CheckCol = Start.second;
574 }
575 
576 static bool IsPartOfWord(char c) {
577  return (isalnum(c) || c == '-' || c == '_');
578 }
579 
581  assert(Count > 0 && "zero and negative counts are not supported");
582  assert((C == 1 || Kind == CheckPlain) &&
583  "count supported only for plain CHECK directives");
584  Count = C;
585  return *this;
586 }
587 
588 // Get a description of the type.
590  switch (Kind) {
591  case Check::CheckNone:
592  return "invalid";
593  case Check::CheckPlain:
594  if (Count > 1)
595  return Prefix.str() + "-COUNT";
596  return Prefix;
597  case Check::CheckNext:
598  return Prefix.str() + "-NEXT";
599  case Check::CheckSame:
600  return Prefix.str() + "-SAME";
601  case Check::CheckNot:
602  return Prefix.str() + "-NOT";
603  case Check::CheckDAG:
604  return Prefix.str() + "-DAG";
605  case Check::CheckLabel:
606  return Prefix.str() + "-LABEL";
607  case Check::CheckEmpty:
608  return Prefix.str() + "-EMPTY";
609  case Check::CheckEOF:
610  return "implicit EOF";
611  case Check::CheckBadNot:
612  return "bad NOT";
614  return "bad COUNT";
615  }
616  llvm_unreachable("unknown FileCheckType");
617 }
618 
619 static std::pair<Check::FileCheckType, StringRef>
621  if (Buffer.size() <= Prefix.size())
622  return {Check::CheckNone, StringRef()};
623 
624  char NextChar = Buffer[Prefix.size()];
625 
626  StringRef Rest = Buffer.drop_front(Prefix.size() + 1);
627  // Verify that the : is present after the prefix.
628  if (NextChar == ':')
629  return {Check::CheckPlain, Rest};
630 
631  if (NextChar != '-')
632  return {Check::CheckNone, StringRef()};
633 
634  if (Rest.consume_front("COUNT-")) {
635  int64_t Count;
636  if (Rest.consumeInteger(10, Count))
637  // Error happened in parsing integer.
638  return {Check::CheckBadCount, Rest};
639  if (Count <= 0 || Count > INT32_MAX)
640  return {Check::CheckBadCount, Rest};
641  if (!Rest.consume_front(":"))
642  return {Check::CheckBadCount, Rest};
643  return {Check::FileCheckType(Check::CheckPlain).setCount(Count), Rest};
644  }
645 
646  if (Rest.consume_front("NEXT:"))
647  return {Check::CheckNext, Rest};
648 
649  if (Rest.consume_front("SAME:"))
650  return {Check::CheckSame, Rest};
651 
652  if (Rest.consume_front("NOT:"))
653  return {Check::CheckNot, Rest};
654 
655  if (Rest.consume_front("DAG:"))
656  return {Check::CheckDAG, Rest};
657 
658  if (Rest.consume_front("LABEL:"))
659  return {Check::CheckLabel, Rest};
660 
661  if (Rest.consume_front("EMPTY:"))
662  return {Check::CheckEmpty, Rest};
663 
664  // You can't combine -NOT with another suffix.
665  if (Rest.startswith("DAG-NOT:") || Rest.startswith("NOT-DAG:") ||
666  Rest.startswith("NEXT-NOT:") || Rest.startswith("NOT-NEXT:") ||
667  Rest.startswith("SAME-NOT:") || Rest.startswith("NOT-SAME:") ||
668  Rest.startswith("EMPTY-NOT:") || Rest.startswith("NOT-EMPTY:"))
669  return {Check::CheckBadNot, Rest};
670 
671  return {Check::CheckNone, Rest};
672 }
673 
674 // From the given position, find the next character after the word.
675 static size_t SkipWord(StringRef Str, size_t Loc) {
676  while (Loc < Str.size() && IsPartOfWord(Str[Loc]))
677  ++Loc;
678  return Loc;
679 }
680 
681 /// Search the buffer for the first prefix in the prefix regular expression.
682 ///
683 /// This searches the buffer using the provided regular expression, however it
684 /// enforces constraints beyond that:
685 /// 1) The found prefix must not be a suffix of something that looks like
686 /// a valid prefix.
687 /// 2) The found prefix must be followed by a valid check type suffix using \c
688 /// FindCheckType above.
689 ///
690 /// Returns a pair of StringRefs into the Buffer, which combines:
691 /// - the first match of the regular expression to satisfy these two is
692 /// returned,
693 /// otherwise an empty StringRef is returned to indicate failure.
694 /// - buffer rewound to the location right after parsed suffix, for parsing
695 /// to continue from
696 ///
697 /// If this routine returns a valid prefix, it will also shrink \p Buffer to
698 /// start at the beginning of the returned prefix, increment \p LineNumber for
699 /// each new line consumed from \p Buffer, and set \p CheckTy to the type of
700 /// check found by examining the suffix.
701 ///
702 /// If no valid prefix is found, the state of Buffer, LineNumber, and CheckTy
703 /// is unspecified.
704 static std::pair<StringRef, StringRef>
706  unsigned &LineNumber, Check::FileCheckType &CheckTy) {
708 
709  while (!Buffer.empty()) {
710  // Find the first (longest) match using the RE.
711  if (!PrefixRE.match(Buffer, &Matches))
712  // No match at all, bail.
713  return {StringRef(), StringRef()};
714 
715  StringRef Prefix = Matches[0];
716  Matches.clear();
717 
718  assert(Prefix.data() >= Buffer.data() &&
719  Prefix.data() < Buffer.data() + Buffer.size() &&
720  "Prefix doesn't start inside of buffer!");
721  size_t Loc = Prefix.data() - Buffer.data();
722  StringRef Skipped = Buffer.substr(0, Loc);
723  Buffer = Buffer.drop_front(Loc);
724  LineNumber += Skipped.count('\n');
725 
726  // Check that the matched prefix isn't a suffix of some other check-like
727  // word.
728  // FIXME: This is a very ad-hoc check. it would be better handled in some
729  // other way. Among other things it seems hard to distinguish between
730  // intentional and unintentional uses of this feature.
731  if (Skipped.empty() || !IsPartOfWord(Skipped.back())) {
732  // Now extract the type.
733  StringRef AfterSuffix;
734  std::tie(CheckTy, AfterSuffix) = FindCheckType(Buffer, Prefix);
735 
736  // If we've found a valid check type for this prefix, we're done.
737  if (CheckTy != Check::CheckNone)
738  return {Prefix, AfterSuffix};
739  }
740 
741  // If we didn't successfully find a prefix, we need to skip this invalid
742  // prefix and continue scanning. We directly skip the prefix that was
743  // matched and any additional parts of that check-like word.
744  Buffer = Buffer.drop_front(SkipWord(Buffer, Prefix.size()));
745  }
746 
747  // We ran out of buffer while skipping partial matches so give up.
748  return {StringRef(), StringRef()};
749 }
750 
751 /// Read the check file, which specifies the sequence of expected strings.
752 ///
753 /// The strings are added to the CheckStrings vector. Returns true in case of
754 /// an error, false otherwise.
756  SourceMgr &SM, StringRef Buffer, Regex &PrefixRE,
757  std::vector<FileCheckString> &CheckStrings) {
758  PatternContext.defineCmdlineVariables(Req.GlobalDefines);
759 
760  std::vector<FileCheckPattern> ImplicitNegativeChecks;
761  for (const auto &PatternString : Req.ImplicitCheckNot) {
762  // Create a buffer with fake command line content in order to display the
763  // command line option responsible for the specific implicit CHECK-NOT.
764  std::string Prefix = "-implicit-check-not='";
765  std::string Suffix = "'";
766  std::unique_ptr<MemoryBuffer> CmdLine = MemoryBuffer::getMemBufferCopy(
767  Prefix + PatternString + Suffix, "command line");
768 
769  StringRef PatternInBuffer =
770  CmdLine->getBuffer().substr(Prefix.size(), PatternString.size());
771  SM.AddNewSourceBuffer(std::move(CmdLine), SMLoc());
772 
773  ImplicitNegativeChecks.push_back(
774  FileCheckPattern(Check::CheckNot, &PatternContext));
775  ImplicitNegativeChecks.back().ParsePattern(PatternInBuffer,
776  "IMPLICIT-CHECK", SM, 0, Req);
777  }
778 
779  std::vector<FileCheckPattern> DagNotMatches = ImplicitNegativeChecks;
780 
781  // LineNumber keeps track of the line on which CheckPrefix instances are
782  // found.
783  unsigned LineNumber = 1;
784 
785  while (1) {
787 
788  // See if a prefix occurs in the memory buffer.
789  StringRef UsedPrefix;
790  StringRef AfterSuffix;
791  std::tie(UsedPrefix, AfterSuffix) =
792  FindFirstMatchingPrefix(PrefixRE, Buffer, LineNumber, CheckTy);
793  if (UsedPrefix.empty())
794  break;
795  assert(UsedPrefix.data() == Buffer.data() &&
796  "Failed to move Buffer's start forward, or pointed prefix outside "
797  "of the buffer!");
798  assert(AfterSuffix.data() >= Buffer.data() &&
799  AfterSuffix.data() < Buffer.data() + Buffer.size() &&
800  "Parsing after suffix doesn't start inside of buffer!");
801 
802  // Location to use for error messages.
803  const char *UsedPrefixStart = UsedPrefix.data();
804 
805  // Skip the buffer to the end of parsed suffix (or just prefix, if no good
806  // suffix was processed).
807  Buffer = AfterSuffix.empty() ? Buffer.drop_front(UsedPrefix.size())
808  : AfterSuffix;
809 
810  // Complain about useful-looking but unsupported suffixes.
811  if (CheckTy == Check::CheckBadNot) {
813  "unsupported -NOT combo on prefix '" + UsedPrefix + "'");
814  return true;
815  }
816 
817  // Complain about invalid count specification.
818  if (CheckTy == Check::CheckBadCount) {
820  "invalid count in -COUNT specification on prefix '" +
821  UsedPrefix + "'");
822  return true;
823  }
824 
825  // Okay, we found the prefix, yay. Remember the rest of the line, but ignore
826  // leading whitespace.
827  if (!(Req.NoCanonicalizeWhiteSpace && Req.MatchFullLines))
828  Buffer = Buffer.substr(Buffer.find_first_not_of(" \t"));
829 
830  // Scan ahead to the end of line.
831  size_t EOL = Buffer.find_first_of("\n\r");
832 
833  // Remember the location of the start of the pattern, for diagnostics.
834  SMLoc PatternLoc = SMLoc::getFromPointer(Buffer.data());
835 
836  // Parse the pattern.
837  FileCheckPattern P(CheckTy, &PatternContext);
838  if (P.ParsePattern(Buffer.substr(0, EOL), UsedPrefix, SM, LineNumber, Req))
839  return true;
840 
841  // Verify that CHECK-LABEL lines do not define or use variables
842  if ((CheckTy == Check::CheckLabel) && P.hasVariable()) {
843  SM.PrintMessage(
844  SMLoc::getFromPointer(UsedPrefixStart), SourceMgr::DK_Error,
845  "found '" + UsedPrefix + "-LABEL:'"
846  " with variable definition or use");
847  return true;
848  }
849 
850  Buffer = Buffer.substr(EOL);
851 
852  // Verify that CHECK-NEXT/SAME/EMPTY lines have at least one CHECK line before them.
853  if ((CheckTy == Check::CheckNext || CheckTy == Check::CheckSame ||
854  CheckTy == Check::CheckEmpty) &&
855  CheckStrings.empty()) {
856  StringRef Type = CheckTy == Check::CheckNext
857  ? "NEXT"
858  : CheckTy == Check::CheckEmpty ? "EMPTY" : "SAME";
859  SM.PrintMessage(SMLoc::getFromPointer(UsedPrefixStart),
861  "found '" + UsedPrefix + "-" + Type +
862  "' without previous '" + UsedPrefix + ": line");
863  return true;
864  }
865 
866  // Handle CHECK-DAG/-NOT.
867  if (CheckTy == Check::CheckDAG || CheckTy == Check::CheckNot) {
868  DagNotMatches.push_back(P);
869  continue;
870  }
871 
872  // Okay, add the string we captured to the output vector and move on.
873  CheckStrings.emplace_back(P, UsedPrefix, PatternLoc);
874  std::swap(DagNotMatches, CheckStrings.back().DagNotStrings);
875  DagNotMatches = ImplicitNegativeChecks;
876  }
877 
878  // Add an EOF pattern for any trailing CHECK-DAG/-NOTs, and use the first
879  // prefix as a filler for the error message.
880  if (!DagNotMatches.empty()) {
881  CheckStrings.emplace_back(
882  FileCheckPattern(Check::CheckEOF, &PatternContext),
883  *Req.CheckPrefixes.begin(), SMLoc::getFromPointer(Buffer.data()));
884  std::swap(DagNotMatches, CheckStrings.back().DagNotStrings);
885  }
886 
887  if (CheckStrings.empty()) {
888  errs() << "error: no check strings found with prefix"
889  << (Req.CheckPrefixes.size() > 1 ? "es " : " ");
890  auto I = Req.CheckPrefixes.begin();
891  auto E = Req.CheckPrefixes.end();
892  if (I != E) {
893  errs() << "\'" << *I << ":'";
894  ++I;
895  }
896  for (; I != E; ++I)
897  errs() << ", \'" << *I << ":'";
898 
899  errs() << '\n';
900  return true;
901  }
902 
903  return false;
904 }
905 
906 static void PrintMatch(bool ExpectedMatch, const SourceMgr &SM,
907  StringRef Prefix, SMLoc Loc, const FileCheckPattern &Pat,
908  int MatchedCount, StringRef Buffer, size_t MatchPos,
909  size_t MatchLen, const FileCheckRequest &Req,
910  std::vector<FileCheckDiag> *Diags) {
911  bool PrintDiag = true;
912  if (ExpectedMatch) {
913  if (!Req.Verbose)
914  return;
915  if (!Req.VerboseVerbose && Pat.getCheckTy() == Check::CheckEOF)
916  return;
917  // Due to their verbosity, we don't print verbose diagnostics here if we're
918  // gathering them for a different rendering, but we always print other
919  // diagnostics.
920  PrintDiag = !Diags;
921  }
922  SMRange MatchRange = ProcessMatchResult(
925  SM, Loc, Pat.getCheckTy(), Buffer, MatchPos, MatchLen, Diags);
926  if (!PrintDiag)
927  return;
928 
929  std::string Message = formatv("{0}: {1} string found in input",
930  Pat.getCheckTy().getDescription(Prefix),
931  (ExpectedMatch ? "expected" : "excluded"))
932  .str();
933  if (Pat.getCount() > 1)
934  Message += formatv(" ({0} out of {1})", MatchedCount, Pat.getCount()).str();
935 
936  SM.PrintMessage(
937  Loc, ExpectedMatch ? SourceMgr::DK_Remark : SourceMgr::DK_Error, Message);
938  SM.PrintMessage(MatchRange.Start, SourceMgr::DK_Note, "found here",
939  {MatchRange});
940  Pat.printVariableUses(SM, Buffer, MatchRange);
941 }
942 
943 static void PrintMatch(bool ExpectedMatch, const SourceMgr &SM,
944  const FileCheckString &CheckStr, int MatchedCount,
945  StringRef Buffer, size_t MatchPos, size_t MatchLen,
946  FileCheckRequest &Req,
947  std::vector<FileCheckDiag> *Diags) {
948  PrintMatch(ExpectedMatch, SM, CheckStr.Prefix, CheckStr.Loc, CheckStr.Pat,
949  MatchedCount, Buffer, MatchPos, MatchLen, Req, Diags);
950 }
951 
952 static void PrintNoMatch(bool ExpectedMatch, const SourceMgr &SM,
953  StringRef Prefix, SMLoc Loc,
954  const FileCheckPattern &Pat, int MatchedCount,
955  StringRef Buffer, bool VerboseVerbose,
956  std::vector<FileCheckDiag> *Diags) {
957  bool PrintDiag = true;
958  if (!ExpectedMatch) {
959  if (!VerboseVerbose)
960  return;
961  // Due to their verbosity, we don't print verbose diagnostics here if we're
962  // gathering them for a different rendering, but we always print other
963  // diagnostics.
964  PrintDiag = !Diags;
965  }
966 
967  // If the current position is at the end of a line, advance to the start of
968  // the next line.
969  Buffer = Buffer.substr(Buffer.find_first_not_of(" \t\n\r"));
970  SMRange SearchRange = ProcessMatchResult(
973  SM, Loc, Pat.getCheckTy(), Buffer, 0, Buffer.size(), Diags);
974  if (!PrintDiag)
975  return;
976 
977  // Print "not found" diagnostic.
978  std::string Message = formatv("{0}: {1} string not found in input",
979  Pat.getCheckTy().getDescription(Prefix),
980  (ExpectedMatch ? "expected" : "excluded"))
981  .str();
982  if (Pat.getCount() > 1)
983  Message += formatv(" ({0} out of {1})", MatchedCount, Pat.getCount()).str();
984  SM.PrintMessage(
985  Loc, ExpectedMatch ? SourceMgr::DK_Error : SourceMgr::DK_Remark, Message);
986 
987  // Print the "scanning from here" line.
988  SM.PrintMessage(SearchRange.Start, SourceMgr::DK_Note, "scanning from here");
989 
990  // Allow the pattern to print additional information if desired.
991  Pat.printVariableUses(SM, Buffer);
992 
993  if (ExpectedMatch)
994  Pat.printFuzzyMatch(SM, Buffer, Diags);
995 }
996 
997 static void PrintNoMatch(bool ExpectedMatch, const SourceMgr &SM,
998  const FileCheckString &CheckStr, int MatchedCount,
999  StringRef Buffer, bool VerboseVerbose,
1000  std::vector<FileCheckDiag> *Diags) {
1001  PrintNoMatch(ExpectedMatch, SM, CheckStr.Prefix, CheckStr.Loc, CheckStr.Pat,
1002  MatchedCount, Buffer, VerboseVerbose, Diags);
1003 }
1004 
1005 /// Count the number of newlines in the specified range.
1006 static unsigned CountNumNewlinesBetween(StringRef Range,
1007  const char *&FirstNewLine) {
1008  unsigned NumNewLines = 0;
1009  while (1) {
1010  // Scan for newline.
1011  Range = Range.substr(Range.find_first_of("\n\r"));
1012  if (Range.empty())
1013  return NumNewLines;
1014 
1015  ++NumNewLines;
1016 
1017  // Handle \n\r and \r\n as a single newline.
1018  if (Range.size() > 1 && (Range[1] == '\n' || Range[1] == '\r') &&
1019  (Range[0] != Range[1]))
1020  Range = Range.substr(1);
1021  Range = Range.substr(1);
1022 
1023  if (NumNewLines == 1)
1024  FirstNewLine = Range.begin();
1025  }
1026 }
1027 
1028 /// Match check string and its "not strings" and/or "dag strings".
1029 size_t FileCheckString::Check(const SourceMgr &SM, StringRef Buffer,
1030  bool IsLabelScanMode, size_t &MatchLen,
1031  FileCheckRequest &Req,
1032  std::vector<FileCheckDiag> *Diags) const {
1033  size_t LastPos = 0;
1034  std::vector<const FileCheckPattern *> NotStrings;
1035 
1036  // IsLabelScanMode is true when we are scanning forward to find CHECK-LABEL
1037  // bounds; we have not processed variable definitions within the bounded block
1038  // yet so cannot handle any final CHECK-DAG yet; this is handled when going
1039  // over the block again (including the last CHECK-LABEL) in normal mode.
1040  if (!IsLabelScanMode) {
1041  // Match "dag strings" (with mixed "not strings" if any).
1042  LastPos = CheckDag(SM, Buffer, NotStrings, Req, Diags);
1043  if (LastPos == StringRef::npos)
1044  return StringRef::npos;
1045  }
1046 
1047  // Match itself from the last position after matching CHECK-DAG.
1048  size_t LastMatchEnd = LastPos;
1049  size_t FirstMatchPos = 0;
1050  // Go match the pattern Count times. Majority of patterns only match with
1051  // count 1 though.
1052  assert(Pat.getCount() != 0 && "pattern count can not be zero");
1053  for (int i = 1; i <= Pat.getCount(); i++) {
1054  StringRef MatchBuffer = Buffer.substr(LastMatchEnd);
1055  size_t CurrentMatchLen;
1056  // get a match at current start point
1057  size_t MatchPos = Pat.match(MatchBuffer, CurrentMatchLen);
1058  if (i == 1)
1059  FirstMatchPos = LastPos + MatchPos;
1060 
1061  // report
1062  if (MatchPos == StringRef::npos) {
1063  PrintNoMatch(true, SM, *this, i, MatchBuffer, Req.VerboseVerbose, Diags);
1064  return StringRef::npos;
1065  }
1066  PrintMatch(true, SM, *this, i, MatchBuffer, MatchPos, CurrentMatchLen, Req,
1067  Diags);
1068 
1069  // move start point after the match
1070  LastMatchEnd += MatchPos + CurrentMatchLen;
1071  }
1072  // Full match len counts from first match pos.
1073  MatchLen = LastMatchEnd - FirstMatchPos;
1074 
1075  // Similar to the above, in "label-scan mode" we can't yet handle CHECK-NEXT
1076  // or CHECK-NOT
1077  if (!IsLabelScanMode) {
1078  size_t MatchPos = FirstMatchPos - LastPos;
1079  StringRef MatchBuffer = Buffer.substr(LastPos);
1080  StringRef SkippedRegion = Buffer.substr(LastPos, MatchPos);
1081 
1082  // If this check is a "CHECK-NEXT", verify that the previous match was on
1083  // the previous line (i.e. that there is one newline between them).
1084  if (CheckNext(SM, SkippedRegion)) {
1086  Pat.getCheckTy(), MatchBuffer, MatchPos, MatchLen,
1087  Diags, Req.Verbose);
1088  return StringRef::npos;
1089  }
1090 
1091  // If this check is a "CHECK-SAME", verify that the previous match was on
1092  // the same line (i.e. that there is no newline between them).
1093  if (CheckSame(SM, SkippedRegion)) {
1095  Pat.getCheckTy(), MatchBuffer, MatchPos, MatchLen,
1096  Diags, Req.Verbose);
1097  return StringRef::npos;
1098  }
1099 
1100  // If this match had "not strings", verify that they don't exist in the
1101  // skipped region.
1102  if (CheckNot(SM, SkippedRegion, NotStrings, Req, Diags))
1103  return StringRef::npos;
1104  }
1105 
1106  return FirstMatchPos;
1107 }
1108 
1109 /// Verify there is a single line in the given buffer.
1110 bool FileCheckString::CheckNext(const SourceMgr &SM, StringRef Buffer) const {
1111  if (Pat.getCheckTy() != Check::CheckNext &&
1112  Pat.getCheckTy() != Check::CheckEmpty)
1113  return false;
1114 
1115  Twine CheckName =
1116  Prefix +
1117  Twine(Pat.getCheckTy() == Check::CheckEmpty ? "-EMPTY" : "-NEXT");
1118 
1119  // Count the number of newlines between the previous match and this one.
1120  const char *FirstNewLine = nullptr;
1121  unsigned NumNewLines = CountNumNewlinesBetween(Buffer, FirstNewLine);
1122 
1123  if (NumNewLines == 0) {
1125  CheckName + ": is on the same line as previous match");
1127  "'next' match was here");
1129  "previous match ended here");
1130  return true;
1131  }
1132 
1133  if (NumNewLines != 1) {
1135  CheckName +
1136  ": is not on the line after the previous match");
1138  "'next' match was here");
1140  "previous match ended here");
1142  "non-matching line after previous match is here");
1143  return true;
1144  }
1145 
1146  return false;
1147 }
1148 
1149 /// Verify there is no newline in the given buffer.
1150 bool FileCheckString::CheckSame(const SourceMgr &SM, StringRef Buffer) const {
1151  if (Pat.getCheckTy() != Check::CheckSame)
1152  return false;
1153 
1154  // Count the number of newlines between the previous match and this one.
1155  const char *FirstNewLine = nullptr;
1156  unsigned NumNewLines = CountNumNewlinesBetween(Buffer, FirstNewLine);
1157 
1158  if (NumNewLines != 0) {
1160  Prefix +
1161  "-SAME: is not on the same line as the previous match");
1163  "'next' match was here");
1165  "previous match ended here");
1166  return true;
1167  }
1168 
1169  return false;
1170 }
1171 
1172 /// Verify there's no "not strings" in the given buffer.
1174  const SourceMgr &SM, StringRef Buffer,
1175  const std::vector<const FileCheckPattern *> &NotStrings,
1176  const FileCheckRequest &Req, std::vector<FileCheckDiag> *Diags) const {
1177  for (const FileCheckPattern *Pat : NotStrings) {
1178  assert((Pat->getCheckTy() == Check::CheckNot) && "Expect CHECK-NOT!");
1179 
1180  size_t MatchLen = 0;
1181  size_t Pos = Pat->match(Buffer, MatchLen);
1182 
1183  if (Pos == StringRef::npos) {
1184  PrintNoMatch(false, SM, Prefix, Pat->getLoc(), *Pat, 1, Buffer,
1185  Req.VerboseVerbose, Diags);
1186  continue;
1187  }
1188 
1189  PrintMatch(false, SM, Prefix, Pat->getLoc(), *Pat, 1, Buffer, Pos, MatchLen,
1190  Req, Diags);
1191 
1192  return true;
1193  }
1194 
1195  return false;
1196 }
1197 
1198 /// Match "dag strings" and their mixed "not strings".
1199 size_t
1201  std::vector<const FileCheckPattern *> &NotStrings,
1202  const FileCheckRequest &Req,
1203  std::vector<FileCheckDiag> *Diags) const {
1204  if (DagNotStrings.empty())
1205  return 0;
1206 
1207  // The start of the search range.
1208  size_t StartPos = 0;
1209 
1210  struct MatchRange {
1211  size_t Pos;
1212  size_t End;
1213  };
1214  // A sorted list of ranges for non-overlapping CHECK-DAG matches. Match
1215  // ranges are erased from this list once they are no longer in the search
1216  // range.
1217  std::list<MatchRange> MatchRanges;
1218 
1219  // We need PatItr and PatEnd later for detecting the end of a CHECK-DAG
1220  // group, so we don't use a range-based for loop here.
1221  for (auto PatItr = DagNotStrings.begin(), PatEnd = DagNotStrings.end();
1222  PatItr != PatEnd; ++PatItr) {
1223  const FileCheckPattern &Pat = *PatItr;
1224  assert((Pat.getCheckTy() == Check::CheckDAG ||
1225  Pat.getCheckTy() == Check::CheckNot) &&
1226  "Invalid CHECK-DAG or CHECK-NOT!");
1227 
1228  if (Pat.getCheckTy() == Check::CheckNot) {
1229  NotStrings.push_back(&Pat);
1230  continue;
1231  }
1232 
1233  assert((Pat.getCheckTy() == Check::CheckDAG) && "Expect CHECK-DAG!");
1234 
1235  // CHECK-DAG always matches from the start.
1236  size_t MatchLen = 0, MatchPos = StartPos;
1237 
1238  // Search for a match that doesn't overlap a previous match in this
1239  // CHECK-DAG group.
1240  for (auto MI = MatchRanges.begin(), ME = MatchRanges.end(); true; ++MI) {
1241  StringRef MatchBuffer = Buffer.substr(MatchPos);
1242  size_t MatchPosBuf = Pat.match(MatchBuffer, MatchLen);
1243  // With a group of CHECK-DAGs, a single mismatching means the match on
1244  // that group of CHECK-DAGs fails immediately.
1245  if (MatchPosBuf == StringRef::npos) {
1246  PrintNoMatch(true, SM, Prefix, Pat.getLoc(), Pat, 1, MatchBuffer,
1247  Req.VerboseVerbose, Diags);
1248  return StringRef::npos;
1249  }
1250  // Re-calc it as the offset relative to the start of the original string.
1251  MatchPos += MatchPosBuf;
1252  if (Req.VerboseVerbose)
1253  PrintMatch(true, SM, Prefix, Pat.getLoc(), Pat, 1, Buffer, MatchPos,
1254  MatchLen, Req, Diags);
1255  MatchRange M{MatchPos, MatchPos + MatchLen};
1256  if (Req.AllowDeprecatedDagOverlap) {
1257  // We don't need to track all matches in this mode, so we just maintain
1258  // one match range that encompasses the current CHECK-DAG group's
1259  // matches.
1260  if (MatchRanges.empty())
1261  MatchRanges.insert(MatchRanges.end(), M);
1262  else {
1263  auto Block = MatchRanges.begin();
1264  Block->Pos = std::min(Block->Pos, M.Pos);
1265  Block->End = std::max(Block->End, M.End);
1266  }
1267  break;
1268  }
1269  // Iterate previous matches until overlapping match or insertion point.
1270  bool Overlap = false;
1271  for (; MI != ME; ++MI) {
1272  if (M.Pos < MI->End) {
1273  // !Overlap => New match has no overlap and is before this old match.
1274  // Overlap => New match overlaps this old match.
1275  Overlap = MI->Pos < M.End;
1276  break;
1277  }
1278  }
1279  if (!Overlap) {
1280  // Insert non-overlapping match into list.
1281  MatchRanges.insert(MI, M);
1282  break;
1283  }
1284  if (Req.VerboseVerbose) {
1285  // Due to their verbosity, we don't print verbose diagnostics here if
1286  // we're gathering them for a different rendering, but we always print
1287  // other diagnostics.
1288  if (!Diags) {
1289  SMLoc OldStart = SMLoc::getFromPointer(Buffer.data() + MI->Pos);
1290  SMLoc OldEnd = SMLoc::getFromPointer(Buffer.data() + MI->End);
1291  SMRange OldRange(OldStart, OldEnd);
1292  SM.PrintMessage(OldStart, SourceMgr::DK_Note,
1293  "match discarded, overlaps earlier DAG match here",
1294  {OldRange});
1295  } else
1296  Diags->rbegin()->MatchTy = FileCheckDiag::MatchFoundButDiscarded;
1297  }
1298  MatchPos = MI->End;
1299  }
1300  if (!Req.VerboseVerbose)
1301  PrintMatch(true, SM, Prefix, Pat.getLoc(), Pat, 1, Buffer, MatchPos,
1302  MatchLen, Req, Diags);
1303 
1304  // Handle the end of a CHECK-DAG group.
1305  if (std::next(PatItr) == PatEnd ||
1306  std::next(PatItr)->getCheckTy() == Check::CheckNot) {
1307  if (!NotStrings.empty()) {
1308  // If there are CHECK-NOTs between two CHECK-DAGs or from CHECK to
1309  // CHECK-DAG, verify that there are no 'not' strings occurred in that
1310  // region.
1311  StringRef SkippedRegion =
1312  Buffer.slice(StartPos, MatchRanges.begin()->Pos);
1313  if (CheckNot(SM, SkippedRegion, NotStrings, Req, Diags))
1314  return StringRef::npos;
1315  // Clear "not strings".
1316  NotStrings.clear();
1317  }
1318  // All subsequent CHECK-DAGs and CHECK-NOTs should be matched from the
1319  // end of this CHECK-DAG group's match range.
1320  StartPos = MatchRanges.rbegin()->End;
1321  // Don't waste time checking for (impossible) overlaps before that.
1322  MatchRanges.clear();
1323  }
1324  }
1325 
1326  return StartPos;
1327 }
1328 
1329 // A check prefix must contain only alphanumeric, hyphens and underscores.
1330 static bool ValidateCheckPrefix(StringRef CheckPrefix) {
1331  Regex Validator("^[a-zA-Z0-9_-]*$");
1332  return Validator.match(CheckPrefix);
1333 }
1334 
1336  StringSet<> PrefixSet;
1337 
1338  for (StringRef Prefix : Req.CheckPrefixes) {
1339  // Reject empty prefixes.
1340  if (Prefix == "")
1341  return false;
1342 
1343  if (!PrefixSet.insert(Prefix).second)
1344  return false;
1345 
1346  if (!ValidateCheckPrefix(Prefix))
1347  return false;
1348  }
1349 
1350  return true;
1351 }
1352 
1353 // Combines the check prefixes into a single regex so that we can efficiently
1354 // scan for any of the set.
1355 //
1356 // The semantics are that the longest-match wins which matches our regex
1357 // library.
1359  // I don't think there's a way to specify an initial value for cl::list,
1360  // so if nothing was specified, add the default
1361  if (Req.CheckPrefixes.empty())
1362  Req.CheckPrefixes.push_back("CHECK");
1363 
1364  // We already validated the contents of CheckPrefixes so just concatenate
1365  // them as alternatives.
1366  SmallString<32> PrefixRegexStr;
1367  for (StringRef Prefix : Req.CheckPrefixes) {
1368  if (Prefix != Req.CheckPrefixes.front())
1369  PrefixRegexStr.push_back('|');
1370 
1371  PrefixRegexStr.append(Prefix);
1372  }
1373 
1374  return Regex(PrefixRegexStr);
1375 }
1376 
1378  std::vector<std::string> &CmdlineDefines) {
1379  assert(GlobalVariableTable.empty() &&
1380  "Overriding defined variable with command-line variable definitions");
1381  for (StringRef CmdlineDef : CmdlineDefines)
1382  GlobalVariableTable.insert(CmdlineDef.split('='));
1383 }
1384 
1386  SmallVector<StringRef, 16> LocalPatternVars, LocalNumericVars;
1387  for (const StringMapEntry<StringRef> &Var : GlobalVariableTable)
1388  if (Var.first()[0] != '$')
1389  LocalPatternVars.push_back(Var.first());
1390 
1391  for (const auto &Var : LocalPatternVars)
1392  GlobalVariableTable.erase(Var);
1393 }
1394 
1395 /// Check the input to FileCheck provided in the \p Buffer against the \p
1396 /// CheckStrings read from the check file.
1397 ///
1398 /// Returns false if the input fails to satisfy the checks.
1400  ArrayRef<FileCheckString> CheckStrings,
1401  std::vector<FileCheckDiag> *Diags) {
1402  bool ChecksFailed = false;
1403 
1404  unsigned i = 0, j = 0, e = CheckStrings.size();
1405  while (true) {
1406  StringRef CheckRegion;
1407  if (j == e) {
1408  CheckRegion = Buffer;
1409  } else {
1410  const FileCheckString &CheckLabelStr = CheckStrings[j];
1411  if (CheckLabelStr.Pat.getCheckTy() != Check::CheckLabel) {
1412  ++j;
1413  continue;
1414  }
1415 
1416  // Scan to next CHECK-LABEL match, ignoring CHECK-NOT and CHECK-DAG
1417  size_t MatchLabelLen = 0;
1418  size_t MatchLabelPos =
1419  CheckLabelStr.Check(SM, Buffer, true, MatchLabelLen, Req, Diags);
1420  if (MatchLabelPos == StringRef::npos)
1421  // Immediately bail if CHECK-LABEL fails, nothing else we can do.
1422  return false;
1423 
1424  CheckRegion = Buffer.substr(0, MatchLabelPos + MatchLabelLen);
1425  Buffer = Buffer.substr(MatchLabelPos + MatchLabelLen);
1426  ++j;
1427  }
1428 
1429  if (Req.EnableVarScope)
1430  PatternContext.clearLocalVars();
1431 
1432  for (; i != j; ++i) {
1433  const FileCheckString &CheckStr = CheckStrings[i];
1434 
1435  // Check each string within the scanned region, including a second check
1436  // of any final CHECK-LABEL (to verify CHECK-NOT and CHECK-DAG)
1437  size_t MatchLen = 0;
1438  size_t MatchPos =
1439  CheckStr.Check(SM, CheckRegion, false, MatchLen, Req, Diags);
1440 
1441  if (MatchPos == StringRef::npos) {
1442  ChecksFailed = true;
1443  i = j;
1444  break;
1445  }
1446 
1447  CheckRegion = CheckRegion.substr(MatchPos + MatchLen);
1448  }
1449 
1450  if (j == e)
1451  break;
1452  }
1453 
1454  // Success if no checks failed.
1455  return !ChecksFailed;
1456 }
llvm::Optional< StringRef > getVarValue(StringRef VarName)
Return the value of variable VarName or None if no such variable has been defined.
Definition: FileCheck.cpp:477
uint64_t CallInst * C
Represents a range in source code.
Definition: SMLoc.h:48
Compile for newline-sensitive matching.
Definition: Regex.h:38
Indicates a good match for an expected pattern.
Definition: FileCheck.h:207
bool ValidateCheckPrefixes()
Definition: FileCheck.cpp:1335
raw_ostream & errs()
This returns a reference to a raw_ostream for standard error.
Indicates no match for an excluded pattern.
Definition: FileCheck.h:216
GCNRegPressure max(const GCNRegPressure &P1, const GCNRegPressure &P2)
LLVM_NODISCARD std::string str() const
str - Get the contents as an std::string.
Definition: StringRef.h:218
This class represents lattice values for constants.
Definition: AllocatorList.h:23
StringMapEntry - This is used to represent one value that is inserted into a StringMap.
Definition: StringMap.h:125
Indicates a discarded match for an expected pattern.
Definition: FileCheck.h:214
amdgpu Simplify well known AMD library false FunctionCallee Value const Twine & Name
LLVM_NODISCARD bool startswith(StringRef Prefix) const
Check if this string starts with the given Prefix.
Definition: StringRef.h:256
void push_back(const T &Elt)
Definition: SmallVector.h:211
bool CheckSame(const SourceMgr &SM, StringRef Buffer) const
Verify there is no newline in the given buffer.
Definition: FileCheck.cpp:1150
static std::string escape(StringRef String)
Turn String into a regex by escaping its special characters.
Definition: Regex.cpp:200
A raw_ostream that writes to an SmallVector or SmallString.
Definition: raw_ostream.h:509
auto formatv(const char *Fmt, Ts &&... Vals) -> formatv_object< decltype(std::make_tuple(detail::build_format_adapter(std::forward< Ts >(Vals))...))>
std::string getDescription(StringRef Prefix) const
Definition: FileCheck.cpp:589
Marks when parsing found a -NOT check combined with another CHECK suffix.
Definition: FileCheck.h:61
size_t getBufferSize() const
Definition: MemoryBuffer.h:61
void reserve(size_type N)
Definition: SmallVector.h:369
std::pair< unsigned, unsigned > getLineAndColumn(SMLoc Loc, unsigned BufferID=0) const
Find the line and column number for the specified location in the specified file. ...
Definition: SourceMgr.cpp:130
LLVM_NODISCARD unsigned edit_distance(StringRef Other, bool AllowReplacements=true, unsigned MaxEditDistance=0) const
Determine the edit distance between this string and another string.
Definition: StringRef.cpp:94
size_t CheckDag(const SourceMgr &SM, StringRef Buffer, std::vector< const FileCheckPattern *> &NotStrings, const FileCheckRequest &Req, std::vector< FileCheckDiag > *Diags) const
Match "dag strings" and their mixed "not strings".
Definition: FileCheck.cpp:1200
bool CheckNext(const SourceMgr &SM, StringRef Buffer) const
Verify there is a single line in the given buffer.
Definition: FileCheck.cpp:1110
SMLoc Start
Definition: SMLoc.h:50
static LLVM_ATTRIBUTE_ALWAYS_INLINE bool CheckSame(const unsigned char *MatcherTable, unsigned &MatcherIndex, SDValue N, const SmallVectorImpl< std::pair< SDValue, SDNode *>> &RecordedNodes)
CheckSame - Implements OP_CheckSame.
StringRef Prefix
Which prefix name this check matched.
Definition: FileCheck.h:245
FileCheckPattern Pat
The pattern to match.
Definition: FileCheck.h:242
void PrintMessage(raw_ostream &OS, SMLoc Loc, DiagKind Kind, const Twine &Msg, ArrayRef< SMRange > Ranges=None, ArrayRef< SMFixIt > FixIts=None, bool ShowColors=true) const
Emit a message about the specified location with the specified string.
Definition: SourceMgr.cpp:247
bool ParsePattern(StringRef PatternStr, StringRef Prefix, SourceMgr &SM, unsigned LineNumber, const FileCheckRequest &Req)
Parses the given string into the Pattern.
Definition: FileCheck.cpp:33
Twine - A lightweight data structure for efficiently representing the concatenation of temporary valu...
Definition: Twine.h:80
LLVM_NODISCARD StringRef drop_front(size_t N=1) const
Return a StringRef equal to &#39;this&#39; but with the first N elements dropped.
Definition: StringRef.h:620
LLVM_NODISCARD size_t count(char C) const
Return the number of occurrences of C in the string.
Definition: StringRef.h:457
SMLoc Loc
The location in the match file that the check string was specified.
Definition: FileCheck.h:248
Check::FileCheckType CheckTy
What is the FileCheck directive for this diagnostic?
Definition: FileCheck.h:188
Indicates a match for an expected pattern, but the match is on the wrong line.
Definition: FileCheck.h:212
LLVM_NODISCARD StringRef slice(size_t Start, size_t End) const
Return a reference to the substring from [Start, End).
Definition: StringRef.h:679
LLVM_NODISCARD StringRef substr(size_t Start, size_t N=npos) const
Return a reference to the substring from [Start, Start + N).
Definition: StringRef.h:578
LLVM_NODISCARD bool empty() const
empty - Check if the string is empty.
Definition: StringRef.h:126
Marks when parsing found a -COUNT directive with invalid count value.
Definition: FileCheck.h:64
std::string itostr(int64_t X)
Definition: StringExtras.h:238
static std::pair< Check::FileCheckType, StringRef > FindCheckType(StringRef Buffer, StringRef Prefix)
Definition: FileCheck.cpp:620
static unsigned CountNumNewlinesBetween(StringRef Range, const char *&FirstNewLine)
Count the number of newlines in the specified range.
Definition: FileCheck.cpp:1006
unsigned AddNewSourceBuffer(std::unique_ptr< MemoryBuffer > F, SMLoc IncludeLoc)
Add a new source buffer to this source manager.
Definition: SourceMgr.h:151
StringRef CanonicalizeFile(MemoryBuffer &MB, SmallVectorImpl< char > &OutputBuffer)
Canonicalize whitespaces in the file.
Definition: FileCheck.cpp:531
size_t match(StringRef Buffer, size_t &MatchLen) const
Matches the pattern string against the input buffer Buffer.
Definition: FileCheck.cpp:275
ArrayRef - Represent a constant reference to an array (0 or more elements consecutively in memory)...
Definition: APInt.h:32
Contains info about various FileCheck options.
Definition: FileCheck.h:26
static bool ValidateCheckPrefix(StringRef CheckPrefix)
Definition: FileCheck.cpp:1330
LLVM_NODISCARD size_t size() const
size - Get the string size.
Definition: StringRef.h:130
size_t Check(const SourceMgr &SM, StringRef Buffer, bool IsLabelScanMode, size_t &MatchLen, FileCheckRequest &Req, std::vector< FileCheckDiag > *Diags) const
Match check string and its "not strings" and/or "dag strings".
Definition: FileCheck.cpp:1029
void append(in_iter S, in_iter E)
Append from an iterator pair.
Definition: SmallString.h:74
unsigned InputEndLine
Definition: FileCheck.h:229
#define P(N)
A check that we found in the input file.
Definition: FileCheck.h:240
void clearLocalVars()
Undefine local variables (variables whose name does not start with a &#39;$&#39; sign), i.e.
Definition: FileCheck.cpp:1385
FileCheckDiag(const SourceMgr &SM, const Check::FileCheckType &CheckTy, SMLoc CheckLoc, MatchType MatchTy, SMRange InputRange)
Definition: FileCheck.cpp:560
The instances of the Type class are immutable: once they are created, they are never changed...
Definition: Type.h:45
bool ReadCheckFile(SourceMgr &SM, StringRef Buffer, Regex &PrefixRE, std::vector< FileCheckString > &CheckStrings)
Read the check file, which specifies the sequence of expected strings.
Definition: FileCheck.cpp:755
size_t size() const
size - Get the array size.
Definition: ArrayRef.h:148
static GCRegistry::Add< CoreCLRGC > E("coreclr", "CoreCLR-compatible GC")
LLVM_NODISCARD size_t find_first_not_of(char C, size_t From=0) const
Find the first character in the string that is not C or npos if not found.
Definition: StringRef.cpp:249
bool CheckInput(SourceMgr &SM, StringRef Buffer, ArrayRef< FileCheckString > CheckStrings, std::vector< FileCheckDiag > *Diags=nullptr)
Check the input to FileCheck provided in the Buffer against the CheckStrings read from the check file...
Definition: FileCheck.cpp:1399
LLVM_NODISCARD size_t find(char C, size_t From=0) const
Search for the first character C in the string.
Definition: StringRef.h:285
unsigned CheckLine
Where is the FileCheck directive for this diagnostic?
Definition: FileCheck.h:190
SMLoc getLoc() const
Returns the location in source code.
Definition: FileCheck.h:151
This owns the files read by a parser, handles include stacks, and handles diagnostic wrangling...
Definition: SourceMgr.h:41
void printVariableUses(const SourceMgr &SM, StringRef Buffer, SMRange MatchRange=None) const
Print value of successful substitutions or name of undefined pattern variables preventing such a succ...
Definition: FileCheck.cpp:368
Check::FileCheckType getCheckTy() const
Definition: FileCheck.h:170
size_t size() const
Definition: SmallVector.h:52
LLVM_NODISCARD char back() const
back - Get the last character in the string.
Definition: StringRef.h:141
#define llvm_unreachable(msg)
Marks that the current location is not supposed to be reachable.
std::pair< typename base::iterator, bool > insert(StringRef Key)
Definition: StringSet.h:38
std::enable_if< std::numeric_limits< T >::is_signed, bool >::type getAsInteger(unsigned Radix, T &Result) const
Parse the current string as an integer of the specified radix.
Definition: StringRef.h:478
Indicates the pattern only matches the end of file.
Definition: FileCheck.h:58
bool isValid() const
Definition: SMLoc.h:59
static size_t SkipWord(StringRef Str, size_t Loc)
Definition: FileCheck.cpp:675
raw_ostream & write_escaped(StringRef Str, bool UseHexEscapes=false)
Output Str, turning &#39;\&#39;, &#39;&#39;, &#39; &#39;, &#39;"&#39;, and anything that doesn&#39;t satisfy llvm::isPrint into an escape...
Indicates a fuzzy match that serves as a suggestion for the next intended match for an expected patte...
Definition: FileCheck.h:223
auto size(R &&Range, typename std::enable_if< std::is_same< typename std::iterator_traits< decltype(Range.begin())>::iterator_category, std::random_access_iterator_tag >::value, void >::type *=nullptr) -> decltype(std::distance(Range.begin(), Range.end()))
Get the size of a range.
Definition: STLExtras.h:1166
static std::pair< StringRef, StringRef > FindFirstMatchingPrefix(Regex &PrefixRE, StringRef &Buffer, unsigned &LineNumber, Check::FileCheckType &CheckTy)
Search the buffer for the first prefix in the prefix regular expression.
Definition: FileCheck.cpp:705
This is a &#39;vector&#39; (really, a variable-sized array), optimized for the case when the array is small...
Definition: SmallVector.h:841
void defineCmdlineVariables(std::vector< std::string > &CmdlineDefines)
Define variables from definitions given on the command line passed as a vector of VAR=VAL strings in ...
Definition: FileCheck.cpp:1377
Indicates no match for an expected pattern, but this might follow good matches when multiple matches ...
Definition: FileCheck.h:220
LLVM_NODISCARD std::pair< StringRef, StringRef > split(char Separator) const
Split into two substrings around the first occurrence of a separator character.
Definition: StringRef.h:696
bool isValid(std::string &Error) const
isValid - returns the error encountered during regex compilation, or matching, if any...
Definition: Regex.cpp:55
StringRef str()
Return a StringRef for the vector contents.
Definition: raw_ostream.h:534
unsigned InputEndCol
Definition: FileCheck.h:230
This interface provides simple read-only access to a block of memory, and provides simple methods for...
Definition: MemoryBuffer.h:41
static void PrintMatch(bool ExpectedMatch, const SourceMgr &SM, StringRef Prefix, SMLoc Loc, const FileCheckPattern &Pat, int MatchedCount, StringRef Buffer, size_t MatchPos, size_t MatchLen, const FileCheckRequest &Req, std::vector< FileCheckDiag > *Diags)
Definition: FileCheck.cpp:906
void swap(llvm::BitVector &LHS, llvm::BitVector &RHS)
Implement std::swap in terms of BitVector swap.
Definition: BitVector.h:940
static std::unique_ptr< MemoryBuffer > getMemBufferCopy(StringRef InputData, const Twine &BufferName="")
Open the specified memory range as a MemoryBuffer, copying the contents and taking ownership of it...
static SMRange ProcessMatchResult(FileCheckDiag::MatchType MatchTy, const SourceMgr &SM, SMLoc Loc, Check::FileCheckType CheckTy, StringRef Buffer, size_t Pos, size_t Len, std::vector< FileCheckDiag > *Diags, bool AdjustPrevDiag=false)
Definition: FileCheck.cpp:411
bool CheckNot(const SourceMgr &SM, StringRef Buffer, const std::vector< const FileCheckPattern *> &NotStrings, const FileCheckRequest &Req, std::vector< FileCheckDiag > *Diags) const
Verify there&#39;s no "not strings" in the given buffer.
Definition: FileCheck.cpp:1173
SMLoc End
Definition: SMLoc.h:50
void printFuzzyMatch(const SourceMgr &SM, StringRef Buffer, std::vector< FileCheckDiag > *Diags) const
Definition: FileCheck.cpp:429
unsigned getNumMatches() const
getNumMatches - In a valid regex, return the number of parenthesized matches it contains.
Definition: Regex.cpp:68
iterator begin() const
Definition: StringRef.h:101
static SMLoc getFromPointer(const char *Ptr)
Definition: SMLoc.h:36
unsigned InputStartLine
The search range if MatchTy is MatchNoneAndExcluded or MatchNoneButExpected, or the match range other...
Definition: FileCheck.h:227
Indicates a match for an excluded pattern.
Definition: FileCheck.h:209
const char * getBufferEnd() const
Definition: MemoryBuffer.h:60
FileCheckType & setCount(int C)
Definition: FileCheck.cpp:580
pointer data()
Return a pointer to the vector&#39;s buffer, even if empty().
Definition: SmallVector.h:144
static const size_t npos
Definition: StringRef.h:50
LLVM_NODISCARD bool empty() const
Definition: SmallVector.h:55
Regex buildCheckPrefixRegex()
Definition: FileCheck.cpp:1358
LLVM_NODISCARD size_t find_first_of(char C, size_t From=0) const
Find the first character in the string that is C, or npos if not found.
Definition: StringRef.h:380
#define I(x, y, z)
Definition: MD5.cpp:58
const char * getBufferStart() const
Definition: MemoryBuffer.h:59
static bool IsPartOfWord(char c)
Definition: FileCheck.cpp:576
LLVM_NODISCARD const char * data() const
data - Get a pointer to the start of the string (which may not be null terminated).
Definition: StringRef.h:122
unsigned InputStartCol
Definition: FileCheck.h:228
assert(ImpDefSCC.getReg()==AMDGPU::SCC &&ImpDefSCC.isDef())
LLVM_NODISCARD char front() const
front - Get the first character in the string.
Definition: StringRef.h:134
LLVM Value Representation.
Definition: Value.h:72
static void PrintNoMatch(bool ExpectedMatch, const SourceMgr &SM, StringRef Prefix, SMLoc Loc, const FileCheckPattern &Pat, int MatchedCount, StringRef Buffer, bool VerboseVerbose, std::vector< FileCheckDiag > *Diags)
Definition: FileCheck.cpp:952
StringSet - A wrapper for StringMap that provides set-like functionality.
Definition: StringSet.h:27
bool match(StringRef String, SmallVectorImpl< StringRef > *Matches=nullptr)
matches - Match the regex against a given String.
Definition: Regex.cpp:72
IRTranslator LLVM IR MI
StringRef - Represent a constant reference to a string, i.e.
Definition: StringRef.h:48
MatchType
What type of match result does this diagnostic describe?
Definition: FileCheck.h:205
int getCount() const
Definition: FileCheck.h:172
Represents a location in source code.
Definition: SMLoc.h:23
iterator end() const
Definition: StringRef.h:103