1 //===- FileCheck.cpp - Check that File's Contents match what is expected --===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
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.
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.
17 //===----------------------------------------------------------------------===//
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/StringExtras.h"
30 #include "llvm/ADT/StringMap.h"
37 static cl::opt<std::string>
38 CheckFilename(cl::Positional, cl::desc("<check-file>"), cl::Required);
40 static cl::opt<std::string>
41 InputFilename("input-file", cl::desc("File to check (defaults to stdin)"),
42 cl::init("-"), cl::value_desc("filename"));
44 static cl::opt<std::string>
45 CheckPrefix("check-prefix", cl::init("CHECK"),
46 cl::desc("Prefix to use from check file (defaults to 'CHECK')"));
49 NoCanonicalizeWhiteSpace("strict-whitespace",
50 cl::desc("Do not treat all horizontal whitespace as equivalent"));
52 //===----------------------------------------------------------------------===//
53 // Pattern Handling Code.
54 //===----------------------------------------------------------------------===//
59 /// MatchEOF - When set, this pattern only matches the end of file. This is
60 /// used for trailing CHECK-NOTs.
63 /// FixedStr - If non-empty, this pattern is a fixed string match with the
64 /// specified fixed string.
67 /// RegEx - If non-empty, this is a regex pattern.
70 /// \brief Contains the number of line this pattern is in.
73 /// VariableUses - Entries in this vector map to uses of a variable in the
74 /// pattern, e.g. "foo[[bar]]baz". In this case, the RegExStr will contain
75 /// "foobaz" and we'll get an entry in this vector that tells us to insert the
76 /// value of bar at offset 3.
77 std::vector<std::pair<StringRef, unsigned> > VariableUses;
79 /// VariableDefs - Maps definitions of variables to their parenthesized
81 /// E.g. for the pattern "foo[[bar:.*]]baz", VariableDefs will map "bar" to 1.
82 std::map<StringRef, unsigned> VariableDefs;
86 Pattern(bool matchEOF = false) : MatchEOF(matchEOF) { }
88 /// ParsePattern - Parse the given string into the Pattern. SM provides the
89 /// SourceMgr used for error reports, and LineNumber is the line number in
90 /// the input file from which the pattern string was read.
91 /// Returns true in case of an error, false otherwise.
92 bool ParsePattern(StringRef PatternStr, SourceMgr &SM, unsigned LineNumber);
94 /// Match - Match the pattern string against the input buffer Buffer. This
95 /// returns the position that is matched or npos if there is no match. If
96 /// there is a match, the size of the matched string is returned in MatchLen.
98 /// The VariableTable StringMap provides the current values of filecheck
99 /// variables and is updated if this match defines new values.
100 size_t Match(StringRef Buffer, size_t &MatchLen,
101 StringMap<StringRef> &VariableTable) const;
103 /// PrintFailureInfo - Print additional information about a failure to match
104 /// involving this pattern.
105 void PrintFailureInfo(const SourceMgr &SM, StringRef Buffer,
106 const StringMap<StringRef> &VariableTable) const;
109 static void AddFixedStringToRegEx(StringRef FixedStr, std::string &TheStr);
110 bool AddRegExToRegEx(StringRef RS, unsigned &CurParen, SourceMgr &SM);
111 void AddBackrefToRegEx(unsigned BackrefNum);
113 /// ComputeMatchDistance - Compute an arbitrary estimate for the quality of
114 /// matching this pattern at the start of \arg Buffer; a distance of zero
115 /// should correspond to a perfect match.
116 unsigned ComputeMatchDistance(StringRef Buffer,
117 const StringMap<StringRef> &VariableTable) const;
119 /// \brief Evaluates expression and stores the result to \p Value.
120 /// \return true on success. false when the expression has invalid syntax.
121 bool EvaluateExpression(StringRef Expr, std::string &Value) const;
123 /// \brief Finds the closing sequence of a regex variable usage or
124 /// definition. Str has to point in the beginning of the definition
125 /// (right after the opening sequence).
126 /// \return offset of the closing sequence within Str, or npos if it was not
128 size_t FindRegexVarEnd(StringRef Str);
132 bool Pattern::ParsePattern(StringRef PatternStr, SourceMgr &SM,
133 unsigned LineNumber) {
134 this->LineNumber = LineNumber;
135 PatternLoc = SMLoc::getFromPointer(PatternStr.data());
137 // Ignore trailing whitespace.
138 while (!PatternStr.empty() &&
139 (PatternStr.back() == ' ' || PatternStr.back() == '\t'))
140 PatternStr = PatternStr.substr(0, PatternStr.size()-1);
142 // Check that there is something on the line.
143 if (PatternStr.empty()) {
144 SM.PrintMessage(PatternLoc, SourceMgr::DK_Error,
145 "found empty check string with prefix '" +
150 // Check to see if this is a fixed string, or if it has regex pieces.
151 if (PatternStr.size() < 2 ||
152 (PatternStr.find("{{") == StringRef::npos &&
153 PatternStr.find("[[") == StringRef::npos)) {
154 FixedStr = PatternStr;
158 // Paren value #0 is for the fully matched string. Any new parenthesized
159 // values add from there.
160 unsigned CurParen = 1;
162 // Otherwise, there is at least one regex piece. Build up the regex pattern
163 // by escaping scary characters in fixed strings, building up one big regex.
164 while (!PatternStr.empty()) {
166 if (PatternStr.startswith("{{")) {
167 // This is the start of a regex match. Scan for the }}.
168 size_t End = PatternStr.find("}}");
169 if (End == StringRef::npos) {
170 SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()),
172 "found start of regex string with no end '}}'");
176 // Enclose {{}} patterns in parens just like [[]] even though we're not
177 // capturing the result for any purpose. This is required in case the
178 // expression contains an alternation like: CHECK: abc{{x|z}}def. We
179 // want this to turn into: "abc(x|z)def" not "abcx|zdef".
183 if (AddRegExToRegEx(PatternStr.substr(2, End-2), CurParen, SM))
187 PatternStr = PatternStr.substr(End+2);
191 // Named RegEx matches. These are of two forms: [[foo:.*]] which matches .*
192 // (or some other regex) and assigns it to the FileCheck variable 'foo'. The
193 // second form is [[foo]] which is a reference to foo. The variable name
194 // itself must be of the form "[a-zA-Z_][0-9a-zA-Z_]*", otherwise we reject
195 // it. This is to catch some common errors.
196 if (PatternStr.startswith("[[")) {
197 // Find the closing bracket pair ending the match. End is going to be an
198 // offset relative to the beginning of the match string.
199 size_t End = FindRegexVarEnd(PatternStr.substr(2));
201 if (End == StringRef::npos) {
202 SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()),
204 "invalid named regex reference, no ]] found");
208 StringRef MatchStr = PatternStr.substr(2, End);
209 PatternStr = PatternStr.substr(End+4);
211 // Get the regex name (e.g. "foo").
212 size_t NameEnd = MatchStr.find(':');
213 StringRef Name = MatchStr.substr(0, NameEnd);
216 SM.PrintMessage(SMLoc::getFromPointer(Name.data()), SourceMgr::DK_Error,
217 "invalid name in named regex: empty name");
221 // Verify that the name/expression is well formed. FileCheck currently
222 // supports @LINE, @LINE+number, @LINE-number expressions. The check here
223 // is relaxed, more strict check is performed in \c EvaluateExpression.
224 bool IsExpression = false;
225 for (unsigned i = 0, e = Name.size(); i != e; ++i) {
226 if (i == 0 && Name[i] == '@') {
227 if (NameEnd != StringRef::npos) {
228 SM.PrintMessage(SMLoc::getFromPointer(Name.data()),
230 "invalid name in named regex definition");
236 if (Name[i] != '_' && !isalnum(Name[i]) &&
237 (!IsExpression || (Name[i] != '+' && Name[i] != '-'))) {
238 SM.PrintMessage(SMLoc::getFromPointer(Name.data()+i),
239 SourceMgr::DK_Error, "invalid name in named regex");
244 // Name can't start with a digit.
245 if (isdigit(Name[0])) {
246 SM.PrintMessage(SMLoc::getFromPointer(Name.data()), SourceMgr::DK_Error,
247 "invalid name in named regex");
252 if (NameEnd == StringRef::npos) {
253 // Handle variables that were defined earlier on the same line by
254 // emitting a backreference.
255 if (VariableDefs.find(Name) != VariableDefs.end()) {
256 unsigned VarParenNum = VariableDefs[Name];
257 if (VarParenNum < 1 || VarParenNum > 9) {
258 SM.PrintMessage(SMLoc::getFromPointer(Name.data()),
260 "Can't back-reference more than 9 variables");
263 AddBackrefToRegEx(VarParenNum);
265 VariableUses.push_back(std::make_pair(Name, RegExStr.size()));
270 // Handle [[foo:.*]].
271 VariableDefs[Name] = CurParen;
275 if (AddRegExToRegEx(MatchStr.substr(NameEnd+1), CurParen, SM))
281 // Handle fixed string matches.
282 // Find the end, which is the start of the next regex.
283 size_t FixedMatchEnd = PatternStr.find("{{");
284 FixedMatchEnd = std::min(FixedMatchEnd, PatternStr.find("[["));
285 AddFixedStringToRegEx(PatternStr.substr(0, FixedMatchEnd), RegExStr);
286 PatternStr = PatternStr.substr(FixedMatchEnd);
292 void Pattern::AddFixedStringToRegEx(StringRef FixedStr, std::string &TheStr) {
293 // Add the characters from FixedStr to the regex, escaping as needed. This
294 // avoids "leaning toothpicks" in common patterns.
295 for (unsigned i = 0, e = FixedStr.size(); i != e; ++i) {
296 switch (FixedStr[i]) {
297 // These are the special characters matched in "p_ere_exp".
313 TheStr += FixedStr[i];
319 bool Pattern::AddRegExToRegEx(StringRef RS, unsigned &CurParen,
323 if (!R.isValid(Error)) {
324 SM.PrintMessage(SMLoc::getFromPointer(RS.data()), SourceMgr::DK_Error,
325 "invalid regex: " + Error);
329 RegExStr += RS.str();
330 CurParen += R.getNumMatches();
334 void Pattern::AddBackrefToRegEx(unsigned BackrefNum) {
335 assert(BackrefNum >= 1 && BackrefNum <= 9 && "Invalid backref number");
336 std::string Backref = std::string("\\") +
337 std::string(1, '0' + BackrefNum);
341 bool Pattern::EvaluateExpression(StringRef Expr, std::string &Value) const {
342 // The only supported expression is @LINE([\+-]\d+)?
343 if (!Expr.startswith("@LINE"))
345 Expr = Expr.substr(StringRef("@LINE").size());
349 Expr = Expr.substr(1);
350 else if (Expr[0] != '-')
352 if (Expr.getAsInteger(10, Offset))
355 Value = llvm::itostr(LineNumber + Offset);
359 /// Match - Match the pattern string against the input buffer Buffer. This
360 /// returns the position that is matched or npos if there is no match. If
361 /// there is a match, the size of the matched string is returned in MatchLen.
362 size_t Pattern::Match(StringRef Buffer, size_t &MatchLen,
363 StringMap<StringRef> &VariableTable) const {
364 // If this is the EOF pattern, match it immediately.
367 return Buffer.size();
370 // If this is a fixed string pattern, just match it now.
371 if (!FixedStr.empty()) {
372 MatchLen = FixedStr.size();
373 return Buffer.find(FixedStr);
378 // If there are variable uses, we need to create a temporary string with the
380 StringRef RegExToMatch = RegExStr;
382 if (!VariableUses.empty()) {
385 unsigned InsertOffset = 0;
386 for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) {
389 if (VariableUses[i].first[0] == '@') {
390 if (!EvaluateExpression(VariableUses[i].first, Value))
391 return StringRef::npos;
393 StringMap<StringRef>::iterator it =
394 VariableTable.find(VariableUses[i].first);
395 // If the variable is undefined, return an error.
396 if (it == VariableTable.end())
397 return StringRef::npos;
399 // Look up the value and escape it so that we can plop it into the regex.
400 AddFixedStringToRegEx(it->second, Value);
403 // Plop it into the regex at the adjusted offset.
404 TmpStr.insert(TmpStr.begin()+VariableUses[i].second+InsertOffset,
405 Value.begin(), Value.end());
406 InsertOffset += Value.size();
409 // Match the newly constructed regex.
410 RegExToMatch = TmpStr;
414 SmallVector<StringRef, 4> MatchInfo;
415 if (!Regex(RegExToMatch, Regex::Newline).match(Buffer, &MatchInfo))
416 return StringRef::npos;
418 // Successful regex match.
419 assert(!MatchInfo.empty() && "Didn't get any match");
420 StringRef FullMatch = MatchInfo[0];
422 // If this defines any variables, remember their values.
423 for (std::map<StringRef, unsigned>::const_iterator I = VariableDefs.begin(),
424 E = VariableDefs.end();
426 assert(I->second < MatchInfo.size() && "Internal paren error");
427 VariableTable[I->first] = MatchInfo[I->second];
430 MatchLen = FullMatch.size();
431 return FullMatch.data()-Buffer.data();
434 unsigned Pattern::ComputeMatchDistance(StringRef Buffer,
435 const StringMap<StringRef> &VariableTable) const {
436 // Just compute the number of matching characters. For regular expressions, we
437 // just compare against the regex itself and hope for the best.
439 // FIXME: One easy improvement here is have the regex lib generate a single
440 // example regular expression which matches, and use that as the example
442 StringRef ExampleString(FixedStr);
443 if (ExampleString.empty())
444 ExampleString = RegExStr;
446 // Only compare up to the first line in the buffer, or the string size.
447 StringRef BufferPrefix = Buffer.substr(0, ExampleString.size());
448 BufferPrefix = BufferPrefix.split('\n').first;
449 return BufferPrefix.edit_distance(ExampleString);
452 void Pattern::PrintFailureInfo(const SourceMgr &SM, StringRef Buffer,
453 const StringMap<StringRef> &VariableTable) const{
454 // If this was a regular expression using variables, print the current
456 if (!VariableUses.empty()) {
457 for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) {
458 SmallString<256> Msg;
459 raw_svector_ostream OS(Msg);
460 StringRef Var = VariableUses[i].first;
463 if (EvaluateExpression(Var, Value)) {
464 OS << "with expression \"";
465 OS.write_escaped(Var) << "\" equal to \"";
466 OS.write_escaped(Value) << "\"";
468 OS << "uses incorrect expression \"";
469 OS.write_escaped(Var) << "\"";
472 StringMap<StringRef>::const_iterator it = VariableTable.find(Var);
474 // Check for undefined variable references.
475 if (it == VariableTable.end()) {
476 OS << "uses undefined variable \"";
477 OS.write_escaped(Var) << "\"";
479 OS << "with variable \"";
480 OS.write_escaped(Var) << "\" equal to \"";
481 OS.write_escaped(it->second) << "\"";
485 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), SourceMgr::DK_Note,
490 // Attempt to find the closest/best fuzzy match. Usually an error happens
491 // because some string in the output didn't exactly match. In these cases, we
492 // would like to show the user a best guess at what "should have" matched, to
493 // save them having to actually check the input manually.
494 size_t NumLinesForward = 0;
495 size_t Best = StringRef::npos;
496 double BestQuality = 0;
498 // Use an arbitrary 4k limit on how far we will search.
499 for (size_t i = 0, e = std::min(size_t(4096), Buffer.size()); i != e; ++i) {
500 if (Buffer[i] == '\n')
503 // Patterns have leading whitespace stripped, so skip whitespace when
504 // looking for something which looks like a pattern.
505 if (Buffer[i] == ' ' || Buffer[i] == '\t')
508 // Compute the "quality" of this match as an arbitrary combination of the
509 // match distance and the number of lines skipped to get to this match.
510 unsigned Distance = ComputeMatchDistance(Buffer.substr(i), VariableTable);
511 double Quality = Distance + (NumLinesForward / 100.);
513 if (Quality < BestQuality || Best == StringRef::npos) {
515 BestQuality = Quality;
519 // Print the "possible intended match here" line if we found something
520 // reasonable and not equal to what we showed in the "scanning from here"
522 if (Best && Best != StringRef::npos && BestQuality < 50) {
523 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data() + Best),
524 SourceMgr::DK_Note, "possible intended match here");
526 // FIXME: If we wanted to be really friendly we would show why the match
527 // failed, as it can be hard to spot simple one character differences.
531 size_t Pattern::FindRegexVarEnd(StringRef Str) {
532 // Offset keeps track of the current offset within the input Str
534 // [...] Nesting depth
535 size_t BracketDepth = 0;
537 while (!Str.empty()) {
538 if (Str.startswith("]]") && BracketDepth == 0)
540 if (Str[0] == '\\') {
541 // Backslash escapes the next char within regexes, so skip them both.
552 assert(BracketDepth > 0 && "Invalid regex");
561 return StringRef::npos;
565 //===----------------------------------------------------------------------===//
567 //===----------------------------------------------------------------------===//
569 /// CheckString - This is a check that we found in the input file.
571 /// Pat - The pattern to match.
574 /// Loc - The location in the match file that the check string was specified.
577 /// IsCheckNext - This is true if this is a CHECK-NEXT: directive (as opposed
578 /// to a CHECK: directive.
581 /// NotStrings - These are all of the strings that are disallowed from
582 /// occurring between this match string and the previous one (or start of
584 std::vector<std::pair<SMLoc, Pattern> > NotStrings;
586 CheckString(const Pattern &P, SMLoc L, bool isCheckNext)
587 : Pat(P), Loc(L), IsCheckNext(isCheckNext) {}
590 /// CanonicalizeInputFile - Remove duplicate horizontal space from the specified
591 /// memory buffer, free it, and return a new one.
592 static MemoryBuffer *CanonicalizeInputFile(MemoryBuffer *MB) {
593 SmallString<128> NewFile;
594 NewFile.reserve(MB->getBufferSize());
596 for (const char *Ptr = MB->getBufferStart(), *End = MB->getBufferEnd();
598 // Eliminate trailing dosish \r.
599 if (Ptr <= End - 2 && Ptr[0] == '\r' && Ptr[1] == '\n') {
603 // If current char is not a horizontal whitespace, dump it to output as is.
604 if (*Ptr != ' ' && *Ptr != '\t') {
605 NewFile.push_back(*Ptr);
609 // Otherwise, add one space and advance over neighboring space.
610 NewFile.push_back(' ');
611 while (Ptr+1 != End &&
612 (Ptr[1] == ' ' || Ptr[1] == '\t'))
616 // Free the old buffer and return a new one.
618 MemoryBuffer::getMemBufferCopy(NewFile.str(), MB->getBufferIdentifier());
625 /// ReadCheckFile - Read the check file, which specifies the sequence of
626 /// expected strings. The strings are added to the CheckStrings vector.
627 /// Returns true in case of an error, false otherwise.
628 static bool ReadCheckFile(SourceMgr &SM,
629 std::vector<CheckString> &CheckStrings) {
630 OwningPtr<MemoryBuffer> File;
632 MemoryBuffer::getFileOrSTDIN(CheckFilename.c_str(), File)) {
633 errs() << "Could not open check file '" << CheckFilename << "': "
634 << ec.message() << '\n';
637 MemoryBuffer *F = File.take();
639 // If we want to canonicalize whitespace, strip excess whitespace from the
640 // buffer containing the CHECK lines.
641 if (!NoCanonicalizeWhiteSpace)
642 F = CanonicalizeInputFile(F);
644 SM.AddNewSourceBuffer(F, SMLoc());
646 // Find all instances of CheckPrefix followed by : in the file.
647 StringRef Buffer = F->getBuffer();
648 std::vector<std::pair<SMLoc, Pattern> > NotMatches;
650 // LineNumber keeps track of the line on which CheckPrefix instances are
652 unsigned LineNumber = 1;
655 // See if Prefix occurs in the memory buffer.
656 size_t PrefixLoc = Buffer.find(CheckPrefix);
657 // If we didn't find a match, we're done.
658 if (PrefixLoc == StringRef::npos)
661 LineNumber += Buffer.substr(0, PrefixLoc).count('\n');
663 Buffer = Buffer.substr(PrefixLoc);
665 const char *CheckPrefixStart = Buffer.data();
667 // When we find a check prefix, keep track of whether we find CHECK: or
669 bool IsCheckNext = false, IsCheckNot = false;
671 // Verify that the : is present after the prefix.
672 if (Buffer[CheckPrefix.size()] == ':') {
673 Buffer = Buffer.substr(CheckPrefix.size()+1);
674 } else if (Buffer.size() > CheckPrefix.size()+6 &&
675 memcmp(Buffer.data()+CheckPrefix.size(), "-NEXT:", 6) == 0) {
676 Buffer = Buffer.substr(CheckPrefix.size()+6);
678 } else if (Buffer.size() > CheckPrefix.size()+5 &&
679 memcmp(Buffer.data()+CheckPrefix.size(), "-NOT:", 5) == 0) {
680 Buffer = Buffer.substr(CheckPrefix.size()+5);
683 Buffer = Buffer.substr(1);
687 // Okay, we found the prefix, yay. Remember the rest of the line, but
688 // ignore leading and trailing whitespace.
689 Buffer = Buffer.substr(Buffer.find_first_not_of(" \t"));
691 // Scan ahead to the end of line.
692 size_t EOL = Buffer.find_first_of("\n\r");
694 // Remember the location of the start of the pattern, for diagnostics.
695 SMLoc PatternLoc = SMLoc::getFromPointer(Buffer.data());
697 // Parse the pattern.
699 if (P.ParsePattern(Buffer.substr(0, EOL), SM, LineNumber))
702 Buffer = Buffer.substr(EOL);
704 // Verify that CHECK-NEXT lines have at least one CHECK line before them.
705 if (IsCheckNext && CheckStrings.empty()) {
706 SM.PrintMessage(SMLoc::getFromPointer(CheckPrefixStart),
708 "found '"+CheckPrefix+"-NEXT:' without previous '"+
709 CheckPrefix+ ": line");
715 NotMatches.push_back(std::make_pair(SMLoc::getFromPointer(Buffer.data()),
720 // Okay, add the string we captured to the output vector and move on.
721 CheckStrings.push_back(CheckString(P,
724 std::swap(NotMatches, CheckStrings.back().NotStrings);
727 // Add an EOF pattern for any trailing CHECK-NOTs.
728 if (!NotMatches.empty()) {
729 CheckStrings.push_back(CheckString(Pattern(true),
730 SMLoc::getFromPointer(Buffer.data()),
732 std::swap(NotMatches, CheckStrings.back().NotStrings);
735 if (CheckStrings.empty()) {
736 errs() << "error: no check strings found with prefix '" << CheckPrefix
744 static void PrintCheckFailed(const SourceMgr &SM, const CheckString &CheckStr,
746 StringMap<StringRef> &VariableTable) {
747 // Otherwise, we have an error, emit an error message.
748 SM.PrintMessage(CheckStr.Loc, SourceMgr::DK_Error,
749 "expected string not found in input");
751 // Print the "scanning from here" line. If the current position is at the
752 // end of a line, advance to the start of the next line.
753 Buffer = Buffer.substr(Buffer.find_first_not_of(" \t\n\r"));
755 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), SourceMgr::DK_Note,
756 "scanning from here");
758 // Allow the pattern to print additional information if desired.
759 CheckStr.Pat.PrintFailureInfo(SM, Buffer, VariableTable);
762 /// CountNumNewlinesBetween - Count the number of newlines in the specified
764 static unsigned CountNumNewlinesBetween(StringRef Range) {
765 unsigned NumNewLines = 0;
768 Range = Range.substr(Range.find_first_of("\n\r"));
769 if (Range.empty()) return NumNewLines;
773 // Handle \n\r and \r\n as a single newline.
774 if (Range.size() > 1 &&
775 (Range[1] == '\n' || Range[1] == '\r') &&
776 (Range[0] != Range[1]))
777 Range = Range.substr(1);
778 Range = Range.substr(1);
782 int main(int argc, char **argv) {
783 sys::PrintStackTraceOnErrorSignal();
784 PrettyStackTraceProgram X(argc, argv);
785 cl::ParseCommandLineOptions(argc, argv);
789 // Read the expected strings from the check file.
790 std::vector<CheckString> CheckStrings;
791 if (ReadCheckFile(SM, CheckStrings))
794 // Open the file to check and add it to SourceMgr.
795 OwningPtr<MemoryBuffer> File;
797 MemoryBuffer::getFileOrSTDIN(InputFilename.c_str(), File)) {
798 errs() << "Could not open input file '" << InputFilename << "': "
799 << ec.message() << '\n';
802 MemoryBuffer *F = File.take();
804 if (F->getBufferSize() == 0) {
805 errs() << "FileCheck error: '" << InputFilename << "' is empty.\n";
809 // Remove duplicate spaces in the input file if requested.
810 if (!NoCanonicalizeWhiteSpace)
811 F = CanonicalizeInputFile(F);
813 SM.AddNewSourceBuffer(F, SMLoc());
815 /// VariableTable - This holds all the current filecheck variables.
816 StringMap<StringRef> VariableTable;
818 // Check that we have all of the expected strings, in order, in the input
820 StringRef Buffer = F->getBuffer();
822 const char *LastMatch = Buffer.data();
824 for (unsigned StrNo = 0, e = CheckStrings.size(); StrNo != e; ++StrNo) {
825 const CheckString &CheckStr = CheckStrings[StrNo];
827 StringRef SearchFrom = Buffer;
829 // Find StrNo in the file.
831 size_t MatchPos = CheckStr.Pat.Match(Buffer, MatchLen, VariableTable);
832 Buffer = Buffer.substr(MatchPos);
834 // If we didn't find a match, reject the input.
835 if (MatchPos == StringRef::npos) {
836 PrintCheckFailed(SM, CheckStr, SearchFrom, VariableTable);
840 StringRef SkippedRegion(LastMatch, Buffer.data()-LastMatch);
842 // If this check is a "CHECK-NEXT", verify that the previous match was on
843 // the previous line (i.e. that there is one newline between them).
844 if (CheckStr.IsCheckNext) {
845 // Count the number of newlines between the previous match and this one.
846 assert(LastMatch != F->getBufferStart() &&
847 "CHECK-NEXT can't be the first check in a file");
849 unsigned NumNewLines = CountNumNewlinesBetween(SkippedRegion);
850 if (NumNewLines == 0) {
851 SM.PrintMessage(CheckStr.Loc, SourceMgr::DK_Error,
852 CheckPrefix+"-NEXT: is on the same line as previous match");
853 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
854 SourceMgr::DK_Note, "'next' match was here");
855 SM.PrintMessage(SMLoc::getFromPointer(LastMatch), SourceMgr::DK_Note,
856 "previous match was here");
860 if (NumNewLines != 1) {
861 SM.PrintMessage(CheckStr.Loc, SourceMgr::DK_Error, CheckPrefix+
862 "-NEXT: is not on the line after the previous match");
863 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
864 SourceMgr::DK_Note, "'next' match was here");
865 SM.PrintMessage(SMLoc::getFromPointer(LastMatch), SourceMgr::DK_Note,
866 "previous match was here");
871 // If this match had "not strings", verify that they don't exist in the
873 for (unsigned ChunkNo = 0, e = CheckStr.NotStrings.size();
874 ChunkNo != e; ++ChunkNo) {
876 size_t Pos = CheckStr.NotStrings[ChunkNo].second.Match(SkippedRegion,
879 if (Pos == StringRef::npos) continue;
881 SM.PrintMessage(SMLoc::getFromPointer(LastMatch+Pos), SourceMgr::DK_Error,
882 CheckPrefix+"-NOT: string occurred!");
883 SM.PrintMessage(CheckStr.NotStrings[ChunkNo].first, SourceMgr::DK_Note,
884 CheckPrefix+"-NOT: pattern specified here");
889 // Otherwise, everything is good. Step over the matched text and remember
890 // the position after the match as the end of the last match.
891 Buffer = Buffer.substr(MatchLen);
892 LastMatch = Buffer.data();