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/Support/CommandLine.h"
20 #include "llvm/Support/MemoryBuffer.h"
21 #include "llvm/Support/PrettyStackTrace.h"
22 #include "llvm/Support/Regex.h"
23 #include "llvm/Support/SourceMgr.h"
24 #include "llvm/Support/raw_ostream.h"
25 #include "llvm/System/Signals.h"
26 #include "llvm/ADT/StringMap.h"
30 static cl::opt<std::string>
31 CheckFilename(cl::Positional, cl::desc("<check-file>"), cl::Required);
33 static cl::opt<std::string>
34 InputFilename("input-file", cl::desc("File to check (defaults to stdin)"),
35 cl::init("-"), cl::value_desc("filename"));
37 static cl::opt<std::string>
38 CheckPrefix("check-prefix", cl::init("CHECK"),
39 cl::desc("Prefix to use from check file (defaults to 'CHECK')"));
42 NoCanonicalizeWhiteSpace("strict-whitespace",
43 cl::desc("Do not treat all horizontal whitespace as equivalent"));
45 //===----------------------------------------------------------------------===//
46 // Pattern Handling Code.
47 //===----------------------------------------------------------------------===//
52 /// FixedStr - If non-empty, this pattern is a fixed string match with the
53 /// specified fixed string.
56 /// RegEx - If non-empty, this is a regex pattern.
59 /// VariableUses - Entries in this vector map to uses of a variable in the
60 /// pattern, e.g. "foo[[bar]]baz". In this case, the RegExStr will contain
61 /// "foobaz" and we'll get an entry in this vector that tells us to insert the
62 /// value of bar at offset 3.
63 std::vector<std::pair<StringRef, unsigned> > VariableUses;
65 /// VariableDefs - Entries in this vector map to definitions of a variable in
66 /// the pattern, e.g. "foo[[bar:.*]]baz". In this case, the RegExStr will
67 /// contain "foo(.*)baz" and VariableDefs will contain the pair "bar",1. The
68 /// index indicates what parenthesized value captures the variable value.
69 std::vector<std::pair<StringRef, unsigned> > VariableDefs;
75 bool ParsePattern(StringRef PatternStr, SourceMgr &SM);
77 /// Match - Match the pattern string against the input buffer Buffer. This
78 /// returns the position that is matched or npos if there is no match. If
79 /// there is a match, the size of the matched string is returned in MatchLen.
81 /// The VariableTable StringMap provides the current values of filecheck
82 /// variables and is updated if this match defines new values.
83 size_t Match(StringRef Buffer, size_t &MatchLen,
84 StringMap<StringRef> &VariableTable) const;
87 static void AddFixedStringToRegEx(StringRef FixedStr, std::string &TheStr);
88 bool AddRegExToRegEx(StringRef RegExStr, unsigned &CurParen, SourceMgr &SM);
92 bool Pattern::ParsePattern(StringRef PatternStr, SourceMgr &SM) {
93 PatternLoc = SMLoc::getFromPointer(PatternStr.data());
95 // Ignore trailing whitespace.
96 while (!PatternStr.empty() &&
97 (PatternStr.back() == ' ' || PatternStr.back() == '\t'))
98 PatternStr = PatternStr.substr(0, PatternStr.size()-1);
100 // Check that there is something on the line.
101 if (PatternStr.empty()) {
102 SM.PrintMessage(PatternLoc, "found empty check string with prefix '" +
103 CheckPrefix+":'", "error");
107 // Check to see if this is a fixed string, or if it has regex pieces.
108 if (PatternStr.size() < 2 ||
109 (PatternStr.find("{{") == StringRef::npos &&
110 PatternStr.find("[[") == StringRef::npos)) {
111 FixedStr = PatternStr;
115 // Paren value #0 is for the fully matched string. Any new parenthesized
116 // values add from their.
117 unsigned CurParen = 1;
119 // Otherwise, there is at least one regex piece. Build up the regex pattern
120 // by escaping scary characters in fixed strings, building up one big regex.
121 while (!PatternStr.empty()) {
123 if (PatternStr.size() >= 2 &&
124 PatternStr[0] == '{' && PatternStr[1] == '{') {
126 // Otherwise, this is the start of a regex match. Scan for the }}.
127 size_t End = PatternStr.find("}}");
128 if (End == StringRef::npos) {
129 SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()),
130 "found start of regex string with no end '}}'", "error");
134 if (AddRegExToRegEx(PatternStr.substr(2, End-2), CurParen, SM))
136 PatternStr = PatternStr.substr(End+2);
140 // Named RegEx matches. These are of two forms: [[foo:.*]] which matches .*
141 // (or some other regex) and assigns it to the FileCheck variable 'foo'. The
142 // second form is [[foo]] which is a reference to foo. The variable name
143 // itself must be of the form "[a-zA-Z][0-9a-zA-Z]*", otherwise we reject
144 // it. This is to catch some common errors.
145 if (PatternStr.size() >= 2 &&
146 PatternStr[0] == '[' && PatternStr[1] == '[') {
147 // Verify that it is terminated properly.
148 size_t End = PatternStr.find("]]");
149 if (End == StringRef::npos) {
150 SM.PrintMessage(SMLoc::getFromPointer(PatternStr.data()),
151 "invalid named regex reference, no ]] found", "error");
155 StringRef MatchStr = PatternStr.substr(2, End-2);
156 PatternStr = PatternStr.substr(End+2);
158 // Get the regex name (e.g. "foo").
159 size_t NameEnd = MatchStr.find(':');
160 StringRef Name = MatchStr.substr(0, NameEnd);
163 SM.PrintMessage(SMLoc::getFromPointer(Name.data()),
164 "invalid name in named regex: empty name", "error");
168 // Verify that the name is well formed.
169 for (unsigned i = 0, e = Name.size(); i != e; ++i)
170 if ((Name[i] < 'a' || Name[i] > 'z') &&
171 (Name[i] < 'A' || Name[i] > 'Z') &&
172 (Name[i] < '0' || Name[i] > '9')) {
173 SM.PrintMessage(SMLoc::getFromPointer(Name.data()+i),
174 "invalid name in named regex", "error");
178 // Name can't start with a digit.
179 if (isdigit(Name[0])) {
180 SM.PrintMessage(SMLoc::getFromPointer(Name.data()),
181 "invalid name in named regex", "error");
186 if (NameEnd == StringRef::npos) {
187 VariableUses.push_back(std::make_pair(Name, RegExStr.size()));
191 // Handle [[foo:.*]].
192 VariableDefs.push_back(std::make_pair(Name, CurParen));
196 if (AddRegExToRegEx(MatchStr.substr(NameEnd+1), CurParen, SM))
202 // Handle fixed string matches.
203 // Find the end, which is the start of the next regex.
204 size_t FixedMatchEnd = PatternStr.find("{{");
205 FixedMatchEnd = std::min(FixedMatchEnd, PatternStr.find("[["));
206 AddFixedStringToRegEx(PatternStr.substr(0, FixedMatchEnd), RegExStr);
207 PatternStr = PatternStr.substr(FixedMatchEnd);
214 void Pattern::AddFixedStringToRegEx(StringRef FixedStr, std::string &TheStr) {
215 // Add the characters from FixedStr to the regex, escaping as needed. This
216 // avoids "leaning toothpicks" in common patterns.
217 for (unsigned i = 0, e = FixedStr.size(); i != e; ++i) {
218 switch (FixedStr[i]) {
219 // These are the special characters matched in "p_ere_exp".
235 TheStr += FixedStr[i];
241 bool Pattern::AddRegExToRegEx(StringRef RegexStr, unsigned &CurParen,
245 if (!R.isValid(Error)) {
246 SM.PrintMessage(SMLoc::getFromPointer(RegexStr.data()),
247 "invalid regex: " + Error, "error");
251 RegExStr += RegexStr.str();
252 CurParen += R.getNumMatches();
256 /// Match - Match the pattern string against the input buffer Buffer. This
257 /// returns the position that is matched or npos if there is no match. If
258 /// there is a match, the size of the matched string is returned in MatchLen.
259 size_t Pattern::Match(StringRef Buffer, size_t &MatchLen,
260 StringMap<StringRef> &VariableTable) const {
261 // If this is a fixed string pattern, just match it now.
262 if (!FixedStr.empty()) {
263 MatchLen = FixedStr.size();
264 return Buffer.find(FixedStr);
269 // If there are variable uses, we need to create a temporary string with the
271 StringRef RegExToMatch = RegExStr;
273 if (!VariableUses.empty()) {
276 unsigned InsertOffset = 0;
277 for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) {
278 // Look up the value and escape it so that we can plop it into the regex.
280 AddFixedStringToRegEx(VariableTable[VariableUses[i].first], Value);
282 // Plop it into the regex at the adjusted offset.
283 TmpStr.insert(TmpStr.begin()+VariableUses[i].second+InsertOffset,
284 Value.begin(), Value.end());
285 InsertOffset += Value.size();
288 // Match the newly constructed regex.
289 RegExToMatch = TmpStr;
293 SmallVector<StringRef, 4> MatchInfo;
294 if (!Regex(RegExToMatch, Regex::Newline).match(Buffer, &MatchInfo))
295 return StringRef::npos;
297 // Successful regex match.
298 assert(!MatchInfo.empty() && "Didn't get any match");
299 StringRef FullMatch = MatchInfo[0];
301 // If this defines any variables, remember their values.
302 for (unsigned i = 0, e = VariableDefs.size(); i != e; ++i) {
303 assert(VariableDefs[i].second < MatchInfo.size() &&
304 "Internal paren error");
305 VariableTable[VariableDefs[i].first] = MatchInfo[VariableDefs[i].second];
308 MatchLen = FullMatch.size();
309 return FullMatch.data()-Buffer.data();
313 //===----------------------------------------------------------------------===//
315 //===----------------------------------------------------------------------===//
317 /// CheckString - This is a check that we found in the input file.
319 /// Pat - The pattern to match.
322 /// Loc - The location in the match file that the check string was specified.
325 /// IsCheckNext - This is true if this is a CHECK-NEXT: directive (as opposed
326 /// to a CHECK: directive.
329 /// NotStrings - These are all of the strings that are disallowed from
330 /// occurring between this match string and the previous one (or start of
332 std::vector<std::pair<SMLoc, Pattern> > NotStrings;
334 CheckString(const Pattern &P, SMLoc L, bool isCheckNext)
335 : Pat(P), Loc(L), IsCheckNext(isCheckNext) {}
338 /// CanonicalizeInputFile - Remove duplicate horizontal space from the specified
339 /// memory buffer, free it, and return a new one.
340 static MemoryBuffer *CanonicalizeInputFile(MemoryBuffer *MB) {
341 SmallVector<char, 16> NewFile;
342 NewFile.reserve(MB->getBufferSize());
344 for (const char *Ptr = MB->getBufferStart(), *End = MB->getBufferEnd();
346 // If C is not a horizontal whitespace, skip it.
347 if (*Ptr != ' ' && *Ptr != '\t') {
348 NewFile.push_back(*Ptr);
352 // Otherwise, add one space and advance over neighboring space.
353 NewFile.push_back(' ');
354 while (Ptr+1 != End &&
355 (Ptr[1] == ' ' || Ptr[1] == '\t'))
359 // Free the old buffer and return a new one.
361 MemoryBuffer::getMemBufferCopy(NewFile.data(),
362 NewFile.data() + NewFile.size(),
363 MB->getBufferIdentifier());
370 /// ReadCheckFile - Read the check file, which specifies the sequence of
371 /// expected strings. The strings are added to the CheckStrings vector.
372 static bool ReadCheckFile(SourceMgr &SM,
373 std::vector<CheckString> &CheckStrings) {
374 // Open the check file, and tell SourceMgr about it.
375 std::string ErrorStr;
377 MemoryBuffer::getFileOrSTDIN(CheckFilename.c_str(), &ErrorStr);
379 errs() << "Could not open check file '" << CheckFilename << "': "
384 // If we want to canonicalize whitespace, strip excess whitespace from the
385 // buffer containing the CHECK lines.
386 if (!NoCanonicalizeWhiteSpace)
387 F = CanonicalizeInputFile(F);
389 SM.AddNewSourceBuffer(F, SMLoc());
391 // Find all instances of CheckPrefix followed by : in the file.
392 StringRef Buffer = F->getBuffer();
394 std::vector<std::pair<SMLoc, Pattern> > NotMatches;
397 // See if Prefix occurs in the memory buffer.
398 Buffer = Buffer.substr(Buffer.find(CheckPrefix));
400 // If we didn't find a match, we're done.
404 const char *CheckPrefixStart = Buffer.data();
406 // When we find a check prefix, keep track of whether we find CHECK: or
408 bool IsCheckNext = false, IsCheckNot = false;
410 // Verify that the : is present after the prefix.
411 if (Buffer[CheckPrefix.size()] == ':') {
412 Buffer = Buffer.substr(CheckPrefix.size()+1);
413 } else if (Buffer.size() > CheckPrefix.size()+6 &&
414 memcmp(Buffer.data()+CheckPrefix.size(), "-NEXT:", 6) == 0) {
415 Buffer = Buffer.substr(CheckPrefix.size()+7);
417 } else if (Buffer.size() > CheckPrefix.size()+5 &&
418 memcmp(Buffer.data()+CheckPrefix.size(), "-NOT:", 5) == 0) {
419 Buffer = Buffer.substr(CheckPrefix.size()+6);
422 Buffer = Buffer.substr(1);
426 // Okay, we found the prefix, yay. Remember the rest of the line, but
427 // ignore leading and trailing whitespace.
428 Buffer = Buffer.substr(Buffer.find_first_not_of(" \t"));
430 // Scan ahead to the end of line.
431 size_t EOL = Buffer.find_first_of("\n\r");
433 // Parse the pattern.
435 if (P.ParsePattern(Buffer.substr(0, EOL), SM))
438 Buffer = Buffer.substr(EOL);
441 // Verify that CHECK-NEXT lines have at least one CHECK line before them.
442 if (IsCheckNext && CheckStrings.empty()) {
443 SM.PrintMessage(SMLoc::getFromPointer(CheckPrefixStart),
444 "found '"+CheckPrefix+"-NEXT:' without previous '"+
445 CheckPrefix+ ": line", "error");
451 NotMatches.push_back(std::make_pair(SMLoc::getFromPointer(Buffer.data()),
457 // Okay, add the string we captured to the output vector and move on.
458 CheckStrings.push_back(CheckString(P,
459 SMLoc::getFromPointer(Buffer.data()),
461 std::swap(NotMatches, CheckStrings.back().NotStrings);
464 if (CheckStrings.empty()) {
465 errs() << "error: no check strings found with prefix '" << CheckPrefix
470 if (!NotMatches.empty()) {
471 errs() << "error: '" << CheckPrefix
472 << "-NOT:' not supported after last check line.\n";
479 static void PrintCheckFailed(const SourceMgr &SM, const CheckString &CheckStr,
481 // Otherwise, we have an error, emit an error message.
482 SM.PrintMessage(CheckStr.Loc, "expected string not found in input",
485 // Print the "scanning from here" line. If the current position is at the
486 // end of a line, advance to the start of the next line.
487 Buffer = Buffer.substr(Buffer.find_first_not_of(" \t\n\r"));
489 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), "scanning from here",
493 /// CountNumNewlinesBetween - Count the number of newlines in the specified
495 static unsigned CountNumNewlinesBetween(StringRef Range) {
496 unsigned NumNewLines = 0;
499 Range = Range.substr(Range.find_first_of("\n\r"));
500 if (Range.empty()) return NumNewLines;
504 // Handle \n\r and \r\n as a single newline.
505 if (Range.size() > 1 &&
506 (Range[1] == '\n' || Range[1] == '\r') &&
507 (Range[0] != Range[1]))
508 Range = Range.substr(1);
509 Range = Range.substr(1);
513 int main(int argc, char **argv) {
514 sys::PrintStackTraceOnErrorSignal();
515 PrettyStackTraceProgram X(argc, argv);
516 cl::ParseCommandLineOptions(argc, argv);
520 // Read the expected strings from the check file.
521 std::vector<CheckString> CheckStrings;
522 if (ReadCheckFile(SM, CheckStrings))
525 // Open the file to check and add it to SourceMgr.
526 std::string ErrorStr;
528 MemoryBuffer::getFileOrSTDIN(InputFilename.c_str(), &ErrorStr);
530 errs() << "Could not open input file '" << InputFilename << "': "
535 // Remove duplicate spaces in the input file if requested.
536 if (!NoCanonicalizeWhiteSpace)
537 F = CanonicalizeInputFile(F);
539 SM.AddNewSourceBuffer(F, SMLoc());
541 /// VariableTable - This holds all the current filecheck variables.
542 StringMap<StringRef> VariableTable;
544 // Check that we have all of the expected strings, in order, in the input
546 StringRef Buffer = F->getBuffer();
548 const char *LastMatch = Buffer.data();
550 for (unsigned StrNo = 0, e = CheckStrings.size(); StrNo != e; ++StrNo) {
551 const CheckString &CheckStr = CheckStrings[StrNo];
553 StringRef SearchFrom = Buffer;
555 // Find StrNo in the file.
557 Buffer = Buffer.substr(CheckStr.Pat.Match(Buffer, MatchLen, VariableTable));
559 // If we didn't find a match, reject the input.
560 if (Buffer.empty()) {
561 PrintCheckFailed(SM, CheckStr, SearchFrom);
565 StringRef SkippedRegion(LastMatch, Buffer.data()-LastMatch);
567 // If this check is a "CHECK-NEXT", verify that the previous match was on
568 // the previous line (i.e. that there is one newline between them).
569 if (CheckStr.IsCheckNext) {
570 // Count the number of newlines between the previous match and this one.
571 assert(LastMatch != F->getBufferStart() &&
572 "CHECK-NEXT can't be the first check in a file");
574 unsigned NumNewLines = CountNumNewlinesBetween(SkippedRegion);
575 if (NumNewLines == 0) {
576 SM.PrintMessage(CheckStr.Loc,
577 CheckPrefix+"-NEXT: is on the same line as previous match",
579 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
580 "'next' match was here", "note");
581 SM.PrintMessage(SMLoc::getFromPointer(LastMatch),
582 "previous match was here", "note");
586 if (NumNewLines != 1) {
587 SM.PrintMessage(CheckStr.Loc,
589 "-NEXT: is not on the line after the previous match",
591 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
592 "'next' match was here", "note");
593 SM.PrintMessage(SMLoc::getFromPointer(LastMatch),
594 "previous match was here", "note");
599 // If this match had "not strings", verify that they don't exist in the
601 for (unsigned ChunkNo = 0, e = CheckStr.NotStrings.size();
602 ChunkNo != e; ++ChunkNo) {
604 size_t Pos = CheckStr.NotStrings[ChunkNo].second.Match(SkippedRegion,
607 if (Pos == StringRef::npos) continue;
609 SM.PrintMessage(SMLoc::getFromPointer(LastMatch+Pos),
610 CheckPrefix+"-NOT: string occurred!", "error");
611 SM.PrintMessage(CheckStr.NotStrings[ChunkNo].first,
612 CheckPrefix+"-NOT: pattern specified here", "note");
617 // Otherwise, everything is good. Step over the matched text and remember
618 // the position after the match as the end of the last match.
619 Buffer = Buffer.substr(MatchLen);
620 LastMatch = Buffer.data();