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] != '_' &&
171 (Name[i] < 'a' || Name[i] > 'z') &&
172 (Name[i] < 'A' || Name[i] > 'Z') &&
173 (Name[i] < '0' || Name[i] > '9')) {
174 SM.PrintMessage(SMLoc::getFromPointer(Name.data()+i),
175 "invalid name in named regex", "error");
179 // Name can't start with a digit.
180 if (isdigit(Name[0])) {
181 SM.PrintMessage(SMLoc::getFromPointer(Name.data()),
182 "invalid name in named regex", "error");
187 if (NameEnd == StringRef::npos) {
188 VariableUses.push_back(std::make_pair(Name, RegExStr.size()));
192 // Handle [[foo:.*]].
193 VariableDefs.push_back(std::make_pair(Name, CurParen));
197 if (AddRegExToRegEx(MatchStr.substr(NameEnd+1), CurParen, SM))
203 // Handle fixed string matches.
204 // Find the end, which is the start of the next regex.
205 size_t FixedMatchEnd = PatternStr.find("{{");
206 FixedMatchEnd = std::min(FixedMatchEnd, PatternStr.find("[["));
207 AddFixedStringToRegEx(PatternStr.substr(0, FixedMatchEnd), RegExStr);
208 PatternStr = PatternStr.substr(FixedMatchEnd);
215 void Pattern::AddFixedStringToRegEx(StringRef FixedStr, std::string &TheStr) {
216 // Add the characters from FixedStr to the regex, escaping as needed. This
217 // avoids "leaning toothpicks" in common patterns.
218 for (unsigned i = 0, e = FixedStr.size(); i != e; ++i) {
219 switch (FixedStr[i]) {
220 // These are the special characters matched in "p_ere_exp".
236 TheStr += FixedStr[i];
242 bool Pattern::AddRegExToRegEx(StringRef RegexStr, unsigned &CurParen,
246 if (!R.isValid(Error)) {
247 SM.PrintMessage(SMLoc::getFromPointer(RegexStr.data()),
248 "invalid regex: " + Error, "error");
252 RegExStr += RegexStr.str();
253 CurParen += R.getNumMatches();
257 /// Match - Match the pattern string against the input buffer Buffer. This
258 /// returns the position that is matched or npos if there is no match. If
259 /// there is a match, the size of the matched string is returned in MatchLen.
260 size_t Pattern::Match(StringRef Buffer, size_t &MatchLen,
261 StringMap<StringRef> &VariableTable) const {
262 // If this is a fixed string pattern, just match it now.
263 if (!FixedStr.empty()) {
264 MatchLen = FixedStr.size();
265 return Buffer.find(FixedStr);
270 // If there are variable uses, we need to create a temporary string with the
272 StringRef RegExToMatch = RegExStr;
274 if (!VariableUses.empty()) {
277 unsigned InsertOffset = 0;
278 for (unsigned i = 0, e = VariableUses.size(); i != e; ++i) {
279 // Look up the value and escape it so that we can plop it into the regex.
281 AddFixedStringToRegEx(VariableTable[VariableUses[i].first], Value);
283 // Plop it into the regex at the adjusted offset.
284 TmpStr.insert(TmpStr.begin()+VariableUses[i].second+InsertOffset,
285 Value.begin(), Value.end());
286 InsertOffset += Value.size();
289 // Match the newly constructed regex.
290 RegExToMatch = TmpStr;
294 SmallVector<StringRef, 4> MatchInfo;
295 if (!Regex(RegExToMatch, Regex::Newline).match(Buffer, &MatchInfo))
296 return StringRef::npos;
298 // Successful regex match.
299 assert(!MatchInfo.empty() && "Didn't get any match");
300 StringRef FullMatch = MatchInfo[0];
302 // If this defines any variables, remember their values.
303 for (unsigned i = 0, e = VariableDefs.size(); i != e; ++i) {
304 assert(VariableDefs[i].second < MatchInfo.size() &&
305 "Internal paren error");
306 VariableTable[VariableDefs[i].first] = MatchInfo[VariableDefs[i].second];
309 MatchLen = FullMatch.size();
310 return FullMatch.data()-Buffer.data();
314 //===----------------------------------------------------------------------===//
316 //===----------------------------------------------------------------------===//
318 /// CheckString - This is a check that we found in the input file.
320 /// Pat - The pattern to match.
323 /// Loc - The location in the match file that the check string was specified.
326 /// IsCheckNext - This is true if this is a CHECK-NEXT: directive (as opposed
327 /// to a CHECK: directive.
330 /// NotStrings - These are all of the strings that are disallowed from
331 /// occurring between this match string and the previous one (or start of
333 std::vector<std::pair<SMLoc, Pattern> > NotStrings;
335 CheckString(const Pattern &P, SMLoc L, bool isCheckNext)
336 : Pat(P), Loc(L), IsCheckNext(isCheckNext) {}
339 /// CanonicalizeInputFile - Remove duplicate horizontal space from the specified
340 /// memory buffer, free it, and return a new one.
341 static MemoryBuffer *CanonicalizeInputFile(MemoryBuffer *MB) {
342 SmallVector<char, 16> NewFile;
343 NewFile.reserve(MB->getBufferSize());
345 for (const char *Ptr = MB->getBufferStart(), *End = MB->getBufferEnd();
347 // If C is not a horizontal whitespace, skip it.
348 if (*Ptr != ' ' && *Ptr != '\t') {
349 NewFile.push_back(*Ptr);
353 // Otherwise, add one space and advance over neighboring space.
354 NewFile.push_back(' ');
355 while (Ptr+1 != End &&
356 (Ptr[1] == ' ' || Ptr[1] == '\t'))
360 // Free the old buffer and return a new one.
362 MemoryBuffer::getMemBufferCopy(NewFile.data(),
363 NewFile.data() + NewFile.size(),
364 MB->getBufferIdentifier());
371 /// ReadCheckFile - Read the check file, which specifies the sequence of
372 /// expected strings. The strings are added to the CheckStrings vector.
373 static bool ReadCheckFile(SourceMgr &SM,
374 std::vector<CheckString> &CheckStrings) {
375 // Open the check file, and tell SourceMgr about it.
376 std::string ErrorStr;
378 MemoryBuffer::getFileOrSTDIN(CheckFilename.c_str(), &ErrorStr);
380 errs() << "Could not open check file '" << CheckFilename << "': "
385 // If we want to canonicalize whitespace, strip excess whitespace from the
386 // buffer containing the CHECK lines.
387 if (!NoCanonicalizeWhiteSpace)
388 F = CanonicalizeInputFile(F);
390 SM.AddNewSourceBuffer(F, SMLoc());
392 // Find all instances of CheckPrefix followed by : in the file.
393 StringRef Buffer = F->getBuffer();
395 std::vector<std::pair<SMLoc, Pattern> > NotMatches;
398 // See if Prefix occurs in the memory buffer.
399 Buffer = Buffer.substr(Buffer.find(CheckPrefix));
401 // If we didn't find a match, we're done.
405 const char *CheckPrefixStart = Buffer.data();
407 // When we find a check prefix, keep track of whether we find CHECK: or
409 bool IsCheckNext = false, IsCheckNot = false;
411 // Verify that the : is present after the prefix.
412 if (Buffer[CheckPrefix.size()] == ':') {
413 Buffer = Buffer.substr(CheckPrefix.size()+1);
414 } else if (Buffer.size() > CheckPrefix.size()+6 &&
415 memcmp(Buffer.data()+CheckPrefix.size(), "-NEXT:", 6) == 0) {
416 Buffer = Buffer.substr(CheckPrefix.size()+7);
418 } else if (Buffer.size() > CheckPrefix.size()+5 &&
419 memcmp(Buffer.data()+CheckPrefix.size(), "-NOT:", 5) == 0) {
420 Buffer = Buffer.substr(CheckPrefix.size()+6);
423 Buffer = Buffer.substr(1);
427 // Okay, we found the prefix, yay. Remember the rest of the line, but
428 // ignore leading and trailing whitespace.
429 Buffer = Buffer.substr(Buffer.find_first_not_of(" \t"));
431 // Scan ahead to the end of line.
432 size_t EOL = Buffer.find_first_of("\n\r");
434 // Parse the pattern.
436 if (P.ParsePattern(Buffer.substr(0, EOL), SM))
439 Buffer = Buffer.substr(EOL);
442 // Verify that CHECK-NEXT lines have at least one CHECK line before them.
443 if (IsCheckNext && CheckStrings.empty()) {
444 SM.PrintMessage(SMLoc::getFromPointer(CheckPrefixStart),
445 "found '"+CheckPrefix+"-NEXT:' without previous '"+
446 CheckPrefix+ ": line", "error");
452 NotMatches.push_back(std::make_pair(SMLoc::getFromPointer(Buffer.data()),
458 // Okay, add the string we captured to the output vector and move on.
459 CheckStrings.push_back(CheckString(P,
460 SMLoc::getFromPointer(Buffer.data()),
462 std::swap(NotMatches, CheckStrings.back().NotStrings);
465 if (CheckStrings.empty()) {
466 errs() << "error: no check strings found with prefix '" << CheckPrefix
471 if (!NotMatches.empty()) {
472 errs() << "error: '" << CheckPrefix
473 << "-NOT:' not supported after last check line.\n";
480 static void PrintCheckFailed(const SourceMgr &SM, const CheckString &CheckStr,
482 // Otherwise, we have an error, emit an error message.
483 SM.PrintMessage(CheckStr.Loc, "expected string not found in input",
486 // Print the "scanning from here" line. If the current position is at the
487 // end of a line, advance to the start of the next line.
488 Buffer = Buffer.substr(Buffer.find_first_not_of(" \t\n\r"));
490 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()), "scanning from here",
494 /// CountNumNewlinesBetween - Count the number of newlines in the specified
496 static unsigned CountNumNewlinesBetween(StringRef Range) {
497 unsigned NumNewLines = 0;
500 Range = Range.substr(Range.find_first_of("\n\r"));
501 if (Range.empty()) return NumNewLines;
505 // Handle \n\r and \r\n as a single newline.
506 if (Range.size() > 1 &&
507 (Range[1] == '\n' || Range[1] == '\r') &&
508 (Range[0] != Range[1]))
509 Range = Range.substr(1);
510 Range = Range.substr(1);
514 int main(int argc, char **argv) {
515 sys::PrintStackTraceOnErrorSignal();
516 PrettyStackTraceProgram X(argc, argv);
517 cl::ParseCommandLineOptions(argc, argv);
521 // Read the expected strings from the check file.
522 std::vector<CheckString> CheckStrings;
523 if (ReadCheckFile(SM, CheckStrings))
526 // Open the file to check and add it to SourceMgr.
527 std::string ErrorStr;
529 MemoryBuffer::getFileOrSTDIN(InputFilename.c_str(), &ErrorStr);
531 errs() << "Could not open input file '" << InputFilename << "': "
536 // Remove duplicate spaces in the input file if requested.
537 if (!NoCanonicalizeWhiteSpace)
538 F = CanonicalizeInputFile(F);
540 SM.AddNewSourceBuffer(F, SMLoc());
542 /// VariableTable - This holds all the current filecheck variables.
543 StringMap<StringRef> VariableTable;
545 // Check that we have all of the expected strings, in order, in the input
547 StringRef Buffer = F->getBuffer();
549 const char *LastMatch = Buffer.data();
551 for (unsigned StrNo = 0, e = CheckStrings.size(); StrNo != e; ++StrNo) {
552 const CheckString &CheckStr = CheckStrings[StrNo];
554 StringRef SearchFrom = Buffer;
556 // Find StrNo in the file.
558 Buffer = Buffer.substr(CheckStr.Pat.Match(Buffer, MatchLen, VariableTable));
560 // If we didn't find a match, reject the input.
561 if (Buffer.empty()) {
562 PrintCheckFailed(SM, CheckStr, SearchFrom);
566 StringRef SkippedRegion(LastMatch, Buffer.data()-LastMatch);
568 // If this check is a "CHECK-NEXT", verify that the previous match was on
569 // the previous line (i.e. that there is one newline between them).
570 if (CheckStr.IsCheckNext) {
571 // Count the number of newlines between the previous match and this one.
572 assert(LastMatch != F->getBufferStart() &&
573 "CHECK-NEXT can't be the first check in a file");
575 unsigned NumNewLines = CountNumNewlinesBetween(SkippedRegion);
576 if (NumNewLines == 0) {
577 SM.PrintMessage(CheckStr.Loc,
578 CheckPrefix+"-NEXT: is on the same line as previous match",
580 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
581 "'next' match was here", "note");
582 SM.PrintMessage(SMLoc::getFromPointer(LastMatch),
583 "previous match was here", "note");
587 if (NumNewLines != 1) {
588 SM.PrintMessage(CheckStr.Loc,
590 "-NEXT: is not on the line after the previous match",
592 SM.PrintMessage(SMLoc::getFromPointer(Buffer.data()),
593 "'next' match was here", "note");
594 SM.PrintMessage(SMLoc::getFromPointer(LastMatch),
595 "previous match was here", "note");
600 // If this match had "not strings", verify that they don't exist in the
602 for (unsigned ChunkNo = 0, e = CheckStr.NotStrings.size();
603 ChunkNo != e; ++ChunkNo) {
605 size_t Pos = CheckStr.NotStrings[ChunkNo].second.Match(SkippedRegion,
608 if (Pos == StringRef::npos) continue;
610 SM.PrintMessage(SMLoc::getFromPointer(LastMatch+Pos),
611 CheckPrefix+"-NOT: string occurred!", "error");
612 SM.PrintMessage(CheckStr.NotStrings[ChunkNo].first,
613 CheckPrefix+"-NOT: pattern specified here", "note");
618 // Otherwise, everything is good. Step over the matched text and remember
619 // the position after the match as the end of the last match.
620 Buffer = Buffer.substr(MatchLen);
621 LastMatch = Buffer.data();