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/SourceMgr.h"
23 #include "llvm/Support/raw_ostream.h"
24 #include "llvm/System/Signals.h"
27 static cl::opt<std::string>
28 CheckFilename(cl::Positional, cl::desc("<check-file>"), cl::Required);
30 static cl::opt<std::string>
31 InputFilename("input-file", cl::desc("File to check (defaults to stdin)"),
32 cl::init("-"), cl::value_desc("filename"));
34 static cl::opt<std::string>
35 CheckPrefix("check-prefix", cl::init("CHECK"),
36 cl::desc("Prefix to use from check file (defaults to 'CHECK')"));
39 NoCanonicalizeWhiteSpace("strict-whitespace",
40 cl::desc("Do not treat all horizontal whitespace as equivalent"));
42 /// CheckString - This is a check that we found in the input file.
44 /// Str - The string to match.
47 /// Loc - The location in the match file that the check string was specified.
50 CheckString(const std::string &S, SMLoc L) : Str(S), Loc(L) {}
54 /// FindFixedStringInBuffer - This works like strstr, except for two things:
55 /// 1) it handles 'nul' characters in memory buffers. 2) it returns the end of
56 /// the memory buffer on match failure instead of null.
57 static const char *FindFixedStringInBuffer(StringRef Str, const char *CurPtr,
58 const MemoryBuffer &MB) {
59 assert(!Str.empty() && "Can't find an empty string");
60 const char *BufEnd = MB.getBufferEnd();
63 // Scan for the first character in the match string.
64 CurPtr = (char*)memchr(CurPtr, Str[0], BufEnd-CurPtr);
66 // If we didn't find the first character of the string, then we failed to
68 if (CurPtr == 0) return BufEnd;
70 // If the match string is one character, then we win.
71 if (Str.size() == 1) return CurPtr;
73 // Otherwise, verify that the rest of the string matches.
74 if (Str.size() <= unsigned(BufEnd-CurPtr) &&
75 memcmp(CurPtr+1, Str.data()+1, Str.size()-1) == 0)
78 // If not, advance past this character and try again.
83 /// ReadCheckFile - Read the check file, which specifies the sequence of
84 /// expected strings. The strings are added to the CheckStrings vector.
85 static bool ReadCheckFile(SourceMgr &SM,
86 std::vector<CheckString> &CheckStrings) {
87 // Open the check file, and tell SourceMgr about it.
90 MemoryBuffer::getFileOrSTDIN(CheckFilename.c_str(), &ErrorStr);
92 errs() << "Could not open check file '" << CheckFilename << "': "
96 SM.AddNewSourceBuffer(F, SMLoc());
98 // Find all instances of CheckPrefix followed by : in the file.
99 const char *CurPtr = F->getBufferStart(), *BufferEnd = F->getBufferEnd();
102 // See if Prefix occurs in the memory buffer.
103 const char *Ptr = FindFixedStringInBuffer(CheckPrefix, CurPtr, *F);
105 // If we didn't find a match, we're done.
106 if (Ptr == BufferEnd)
109 // Verify that the : is present after the prefix.
110 if (Ptr[CheckPrefix.size()] != ':') {
115 // Okay, we found the prefix, yay. Remember the rest of the line, but
116 // ignore leading and trailing whitespace.
117 Ptr += CheckPrefix.size()+1;
119 while (*Ptr == ' ' || *Ptr == '\t')
122 // Scan ahead to the end of line.
124 while (CurPtr != BufferEnd && *CurPtr != '\n' && *CurPtr != '\r')
127 // Ignore trailing whitespace.
128 while (CurPtr[-1] == ' ' || CurPtr[-1] == '\t')
131 // Check that there is something on the line.
133 SM.PrintMessage(SMLoc::getFromPointer(CurPtr),
134 "found empty check string with prefix '"+CheckPrefix+":'",
139 // Okay, add the string we captured to the output vector and move on.
140 CheckStrings.push_back(CheckString(std::string(Ptr, CurPtr),
141 SMLoc::getFromPointer(Ptr)));
144 if (CheckStrings.empty()) {
145 errs() << "error: no check strings found with prefix '" << CheckPrefix
153 // CanonicalizeCheckStrings - Replace all sequences of horizontal whitespace in
154 // the check strings with a single space.
155 static void CanonicalizeCheckStrings(std::vector<CheckString> &CheckStrings) {
156 for (unsigned i = 0, e = CheckStrings.size(); i != e; ++i) {
157 std::string &Str = CheckStrings[i].Str;
159 for (unsigned C = 0; C != Str.size(); ++C) {
160 // If C is not a horizontal whitespace, skip it.
161 if (Str[C] != ' ' && Str[C] != '\t')
164 // Replace the character with space, then remove any other space
165 // characters after it.
168 while (C+1 != Str.size() &&
169 (Str[C+1] == ' ' || Str[C+1] == '\t'))
170 Str.erase(Str.begin()+C+1);
175 /// CanonicalizeInputFile - Remove duplicate horizontal space from the specified
176 /// memory buffer, free it, and return a new one.
177 static MemoryBuffer *CanonicalizeInputFile(MemoryBuffer *MB) {
178 SmallVector<char, 16> NewFile;
179 NewFile.reserve(MB->getBufferSize());
181 for (const char *Ptr = MB->getBufferStart(), *End = MB->getBufferEnd();
183 // If C is not a horizontal whitespace, skip it.
184 if (*Ptr != ' ' && *Ptr != '\t') {
185 NewFile.push_back(*Ptr);
189 // Otherwise, add one space and advance over neighboring space.
190 NewFile.push_back(' ');
191 while (Ptr+1 != End &&
192 (Ptr[1] == ' ' || Ptr[1] == '\t'))
196 // Free the old buffer and return a new one.
198 MemoryBuffer::getMemBufferCopy(NewFile.data(),
199 NewFile.data() + NewFile.size(),
200 MB->getBufferIdentifier());
207 int main(int argc, char **argv) {
208 sys::PrintStackTraceOnErrorSignal();
209 PrettyStackTraceProgram X(argc, argv);
210 cl::ParseCommandLineOptions(argc, argv);
214 // Read the expected strings from the check file.
215 std::vector<CheckString> CheckStrings;
216 if (ReadCheckFile(SM, CheckStrings))
219 // Remove duplicate spaces in the check strings if requested.
220 if (!NoCanonicalizeWhiteSpace)
221 CanonicalizeCheckStrings(CheckStrings);
223 // Open the file to check and add it to SourceMgr.
224 std::string ErrorStr;
226 MemoryBuffer::getFileOrSTDIN(InputFilename.c_str(), &ErrorStr);
228 errs() << "Could not open input file '" << InputFilename << "': "
233 // Remove duplicate spaces in the input file if requested.
234 if (!NoCanonicalizeWhiteSpace)
235 F = CanonicalizeInputFile(F);
237 SM.AddNewSourceBuffer(F, SMLoc());
239 // Check that we have all of the expected strings, in order, in the input
241 const char *CurPtr = F->getBufferStart(), *BufferEnd = F->getBufferEnd();
243 for (unsigned StrNo = 0, e = CheckStrings.size(); StrNo != e; ++StrNo) {
244 const CheckString &CheckStr = CheckStrings[StrNo];
246 // Find StrNo in the file.
247 const char *Ptr = FindFixedStringInBuffer(CheckStr.Str, CurPtr, *F);
249 // If we found a match, we're done, move on.
250 if (Ptr != BufferEnd) {
251 CurPtr = Ptr + CheckStr.Str.size();
255 // Otherwise, we have an error, emit an error message.
256 SM.PrintMessage(CheckStr.Loc, "expected string not found in input",
259 // Print the "scanning from here" line. If the current position is at the
260 // end of a line, advance to the start of the next line.
261 const char *Scan = CurPtr;
262 while (Scan != BufferEnd &&
263 (*Scan == ' ' || *Scan == '\t'))
265 if (*Scan == '\n' || *Scan == '\r')
269 SM.PrintMessage(SMLoc::getFromPointer(CurPtr), "scanning from here",