FileCheck.cpp revision 207e1bcf897d1c732f717b9773029651ecc59ab2 
1//===- FileCheck.cpp - Check that File's Contents match what is expected --===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
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.
12//
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.
16//
17//===----------------------------------------------------------------------===//
18
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"
25using namespace llvm;
26
27static cl::opt<std::string>
28CheckFilename(cl::Positional, cl::desc("<check-file>"), cl::Required);
29
30static cl::opt<std::string>
31InputFilename("input-file", cl::desc("File to check (defaults to stdin)"),
32              cl::init("-"), cl::value_desc("filename"));
33
34static cl::opt<std::string>
35CheckPrefix("check-prefix", cl::init("CHECK"),
36            cl::desc("Prefix to use from check file (defaults to 'CHECK')"));
37
38static cl::opt<bool>
39NoCanonicalizeWhiteSpace("strict-whitespace",
40              cl::desc("Do not treat all horizontal whitespace as equivalent"));
41
42/// CheckString - This is a check that we found in the input file.
43struct CheckString {
44  /// Str - The string to match.
45  std::string Str;
46
47  /// Loc - The location in the match file that the check string was specified.
48  SMLoc Loc;
49
50  CheckString(const std::string &S, SMLoc L) : Str(S), Loc(L) {}
51};
52
53
54/// FindStringInBuffer - This is basically just a strstr wrapper that differs in
55/// two ways: first it handles 'nul' characters in memory buffers, second, it
56/// returns the end of the memory buffer on match failure.
57static const char *FindStringInBuffer(const char *Str, const char *CurPtr,
58                                      const MemoryBuffer &MB) {
59  // Check to see if we have a match.  If so, just return it.
60  if (const char *Res = strstr(CurPtr, Str))
61    return Res;
62
63  // If not, check to make sure we didn't just find an embedded nul in the
64  // memory buffer.
65  const char *Ptr = CurPtr + strlen(CurPtr);
66
67  // If we really reached the end of the file, return it.
68  if (Ptr == MB.getBufferEnd())
69    return Ptr;
70
71  // Otherwise, just skip this section of the file, including the nul.
72  return FindStringInBuffer(Str, Ptr+1, MB);
73}
74
75/// ReadCheckFile - Read the check file, which specifies the sequence of
76/// expected strings.  The strings are added to the CheckStrings vector.
77static bool ReadCheckFile(SourceMgr &SM,
78                          std::vector<CheckString> &CheckStrings) {
79  // Open the check file, and tell SourceMgr about it.
80  std::string ErrorStr;
81  MemoryBuffer *F =
82    MemoryBuffer::getFileOrSTDIN(CheckFilename.c_str(), &ErrorStr);
83  if (F == 0) {
84    errs() << "Could not open check file '" << CheckFilename << "': "
85           << ErrorStr << '\n';
86    return true;
87  }
88  SM.AddNewSourceBuffer(F, SMLoc());
89
90  // Find all instances of CheckPrefix followed by : in the file.  The
91  // MemoryBuffer is guaranteed to be nul terminated, but may have nul's
92  // embedded into it.  We don't support check strings with embedded nuls.
93  std::string Prefix = CheckPrefix + ":";
94  const char *CurPtr = F->getBufferStart(), *BufferEnd = F->getBufferEnd();
95
96  while (1) {
97    // See if Prefix occurs in the memory buffer.
98    const char *Ptr = FindStringInBuffer(Prefix.c_str(), CurPtr, *F);
99
100    // If we didn't find a match, we're done.
101    if (Ptr == BufferEnd)
102      break;
103
104    // Okay, we found the prefix, yay.  Remember the rest of the line, but
105    // ignore leading and trailing whitespace.
106    Ptr += Prefix.size();
107    while (*Ptr == ' ' || *Ptr == '\t')
108      ++Ptr;
109
110    // Scan ahead to the end of line.
111    CurPtr = Ptr;
112    while (CurPtr != BufferEnd && *CurPtr != '\n' && *CurPtr != '\r')
113      ++CurPtr;
114
115    // Ignore trailing whitespace.
116    while (CurPtr[-1] == ' ' || CurPtr[-1] == '\t')
117      --CurPtr;
118
119    // Check that there is something on the line.
120    if (Ptr >= CurPtr) {
121      SM.PrintMessage(SMLoc::getFromPointer(CurPtr),
122                      "found empty check string with prefix '"+Prefix+"'",
123                      "error");
124      return true;
125    }
126
127    // Okay, add the string we captured to the output vector and move on.
128    CheckStrings.push_back(CheckString(std::string(Ptr, CurPtr),
129                                       SMLoc::getFromPointer(Ptr)));
130  }
131
132  if (CheckStrings.empty()) {
133    errs() << "error: no check strings found with prefix '" << Prefix << "'\n";
134    return true;
135  }
136
137  return false;
138}
139
140// CanonicalizeCheckStrings - Replace all sequences of horizontal whitespace in
141// the check strings with a single space.
142static void CanonicalizeCheckStrings(std::vector<CheckString> &CheckStrings) {
143  for (unsigned i = 0, e = CheckStrings.size(); i != e; ++i) {
144    std::string &Str = CheckStrings[i].Str;
145
146    for (unsigned C = 0; C != Str.size(); ++C) {
147      // If C is not a horizontal whitespace, skip it.
148      if (Str[C] != ' ' && Str[C] != '\t')
149        continue;
150
151      // Replace the character with space, then remove any other space
152      // characters after it.
153      Str[C] = ' ';
154
155      while (C+1 != Str.size() &&
156             (Str[C+1] == ' ' || Str[C+1] == '\t'))
157        Str.erase(Str.begin()+C+1);
158    }
159  }
160}
161
162/// CanonicalizeInputFile - Remove duplicate horizontal space from the specified
163/// memory buffer, free it, and return a new one.
164static MemoryBuffer *CanonicalizeInputFile(MemoryBuffer *MB) {
165  SmallVector<char, 16> NewFile;
166  NewFile.reserve(MB->getBufferSize());
167
168  for (const char *Ptr = MB->getBufferStart(), *End = MB->getBufferEnd();
169       Ptr != End; ++Ptr) {
170    // If C is not a horizontal whitespace, skip it.
171    if (*Ptr != ' ' && *Ptr != '\t') {
172      NewFile.push_back(*Ptr);
173      continue;
174    }
175
176    // Otherwise, add one space and advance over neighboring space.
177    NewFile.push_back(' ');
178    while (Ptr+1 != End &&
179           (Ptr[1] == ' ' || Ptr[1] == '\t'))
180      ++Ptr;
181  }
182
183  // Free the old buffer and return a new one.
184  MemoryBuffer *MB2 =
185    MemoryBuffer::getMemBufferCopy(NewFile.data(),
186                                   NewFile.data() + NewFile.size(),
187                                   MB->getBufferIdentifier());
188
189  delete MB;
190  return MB2;
191}
192
193
194int main(int argc, char **argv) {
195  sys::PrintStackTraceOnErrorSignal();
196  PrettyStackTraceProgram X(argc, argv);
197  cl::ParseCommandLineOptions(argc, argv);
198
199  SourceMgr SM;
200
201  // Read the expected strings from the check file.
202  std::vector<CheckString> CheckStrings;
203  if (ReadCheckFile(SM, CheckStrings))
204    return 2;
205
206  // Remove duplicate spaces in the check strings if requested.
207  if (!NoCanonicalizeWhiteSpace)
208    CanonicalizeCheckStrings(CheckStrings);
209
210  // Open the file to check and add it to SourceMgr.
211  std::string ErrorStr;
212  MemoryBuffer *F =
213    MemoryBuffer::getFileOrSTDIN(InputFilename.c_str(), &ErrorStr);
214  if (F == 0) {
215    errs() << "Could not open input file '" << InputFilename << "': "
216           << ErrorStr << '\n';
217    return true;
218  }
219
220  // Remove duplicate spaces in the input file if requested.
221  if (!NoCanonicalizeWhiteSpace)
222    F = CanonicalizeInputFile(F);
223
224  SM.AddNewSourceBuffer(F, SMLoc());
225
226  // Check that we have all of the expected strings, in order, in the input
227  // file.
228  const char *CurPtr = F->getBufferStart(), *BufferEnd = F->getBufferEnd();
229
230  for (unsigned StrNo = 0, e = CheckStrings.size(); StrNo != e; ++StrNo) {
231    const CheckString &CheckStr = CheckStrings[StrNo];
232
233    // Find StrNo in the file.
234    const char *Ptr = FindStringInBuffer(CheckStr.Str.c_str(), CurPtr, *F);
235
236    // If we found a match, we're done, move on.
237    if (Ptr != BufferEnd) {
238      CurPtr = Ptr + CheckStr.Str.size();
239      continue;
240    }
241
242    // Otherwise, we have an error, emit an error message.
243    SM.PrintMessage(CheckStr.Loc, "expected string not found in input",
244                    "error");
245
246    // Print the "scanning from here" line.  If the current position is at the
247    // end of a line, advance to the start of the next line.
248    const char *Scan = CurPtr;
249    while (Scan != BufferEnd &&
250           (*Scan == ' ' || *Scan == '\t'))
251      ++Scan;
252    if (*Scan == '\n' || *Scan == '\r')
253      CurPtr = Scan+1;
254
255
256    SM.PrintMessage(SMLoc::getFromPointer(CurPtr), "scanning from here",
257                    "note");
258    return 1;
259  }
260
261  return 0;
262}
263