1ee3c74fbSChris Lattner //===- FileCheck.cpp - Check that File's Contents match what is expected --===//
2ee3c74fbSChris Lattner //
3ee3c74fbSChris Lattner //                     The LLVM Compiler Infrastructure
4ee3c74fbSChris Lattner //
5ee3c74fbSChris Lattner // This file is distributed under the University of Illinois Open Source
6ee3c74fbSChris Lattner // License. See LICENSE.TXT for details.
7ee3c74fbSChris Lattner //
8ee3c74fbSChris Lattner //===----------------------------------------------------------------------===//
9ee3c74fbSChris Lattner //
10ee3c74fbSChris Lattner // FileCheck does a line-by line check of a file that validates whether it
11ee3c74fbSChris Lattner // contains the expected content.  This is useful for regression tests etc.
12ee3c74fbSChris Lattner //
13ee3c74fbSChris Lattner // This program exits with an error status of 2 on error, exit status of 0 if
14ee3c74fbSChris Lattner // the file matched the expected contents, and exit status of 1 if it did not
15ee3c74fbSChris Lattner // contain the expected contents.
16ee3c74fbSChris Lattner //
17ee3c74fbSChris Lattner //===----------------------------------------------------------------------===//
18ee3c74fbSChris Lattner 
19ee3c74fbSChris Lattner #include "llvm/Support/CommandLine.h"
20ee3c74fbSChris Lattner #include "llvm/Support/MemoryBuffer.h"
21ee3c74fbSChris Lattner #include "llvm/Support/PrettyStackTrace.h"
22ee3c74fbSChris Lattner #include "llvm/Support/SourceMgr.h"
23ee3c74fbSChris Lattner #include "llvm/Support/raw_ostream.h"
24ee3c74fbSChris Lattner #include "llvm/System/Signals.h"
25ee3c74fbSChris Lattner using namespace llvm;
26ee3c74fbSChris Lattner 
27ee3c74fbSChris Lattner static cl::opt<std::string>
28ee3c74fbSChris Lattner CheckFilename(cl::Positional, cl::desc("<check-file>"), cl::Required);
29ee3c74fbSChris Lattner 
30ee3c74fbSChris Lattner static cl::opt<std::string>
31ee3c74fbSChris Lattner InputFilename("input-file", cl::desc("File to check (defaults to stdin)"),
32ee3c74fbSChris Lattner               cl::init("-"), cl::value_desc("filename"));
33ee3c74fbSChris Lattner 
34ee3c74fbSChris Lattner static cl::opt<std::string>
35ee3c74fbSChris Lattner CheckPrefix("check-prefix", cl::init("CHECK"),
36ee3c74fbSChris Lattner             cl::desc("Prefix to use from check file (defaults to 'CHECK')"));
37ee3c74fbSChris Lattner 
382c3e5cdfSChris Lattner static cl::opt<bool>
392c3e5cdfSChris Lattner NoCanonicalizeWhiteSpace("strict-whitespace",
402c3e5cdfSChris Lattner               cl::desc("Do not treat all horizontal whitespace as equivalent"));
412c3e5cdfSChris Lattner 
42*26cccfe1SChris Lattner /// CheckString - This is a check that we found in the input file.
43*26cccfe1SChris Lattner struct CheckString {
44*26cccfe1SChris Lattner   /// Str - The string to match.
45*26cccfe1SChris Lattner   std::string Str;
46*26cccfe1SChris Lattner 
47*26cccfe1SChris Lattner   /// Loc - The location in the match file that the check string was specified.
48*26cccfe1SChris Lattner   SMLoc Loc;
49*26cccfe1SChris Lattner 
50*26cccfe1SChris Lattner   CheckString(const std::string &S, SMLoc L) : Str(S), Loc(L) {}
51*26cccfe1SChris Lattner };
52*26cccfe1SChris Lattner 
53ee3c74fbSChris Lattner 
54ee3c74fbSChris Lattner /// FindStringInBuffer - This is basically just a strstr wrapper that differs in
55ee3c74fbSChris Lattner /// two ways: first it handles 'nul' characters in memory buffers, second, it
56ee3c74fbSChris Lattner /// returns the end of the memory buffer on match failure.
57ee3c74fbSChris Lattner static const char *FindStringInBuffer(const char *Str, const char *CurPtr,
58ee3c74fbSChris Lattner                                       const MemoryBuffer &MB) {
59ee3c74fbSChris Lattner   // Check to see if we have a match.  If so, just return it.
60ee3c74fbSChris Lattner   if (const char *Res = strstr(CurPtr, Str))
61ee3c74fbSChris Lattner     return Res;
62ee3c74fbSChris Lattner 
63ee3c74fbSChris Lattner   // If not, check to make sure we didn't just find an embedded nul in the
64ee3c74fbSChris Lattner   // memory buffer.
65ee3c74fbSChris Lattner   const char *Ptr = CurPtr + strlen(CurPtr);
66ee3c74fbSChris Lattner 
67ee3c74fbSChris Lattner   // If we really reached the end of the file, return it.
68ee3c74fbSChris Lattner   if (Ptr == MB.getBufferEnd())
69ee3c74fbSChris Lattner     return Ptr;
70ee3c74fbSChris Lattner 
71ee3c74fbSChris Lattner   // Otherwise, just skip this section of the file, including the nul.
72ee3c74fbSChris Lattner   return FindStringInBuffer(Str, Ptr+1, MB);
73ee3c74fbSChris Lattner }
74ee3c74fbSChris Lattner 
75ee3c74fbSChris Lattner /// ReadCheckFile - Read the check file, which specifies the sequence of
76ee3c74fbSChris Lattner /// expected strings.  The strings are added to the CheckStrings vector.
77ee3c74fbSChris Lattner static bool ReadCheckFile(SourceMgr &SM,
78*26cccfe1SChris Lattner                           std::vector<CheckString> &CheckStrings) {
79ee3c74fbSChris Lattner   // Open the check file, and tell SourceMgr about it.
80ee3c74fbSChris Lattner   std::string ErrorStr;
81ee3c74fbSChris Lattner   MemoryBuffer *F =
82ee3c74fbSChris Lattner     MemoryBuffer::getFileOrSTDIN(CheckFilename.c_str(), &ErrorStr);
83ee3c74fbSChris Lattner   if (F == 0) {
84ee3c74fbSChris Lattner     errs() << "Could not open check file '" << CheckFilename << "': "
85ee3c74fbSChris Lattner            << ErrorStr << '\n';
86ee3c74fbSChris Lattner     return true;
87ee3c74fbSChris Lattner   }
88ee3c74fbSChris Lattner   SM.AddNewSourceBuffer(F, SMLoc());
89ee3c74fbSChris Lattner 
90ee3c74fbSChris Lattner   // Find all instances of CheckPrefix followed by : in the file.  The
91ee3c74fbSChris Lattner   // MemoryBuffer is guaranteed to be nul terminated, but may have nul's
92ee3c74fbSChris Lattner   // embedded into it.  We don't support check strings with embedded nuls.
93ee3c74fbSChris Lattner   std::string Prefix = CheckPrefix + ":";
94ee3c74fbSChris Lattner   const char *CurPtr = F->getBufferStart(), *BufferEnd = F->getBufferEnd();
95ee3c74fbSChris Lattner 
96ee3c74fbSChris Lattner   while (1) {
97ee3c74fbSChris Lattner     // See if Prefix occurs in the memory buffer.
98ee3c74fbSChris Lattner     const char *Ptr = FindStringInBuffer(Prefix.c_str(), CurPtr, *F);
99ee3c74fbSChris Lattner 
100ee3c74fbSChris Lattner     // If we didn't find a match, we're done.
101ee3c74fbSChris Lattner     if (Ptr == BufferEnd)
102ee3c74fbSChris Lattner       break;
103ee3c74fbSChris Lattner 
104ee3c74fbSChris Lattner     // Okay, we found the prefix, yay.  Remember the rest of the line, but
105ee3c74fbSChris Lattner     // ignore leading and trailing whitespace.
106ee3c74fbSChris Lattner     Ptr += Prefix.size();
107ee3c74fbSChris Lattner     while (*Ptr == ' ' || *Ptr == '\t')
108ee3c74fbSChris Lattner       ++Ptr;
109ee3c74fbSChris Lattner 
110ee3c74fbSChris Lattner     // Scan ahead to the end of line.
111ee3c74fbSChris Lattner     CurPtr = Ptr;
112ee3c74fbSChris Lattner     while (CurPtr != BufferEnd && *CurPtr != '\n' && *CurPtr != '\r')
113ee3c74fbSChris Lattner       ++CurPtr;
114ee3c74fbSChris Lattner 
115ee3c74fbSChris Lattner     // Ignore trailing whitespace.
116ee3c74fbSChris Lattner     while (CurPtr[-1] == ' ' || CurPtr[-1] == '\t')
117ee3c74fbSChris Lattner       --CurPtr;
118ee3c74fbSChris Lattner 
119ee3c74fbSChris Lattner     // Check that there is something on the line.
120ee3c74fbSChris Lattner     if (Ptr >= CurPtr) {
121ee3c74fbSChris Lattner       SM.PrintMessage(SMLoc::getFromPointer(CurPtr),
122ee3c74fbSChris Lattner                       "found empty check string with prefix '"+Prefix+"'",
123ee3c74fbSChris Lattner                       "error");
124ee3c74fbSChris Lattner       return true;
125ee3c74fbSChris Lattner     }
126ee3c74fbSChris Lattner 
127ee3c74fbSChris Lattner     // Okay, add the string we captured to the output vector and move on.
128*26cccfe1SChris Lattner     CheckStrings.push_back(CheckString(std::string(Ptr, CurPtr),
129ee3c74fbSChris Lattner                                        SMLoc::getFromPointer(Ptr)));
130ee3c74fbSChris Lattner   }
131ee3c74fbSChris Lattner 
132ee3c74fbSChris Lattner   if (CheckStrings.empty()) {
133ee3c74fbSChris Lattner     errs() << "error: no check strings found with prefix '" << Prefix << "'\n";
134ee3c74fbSChris Lattner     return true;
135ee3c74fbSChris Lattner   }
136ee3c74fbSChris Lattner 
137ee3c74fbSChris Lattner   return false;
138ee3c74fbSChris Lattner }
139ee3c74fbSChris Lattner 
1402c3e5cdfSChris Lattner // CanonicalizeCheckStrings - Replace all sequences of horizontal whitespace in
1412c3e5cdfSChris Lattner // the check strings with a single space.
142*26cccfe1SChris Lattner static void CanonicalizeCheckStrings(std::vector<CheckString> &CheckStrings) {
1432c3e5cdfSChris Lattner   for (unsigned i = 0, e = CheckStrings.size(); i != e; ++i) {
144*26cccfe1SChris Lattner     std::string &Str = CheckStrings[i].Str;
1452c3e5cdfSChris Lattner 
1462c3e5cdfSChris Lattner     for (unsigned C = 0; C != Str.size(); ++C) {
1472c3e5cdfSChris Lattner       // If C is not a horizontal whitespace, skip it.
1482c3e5cdfSChris Lattner       if (Str[C] != ' ' && Str[C] != '\t')
1492c3e5cdfSChris Lattner         continue;
1502c3e5cdfSChris Lattner 
1512c3e5cdfSChris Lattner       // Replace the character with space, then remove any other space
1522c3e5cdfSChris Lattner       // characters after it.
1532c3e5cdfSChris Lattner       Str[C] = ' ';
1542c3e5cdfSChris Lattner 
1552c3e5cdfSChris Lattner       while (C+1 != Str.size() &&
1562c3e5cdfSChris Lattner              (Str[C+1] == ' ' || Str[C+1] == '\t'))
1572c3e5cdfSChris Lattner         Str.erase(Str.begin()+C+1);
1582c3e5cdfSChris Lattner     }
1592c3e5cdfSChris Lattner   }
1602c3e5cdfSChris Lattner }
1612c3e5cdfSChris Lattner 
1622c3e5cdfSChris Lattner /// CanonicalizeInputFile - Remove duplicate horizontal space from the specified
1632c3e5cdfSChris Lattner /// memory buffer, free it, and return a new one.
1642c3e5cdfSChris Lattner static MemoryBuffer *CanonicalizeInputFile(MemoryBuffer *MB) {
1656eca7ab9SDaniel Dunbar   SmallVector<char, 16> NewFile;
1662c3e5cdfSChris Lattner   NewFile.reserve(MB->getBufferSize());
1672c3e5cdfSChris Lattner 
1682c3e5cdfSChris Lattner   for (const char *Ptr = MB->getBufferStart(), *End = MB->getBufferEnd();
1692c3e5cdfSChris Lattner        Ptr != End; ++Ptr) {
1702c3e5cdfSChris Lattner     // If C is not a horizontal whitespace, skip it.
1712c3e5cdfSChris Lattner     if (*Ptr != ' ' && *Ptr != '\t') {
1722c3e5cdfSChris Lattner       NewFile.push_back(*Ptr);
1732c3e5cdfSChris Lattner       continue;
1742c3e5cdfSChris Lattner     }
1752c3e5cdfSChris Lattner 
1762c3e5cdfSChris Lattner     // Otherwise, add one space and advance over neighboring space.
1772c3e5cdfSChris Lattner     NewFile.push_back(' ');
1782c3e5cdfSChris Lattner     while (Ptr+1 != End &&
1792c3e5cdfSChris Lattner            (Ptr[1] == ' ' || Ptr[1] == '\t'))
1802c3e5cdfSChris Lattner       ++Ptr;
1812c3e5cdfSChris Lattner   }
1822c3e5cdfSChris Lattner 
1832c3e5cdfSChris Lattner   // Free the old buffer and return a new one.
1842c3e5cdfSChris Lattner   MemoryBuffer *MB2 =
1856eca7ab9SDaniel Dunbar     MemoryBuffer::getMemBufferCopy(NewFile.data(),
1866eca7ab9SDaniel Dunbar                                    NewFile.data() + NewFile.size(),
1872c3e5cdfSChris Lattner                                    MB->getBufferIdentifier());
1882c3e5cdfSChris Lattner 
1892c3e5cdfSChris Lattner   delete MB;
1902c3e5cdfSChris Lattner   return MB2;
1912c3e5cdfSChris Lattner }
1922c3e5cdfSChris Lattner 
193ee3c74fbSChris Lattner 
194ee3c74fbSChris Lattner int main(int argc, char **argv) {
195ee3c74fbSChris Lattner   sys::PrintStackTraceOnErrorSignal();
196ee3c74fbSChris Lattner   PrettyStackTraceProgram X(argc, argv);
197ee3c74fbSChris Lattner   cl::ParseCommandLineOptions(argc, argv);
198ee3c74fbSChris Lattner 
199ee3c74fbSChris Lattner   SourceMgr SM;
200ee3c74fbSChris Lattner 
201ee3c74fbSChris Lattner   // Read the expected strings from the check file.
202*26cccfe1SChris Lattner   std::vector<CheckString> CheckStrings;
203ee3c74fbSChris Lattner   if (ReadCheckFile(SM, CheckStrings))
204ee3c74fbSChris Lattner     return 2;
205ee3c74fbSChris Lattner 
2062c3e5cdfSChris Lattner   // Remove duplicate spaces in the check strings if requested.
2072c3e5cdfSChris Lattner   if (!NoCanonicalizeWhiteSpace)
2082c3e5cdfSChris Lattner     CanonicalizeCheckStrings(CheckStrings);
2092c3e5cdfSChris Lattner 
210ee3c74fbSChris Lattner   // Open the file to check and add it to SourceMgr.
211ee3c74fbSChris Lattner   std::string ErrorStr;
212ee3c74fbSChris Lattner   MemoryBuffer *F =
213ee3c74fbSChris Lattner     MemoryBuffer::getFileOrSTDIN(InputFilename.c_str(), &ErrorStr);
214ee3c74fbSChris Lattner   if (F == 0) {
215ee3c74fbSChris Lattner     errs() << "Could not open input file '" << InputFilename << "': "
216ee3c74fbSChris Lattner            << ErrorStr << '\n';
217ee3c74fbSChris Lattner     return true;
218ee3c74fbSChris Lattner   }
2192c3e5cdfSChris Lattner 
2202c3e5cdfSChris Lattner   // Remove duplicate spaces in the input file if requested.
2212c3e5cdfSChris Lattner   if (!NoCanonicalizeWhiteSpace)
2222c3e5cdfSChris Lattner     F = CanonicalizeInputFile(F);
2232c3e5cdfSChris Lattner 
224ee3c74fbSChris Lattner   SM.AddNewSourceBuffer(F, SMLoc());
225ee3c74fbSChris Lattner 
226ee3c74fbSChris Lattner   // Check that we have all of the expected strings, in order, in the input
227ee3c74fbSChris Lattner   // file.
228ee3c74fbSChris Lattner   const char *CurPtr = F->getBufferStart(), *BufferEnd = F->getBufferEnd();
229ee3c74fbSChris Lattner 
230ee3c74fbSChris Lattner   for (unsigned StrNo = 0, e = CheckStrings.size(); StrNo != e; ++StrNo) {
231*26cccfe1SChris Lattner     const CheckString &CheckStr = CheckStrings[StrNo];
232ee3c74fbSChris Lattner 
233ee3c74fbSChris Lattner     // Find StrNo in the file.
234*26cccfe1SChris Lattner     const char *Ptr = FindStringInBuffer(CheckStr.Str.c_str(), CurPtr, *F);
235ee3c74fbSChris Lattner 
236ee3c74fbSChris Lattner     // If we found a match, we're done, move on.
237ee3c74fbSChris Lattner     if (Ptr != BufferEnd) {
238*26cccfe1SChris Lattner       CurPtr = Ptr + CheckStr.Str.size();
239ee3c74fbSChris Lattner       continue;
240ee3c74fbSChris Lattner     }
241ee3c74fbSChris Lattner 
242ee3c74fbSChris Lattner     // Otherwise, we have an error, emit an error message.
243*26cccfe1SChris Lattner     SM.PrintMessage(CheckStr.Loc, "expected string not found in input",
244ee3c74fbSChris Lattner                     "error");
2450029c0e5SChris Lattner 
2468bda49a8SDaniel Dunbar     // Print the "scanning from here" line.  If the current position is at the
2478bda49a8SDaniel Dunbar     // end of a line, advance to the start of the next line.
2480029c0e5SChris Lattner     const char *Scan = CurPtr;
2490029c0e5SChris Lattner     while (Scan != BufferEnd &&
2500029c0e5SChris Lattner            (*Scan == ' ' || *Scan == '\t'))
2510029c0e5SChris Lattner       ++Scan;
2520029c0e5SChris Lattner     if (*Scan == '\n' || *Scan == '\r')
2530029c0e5SChris Lattner       CurPtr = Scan+1;
2540029c0e5SChris Lattner 
2550029c0e5SChris Lattner 
256ee3c74fbSChris Lattner     SM.PrintMessage(SMLoc::getFromPointer(CurPtr), "scanning from here",
257ee3c74fbSChris Lattner                     "note");
258aedd8185SChris Lattner     return 1;
259ee3c74fbSChris Lattner   }
260ee3c74fbSChris Lattner 
261ee3c74fbSChris Lattner   return 0;
262ee3c74fbSChris Lattner }
263