ScanfFormatString.cpp revision 71b86be7f07080a3f6acd8e08647813bde2d2a3b 
1//= ScanfFormatString.cpp - Analysis of printf format strings --*- C++ -*-===//
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// Handling of format string in scanf and friends.  The structure of format
11// strings for fscanf() are described in C99 7.19.6.2.
12//
13//===----------------------------------------------------------------------===//
14
15#include "clang/Analysis/Analyses/FormatString.h"
16#include "FormatStringParsing.h"
17
18using clang::analyze_format_string::ArgTypeResult;
19using clang::analyze_format_string::FormatStringHandler;
20using clang::analyze_format_string::LengthModifier;
21using clang::analyze_format_string::OptionalAmount;
22using clang::analyze_scanf::ConversionSpecifier;
23using clang::analyze_scanf::ScanfSpecifier;
24
25typedef clang::analyze_format_string::SpecifierResult<ScanfSpecifier>
26        ScanfSpecifierResult;
27
28static bool ParseScanList(FormatStringHandler &H,
29                          ConversionSpecifier &CS,
30                          const char *&Beg, const char *E) {
31  const char *I = Beg;
32  const char *start = I - 1;
33  UpdateOnReturn <const char*> UpdateBeg(Beg, I);
34
35  // No more characters?
36  if (I == E) {
37    H.HandleIncompleteScanList(start, I);
38    return true;
39  }
40
41  // Special case: ']' is the first character.
42  if (*I == ']') {
43    if (++I == E) {
44      H.HandleIncompleteScanList(start, I - 1);
45      return true;
46    }
47  }
48
49  // Look for a ']' character which denotes the end of the scan list.
50  while (*I != ']') {
51    if (++I == E) {
52      H.HandleIncompleteScanList(start, I - 1);
53      return true;
54    }
55  }
56
57  CS.setEndScanList(I);
58  return false;
59}
60
61// FIXME: Much of this is copy-paste from ParsePrintfSpecifier.
62// We can possibly refactor.
63static ScanfSpecifierResult ParseScanfSpecifier(FormatStringHandler &H,
64                                                const char *&Beg,
65                                                const char *E,
66                                                unsigned &argIndex) {
67
68  using namespace clang::analyze_scanf;
69  const char *I = Beg;
70  const char *Start = 0;
71  UpdateOnReturn <const char*> UpdateBeg(Beg, I);
72
73    // Look for a '%' character that indicates the start of a format specifier.
74  for ( ; I != E ; ++I) {
75    char c = *I;
76    if (c == '\0') {
77        // Detect spurious null characters, which are likely errors.
78      H.HandleNullChar(I);
79      return true;
80    }
81    if (c == '%') {
82      Start = I++;  // Record the start of the format specifier.
83      break;
84    }
85  }
86
87    // No format specifier found?
88  if (!Start)
89    return false;
90
91  if (I == E) {
92      // No more characters left?
93    H.HandleIncompleteSpecifier(Start, E - Start);
94    return true;
95  }
96
97  ScanfSpecifier FS;
98  if (ParseArgPosition(H, FS, Start, I, E))
99    return true;
100
101  if (I == E) {
102      // No more characters left?
103    H.HandleIncompleteSpecifier(Start, E - Start);
104    return true;
105  }
106
107  // Look for '*' flag if it is present.
108  if (*I == '*') {
109    FS.setSuppressAssignment(I);
110    if (++I == E) {
111      H.HandleIncompleteSpecifier(Start, E - Start);
112      return true;
113    }
114  }
115
116  // Look for the field width (if any).  Unlike printf, this is either
117  // a fixed integer or isn't present.
118  const OptionalAmount &Amt = clang::analyze_format_string::ParseAmount(I, E);
119  if (Amt.getHowSpecified() != OptionalAmount::NotSpecified) {
120    assert(Amt.getHowSpecified() == OptionalAmount::Constant);
121    FS.setFieldWidth(Amt);
122
123    if (I == E) {
124      // No more characters left?
125      H.HandleIncompleteSpecifier(Start, E - Start);
126      return true;
127    }
128  }
129
130  // Look for the length modifier.
131  if (ParseLengthModifier(FS, I, E) && I == E) {
132      // No more characters left?
133    H.HandleIncompleteSpecifier(Start, E - Start);
134    return true;
135  }
136
137  // Detect spurious null characters, which are likely errors.
138  if (*I == '\0') {
139    H.HandleNullChar(I);
140    return true;
141  }
142
143  // Finally, look for the conversion specifier.
144  const char *conversionPosition = I++;
145  ConversionSpecifier::Kind k = ConversionSpecifier::InvalidSpecifier;
146  switch (*conversionPosition) {
147    default:
148      break;
149    case '%': k = ConversionSpecifier::PercentArg;   break;
150    case 'A': k = ConversionSpecifier::AArg; break;
151    case 'E': k = ConversionSpecifier::EArg; break;
152    case 'F': k = ConversionSpecifier::FArg; break;
153    case 'G': k = ConversionSpecifier::GArg; break;
154    case 'X': k = ConversionSpecifier::XArg; break;
155    case 'a': k = ConversionSpecifier::aArg; break;
156    case 'd': k = ConversionSpecifier::dArg; break;
157    case 'e': k = ConversionSpecifier::eArg; break;
158    case 'f': k = ConversionSpecifier::fArg; break;
159    case 'g': k = ConversionSpecifier::gArg; break;
160    case 'i': k = ConversionSpecifier::iArg; break;
161    case 'n': k = ConversionSpecifier::ConsumedSoFarArg; break;
162    case 'c': k = ConversionSpecifier::cArg; break;
163    case 'C': k = ConversionSpecifier::CArg; break;
164    case 'S': k = ConversionSpecifier::SArg; break;
165    case '[': k = ConversionSpecifier::ScanListArg; break;
166  }
167  ConversionSpecifier CS(conversionPosition, k);
168  if (k == ConversionSpecifier::ScanListArg) {
169    if (!ParseScanList(H, CS, I, E))
170      return true;
171  }
172  FS.setConversionSpecifier(CS);
173  if (CS.consumesDataArgument() && !FS.getSuppressAssignment()
174      && !FS.usesPositionalArg())
175    FS.setArgIndex(argIndex++);
176
177  // FIXME: '%' and '*' doesn't make sense.  Issue a warning.
178  // FIXME: 'ConsumedSoFar' and '*' doesn't make sense.
179
180  if (k == ConversionSpecifier::InvalidSpecifier) {
181    // Assume the conversion takes one argument.
182    return !H.HandleInvalidScanfConversionSpecifier(FS, Beg, I - Beg);
183  }
184  return ScanfSpecifierResult(Start, FS);
185}
186
187bool clang::analyze_format_string::ParseScanfString(FormatStringHandler &H,
188                                                    const char *I,
189                                                    const char *E) {
190
191  unsigned argIndex = 0;
192
193  // Keep looking for a format specifier until we have exhausted the string.
194  while (I != E) {
195    const ScanfSpecifierResult &FSR = ParseScanfSpecifier(H, I, E, argIndex);
196    // Did a fail-stop error of any kind occur when parsing the specifier?
197    // If so, don't do any more processing.
198    if (FSR.shouldStop())
199      return true;;
200      // Did we exhaust the string or encounter an error that
201      // we can recover from?
202    if (!FSR.hasValue())
203      continue;
204      // We have a format specifier.  Pass it to the callback.
205    if (!H.HandleScanfSpecifier(FSR.getValue(), FSR.getStart(),
206                                I - FSR.getStart())) {
207      return true;
208    }
209  }
210  assert(I == E && "Format string not exhausted");
211  return false;
212}
213
214
215