Regex.cpp revision 263508 
1//===-- Regex.cpp - Regular Expression matcher implementation -------------===//
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// This file implements a POSIX regular expression matcher.
11//
12//===----------------------------------------------------------------------===//
13
14#include "llvm/Support/Regex.h"
15#include "regex_impl.h"
16#include "llvm/ADT/SmallVector.h"
17#include "llvm/Support/ErrorHandling.h"
18#include "llvm/Support/raw_ostream.h"
19#include <string>
20using namespace llvm;
21
22Regex::Regex(StringRef regex, unsigned Flags) {
23  unsigned flags = 0;
24  preg = new llvm_regex();
25  preg->re_endp = regex.end();
26  if (Flags & IgnoreCase)
27    flags |= REG_ICASE;
28  if (Flags & Newline)
29    flags |= REG_NEWLINE;
30  if (!(Flags & BasicRegex))
31    flags |= REG_EXTENDED;
32  error = llvm_regcomp(preg, regex.data(), flags|REG_PEND);
33}
34
35Regex::~Regex() {
36  llvm_regfree(preg);
37  delete preg;
38}
39
40bool Regex::isValid(std::string &Error) {
41  if (!error)
42    return true;
43
44  size_t len = llvm_regerror(error, preg, NULL, 0);
45
46  Error.resize(len - 1);
47  llvm_regerror(error, preg, &Error[0], len);
48  return false;
49}
50
51/// getNumMatches - In a valid regex, return the number of parenthesized
52/// matches it contains.
53unsigned Regex::getNumMatches() const {
54  return preg->re_nsub;
55}
56
57bool Regex::match(StringRef String, SmallVectorImpl<StringRef> *Matches){
58  unsigned nmatch = Matches ? preg->re_nsub+1 : 0;
59
60  // pmatch needs to have at least one element.
61  SmallVector<llvm_regmatch_t, 8> pm;
62  pm.resize(nmatch > 0 ? nmatch : 1);
63  pm[0].rm_so = 0;
64  pm[0].rm_eo = String.size();
65
66  int rc = llvm_regexec(preg, String.data(), nmatch, pm.data(), REG_STARTEND);
67
68  if (rc == REG_NOMATCH)
69    return false;
70  if (rc != 0) {
71    // regexec can fail due to invalid pattern or running out of memory.
72    error = rc;
73    return false;
74  }
75
76  // There was a match.
77
78  if (Matches) { // match position requested
79    Matches->clear();
80
81    for (unsigned i = 0; i != nmatch; ++i) {
82      if (pm[i].rm_so == -1) {
83        // this group didn't match
84        Matches->push_back(StringRef());
85        continue;
86      }
87      assert(pm[i].rm_eo >= pm[i].rm_so);
88      Matches->push_back(StringRef(String.data()+pm[i].rm_so,
89                                   pm[i].rm_eo-pm[i].rm_so));
90    }
91  }
92
93  return true;
94}
95
96std::string Regex::sub(StringRef Repl, StringRef String,
97                       std::string *Error) {
98  SmallVector<StringRef, 8> Matches;
99
100  // Reset error, if given.
101  if (Error && !Error->empty()) *Error = "";
102
103  // Return the input if there was no match.
104  if (!match(String, &Matches))
105    return String;
106
107  // Otherwise splice in the replacement string, starting with the prefix before
108  // the match.
109  std::string Res(String.begin(), Matches[0].begin());
110
111  // Then the replacement string, honoring possible substitutions.
112  while (!Repl.empty()) {
113    // Skip to the next escape.
114    std::pair<StringRef, StringRef> Split = Repl.split('\\');
115
116    // Add the skipped substring.
117    Res += Split.first;
118
119    // Check for terminimation and trailing backslash.
120    if (Split.second.empty()) {
121      if (Repl.size() != Split.first.size() &&
122          Error && Error->empty())
123        *Error = "replacement string contained trailing backslash";
124      break;
125    }
126
127    // Otherwise update the replacement string and interpret escapes.
128    Repl = Split.second;
129
130    // FIXME: We should have a StringExtras function for mapping C99 escapes.
131    switch (Repl[0]) {
132      // Treat all unrecognized characters as self-quoting.
133    default:
134      Res += Repl[0];
135      Repl = Repl.substr(1);
136      break;
137
138      // Single character escapes.
139    case 't':
140      Res += '\t';
141      Repl = Repl.substr(1);
142      break;
143    case 'n':
144      Res += '\n';
145      Repl = Repl.substr(1);
146      break;
147
148      // Decimal escapes are backreferences.
149    case '0': case '1': case '2': case '3': case '4':
150    case '5': case '6': case '7': case '8': case '9': {
151      // Extract the backreference number.
152      StringRef Ref = Repl.slice(0, Repl.find_first_not_of("0123456789"));
153      Repl = Repl.substr(Ref.size());
154
155      unsigned RefValue;
156      if (!Ref.getAsInteger(10, RefValue) &&
157          RefValue < Matches.size())
158        Res += Matches[RefValue];
159      else if (Error && Error->empty())
160        *Error = "invalid backreference string '" + Ref.str() + "'";
161      break;
162    }
163    }
164  }
165
166  // And finally the suffix.
167  Res += StringRef(Matches[0].end(), String.end() - Matches[0].end());
168
169  return Res;
170}
171
172bool Regex::isLiteralERE(StringRef Str) {
173  // Check for regex metacharacters.  This list was derived from our regex
174  // implementation in regcomp.c and double checked against the POSIX extended
175  // regular expression specification.
176  return Str.find_first_of("()^$|*+?.[]\\{}") == StringRef::npos;
177}
178