lib/Support/Regex.cpp

   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 "llvm/ADT/SmallVector.h"
  16 #include "llvm/ADT/StringRef.h"
  17 #include "llvm/ADT/Twine.h"
  18 #include <string>
  19
  20 // Important this comes last because it defines "_REGEX_H_". At least on
  21 // Darwin, if included before any header that (transitively) includes
  22 // xlocale.h, this will cause trouble, because of missing regex-related types.
  23 #include "regex_impl.h"
  24
  25 using namespace llvm;
  26
  27 Regex::Regex() : preg(nullptr), error(REG_BADPAT) {}
  28
  29 Regex::Regex(StringRef regex, unsigned Flags) {
  30   unsigned flags = 0;
  31   preg = new llvm_regex();
  32   preg->re_endp = regex.end();
  33   if (Flags & IgnoreCase)
  34     flags |= REG_ICASE;
  35   if (Flags & Newline)
  36     flags |= REG_NEWLINE;
  37   if (!(Flags & BasicRegex))
  38     flags |= REG_EXTENDED;
  39   error = llvm_regcomp(preg, regex.data(), flags|REG_PEND);
  40 }
  41
  42 Regex::Regex(Regex &&regex) {
  43   preg = regex.preg;
  44   error = regex.error;
  45   regex.preg = nullptr;
  46   regex.error = REG_BADPAT;
  47 }
  48
  49 Regex::~Regex() {
  50   if (preg) {
  51     llvm_regfree(preg);
  52     delete preg;
  53   }
  54 }
  55
  56 bool Regex::isValid(std::string &Error) const {
  57   if (!error)
  58     return true;
  59
  60   size_t len = llvm_regerror(error, preg, nullptr, 0);
  61
  62   Error.resize(len - 1);
  63   llvm_regerror(error, preg, &Error[0], len);
  64   return false;
  65 }
  66
  67 /// getNumMatches - In a valid regex, return the number of parenthesized
  68 /// matches it contains.
  69 unsigned Regex::getNumMatches() const {
  70   return preg->re_nsub;
  71 }
  72
  73 bool Regex::match(StringRef String, SmallVectorImpl<StringRef> *Matches){
  74   if (error)
  75     return false;
  76
  77   unsigned nmatch = Matches ? preg->re_nsub+1 : 0;
  78
  79   // pmatch needs to have at least one element.
  80   SmallVector<llvm_regmatch_t, 8> pm;
  81   pm.resize(nmatch > 0 ? nmatch : 1);
  82   pm[0].rm_so = 0;
  83   pm[0].rm_eo = String.size();
  84
  85   int rc = llvm_regexec(preg, String.data(), nmatch, pm.data(), REG_STARTEND);
  86
  87   if (rc == REG_NOMATCH)
  88     return false;
  89   if (rc != 0) {
  90     // regexec can fail due to invalid pattern or running out of memory.
  91     error = rc;
  92     return false;
  93   }
  94
  95   // There was a match.
  96
  97   if (Matches) { // match position requested
  98     Matches->clear();
  99
 100     for (unsigned i = 0; i != nmatch; ++i) {
 101       if (pm[i].rm_so == -1) {
 102         // this group didn't match
 103         Matches->push_back(StringRef());
 104         continue;
 105       }
 106       assert(pm[i].rm_eo >= pm[i].rm_so);
 107       Matches->push_back(StringRef(String.data()+pm[i].rm_so,
 108                                    pm[i].rm_eo-pm[i].rm_so));
 109     }
 110   }
 111
 112   return true;
 113 }
 114
 115 std::string Regex::sub(StringRef Repl, StringRef String,
 116                        std::string *Error) {
 117   SmallVector<StringRef, 8> Matches;
 118
 119   // Reset error, if given.
 120   if (Error && !Error->empty()) *Error = "";
 121
 122   // Return the input if there was no match.
 123   if (!match(String, &Matches))
 124     return String;
 125
 126   // Otherwise splice in the replacement string, starting with the prefix before
 127   // the match.
 128   std::string Res(String.begin(), Matches[0].begin());
 129
 130   // Then the replacement string, honoring possible substitutions.
 131   while (!Repl.empty()) {
 132     // Skip to the next escape.
 133     std::pair<StringRef, StringRef> Split = Repl.split('\\');
 134
 135     // Add the skipped substring.
 136     Res += Split.first;
 137
 138     // Check for terminimation and trailing backslash.
 139     if (Split.second.empty()) {
 140       if (Repl.size() != Split.first.size() &&
 141           Error && Error->empty())
 142         *Error = "replacement string contained trailing backslash";
 143       break;
 144     }
 145
 146     // Otherwise update the replacement string and interpret escapes.
 147     Repl = Split.second;
 148
 149     // FIXME: We should have a StringExtras function for mapping C99 escapes.
 150     switch (Repl[0]) {
 151       // Treat all unrecognized characters as self-quoting.
 152     default:
 153       Res += Repl[0];
 154       Repl = Repl.substr(1);
 155       break;
 156
 157       // Single character escapes.
 158     case 't':
 159       Res += '\t';
 160       Repl = Repl.substr(1);
 161       break;
 162     case 'n':
 163       Res += '\n';
 164       Repl = Repl.substr(1);
 165       break;
 166
 167       // Decimal escapes are backreferences.
 168     case '0': case '1': case '2': case '3': case '4':
 169     case '5': case '6': case '7': case '8': case '9': {
 170       // Extract the backreference number.
 171       StringRef Ref = Repl.slice(0, Repl.find_first_not_of("0123456789"));
 172       Repl = Repl.substr(Ref.size());
 173
 174       unsigned RefValue;
 175       if (!Ref.getAsInteger(10, RefValue) &&
 176           RefValue < Matches.size())
 177         Res += Matches[RefValue];
 178       else if (Error && Error->empty())
 179         *Error = ("invalid backreference string '" + Twine(Ref) + "'").str();
 180       break;
 181     }
 182     }
 183   }
 184
 185   // And finally the suffix.
 186   Res += StringRef(Matches[0].end(), String.end() - Matches[0].end());
 187
 188   return Res;
 189 }
 190
 191 // These are the special characters matched in functions like "p_ere_exp".
 192 static const char RegexMetachars[] = "()^$|*+?.[]\\{}";
 193
 194 bool Regex::isLiteralERE(StringRef Str) {
 195   // Check for regex metacharacters.  This list was derived from our regex
 196   // implementation in regcomp.c and double checked against the POSIX extended
 197   // regular expression specification.
 198   return Str.find_first_of(RegexMetachars) == StringRef::npos;
 199 }
 200
 201 std::string Regex::escape(StringRef String) {
 202   std::string RegexStr;
 203   for (unsigned i = 0, e = String.size(); i != e; ++i) {
 204     if (strchr(RegexMetachars, String[i]))
 205       RegexStr += '\\';
 206     RegexStr += String[i];
 207   }
 208
 209   return RegexStr;
 210 }