tools/pcre/pcrecpp.cc
author David Anderson <dvander@alliedmods.net>
Thu Oct 29 00:06:12 2009 -0700 (2009-10-29)
changeset 2844 87f951bece2d
parent 2762 d2d467bf1b5a
child 3831 babd0eee3c5e
permissions -rw-r--r--
Fix iloop race when using AddCommandListener and SourceMod unloads (bug 4077, r=pred).
dvander@2762
     1
// Copyright (c) 2005, Google Inc.
dvander@2762
     2
// All rights reserved.
dvander@2762
     3
//
dvander@2762
     4
// Redistribution and use in source and binary forms, with or without
dvander@2762
     5
// modification, are permitted provided that the following conditions are
dvander@2762
     6
// met:
dvander@2762
     7
//
dvander@2762
     8
//     * Redistributions of source code must retain the above copyright
dvander@2762
     9
// notice, this list of conditions and the following disclaimer.
dvander@2762
    10
//     * Redistributions in binary form must reproduce the above
dvander@2762
    11
// copyright notice, this list of conditions and the following disclaimer
dvander@2762
    12
// in the documentation and/or other materials provided with the
dvander@2762
    13
// distribution.
dvander@2762
    14
//     * Neither the name of Google Inc. nor the names of its
dvander@2762
    15
// contributors may be used to endorse or promote products derived from
dvander@2762
    16
// this software without specific prior written permission.
dvander@2762
    17
//
dvander@2762
    18
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
dvander@2762
    19
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
dvander@2762
    20
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
dvander@2762
    21
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
dvander@2762
    22
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
dvander@2762
    23
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
dvander@2762
    24
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
dvander@2762
    25
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
dvander@2762
    26
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
dvander@2762
    27
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
dvander@2762
    28
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
dvander@2762
    29
//
dvander@2762
    30
// Author: Sanjay Ghemawat
dvander@2762
    31
dvander@2762
    32
#ifdef HAVE_CONFIG_H
dvander@2762
    33
#include "config.h"
dvander@2762
    34
#endif
dvander@2762
    35
dvander@2762
    36
#include <stdlib.h>
dvander@2762
    37
#include <stdio.h>
dvander@2762
    38
#include <ctype.h>
dvander@2762
    39
#include <limits.h>      /* for SHRT_MIN, USHRT_MAX, etc */
dvander@2762
    40
#include <assert.h>
dvander@2762
    41
#include <errno.h>
dvander@2762
    42
#include <string>
dvander@2762
    43
#include <algorithm>
dvander@2762
    44
dvander@2762
    45
#include "pcrecpp_internal.h"
dvander@2762
    46
#include "pcre.h"
dvander@2762
    47
#include "pcrecpp.h"
dvander@2762
    48
#include "pcre_stringpiece.h"
dvander@2762
    49
dvander@2762
    50
dvander@2762
    51
namespace pcrecpp {
dvander@2762
    52
dvander@2762
    53
// Maximum number of args we can set
dvander@2762
    54
static const int kMaxArgs = 16;
dvander@2762
    55
static const int kVecSize = (1 + kMaxArgs) * 3;  // results + PCRE workspace
dvander@2762
    56
dvander@2762
    57
// Special object that stands-in for no argument
dvander@2762
    58
Arg RE::no_arg((void*)NULL);
dvander@2762
    59
dvander@2762
    60
// This is for ABI compatibility with old versions of pcre (pre-7.6),
dvander@2762
    61
// which defined a global no_arg variable instead of putting it in the
dvander@2762
    62
// RE class.  This works on GCC >= 3, at least.  It definitely works
dvander@2762
    63
// for ELF, but may not for other object formats (Mach-O, for
dvander@2762
    64
// instance, does not support aliases.)  We could probably have a more
dvander@2762
    65
// inclusive test if we ever needed it.  (Note that not only the
dvander@2762
    66
// __attribute__ syntax, but also __USER_LABEL_PREFIX__, are
dvander@2762
    67
// gnu-specific.)
dvander@2762
    68
#if defined(__GNUC__) && __GNUC__ >= 3 && defined(__ELF__)
dvander@2762
    69
# define ULP_AS_STRING(x)            ULP_AS_STRING_INTERNAL(x)
dvander@2762
    70
# define ULP_AS_STRING_INTERNAL(x)   #x
dvander@2762
    71
# define USER_LABEL_PREFIX_STR       ULP_AS_STRING(__USER_LABEL_PREFIX__)
dvander@2762
    72
extern Arg no_arg
dvander@2762
    73
  __attribute__((alias(USER_LABEL_PREFIX_STR "_ZN7pcrecpp2RE6no_argE")));
dvander@2762
    74
#endif
dvander@2762
    75
dvander@2762
    76
// If a regular expression has no error, its error_ field points here
dvander@2762
    77
static const string empty_string;
dvander@2762
    78
dvander@2762
    79
// If the user doesn't ask for any options, we just use this one
dvander@2762
    80
static RE_Options default_options;
dvander@2762
    81
dvander@2762
    82
void RE::Init(const string& pat, const RE_Options* options) {
dvander@2762
    83
  pattern_ = pat;
dvander@2762
    84
  if (options == NULL) {
dvander@2762
    85
    options_ = default_options;
dvander@2762
    86
  } else {
dvander@2762
    87
    options_ = *options;
dvander@2762
    88
  }
dvander@2762
    89
  error_ = &empty_string;
dvander@2762
    90
  re_full_ = NULL;
dvander@2762
    91
  re_partial_ = NULL;
dvander@2762
    92
dvander@2762
    93
  re_partial_ = Compile(UNANCHORED);
dvander@2762
    94
  if (re_partial_ != NULL) {
dvander@2762
    95
    re_full_ = Compile(ANCHOR_BOTH);
dvander@2762
    96
  }
dvander@2762
    97
}
dvander@2762
    98
dvander@2762
    99
void RE::Cleanup() {
dvander@2762
   100
  if (re_full_ != NULL)         (*pcre_free)(re_full_);
dvander@2762
   101
  if (re_partial_ != NULL)      (*pcre_free)(re_partial_);
dvander@2762
   102
  if (error_ != &empty_string)  delete error_;
dvander@2762
   103
}
dvander@2762
   104
dvander@2762
   105
dvander@2762
   106
RE::~RE() {
dvander@2762
   107
  Cleanup();
dvander@2762
   108
}
dvander@2762
   109
dvander@2762
   110
dvander@2762
   111
pcre* RE::Compile(Anchor anchor) {
dvander@2762
   112
  // First, convert RE_Options into pcre options
dvander@2762
   113
  int pcre_options = 0;
dvander@2762
   114
  pcre_options = options_.all_options();
dvander@2762
   115
dvander@2762
   116
  // Special treatment for anchoring.  This is needed because at
dvander@2762
   117
  // runtime pcre only provides an option for anchoring at the
dvander@2762
   118
  // beginning of a string (unless you use offset).
dvander@2762
   119
  //
dvander@2762
   120
  // There are three types of anchoring we want:
dvander@2762
   121
  //    UNANCHORED      Compile the original pattern, and use
dvander@2762
   122
  //                    a pcre unanchored match.
dvander@2762
   123
  //    ANCHOR_START    Compile the original pattern, and use
dvander@2762
   124
  //                    a pcre anchored match.
dvander@2762
   125
  //    ANCHOR_BOTH     Tack a "\z" to the end of the original pattern
dvander@2762
   126
  //                    and use a pcre anchored match.
dvander@2762
   127
dvander@2762
   128
  const char* compile_error;
dvander@2762
   129
  int eoffset;
dvander@2762
   130
  pcre* re;
dvander@2762
   131
  if (anchor != ANCHOR_BOTH) {
dvander@2762
   132
    re = pcre_compile(pattern_.c_str(), pcre_options,
dvander@2762
   133
                      &compile_error, &eoffset, NULL);
dvander@2762
   134
  } else {
dvander@2762
   135
    // Tack a '\z' at the end of RE.  Parenthesize it first so that
dvander@2762
   136
    // the '\z' applies to all top-level alternatives in the regexp.
dvander@2762
   137
    string wrapped = "(?:";  // A non-counting grouping operator
dvander@2762
   138
    wrapped += pattern_;
dvander@2762
   139
    wrapped += ")\\z";
dvander@2762
   140
    re = pcre_compile(wrapped.c_str(), pcre_options,
dvander@2762
   141
                      &compile_error, &eoffset, NULL);
dvander@2762
   142
  }
dvander@2762
   143
  if (re == NULL) {
dvander@2762
   144
    if (error_ == &empty_string) error_ = new string(compile_error);
dvander@2762
   145
  }
dvander@2762
   146
  return re;
dvander@2762
   147
}
dvander@2762
   148
dvander@2762
   149
/***** Matching interfaces *****/
dvander@2762
   150
dvander@2762
   151
bool RE::FullMatch(const StringPiece& text,
dvander@2762
   152
                   const Arg& ptr1,
dvander@2762
   153
                   const Arg& ptr2,
dvander@2762
   154
                   const Arg& ptr3,
dvander@2762
   155
                   const Arg& ptr4,
dvander@2762
   156
                   const Arg& ptr5,
dvander@2762
   157
                   const Arg& ptr6,
dvander@2762
   158
                   const Arg& ptr7,
dvander@2762
   159
                   const Arg& ptr8,
dvander@2762
   160
                   const Arg& ptr9,
dvander@2762
   161
                   const Arg& ptr10,
dvander@2762
   162
                   const Arg& ptr11,
dvander@2762
   163
                   const Arg& ptr12,
dvander@2762
   164
                   const Arg& ptr13,
dvander@2762
   165
                   const Arg& ptr14,
dvander@2762
   166
                   const Arg& ptr15,
dvander@2762
   167
                   const Arg& ptr16) const {
dvander@2762
   168
  const Arg* args[kMaxArgs];
dvander@2762
   169
  int n = 0;
dvander@2762
   170
  if (&ptr1  == &no_arg) goto done; args[n++] = &ptr1;
dvander@2762
   171
  if (&ptr2  == &no_arg) goto done; args[n++] = &ptr2;
dvander@2762
   172
  if (&ptr3  == &no_arg) goto done; args[n++] = &ptr3;
dvander@2762
   173
  if (&ptr4  == &no_arg) goto done; args[n++] = &ptr4;
dvander@2762
   174
  if (&ptr5  == &no_arg) goto done; args[n++] = &ptr5;
dvander@2762
   175
  if (&ptr6  == &no_arg) goto done; args[n++] = &ptr6;
dvander@2762
   176
  if (&ptr7  == &no_arg) goto done; args[n++] = &ptr7;
dvander@2762
   177
  if (&ptr8  == &no_arg) goto done; args[n++] = &ptr8;
dvander@2762
   178
  if (&ptr9  == &no_arg) goto done; args[n++] = &ptr9;
dvander@2762
   179
  if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
dvander@2762
   180
  if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
dvander@2762
   181
  if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
dvander@2762
   182
  if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
dvander@2762
   183
  if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
dvander@2762
   184
  if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
dvander@2762
   185
  if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
dvander@2762
   186
 done:
dvander@2762
   187
dvander@2762
   188
  int consumed;
dvander@2762
   189
  int vec[kVecSize];
dvander@2762
   190
  return DoMatchImpl(text, ANCHOR_BOTH, &consumed, args, n, vec, kVecSize);
dvander@2762
   191
}
dvander@2762
   192
dvander@2762
   193
bool RE::PartialMatch(const StringPiece& text,
dvander@2762
   194
                      const Arg& ptr1,
dvander@2762
   195
                      const Arg& ptr2,
dvander@2762
   196
                      const Arg& ptr3,
dvander@2762
   197
                      const Arg& ptr4,
dvander@2762
   198
                      const Arg& ptr5,
dvander@2762
   199
                      const Arg& ptr6,
dvander@2762
   200
                      const Arg& ptr7,
dvander@2762
   201
                      const Arg& ptr8,
dvander@2762
   202
                      const Arg& ptr9,
dvander@2762
   203
                      const Arg& ptr10,
dvander@2762
   204
                      const Arg& ptr11,
dvander@2762
   205
                      const Arg& ptr12,
dvander@2762
   206
                      const Arg& ptr13,
dvander@2762
   207
                      const Arg& ptr14,
dvander@2762
   208
                      const Arg& ptr15,
dvander@2762
   209
                      const Arg& ptr16) const {
dvander@2762
   210
  const Arg* args[kMaxArgs];
dvander@2762
   211
  int n = 0;
dvander@2762
   212
  if (&ptr1  == &no_arg) goto done; args[n++] = &ptr1;
dvander@2762
   213
  if (&ptr2  == &no_arg) goto done; args[n++] = &ptr2;
dvander@2762
   214
  if (&ptr3  == &no_arg) goto done; args[n++] = &ptr3;
dvander@2762
   215
  if (&ptr4  == &no_arg) goto done; args[n++] = &ptr4;
dvander@2762
   216
  if (&ptr5  == &no_arg) goto done; args[n++] = &ptr5;
dvander@2762
   217
  if (&ptr6  == &no_arg) goto done; args[n++] = &ptr6;
dvander@2762
   218
  if (&ptr7  == &no_arg) goto done; args[n++] = &ptr7;
dvander@2762
   219
  if (&ptr8  == &no_arg) goto done; args[n++] = &ptr8;
dvander@2762
   220
  if (&ptr9  == &no_arg) goto done; args[n++] = &ptr9;
dvander@2762
   221
  if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
dvander@2762
   222
  if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
dvander@2762
   223
  if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
dvander@2762
   224
  if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
dvander@2762
   225
  if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
dvander@2762
   226
  if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
dvander@2762
   227
  if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
dvander@2762
   228
 done:
dvander@2762
   229
dvander@2762
   230
  int consumed;
dvander@2762
   231
  int vec[kVecSize];
dvander@2762
   232
  return DoMatchImpl(text, UNANCHORED, &consumed, args, n, vec, kVecSize);
dvander@2762
   233
}
dvander@2762
   234
dvander@2762
   235
bool RE::Consume(StringPiece* input,
dvander@2762
   236
                 const Arg& ptr1,
dvander@2762
   237
                 const Arg& ptr2,
dvander@2762
   238
                 const Arg& ptr3,
dvander@2762
   239
                 const Arg& ptr4,
dvander@2762
   240
                 const Arg& ptr5,
dvander@2762
   241
                 const Arg& ptr6,
dvander@2762
   242
                 const Arg& ptr7,
dvander@2762
   243
                 const Arg& ptr8,
dvander@2762
   244
                 const Arg& ptr9,
dvander@2762
   245
                 const Arg& ptr10,
dvander@2762
   246
                 const Arg& ptr11,
dvander@2762
   247
                 const Arg& ptr12,
dvander@2762
   248
                 const Arg& ptr13,
dvander@2762
   249
                 const Arg& ptr14,
dvander@2762
   250
                 const Arg& ptr15,
dvander@2762
   251
                 const Arg& ptr16) const {
dvander@2762
   252
  const Arg* args[kMaxArgs];
dvander@2762
   253
  int n = 0;
dvander@2762
   254
  if (&ptr1  == &no_arg) goto done; args[n++] = &ptr1;
dvander@2762
   255
  if (&ptr2  == &no_arg) goto done; args[n++] = &ptr2;
dvander@2762
   256
  if (&ptr3  == &no_arg) goto done; args[n++] = &ptr3;
dvander@2762
   257
  if (&ptr4  == &no_arg) goto done; args[n++] = &ptr4;
dvander@2762
   258
  if (&ptr5  == &no_arg) goto done; args[n++] = &ptr5;
dvander@2762
   259
  if (&ptr6  == &no_arg) goto done; args[n++] = &ptr6;
dvander@2762
   260
  if (&ptr7  == &no_arg) goto done; args[n++] = &ptr7;
dvander@2762
   261
  if (&ptr8  == &no_arg) goto done; args[n++] = &ptr8;
dvander@2762
   262
  if (&ptr9  == &no_arg) goto done; args[n++] = &ptr9;
dvander@2762
   263
  if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
dvander@2762
   264
  if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
dvander@2762
   265
  if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
dvander@2762
   266
  if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
dvander@2762
   267
  if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
dvander@2762
   268
  if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
dvander@2762
   269
  if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
dvander@2762
   270
 done:
dvander@2762
   271
dvander@2762
   272
  int consumed;
dvander@2762
   273
  int vec[kVecSize];
dvander@2762
   274
  if (DoMatchImpl(*input, ANCHOR_START, &consumed,
dvander@2762
   275
                  args, n, vec, kVecSize)) {
dvander@2762
   276
    input->remove_prefix(consumed);
dvander@2762
   277
    return true;
dvander@2762
   278
  } else {
dvander@2762
   279
    return false;
dvander@2762
   280
  }
dvander@2762
   281
}
dvander@2762
   282
dvander@2762
   283
bool RE::FindAndConsume(StringPiece* input,
dvander@2762
   284
                        const Arg& ptr1,
dvander@2762
   285
                        const Arg& ptr2,
dvander@2762
   286
                        const Arg& ptr3,
dvander@2762
   287
                        const Arg& ptr4,
dvander@2762
   288
                        const Arg& ptr5,
dvander@2762
   289
                        const Arg& ptr6,
dvander@2762
   290
                        const Arg& ptr7,
dvander@2762
   291
                        const Arg& ptr8,
dvander@2762
   292
                        const Arg& ptr9,
dvander@2762
   293
                        const Arg& ptr10,
dvander@2762
   294
                        const Arg& ptr11,
dvander@2762
   295
                        const Arg& ptr12,
dvander@2762
   296
                        const Arg& ptr13,
dvander@2762
   297
                        const Arg& ptr14,
dvander@2762
   298
                        const Arg& ptr15,
dvander@2762
   299
                        const Arg& ptr16) const {
dvander@2762
   300
  const Arg* args[kMaxArgs];
dvander@2762
   301
  int n = 0;
dvander@2762
   302
  if (&ptr1  == &no_arg) goto done; args[n++] = &ptr1;
dvander@2762
   303
  if (&ptr2  == &no_arg) goto done; args[n++] = &ptr2;
dvander@2762
   304
  if (&ptr3  == &no_arg) goto done; args[n++] = &ptr3;
dvander@2762
   305
  if (&ptr4  == &no_arg) goto done; args[n++] = &ptr4;
dvander@2762
   306
  if (&ptr5  == &no_arg) goto done; args[n++] = &ptr5;
dvander@2762
   307
  if (&ptr6  == &no_arg) goto done; args[n++] = &ptr6;
dvander@2762
   308
  if (&ptr7  == &no_arg) goto done; args[n++] = &ptr7;
dvander@2762
   309
  if (&ptr8  == &no_arg) goto done; args[n++] = &ptr8;
dvander@2762
   310
  if (&ptr9  == &no_arg) goto done; args[n++] = &ptr9;
dvander@2762
   311
  if (&ptr10 == &no_arg) goto done; args[n++] = &ptr10;
dvander@2762
   312
  if (&ptr11 == &no_arg) goto done; args[n++] = &ptr11;
dvander@2762
   313
  if (&ptr12 == &no_arg) goto done; args[n++] = &ptr12;
dvander@2762
   314
  if (&ptr13 == &no_arg) goto done; args[n++] = &ptr13;
dvander@2762
   315
  if (&ptr14 == &no_arg) goto done; args[n++] = &ptr14;
dvander@2762
   316
  if (&ptr15 == &no_arg) goto done; args[n++] = &ptr15;
dvander@2762
   317
  if (&ptr16 == &no_arg) goto done; args[n++] = &ptr16;
dvander@2762
   318
 done:
dvander@2762
   319
dvander@2762
   320
  int consumed;
dvander@2762
   321
  int vec[kVecSize];
dvander@2762
   322
  if (DoMatchImpl(*input, UNANCHORED, &consumed,
dvander@2762
   323
                  args, n, vec, kVecSize)) {
dvander@2762
   324
    input->remove_prefix(consumed);
dvander@2762
   325
    return true;
dvander@2762
   326
  } else {
dvander@2762
   327
    return false;
dvander@2762
   328
  }
dvander@2762
   329
}
dvander@2762
   330
dvander@2762
   331
bool RE::Replace(const StringPiece& rewrite,
dvander@2762
   332
                 string *str) const {
dvander@2762
   333
  int vec[kVecSize];
dvander@2762
   334
  int matches = TryMatch(*str, 0, UNANCHORED, vec, kVecSize);
dvander@2762
   335
  if (matches == 0)
dvander@2762
   336
    return false;
dvander@2762
   337
dvander@2762
   338
  string s;
dvander@2762
   339
  if (!Rewrite(&s, rewrite, *str, vec, matches))
dvander@2762
   340
    return false;
dvander@2762
   341
dvander@2762
   342
  assert(vec[0] >= 0);
dvander@2762
   343
  assert(vec[1] >= 0);
dvander@2762
   344
  str->replace(vec[0], vec[1] - vec[0], s);
dvander@2762
   345
  return true;
dvander@2762
   346
}
dvander@2762
   347
dvander@2762
   348
// Returns PCRE_NEWLINE_CRLF, PCRE_NEWLINE_CR, or PCRE_NEWLINE_LF.
dvander@2762
   349
// Note that PCRE_NEWLINE_CRLF is defined to be P_N_CR | P_N_LF.
dvander@2762
   350
// Modified by PH to add PCRE_NEWLINE_ANY and PCRE_NEWLINE_ANYCRLF.
dvander@2762
   351
dvander@2762
   352
static int NewlineMode(int pcre_options) {
dvander@2762
   353
  // TODO: if we can make it threadsafe, cache this var
dvander@2762
   354
  int newline_mode = 0;
dvander@2762
   355
  /* if (newline_mode) return newline_mode; */  // do this once it's cached
dvander@2762
   356
  if (pcre_options & (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|
dvander@2762
   357
                      PCRE_NEWLINE_ANY|PCRE_NEWLINE_ANYCRLF)) {
dvander@2762
   358
    newline_mode = (pcre_options &
dvander@2762
   359
                    (PCRE_NEWLINE_CRLF|PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|
dvander@2762
   360
                     PCRE_NEWLINE_ANY|PCRE_NEWLINE_ANYCRLF));
dvander@2762
   361
  } else {
dvander@2762
   362
    int newline;
dvander@2762
   363
    pcre_config(PCRE_CONFIG_NEWLINE, &newline);
dvander@2762
   364
    if (newline == 10)
dvander@2762
   365
      newline_mode = PCRE_NEWLINE_LF;
dvander@2762
   366
    else if (newline == 13)
dvander@2762
   367
      newline_mode = PCRE_NEWLINE_CR;
dvander@2762
   368
    else if (newline == 3338)
dvander@2762
   369
      newline_mode = PCRE_NEWLINE_CRLF;
dvander@2762
   370
    else if (newline == -1)
dvander@2762
   371
      newline_mode = PCRE_NEWLINE_ANY;
dvander@2762
   372
    else if (newline == -2)
dvander@2762
   373
      newline_mode = PCRE_NEWLINE_ANYCRLF;
dvander@2762
   374
    else
dvander@2762
   375
      assert(NULL == "Unexpected return value from pcre_config(NEWLINE)");
dvander@2762
   376
  }
dvander@2762
   377
  return newline_mode;
dvander@2762
   378
}
dvander@2762
   379
dvander@2762
   380
int RE::GlobalReplace(const StringPiece& rewrite,
dvander@2762
   381
                      string *str) const {
dvander@2762
   382
  int count = 0;
dvander@2762
   383
  int vec[kVecSize];
dvander@2762
   384
  string out;
dvander@2762
   385
  int start = 0;
dvander@2762
   386
  int lastend = -1;
dvander@2762
   387
dvander@2762
   388
  while (start <= static_cast<int>(str->length())) {
dvander@2762
   389
    int matches = TryMatch(*str, start, UNANCHORED, vec, kVecSize);
dvander@2762
   390
    if (matches <= 0)
dvander@2762
   391
      break;
dvander@2762
   392
    int matchstart = vec[0], matchend = vec[1];
dvander@2762
   393
    assert(matchstart >= start);
dvander@2762
   394
    assert(matchend >= matchstart);
dvander@2762
   395
    if (matchstart == matchend && matchstart == lastend) {
dvander@2762
   396
      // advance one character if we matched an empty string at the same
dvander@2762
   397
      // place as the last match occurred
dvander@2762
   398
      matchend = start + 1;
dvander@2762
   399
      // If the current char is CR and we're in CRLF mode, skip LF too.
dvander@2762
   400
      // Note it's better to call pcre_fullinfo() than to examine
dvander@2762
   401
      // all_options(), since options_ could have changed bewteen
dvander@2762
   402
      // compile-time and now, but this is simpler and safe enough.
dvander@2762
   403
      // Modified by PH to add ANY and ANYCRLF.
dvander@2762
   404
      if (start+1 < static_cast<int>(str->length()) &&
dvander@2762
   405
          (*str)[start] == '\r' && (*str)[start+1] == '\n' &&
dvander@2762
   406
          (NewlineMode(options_.all_options()) == PCRE_NEWLINE_CRLF ||
dvander@2762
   407
           NewlineMode(options_.all_options()) == PCRE_NEWLINE_ANY ||
dvander@2762
   408
           NewlineMode(options_.all_options()) == PCRE_NEWLINE_ANYCRLF)
dvander@2762
   409
          ) {
dvander@2762
   410
        matchend++;
dvander@2762
   411
      }
dvander@2762
   412
      // We also need to advance more than one char if we're in utf8 mode.
dvander@2762
   413
#ifdef SUPPORT_UTF8
dvander@2762
   414
      if (options_.utf8()) {
dvander@2762
   415
        while (matchend < static_cast<int>(str->length()) &&
dvander@2762
   416
               ((*str)[matchend] & 0xc0) == 0x80)
dvander@2762
   417
          matchend++;
dvander@2762
   418
      }
dvander@2762
   419
#endif
dvander@2762
   420
      if (matchend <= static_cast<int>(str->length()))
dvander@2762
   421
        out.append(*str, start, matchend - start);
dvander@2762
   422
      start = matchend;
dvander@2762
   423
    } else {
dvander@2762
   424
      out.append(*str, start, matchstart - start);
dvander@2762
   425
      Rewrite(&out, rewrite, *str, vec, matches);
dvander@2762
   426
      start = matchend;
dvander@2762
   427
      lastend = matchend;
dvander@2762
   428
      count++;
dvander@2762
   429
    }
dvander@2762
   430
  }
dvander@2762
   431
dvander@2762
   432
  if (count == 0)
dvander@2762
   433
    return 0;
dvander@2762
   434
dvander@2762
   435
  if (start < static_cast<int>(str->length()))
dvander@2762
   436
    out.append(*str, start, str->length() - start);
dvander@2762
   437
  swap(out, *str);
dvander@2762
   438
  return count;
dvander@2762
   439
}
dvander@2762
   440
dvander@2762
   441
bool RE::Extract(const StringPiece& rewrite,
dvander@2762
   442
                 const StringPiece& text,
dvander@2762
   443
                 string *out) const {
dvander@2762
   444
  int vec[kVecSize];
dvander@2762
   445
  int matches = TryMatch(text, 0, UNANCHORED, vec, kVecSize);
dvander@2762
   446
  if (matches == 0)
dvander@2762
   447
    return false;
dvander@2762
   448
  out->erase();
dvander@2762
   449
  return Rewrite(out, rewrite, text, vec, matches);
dvander@2762
   450
}
dvander@2762
   451
dvander@2762
   452
/*static*/ string RE::QuoteMeta(const StringPiece& unquoted) {
dvander@2762
   453
  string result;
dvander@2762
   454
dvander@2762
   455
  // Escape any ascii character not in [A-Za-z_0-9].
dvander@2762
   456
  //
dvander@2762
   457
  // Note that it's legal to escape a character even if it has no
dvander@2762
   458
  // special meaning in a regular expression -- so this function does
dvander@2762
   459
  // that.  (This also makes it identical to the perl function of the
dvander@2762
   460
  // same name; see `perldoc -f quotemeta`.)  The one exception is
dvander@2762
   461
  // escaping NUL: rather than doing backslash + NUL, like perl does,
dvander@2762
   462
  // we do '\0', because pcre itself doesn't take embedded NUL chars.
dvander@2762
   463
  for (int ii = 0; ii < unquoted.size(); ++ii) {
dvander@2762
   464
    // Note that using 'isalnum' here raises the benchmark time from
dvander@2762
   465
    // 32ns to 58ns:
dvander@2762
   466
    if (unquoted[ii] == '\0') {
dvander@2762
   467
      result += "\\0";
dvander@2762
   468
    } else if ((unquoted[ii] < 'a' || unquoted[ii] > 'z') &&
dvander@2762
   469
               (unquoted[ii] < 'A' || unquoted[ii] > 'Z') &&
dvander@2762
   470
               (unquoted[ii] < '0' || unquoted[ii] > '9') &&
dvander@2762
   471
               unquoted[ii] != '_' &&
dvander@2762
   472
               // If this is the part of a UTF8 or Latin1 character, we need
dvander@2762
   473
               // to copy this byte without escaping.  Experimentally this is
dvander@2762
   474
               // what works correctly with the regexp library.
dvander@2762
   475
               !(unquoted[ii] & 128)) {
dvander@2762
   476
      result += '\\';
dvander@2762
   477
      result += unquoted[ii];
dvander@2762
   478
    } else {
dvander@2762
   479
      result += unquoted[ii];
dvander@2762
   480
    }
dvander@2762
   481
  }
dvander@2762
   482
dvander@2762
   483
  return result;
dvander@2762
   484
}
dvander@2762
   485
dvander@2762
   486
/***** Actual matching and rewriting code *****/
dvander@2762
   487
dvander@2762
   488
int RE::TryMatch(const StringPiece& text,
dvander@2762
   489
                 int startpos,
dvander@2762
   490
                 Anchor anchor,
dvander@2762
   491
                 int *vec,
dvander@2762
   492
                 int vecsize) const {
dvander@2762
   493
  pcre* re = (anchor == ANCHOR_BOTH) ? re_full_ : re_partial_;
dvander@2762
   494
  if (re == NULL) {
dvander@2762
   495
    //fprintf(stderr, "Matching against invalid re: %s\n", error_->c_str());
dvander@2762
   496
    return 0;
dvander@2762
   497
  }
dvander@2762
   498
dvander@2762
   499
  pcre_extra extra = { 0, 0, 0, 0, 0, 0 };
dvander@2762
   500
  if (options_.match_limit() > 0) {
dvander@2762
   501
    extra.flags |= PCRE_EXTRA_MATCH_LIMIT;
dvander@2762
   502
    extra.match_limit = options_.match_limit();
dvander@2762
   503
  }
dvander@2762
   504
  if (options_.match_limit_recursion() > 0) {
dvander@2762
   505
    extra.flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION;
dvander@2762
   506
    extra.match_limit_recursion = options_.match_limit_recursion();
dvander@2762
   507
  }
dvander@2762
   508
  int rc = pcre_exec(re,              // The regular expression object
dvander@2762
   509
                     &extra,
dvander@2762
   510
                     (text.data() == NULL) ? "" : text.data(),
dvander@2762
   511
                     text.size(),
dvander@2762
   512
                     startpos,
dvander@2762
   513
                     (anchor == UNANCHORED) ? 0 : PCRE_ANCHORED,
dvander@2762
   514
                     vec,
dvander@2762
   515
                     vecsize);
dvander@2762
   516
dvander@2762
   517
  // Handle errors
dvander@2762
   518
  if (rc == PCRE_ERROR_NOMATCH) {
dvander@2762
   519
    return 0;
dvander@2762
   520
  } else if (rc < 0) {
dvander@2762
   521
    //fprintf(stderr, "Unexpected return code: %d when matching '%s'\n",
dvander@2762
   522
    //        re, pattern_.c_str());
dvander@2762
   523
    return 0;
dvander@2762
   524
  } else if (rc == 0) {
dvander@2762
   525
    // pcre_exec() returns 0 as a special case when the number of
dvander@2762
   526
    // capturing subpatterns exceeds the size of the vector.
dvander@2762
   527
    // When this happens, there is a match and the output vector
dvander@2762
   528
    // is filled, but we miss out on the positions of the extra subpatterns.
dvander@2762
   529
    rc = vecsize / 2;
dvander@2762
   530
  }
dvander@2762
   531
dvander@2762
   532
  return rc;
dvander@2762
   533
}
dvander@2762
   534
dvander@2762
   535
bool RE::DoMatchImpl(const StringPiece& text,
dvander@2762
   536
                     Anchor anchor,
dvander@2762
   537
                     int* consumed,
dvander@2762
   538
                     const Arg* const* args,
dvander@2762
   539
                     int n,
dvander@2762
   540
                     int* vec,
dvander@2762
   541
                     int vecsize) const {
dvander@2762
   542
  assert((1 + n) * 3 <= vecsize);  // results + PCRE workspace
dvander@2762
   543
  int matches = TryMatch(text, 0, anchor, vec, vecsize);
dvander@2762
   544
  assert(matches >= 0);  // TryMatch never returns negatives
dvander@2762
   545
  if (matches == 0)
dvander@2762
   546
    return false;
dvander@2762
   547
dvander@2762
   548
  *consumed = vec[1];
dvander@2762
   549
dvander@2762
   550
  if (n == 0 || args == NULL) {
dvander@2762
   551
    // We are not interested in results
dvander@2762
   552
    return true;
dvander@2762
   553
  }
dvander@2762
   554
dvander@2762
   555
  if (NumberOfCapturingGroups() < n) {
dvander@2762
   556
    // RE has fewer capturing groups than number of arg pointers passed in
dvander@2762
   557
    return false;
dvander@2762
   558
  }
dvander@2762
   559
dvander@2762
   560
  // If we got here, we must have matched the whole pattern.
dvander@2762
   561
  // We do not need (can not do) any more checks on the value of 'matches' here
dvander@2762
   562
  // -- see the comment for TryMatch.
dvander@2762
   563
  for (int i = 0; i < n; i++) {
dvander@2762
   564
    const int start = vec[2*(i+1)];
dvander@2762
   565
    const int limit = vec[2*(i+1)+1];
dvander@2762
   566
    if (!args[i]->Parse(text.data() + start, limit-start)) {
dvander@2762
   567
      // TODO: Should we indicate what the error was?
dvander@2762
   568
      return false;
dvander@2762
   569
    }
dvander@2762
   570
  }
dvander@2762
   571
dvander@2762
   572
  return true;
dvander@2762
   573
}
dvander@2762
   574
dvander@2762
   575
bool RE::DoMatch(const StringPiece& text,
dvander@2762
   576
                 Anchor anchor,
dvander@2762
   577
                 int* consumed,
dvander@2762
   578
                 const Arg* const args[],
dvander@2762
   579
                 int n) const {
dvander@2762
   580
  assert(n >= 0);
dvander@2762
   581
  size_t const vecsize = (1 + n) * 3;  // results + PCRE workspace
dvander@2762
   582
                                       // (as for kVecSize)
dvander@2762
   583
  int space[21];   // use stack allocation for small vecsize (common case)
dvander@2762
   584
  int* vec = vecsize <= 21 ? space : new int[vecsize];
dvander@2762
   585
  bool retval = DoMatchImpl(text, anchor, consumed, args, n, vec, vecsize);
dvander@2762
   586
  if (vec != space) delete [] vec;
dvander@2762
   587
  return retval;
dvander@2762
   588
}
dvander@2762
   589
dvander@2762
   590
bool RE::Rewrite(string *out, const StringPiece &rewrite,
dvander@2762
   591
                 const StringPiece &text, int *vec, int veclen) const {
dvander@2762
   592
  for (const char *s = rewrite.data(), *end = s + rewrite.size();
dvander@2762
   593
       s < end; s++) {
dvander@2762
   594
    int c = *s;
dvander@2762
   595
    if (c == '\\') {
dvander@2762
   596
      c = *++s;
dvander@2762
   597
      if (isdigit(c)) {
dvander@2762
   598
        int n = (c - '0');
dvander@2762
   599
        if (n >= veclen) {
dvander@2762
   600
          //fprintf(stderr, requested group %d in regexp %.*s\n",
dvander@2762
   601
          //        n, rewrite.size(), rewrite.data());
dvander@2762
   602
          return false;
dvander@2762
   603
        }
dvander@2762
   604
        int start = vec[2 * n];
dvander@2762
   605
        if (start >= 0)
dvander@2762
   606
          out->append(text.data() + start, vec[2 * n + 1] - start);
dvander@2762
   607
      } else if (c == '\\') {
dvander@2762
   608
        *out += '\\';
dvander@2762
   609
      } else {
dvander@2762
   610
        //fprintf(stderr, "invalid rewrite pattern: %.*s\n",
dvander@2762
   611
        //        rewrite.size(), rewrite.data());
dvander@2762
   612
        return false;
dvander@2762
   613
      }
dvander@2762
   614
    } else {
dvander@2762
   615
      *out += c;
dvander@2762
   616
    }
dvander@2762
   617
  }
dvander@2762
   618
  return true;
dvander@2762
   619
}
dvander@2762
   620
dvander@2762
   621
// Return the number of capturing subpatterns, or -1 if the
dvander@2762
   622
// regexp wasn't valid on construction.
dvander@2762
   623
int RE::NumberOfCapturingGroups() const {
dvander@2762
   624
  if (re_partial_ == NULL) return -1;
dvander@2762
   625
dvander@2762
   626
  int result;
dvander@2762
   627
  int pcre_retval = pcre_fullinfo(re_partial_,  // The regular expression object
dvander@2762
   628
                                  NULL,         // We did not study the pattern
dvander@2762
   629
                                  PCRE_INFO_CAPTURECOUNT,
dvander@2762
   630
                                  &result);
dvander@2762
   631
  assert(pcre_retval == 0);
dvander@2762
   632
  return result;
dvander@2762
   633
}
dvander@2762
   634
dvander@2762
   635
/***** Parsers for various types *****/
dvander@2762
   636
dvander@2762
   637
bool Arg::parse_null(const char* str, int n, void* dest) {
dvander@2762
   638
  // We fail if somebody asked us to store into a non-NULL void* pointer
dvander@2762
   639
  return (dest == NULL);
dvander@2762
   640
}
dvander@2762
   641
dvander@2762
   642
bool Arg::parse_string(const char* str, int n, void* dest) {
dvander@2762
   643
  if (dest == NULL) return true;
dvander@2762
   644
  reinterpret_cast<string*>(dest)->assign(str, n);
dvander@2762
   645
  return true;
dvander@2762
   646
}
dvander@2762
   647
dvander@2762
   648
bool Arg::parse_stringpiece(const char* str, int n, void* dest) {
dvander@2762
   649
  if (dest == NULL) return true;
dvander@2762
   650
  reinterpret_cast<StringPiece*>(dest)->set(str, n);
dvander@2762
   651
  return true;
dvander@2762
   652
}
dvander@2762
   653
dvander@2762
   654
bool Arg::parse_char(const char* str, int n, void* dest) {
dvander@2762
   655
  if (n != 1) return false;
dvander@2762
   656
  if (dest == NULL) return true;
dvander@2762
   657
  *(reinterpret_cast<char*>(dest)) = str[0];
dvander@2762
   658
  return true;
dvander@2762
   659
}
dvander@2762
   660
dvander@2762
   661
bool Arg::parse_uchar(const char* str, int n, void* dest) {
dvander@2762
   662
  if (n != 1) return false;
dvander@2762
   663
  if (dest == NULL) return true;
dvander@2762
   664
  *(reinterpret_cast<unsigned char*>(dest)) = str[0];
dvander@2762
   665
  return true;
dvander@2762
   666
}
dvander@2762
   667
dvander@2762
   668
// Largest number spec that we are willing to parse
dvander@2762
   669
static const int kMaxNumberLength = 32;
dvander@2762
   670
dvander@2762
   671
// REQUIRES "buf" must have length at least kMaxNumberLength+1
dvander@2762
   672
// REQUIRES "n > 0"
dvander@2762
   673
// Copies "str" into "buf" and null-terminates if necessary.
dvander@2762
   674
// Returns one of:
dvander@2762
   675
//      a. "str" if no termination is needed
dvander@2762
   676
//      b. "buf" if the string was copied and null-terminated
dvander@2762
   677
//      c. "" if the input was invalid and has no hope of being parsed
dvander@2762
   678
static const char* TerminateNumber(char* buf, const char* str, int n) {
dvander@2762
   679
  if ((n > 0) && isspace(*str)) {
dvander@2762
   680
    // We are less forgiving than the strtoxxx() routines and do not
dvander@2762
   681
    // allow leading spaces.
dvander@2762
   682
    return "";
dvander@2762
   683
  }
dvander@2762
   684
dvander@2762
   685
  // See if the character right after the input text may potentially
dvander@2762
   686
  // look like a digit.
dvander@2762
   687
  if (isdigit(str[n]) ||
dvander@2762
   688
      ((str[n] >= 'a') && (str[n] <= 'f')) ||
dvander@2762
   689
      ((str[n] >= 'A') && (str[n] <= 'F'))) {
dvander@2762
   690
    if (n > kMaxNumberLength) return ""; // Input too big to be a valid number
dvander@2762
   691
    memcpy(buf, str, n);
dvander@2762
   692
    buf[n] = '\0';
dvander@2762
   693
    return buf;
dvander@2762
   694
  } else {
dvander@2762
   695
    // We can parse right out of the supplied string, so return it.
dvander@2762
   696
    return str;
dvander@2762
   697
  }
dvander@2762
   698
}
dvander@2762
   699
dvander@2762
   700
bool Arg::parse_long_radix(const char* str,
dvander@2762
   701
                           int n,
dvander@2762
   702
                           void* dest,
dvander@2762
   703
                           int radix) {
dvander@2762
   704
  if (n == 0) return false;
dvander@2762
   705
  char buf[kMaxNumberLength+1];
dvander@2762
   706
  str = TerminateNumber(buf, str, n);
dvander@2762
   707
  char* end;
dvander@2762
   708
  errno = 0;
dvander@2762
   709
  long r = strtol(str, &end, radix);
dvander@2762
   710
  if (end != str + n) return false;   // Leftover junk
dvander@2762
   711
  if (errno) return false;
dvander@2762
   712
  if (dest == NULL) return true;
dvander@2762
   713
  *(reinterpret_cast<long*>(dest)) = r;
dvander@2762
   714
  return true;
dvander@2762
   715
}
dvander@2762
   716
dvander@2762
   717
bool Arg::parse_ulong_radix(const char* str,
dvander@2762
   718
                            int n,
dvander@2762
   719
                            void* dest,
dvander@2762
   720
                            int radix) {
dvander@2762
   721
  if (n == 0) return false;
dvander@2762
   722
  char buf[kMaxNumberLength+1];
dvander@2762
   723
  str = TerminateNumber(buf, str, n);
dvander@2762
   724
  if (str[0] == '-') return false;    // strtoul() on a negative number?!
dvander@2762
   725
  char* end;
dvander@2762
   726
  errno = 0;
dvander@2762
   727
  unsigned long r = strtoul(str, &end, radix);
dvander@2762
   728
  if (end != str + n) return false;   // Leftover junk
dvander@2762
   729
  if (errno) return false;
dvander@2762
   730
  if (dest == NULL) return true;
dvander@2762
   731
  *(reinterpret_cast<unsigned long*>(dest)) = r;
dvander@2762
   732
  return true;
dvander@2762
   733
}
dvander@2762
   734
dvander@2762
   735
bool Arg::parse_short_radix(const char* str,
dvander@2762
   736
                            int n,
dvander@2762
   737
                            void* dest,
dvander@2762
   738
                            int radix) {
dvander@2762
   739
  long r;
dvander@2762
   740
  if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse
dvander@2762
   741
  if (r < SHRT_MIN || r > SHRT_MAX) return false;       // Out of range
dvander@2762
   742
  if (dest == NULL) return true;
dvander@2762
   743
  *(reinterpret_cast<short*>(dest)) = static_cast<short>(r);
dvander@2762
   744
  return true;
dvander@2762
   745
}
dvander@2762
   746
dvander@2762
   747
bool Arg::parse_ushort_radix(const char* str,
dvander@2762
   748
                             int n,
dvander@2762
   749
                             void* dest,
dvander@2762
   750
                             int radix) {
dvander@2762
   751
  unsigned long r;
dvander@2762
   752
  if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse
dvander@2762
   753
  if (r > USHRT_MAX) return false;                      // Out of range
dvander@2762
   754
  if (dest == NULL) return true;
dvander@2762
   755
  *(reinterpret_cast<unsigned short*>(dest)) = static_cast<unsigned short>(r);
dvander@2762
   756
  return true;
dvander@2762
   757
}
dvander@2762
   758
dvander@2762
   759
bool Arg::parse_int_radix(const char* str,
dvander@2762
   760
                          int n,
dvander@2762
   761
                          void* dest,
dvander@2762
   762
                          int radix) {
dvander@2762
   763
  long r;
dvander@2762
   764
  if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse
dvander@2762
   765
  if (r < INT_MIN || r > INT_MAX) return false;         // Out of range
dvander@2762
   766
  if (dest == NULL) return true;
dvander@2762
   767
  *(reinterpret_cast<int*>(dest)) = r;
dvander@2762
   768
  return true;
dvander@2762
   769
}
dvander@2762
   770
dvander@2762
   771
bool Arg::parse_uint_radix(const char* str,
dvander@2762
   772
                           int n,
dvander@2762
   773
                           void* dest,
dvander@2762
   774
                           int radix) {
dvander@2762
   775
  unsigned long r;
dvander@2762
   776
  if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse
dvander@2762
   777
  if (r > UINT_MAX) return false;                       // Out of range
dvander@2762
   778
  if (dest == NULL) return true;
dvander@2762
   779
  *(reinterpret_cast<unsigned int*>(dest)) = r;
dvander@2762
   780
  return true;
dvander@2762
   781
}
dvander@2762
   782
dvander@2762
   783
bool Arg::parse_longlong_radix(const char* str,
dvander@2762
   784
                               int n,
dvander@2762
   785
                               void* dest,
dvander@2762
   786
                               int radix) {
dvander@2762
   787
#ifndef HAVE_LONG_LONG
dvander@2762
   788
  return false;
dvander@2762
   789
#else
dvander@2762
   790
  if (n == 0) return false;
dvander@2762
   791
  char buf[kMaxNumberLength+1];
dvander@2762
   792
  str = TerminateNumber(buf, str, n);
dvander@2762
   793
  char* end;
dvander@2762
   794
  errno = 0;
dvander@2762
   795
#if defined HAVE_STRTOQ
dvander@2762
   796
  long long r = strtoq(str, &end, radix);
dvander@2762
   797
#elif defined HAVE_STRTOLL
dvander@2762
   798
  long long r = strtoll(str, &end, radix);
dvander@2762
   799
#elif defined HAVE__STRTOI64
dvander@2762
   800
  long long r = _strtoi64(str, &end, radix);
dvander@2762
   801
#else
dvander@2762
   802
#error parse_longlong_radix: cannot convert input to a long-long
dvander@2762
   803
#endif
dvander@2762
   804
  if (end != str + n) return false;   // Leftover junk
dvander@2762
   805
  if (errno) return false;
dvander@2762
   806
  if (dest == NULL) return true;
dvander@2762
   807
  *(reinterpret_cast<long long*>(dest)) = r;
dvander@2762
   808
  return true;
dvander@2762
   809
#endif   /* HAVE_LONG_LONG */
dvander@2762
   810
}
dvander@2762
   811
dvander@2762
   812
bool Arg::parse_ulonglong_radix(const char* str,
dvander@2762
   813
                                int n,
dvander@2762
   814
                                void* dest,
dvander@2762
   815
                                int radix) {
dvander@2762
   816
#ifndef HAVE_UNSIGNED_LONG_LONG
dvander@2762
   817
  return false;
dvander@2762
   818
#else
dvander@2762
   819
  if (n == 0) return false;
dvander@2762
   820
  char buf[kMaxNumberLength+1];
dvander@2762
   821
  str = TerminateNumber(buf, str, n);
dvander@2762
   822
  if (str[0] == '-') return false;    // strtoull() on a negative number?!
dvander@2762
   823
  char* end;
dvander@2762
   824
  errno = 0;
dvander@2762
   825
#if defined HAVE_STRTOQ
dvander@2762
   826
  unsigned long long r = strtouq(str, &end, radix);
dvander@2762
   827
#elif defined HAVE_STRTOLL
dvander@2762
   828
  unsigned long long r = strtoull(str, &end, radix);
dvander@2762
   829
#elif defined HAVE__STRTOI64
dvander@2762
   830
  unsigned long long r = _strtoui64(str, &end, radix);
dvander@2762
   831
#else
dvander@2762
   832
#error parse_ulonglong_radix: cannot convert input to a long-long
dvander@2762
   833
#endif
dvander@2762
   834
  if (end != str + n) return false;   // Leftover junk
dvander@2762
   835
  if (errno) return false;
dvander@2762
   836
  if (dest == NULL) return true;
dvander@2762
   837
  *(reinterpret_cast<unsigned long long*>(dest)) = r;
dvander@2762
   838
  return true;
dvander@2762
   839
#endif   /* HAVE_UNSIGNED_LONG_LONG */
dvander@2762
   840
}
dvander@2762
   841
dvander@2762
   842
bool Arg::parse_double(const char* str, int n, void* dest) {
dvander@2762
   843
  if (n == 0) return false;
dvander@2762
   844
  static const int kMaxLength = 200;
dvander@2762
   845
  char buf[kMaxLength];
dvander@2762
   846
  if (n >= kMaxLength) return false;
dvander@2762
   847
  memcpy(buf, str, n);
dvander@2762
   848
  buf[n] = '\0';
dvander@2762
   849
  errno = 0;
dvander@2762
   850
  char* end;
dvander@2762
   851
  double r = strtod(buf, &end);
dvander@2762
   852
  if (end != buf + n) return false;   // Leftover junk
dvander@2762
   853
  if (errno) return false;
dvander@2762
   854
  if (dest == NULL) return true;
dvander@2762
   855
  *(reinterpret_cast<double*>(dest)) = r;
dvander@2762
   856
  return true;
dvander@2762
   857
}
dvander@2762
   858
dvander@2762
   859
bool Arg::parse_float(const char* str, int n, void* dest) {
dvander@2762
   860
  double r;
dvander@2762
   861
  if (!parse_double(str, n, &r)) return false;
dvander@2762
   862
  if (dest == NULL) return true;
dvander@2762
   863
  *(reinterpret_cast<float*>(dest)) = static_cast<float>(r);
dvander@2762
   864
  return true;
dvander@2762
   865
}
dvander@2762
   866
dvander@2762
   867
dvander@2762
   868
#define DEFINE_INTEGER_PARSERS(name)                                    \
dvander@2762
   869
  bool Arg::parse_##name(const char* str, int n, void* dest) {          \
dvander@2762
   870
    return parse_##name##_radix(str, n, dest, 10);                      \
dvander@2762
   871
  }                                                                     \
dvander@2762
   872
  bool Arg::parse_##name##_hex(const char* str, int n, void* dest) {    \
dvander@2762
   873
    return parse_##name##_radix(str, n, dest, 16);                      \
dvander@2762
   874
  }                                                                     \
dvander@2762
   875
  bool Arg::parse_##name##_octal(const char* str, int n, void* dest) {  \
dvander@2762
   876
    return parse_##name##_radix(str, n, dest, 8);                       \
dvander@2762
   877
  }                                                                     \
dvander@2762
   878
  bool Arg::parse_##name##_cradix(const char* str, int n, void* dest) { \
dvander@2762
   879
    return parse_##name##_radix(str, n, dest, 0);                       \
dvander@2762
   880
  }
dvander@2762
   881
dvander@2762
   882
DEFINE_INTEGER_PARSERS(short)      /*                                   */
dvander@2762
   883
DEFINE_INTEGER_PARSERS(ushort)     /*                                   */
dvander@2762
   884
DEFINE_INTEGER_PARSERS(int)        /* Don't use semicolons after these  */
dvander@2762
   885
DEFINE_INTEGER_PARSERS(uint)       /* statements because they can cause */
dvander@2762
   886
DEFINE_INTEGER_PARSERS(long)       /* compiler warnings if the checking */
dvander@2762
   887
DEFINE_INTEGER_PARSERS(ulong)      /* level is turned up high enough.   */
dvander@2762
   888
DEFINE_INTEGER_PARSERS(longlong)   /*                                   */
dvander@2762
   889
DEFINE_INTEGER_PARSERS(ulonglong)  /*                                   */
dvander@2762
   890
dvander@2762
   891
#undef DEFINE_INTEGER_PARSERS
dvander@2762
   892
dvander@2762
   893
}   // namespace pcrecpp