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