Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 1 | // Copyright (c) 2010, 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 | |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 32 | #include <stdlib.h> |
| 33 | #include <stdio.h> |
| 34 | #include <ctype.h> |
| 35 | #include <limits.h> /* for SHRT_MIN, USHRT_MAX, etc */ |
| 36 | #include <string.h> /* for memcpy */ |
| 37 | #include <assert.h> |
| 38 | #include <errno.h> |
| 39 | #include <string> |
| 40 | #include <algorithm> |
| 41 | |
| 42 | #include "pcrecpp_internal.h" |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 43 | #include "pcre2.h" |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 44 | #include "pcrecpp.h" |
| 45 | #include "pcre_stringpiece.h" |
| 46 | |
| 47 | |
| 48 | namespace pcrecpp { |
| 49 | |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 50 | // If the user doesn't ask for any options, we just use this one |
| 51 | static RE_Options default_options; |
| 52 | |
| 53 | void RE::Init(const string& pat, const RE_Options* options) { |
| 54 | pattern_ = pat; |
| 55 | if (options == NULL) { |
| 56 | options_ = default_options; |
| 57 | } else { |
| 58 | options_ = *options; |
| 59 | } |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 60 | error_ = ""; |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 61 | re_full_ = NULL; |
| 62 | re_partial_ = NULL; |
| 63 | |
| 64 | re_partial_ = Compile(UNANCHORED); |
| 65 | if (re_partial_ != NULL) { |
| 66 | re_full_ = Compile(ANCHOR_BOTH); |
| 67 | } |
| 68 | } |
| 69 | |
| 70 | void RE::Cleanup() { |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 71 | if (re_full_ != NULL) pcre2_code_free(re_full_); |
| 72 | if (re_partial_ != NULL) pcre2_code_free(re_partial_); |
| 73 | error_ = ""; |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 74 | } |
| 75 | |
| 76 | |
| 77 | RE::~RE() { |
| 78 | Cleanup(); |
| 79 | } |
| 80 | |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 81 | static void format_pcre_error(int error, string & str) { |
| 82 | PCRE2_UCHAR8 buffer[256]; |
| 83 | auto rc = pcre2_get_error_message(error, buffer, 256); |
| 84 | str.assign(reinterpret_cast<string::value_type*>(buffer)); |
| 85 | if (rc == PCRE2_ERROR_NOMEMORY) { |
| 86 | str.append("..."); |
| 87 | } |
| 88 | } |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 89 | |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 90 | pcre2_code* RE::Compile(Anchor anchor) { |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 91 | // First, convert RE_Options into pcre options |
| 92 | int pcre_options = 0; |
| 93 | pcre_options = options_.all_options(); |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 94 | typedef std::unique_ptr<pcre2_compile_context, |
| 95 | decltype(pcre2_compile_context_free)*> compile_context_ptr; |
| 96 | compile_context_ptr compile_context(NULL, pcre2_compile_context_free); |
| 97 | |
| 98 | // As of pcre2 the newline mode must be passed through the compile context. |
| 99 | // So we only need one if the newline mode is actually set. |
| 100 | if (options_.newline_mode()) { |
| 101 | compile_context = compile_context_ptr(pcre2_compile_context_create(NULL), |
| 102 | pcre2_compile_context_free); |
| 103 | if (!compile_context) { |
| 104 | error_ = "Unable to allocate memory for pcre2_compile_congext"; |
| 105 | return NULL; |
| 106 | } |
| 107 | if (pcre2_set_newline(compile_context.get(), |
| 108 | options_.newline_mode()) == PCRE2_ERROR_BADDATA) { |
| 109 | error_ = "REOptions: bad newline mode given"; |
| 110 | return NULL; |
| 111 | } |
| 112 | } |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 113 | |
| 114 | // Special treatment for anchoring. This is needed because at |
| 115 | // runtime pcre only provides an option for anchoring at the |
| 116 | // beginning of a string (unless you use offset). |
| 117 | // |
| 118 | // There are three types of anchoring we want: |
| 119 | // UNANCHORED Compile the original pattern, and use |
| 120 | // a pcre unanchored match. |
| 121 | // ANCHOR_START Compile the original pattern, and use |
| 122 | // a pcre anchored match. |
| 123 | // ANCHOR_BOTH Tack a "\z" to the end of the original pattern |
| 124 | // and use a pcre anchored match. |
| 125 | |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 126 | int compile_error; |
| 127 | PCRE2_SIZE eoffset; |
| 128 | pcre2_code* re; |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 129 | if (anchor != ANCHOR_BOTH) { |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 130 | re = pcre2_compile(reinterpret_cast<PCRE2_SPTR>(pattern_.c_str()), |
| 131 | pattern_.length(), pcre_options, &compile_error, |
| 132 | &eoffset, compile_context.get()); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 133 | } else { |
| 134 | // Tack a '\z' at the end of RE. Parenthesize it first so that |
| 135 | // the '\z' applies to all top-level alternatives in the regexp. |
| 136 | string wrapped = "(?:"; // A non-counting grouping operator |
| 137 | wrapped += pattern_; |
| 138 | wrapped += ")\\z"; |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 139 | re = pcre2_compile(reinterpret_cast<PCRE2_SPTR>(wrapped.c_str()), |
| 140 | wrapped.length(), pcre_options, &compile_error, &eoffset, |
| 141 | compile_context.get()); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 142 | } |
| 143 | if (re == NULL) { |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 144 | format_pcre_error(compile_error, error_); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 145 | } |
| 146 | return re; |
| 147 | } |
| 148 | |
| 149 | /***** Matching interfaces *****/ |
| 150 | |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 151 | bool RE::Replace(const StringPiece& rewrite, |
| 152 | string *str) const { |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 153 | pcre2_match_data_ptr match_data; |
| 154 | int matches = TryMatch(*str, 0, UNANCHORED, true, match_data); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 155 | if (matches == 0) |
| 156 | return false; |
| 157 | |
| 158 | string s; |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 159 | if (!Rewrite(&s, rewrite, *str, match_data)) |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 160 | return false; |
| 161 | |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 162 | auto vec = pcre2_get_ovector_pointer(match_data.get()); |
| 163 | |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 164 | assert(vec[0] >= 0); |
| 165 | assert(vec[1] >= 0); |
| 166 | str->replace(vec[0], vec[1] - vec[0], s); |
| 167 | return true; |
| 168 | } |
| 169 | |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 170 | static bool is_multi_char_newline_mode(int value) { |
| 171 | switch (value) { |
| 172 | case PCRE2_NEWLINE_CR: |
| 173 | case PCRE2_NEWLINE_LF: |
| 174 | return false; |
| 175 | case PCRE2_NEWLINE_CRLF: |
| 176 | case PCRE2_NEWLINE_ANY: |
| 177 | case PCRE2_NEWLINE_ANYCRLF: |
| 178 | return true; |
| 179 | default: |
| 180 | return false; |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 181 | } |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 182 | } |
| 183 | |
| 184 | int RE::GlobalReplace(const StringPiece& rewrite, |
| 185 | string *str) const { |
| 186 | int count = 0; |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 187 | string out; |
| 188 | int start = 0; |
| 189 | bool last_match_was_empty_string = false; |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 190 | pcre2_match_data_ptr match_data; |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 191 | |
| 192 | while (start <= static_cast<int>(str->length())) { |
| 193 | // If the previous match was for the empty string, we shouldn't |
| 194 | // just match again: we'll match in the same way and get an |
| 195 | // infinite loop. Instead, we do the match in a special way: |
| 196 | // anchored -- to force another try at the same position -- |
| 197 | // and with a flag saying that this time, ignore empty matches. |
| 198 | // If this special match returns, that means there's a non-empty |
| 199 | // match at this position as well, and we can continue. If not, |
| 200 | // we do what perl does, and just advance by one. |
| 201 | // Notice that perl prints '@@@' for this; |
| 202 | // perl -le '$_ = "aa"; s/b*|aa/@/g; print' |
| 203 | int matches; |
| 204 | if (last_match_was_empty_string) { |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 205 | matches = TryMatch(*str, start, ANCHOR_START, false, match_data); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 206 | if (matches <= 0) { |
| 207 | int matchend = start + 1; // advance one character. |
| 208 | // If the current char is CR and we're in CRLF mode, skip LF too. |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 209 | // Note it's better to call pcre2_pattern_info() than to examine |
| 210 | // all_options(), since options_ could have changed between |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 211 | // compile-time and now, but this is simpler and safe enough. |
| 212 | // Modified by PH to add ANY and ANYCRLF. |
| 213 | if (matchend < static_cast<int>(str->length()) && |
| 214 | (*str)[start] == '\r' && (*str)[matchend] == '\n' && |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 215 | is_multi_char_newline_mode(options_.newline_mode())) { |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 216 | matchend++; |
| 217 | } |
| 218 | // We also need to advance more than one char if we're in utf8 mode. |
| 219 | #ifdef SUPPORT_UTF8 |
| 220 | if (options_.utf8()) { |
| 221 | while (matchend < static_cast<int>(str->length()) && |
| 222 | ((*str)[matchend] & 0xc0) == 0x80) |
| 223 | matchend++; |
| 224 | } |
| 225 | #endif |
| 226 | if (start < static_cast<int>(str->length())) |
| 227 | out.append(*str, start, matchend - start); |
| 228 | start = matchend; |
| 229 | last_match_was_empty_string = false; |
| 230 | continue; |
| 231 | } |
| 232 | } else { |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 233 | matches = TryMatch(*str, start, UNANCHORED, true, match_data); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 234 | if (matches <= 0) |
| 235 | break; |
| 236 | } |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 237 | auto vec = pcre2_get_ovector_pointer(match_data.get()); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 238 | int matchstart = vec[0], matchend = vec[1]; |
| 239 | assert(matchstart >= start); |
| 240 | assert(matchend >= matchstart); |
| 241 | out.append(*str, start, matchstart - start); |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 242 | Rewrite(&out, rewrite, *str, match_data); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 243 | start = matchend; |
| 244 | count++; |
| 245 | last_match_was_empty_string = (matchstart == matchend); |
| 246 | } |
| 247 | |
| 248 | if (count == 0) |
| 249 | return 0; |
| 250 | |
| 251 | if (start < static_cast<int>(str->length())) |
| 252 | out.append(*str, start, str->length() - start); |
| 253 | swap(out, *str); |
| 254 | return count; |
| 255 | } |
| 256 | |
| 257 | bool RE::Extract(const StringPiece& rewrite, |
| 258 | const StringPiece& text, |
| 259 | string *out) const { |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 260 | pcre2_match_data_ptr match_data; |
| 261 | int matches = TryMatch(text, 0, UNANCHORED, true, match_data); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 262 | if (matches == 0) |
| 263 | return false; |
| 264 | out->erase(); |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 265 | return Rewrite(out, rewrite, text, match_data); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 266 | } |
| 267 | |
| 268 | /*static*/ string RE::QuoteMeta(const StringPiece& unquoted) { |
| 269 | string result; |
| 270 | |
| 271 | // Escape any ascii character not in [A-Za-z_0-9]. |
| 272 | // |
| 273 | // Note that it's legal to escape a character even if it has no |
| 274 | // special meaning in a regular expression -- so this function does |
| 275 | // that. (This also makes it identical to the perl function of the |
| 276 | // same name; see `perldoc -f quotemeta`.) The one exception is |
| 277 | // escaping NUL: rather than doing backslash + NUL, like perl does, |
| 278 | // we do '\0', because pcre itself doesn't take embedded NUL chars. |
| 279 | for (int ii = 0; ii < unquoted.size(); ++ii) { |
| 280 | // Note that using 'isalnum' here raises the benchmark time from |
| 281 | // 32ns to 58ns: |
| 282 | if (unquoted[ii] == '\0') { |
| 283 | result += "\\0"; |
| 284 | } else if ((unquoted[ii] < 'a' || unquoted[ii] > 'z') && |
| 285 | (unquoted[ii] < 'A' || unquoted[ii] > 'Z') && |
| 286 | (unquoted[ii] < '0' || unquoted[ii] > '9') && |
| 287 | unquoted[ii] != '_' && |
| 288 | // If this is the part of a UTF8 or Latin1 character, we need |
| 289 | // to copy this byte without escaping. Experimentally this is |
| 290 | // what works correctly with the regexp library. |
| 291 | !(unquoted[ii] & 128)) { |
| 292 | result += '\\'; |
| 293 | result += unquoted[ii]; |
| 294 | } else { |
| 295 | result += unquoted[ii]; |
| 296 | } |
| 297 | } |
| 298 | |
| 299 | return result; |
| 300 | } |
| 301 | |
| 302 | /***** Actual matching and rewriting code *****/ |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 303 | int RE::TryMatch(const StringPiece& text, |
| 304 | int startpos, |
| 305 | Anchor anchor, |
| 306 | bool empty_ok, |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 307 | pcre2_match_data_ptr & match_data) const { |
| 308 | typedef std::unique_ptr<pcre2_match_context, |
| 309 | decltype(pcre2_match_context_free)*> match_context_ptr; |
| 310 | |
| 311 | pcre2_code* re = (anchor == ANCHOR_BOTH) ? re_full_ : re_partial_; |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 312 | if (re == NULL) { |
| 313 | //fprintf(stderr, "Matching against invalid re: %s\n", error_->c_str()); |
| 314 | return 0; |
| 315 | } |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 316 | match_context_ptr match_context = match_context_ptr( |
| 317 | pcre2_match_context_create(NULL), |
| 318 | pcre2_match_context_free); |
| 319 | if (!match_context) |
| 320 | return 0; |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 321 | |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 322 | if (options_.match_limit() > 0) { |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 323 | pcre2_set_match_limit(match_context.get(), options_.match_limit()); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 324 | } |
| 325 | if (options_.match_limit_recursion() > 0) { |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 326 | pcre2_set_recursion_limit(match_context.get(), |
| 327 | options_.match_limit_recursion()); |
| 328 | } |
| 329 | |
| 330 | match_data = pcre2_match_data_ptr( |
| 331 | pcre2_match_data_create_from_pattern(re, NULL), |
| 332 | pcre2_match_data_free); |
| 333 | if (!match_data) { |
| 334 | return 0; |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 335 | } |
| 336 | |
| 337 | // int options = 0; |
| 338 | // Changed by PH as a result of bugzilla #1288 |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 339 | int options = (options_.all_options() & PCRE2_NO_UTF_CHECK); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 340 | |
| 341 | if (anchor != UNANCHORED) |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 342 | options |= PCRE2_ANCHORED; |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 343 | if (!empty_ok) |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 344 | options |= PCRE2_NOTEMPTY; |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 345 | |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 346 | int rc = pcre2_match( |
| 347 | re, reinterpret_cast<PCRE2_SPTR>((text.empty()) ? "" : text.data()), |
| 348 | text.size(), startpos, options, match_data.get(), match_context.get()); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 349 | |
| 350 | // Handle errors |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 351 | if (rc == PCRE2_ERROR_NOMATCH) { |
| 352 | return 0; |
| 353 | } |
| 354 | if (rc == PCRE2_ERROR_PARTIAL) { |
| 355 | // not sure what to do with partial yet |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 356 | return 0; |
| 357 | } else if (rc < 0) { |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 358 | // For any other error condition also return 0. |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 359 | return 0; |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 360 | } |
| 361 | |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 362 | return rc; // return number of matches found |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 363 | } |
| 364 | |
| 365 | bool RE::DoMatchImpl(const StringPiece& text, |
| 366 | Anchor anchor, |
| 367 | int* consumed, |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 368 | const Arg* args, |
| 369 | int n) const { |
| 370 | pcre2_match_data_ptr match_data; |
| 371 | int matches = TryMatch(text, 0, anchor, true, match_data); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 372 | assert(matches >= 0); // TryMatch never returns negatives |
| 373 | if (matches == 0) |
| 374 | return false; |
| 375 | |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 376 | auto vec = pcre2_get_ovector_pointer(match_data.get()); |
| 377 | |
| 378 | // allow for NULL |
| 379 | if (consumed != NULL) |
| 380 | *consumed = vec[1]; |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 381 | |
| 382 | if (n == 0 || args == NULL) { |
| 383 | // We are not interested in results |
| 384 | return true; |
| 385 | } |
| 386 | |
| 387 | if (NumberOfCapturingGroups() < n) { |
| 388 | // RE has fewer capturing groups than number of arg pointers passed in |
| 389 | return false; |
| 390 | } |
| 391 | |
| 392 | // If we got here, we must have matched the whole pattern. |
| 393 | // We do not need (can not do) any more checks on the value of 'matches' here |
| 394 | // -- see the comment for TryMatch. |
| 395 | for (int i = 0; i < n; i++) { |
| 396 | const int start = vec[2*(i+1)]; |
| 397 | const int limit = vec[2*(i+1)+1]; |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 398 | if (!args[i].Parse(text.data() + start, limit - start)) { |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 399 | // TODO: Should we indicate what the error was? |
| 400 | return false; |
| 401 | } |
| 402 | } |
| 403 | |
| 404 | return true; |
| 405 | } |
| 406 | |
| 407 | bool RE::DoMatch(const StringPiece& text, |
| 408 | Anchor anchor, |
| 409 | int* consumed, |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 410 | Arg const args[], |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 411 | int n) const { |
| 412 | assert(n >= 0); |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 413 | bool retval = DoMatchImpl(text, anchor, consumed, args, n); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 414 | return retval; |
| 415 | } |
| 416 | |
| 417 | bool RE::Rewrite(string *out, const StringPiece &rewrite, |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 418 | const StringPiece &text, |
| 419 | pcre2_match_data_ptr const & match_data) const { |
| 420 | auto veclen = pcre2_get_ovector_count(match_data.get()); |
| 421 | auto vec = pcre2_get_ovector_pointer(match_data.get()); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 422 | for (const char *s = rewrite.data(), *end = s + rewrite.size(); |
| 423 | s < end; s++) { |
| 424 | int c = *s; |
| 425 | if (c == '\\') { |
| 426 | c = *++s; |
| 427 | if (isdigit(c)) { |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 428 | decltype(veclen) n = (c - '0'); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 429 | if (n >= veclen) { |
| 430 | //fprintf(stderr, requested group %d in regexp %.*s\n", |
| 431 | // n, rewrite.size(), rewrite.data()); |
| 432 | return false; |
| 433 | } |
| 434 | int start = vec[2 * n]; |
| 435 | if (start >= 0) |
| 436 | out->append(text.data() + start, vec[2 * n + 1] - start); |
| 437 | } else if (c == '\\') { |
| 438 | *out += '\\'; |
| 439 | } else { |
| 440 | //fprintf(stderr, "invalid rewrite pattern: %.*s\n", |
| 441 | // rewrite.size(), rewrite.data()); |
| 442 | return false; |
| 443 | } |
| 444 | } else { |
| 445 | *out += c; |
| 446 | } |
| 447 | } |
| 448 | return true; |
| 449 | } |
| 450 | |
| 451 | // Return the number of capturing subpatterns, or -1 if the |
| 452 | // regexp wasn't valid on construction. |
| 453 | int RE::NumberOfCapturingGroups() const { |
| 454 | if (re_partial_ == NULL) return -1; |
| 455 | |
| 456 | int result; |
Janis Danisevskis | 1fd6dd5 | 2016-04-01 12:25:25 +0100 | [diff] [blame] | 457 | int pcre_retval = pcre2_pattern_info(re_partial_, PCRE2_INFO_CAPTURECOUNT, |
| 458 | &result); |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 459 | assert(pcre_retval == 0); |
| 460 | return result; |
| 461 | } |
| 462 | |
| 463 | /***** Parsers for various types *****/ |
| 464 | |
| 465 | bool Arg::parse_null(const char* str, int n, void* dest) { |
| 466 | (void)str; |
| 467 | (void)n; |
| 468 | // We fail if somebody asked us to store into a non-NULL void* pointer |
| 469 | return (dest == NULL); |
| 470 | } |
| 471 | |
| 472 | bool Arg::parse_string(const char* str, int n, void* dest) { |
| 473 | if (dest == NULL) return true; |
| 474 | reinterpret_cast<string*>(dest)->assign(str, n); |
| 475 | return true; |
| 476 | } |
| 477 | |
| 478 | bool Arg::parse_stringpiece(const char* str, int n, void* dest) { |
| 479 | if (dest == NULL) return true; |
| 480 | reinterpret_cast<StringPiece*>(dest)->set(str, n); |
| 481 | return true; |
| 482 | } |
| 483 | |
| 484 | bool Arg::parse_char(const char* str, int n, void* dest) { |
| 485 | if (n != 1) return false; |
| 486 | if (dest == NULL) return true; |
| 487 | *(reinterpret_cast<char*>(dest)) = str[0]; |
| 488 | return true; |
| 489 | } |
| 490 | |
| 491 | bool Arg::parse_uchar(const char* str, int n, void* dest) { |
| 492 | if (n != 1) return false; |
| 493 | if (dest == NULL) return true; |
| 494 | *(reinterpret_cast<unsigned char*>(dest)) = str[0]; |
| 495 | return true; |
| 496 | } |
| 497 | |
| 498 | // Largest number spec that we are willing to parse |
| 499 | static const int kMaxNumberLength = 32; |
| 500 | |
| 501 | // REQUIRES "buf" must have length at least kMaxNumberLength+1 |
| 502 | // REQUIRES "n > 0" |
| 503 | // Copies "str" into "buf" and null-terminates if necessary. |
| 504 | // Returns one of: |
| 505 | // a. "str" if no termination is needed |
| 506 | // b. "buf" if the string was copied and null-terminated |
| 507 | // c. "" if the input was invalid and has no hope of being parsed |
| 508 | static const char* TerminateNumber(char* buf, const char* str, int n) { |
| 509 | if ((n > 0) && isspace(*str)) { |
| 510 | // We are less forgiving than the strtoxxx() routines and do not |
| 511 | // allow leading spaces. |
| 512 | return ""; |
| 513 | } |
| 514 | |
| 515 | // See if the character right after the input text may potentially |
| 516 | // look like a digit. |
| 517 | if (isdigit(str[n]) || |
| 518 | ((str[n] >= 'a') && (str[n] <= 'f')) || |
| 519 | ((str[n] >= 'A') && (str[n] <= 'F'))) { |
| 520 | if (n > kMaxNumberLength) return ""; // Input too big to be a valid number |
| 521 | memcpy(buf, str, n); |
| 522 | buf[n] = '\0'; |
| 523 | return buf; |
| 524 | } else { |
| 525 | // We can parse right out of the supplied string, so return it. |
| 526 | return str; |
| 527 | } |
| 528 | } |
| 529 | |
| 530 | bool Arg::parse_long_radix(const char* str, |
| 531 | int n, |
| 532 | void* dest, |
| 533 | int radix) { |
| 534 | if (n == 0) return false; |
| 535 | char buf[kMaxNumberLength+1]; |
| 536 | str = TerminateNumber(buf, str, n); |
| 537 | char* end; |
| 538 | errno = 0; |
| 539 | long r = strtol(str, &end, radix); |
| 540 | if (end != str + n) return false; // Leftover junk |
| 541 | if (errno) return false; |
| 542 | if (dest == NULL) return true; |
| 543 | *(reinterpret_cast<long*>(dest)) = r; |
| 544 | return true; |
| 545 | } |
| 546 | |
| 547 | bool Arg::parse_ulong_radix(const char* str, |
| 548 | int n, |
| 549 | void* dest, |
| 550 | int radix) { |
| 551 | if (n == 0) return false; |
| 552 | char buf[kMaxNumberLength+1]; |
| 553 | str = TerminateNumber(buf, str, n); |
| 554 | if (str[0] == '-') return false; // strtoul() on a negative number?! |
| 555 | char* end; |
| 556 | errno = 0; |
| 557 | unsigned long r = strtoul(str, &end, radix); |
| 558 | if (end != str + n) return false; // Leftover junk |
| 559 | if (errno) return false; |
| 560 | if (dest == NULL) return true; |
| 561 | *(reinterpret_cast<unsigned long*>(dest)) = r; |
| 562 | return true; |
| 563 | } |
| 564 | |
| 565 | bool Arg::parse_short_radix(const char* str, |
| 566 | int n, |
| 567 | void* dest, |
| 568 | int radix) { |
| 569 | long r; |
| 570 | if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse |
| 571 | if (r < SHRT_MIN || r > SHRT_MAX) return false; // Out of range |
| 572 | if (dest == NULL) return true; |
| 573 | *(reinterpret_cast<short*>(dest)) = static_cast<short>(r); |
| 574 | return true; |
| 575 | } |
| 576 | |
| 577 | bool Arg::parse_ushort_radix(const char* str, |
| 578 | int n, |
| 579 | void* dest, |
| 580 | int radix) { |
| 581 | unsigned long r; |
| 582 | if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse |
| 583 | if (r > USHRT_MAX) return false; // Out of range |
| 584 | if (dest == NULL) return true; |
| 585 | *(reinterpret_cast<unsigned short*>(dest)) = static_cast<unsigned short>(r); |
| 586 | return true; |
| 587 | } |
| 588 | |
| 589 | bool Arg::parse_int_radix(const char* str, |
| 590 | int n, |
| 591 | void* dest, |
| 592 | int radix) { |
| 593 | long r; |
| 594 | if (!parse_long_radix(str, n, &r, radix)) return false; // Could not parse |
| 595 | if (r < INT_MIN || r > INT_MAX) return false; // Out of range |
| 596 | if (dest == NULL) return true; |
| 597 | *(reinterpret_cast<int*>(dest)) = r; |
| 598 | return true; |
| 599 | } |
| 600 | |
| 601 | bool Arg::parse_uint_radix(const char* str, |
| 602 | int n, |
| 603 | void* dest, |
| 604 | int radix) { |
| 605 | unsigned long r; |
| 606 | if (!parse_ulong_radix(str, n, &r, radix)) return false; // Could not parse |
| 607 | if (r > UINT_MAX) return false; // Out of range |
| 608 | if (dest == NULL) return true; |
| 609 | *(reinterpret_cast<unsigned int*>(dest)) = r; |
| 610 | return true; |
| 611 | } |
| 612 | |
| 613 | bool Arg::parse_longlong_radix(const char* str, |
| 614 | int n, |
| 615 | void* dest, |
| 616 | int radix) { |
| 617 | #ifndef HAVE_LONG_LONG |
| 618 | return false; |
| 619 | #else |
| 620 | if (n == 0) return false; |
| 621 | char buf[kMaxNumberLength+1]; |
| 622 | str = TerminateNumber(buf, str, n); |
| 623 | char* end; |
| 624 | errno = 0; |
| 625 | #if defined HAVE_STRTOQ |
| 626 | long long r = strtoq(str, &end, radix); |
| 627 | #elif defined HAVE_STRTOLL |
| 628 | long long r = strtoll(str, &end, radix); |
| 629 | #elif defined HAVE__STRTOI64 |
| 630 | long long r = _strtoi64(str, &end, radix); |
| 631 | #elif defined HAVE_STRTOIMAX |
| 632 | long long r = strtoimax(str, &end, radix); |
| 633 | #else |
| 634 | #error parse_longlong_radix: cannot convert input to a long-long |
| 635 | #endif |
| 636 | if (end != str + n) return false; // Leftover junk |
| 637 | if (errno) return false; |
| 638 | if (dest == NULL) return true; |
| 639 | *(reinterpret_cast<long long*>(dest)) = r; |
| 640 | return true; |
| 641 | #endif /* HAVE_LONG_LONG */ |
| 642 | } |
| 643 | |
| 644 | bool Arg::parse_ulonglong_radix(const char* str, |
| 645 | int n, |
| 646 | void* dest, |
| 647 | int radix) { |
| 648 | #ifndef HAVE_UNSIGNED_LONG_LONG |
Nick Kralevich | f73ff17 | 2014-09-27 12:41:49 -0700 | [diff] [blame] | 649 | return false; |
| 650 | #else |
| 651 | if (n == 0) return false; |
| 652 | char buf[kMaxNumberLength+1]; |
| 653 | str = TerminateNumber(buf, str, n); |
| 654 | if (str[0] == '-') return false; // strtoull() on a negative number?! |
| 655 | char* end; |
| 656 | errno = 0; |
| 657 | #if defined HAVE_STRTOQ |
| 658 | unsigned long long r = strtouq(str, &end, radix); |
| 659 | #elif defined HAVE_STRTOLL |
| 660 | unsigned long long r = strtoull(str, &end, radix); |
| 661 | #elif defined HAVE__STRTOI64 |
| 662 | unsigned long long r = _strtoui64(str, &end, radix); |
| 663 | #elif defined HAVE_STRTOIMAX |
| 664 | unsigned long long r = strtoumax(str, &end, radix); |
| 665 | #else |
| 666 | #error parse_ulonglong_radix: cannot convert input to a long-long |
| 667 | #endif |
| 668 | if (end != str + n) return false; // Leftover junk |
| 669 | if (errno) return false; |
| 670 | if (dest == NULL) return true; |
| 671 | *(reinterpret_cast<unsigned long long*>(dest)) = r; |
| 672 | return true; |
| 673 | #endif /* HAVE_UNSIGNED_LONG_LONG */ |
| 674 | } |
| 675 | |
| 676 | bool Arg::parse_double(const char* str, int n, void* dest) { |
| 677 | if (n == 0) return false; |
| 678 | static const int kMaxLength = 200; |
| 679 | char buf[kMaxLength]; |
| 680 | if (n >= kMaxLength) return false; |
| 681 | memcpy(buf, str, n); |
| 682 | buf[n] = '\0'; |
| 683 | errno = 0; |
| 684 | char* end; |
| 685 | double r = strtod(buf, &end); |
| 686 | if (end != buf + n) return false; // Leftover junk |
| 687 | if (errno) return false; |
| 688 | if (dest == NULL) return true; |
| 689 | *(reinterpret_cast<double*>(dest)) = r; |
| 690 | return true; |
| 691 | } |
| 692 | |
| 693 | bool Arg::parse_float(const char* str, int n, void* dest) { |
| 694 | double r; |
| 695 | if (!parse_double(str, n, &r)) return false; |
| 696 | if (dest == NULL) return true; |
| 697 | *(reinterpret_cast<float*>(dest)) = static_cast<float>(r); |
| 698 | return true; |
| 699 | } |
| 700 | |
| 701 | |
| 702 | #define DEFINE_INTEGER_PARSERS(name) \ |
| 703 | bool Arg::parse_##name(const char* str, int n, void* dest) { \ |
| 704 | return parse_##name##_radix(str, n, dest, 10); \ |
| 705 | } \ |
| 706 | bool Arg::parse_##name##_hex(const char* str, int n, void* dest) { \ |
| 707 | return parse_##name##_radix(str, n, dest, 16); \ |
| 708 | } \ |
| 709 | bool Arg::parse_##name##_octal(const char* str, int n, void* dest) { \ |
| 710 | return parse_##name##_radix(str, n, dest, 8); \ |
| 711 | } \ |
| 712 | bool Arg::parse_##name##_cradix(const char* str, int n, void* dest) { \ |
| 713 | return parse_##name##_radix(str, n, dest, 0); \ |
| 714 | } |
| 715 | |
| 716 | DEFINE_INTEGER_PARSERS(short) /* */ |
| 717 | DEFINE_INTEGER_PARSERS(ushort) /* */ |
| 718 | DEFINE_INTEGER_PARSERS(int) /* Don't use semicolons after these */ |
| 719 | DEFINE_INTEGER_PARSERS(uint) /* statements because they can cause */ |
| 720 | DEFINE_INTEGER_PARSERS(long) /* compiler warnings if the checking */ |
| 721 | DEFINE_INTEGER_PARSERS(ulong) /* level is turned up high enough. */ |
| 722 | DEFINE_INTEGER_PARSERS(longlong) /* */ |
| 723 | DEFINE_INTEGER_PARSERS(ulonglong) /* */ |
| 724 | |
| 725 | #undef DEFINE_INTEGER_PARSERS |
| 726 | |
| 727 | } // namespace pcrecpp |