1 // class template regex -*- C++ -*-
3 // Copyright (C) 2013-2014 Free Software Foundation, Inc.
5 // This file is part of the GNU ISO C++ Library. This library is free
6 // software; you can redistribute it and/or modify it under the
7 // terms of the GNU General Public License as published by the
8 // Free Software Foundation; either version 3, or (at your option)
11 // This library is distributed in the hope that it will be useful,
12 // but WITHOUT ANY WARRANTY; without even the implied warranty of
13 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 // GNU General Public License for more details.
16 // Under Section 7 of GPL version 3, you are granted additional
17 // permissions described in the GCC Runtime Library Exception, version
18 // 3.1, as published by the Free Software Foundation.
20 // You should have received a copy of the GNU General Public License and
21 // a copy of the GCC Runtime Library Exception along with this program;
22 // see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
23 // <http://www.gnu.org/licenses/>.
26 * @file bits/regex_scanner.tcc
27 * This is an internal header file, included by other library headers.
28 * Do not attempt to use it directly. @headername{regex}
31 // FIXME make comments doxygen format.
33 // N3376 specified 6 regex styles: ECMAScript, basic, extended, grep, egrep
35 // 1) grep is basic except '\n' is treated as '|'
36 // 2) egrep is extended except '\n' is treated as '|'
37 // 3) awk is extended except special escaping rules, and there's no
42 // ECMAScript: ECMA-262 15.10
45 // http://pubs.opengroup.org/onlinepubs/009695399/basedefs/xbd_chap09.html
47 // awk: http://pubs.opengroup.org/onlinepubs/000095399/utilities/awk.html
49 namespace std _GLIBCXX_VISIBILITY(default)
53 _GLIBCXX_BEGIN_NAMESPACE_VERSION
55 template<typename _CharT>
57 _Scanner(typename _Scanner::_IterT __begin,
58 typename _Scanner::_IterT __end,
59 _FlagT __flags, std::locale __loc)
60 : _ScannerBase(__flags),
61 _M_current(__begin), _M_end(__end),
62 _M_ctype(std::use_facet<_CtypeT>(__loc)),
63 _M_eat_escape(_M_is_ecma()
64 ? &_Scanner::_M_eat_escape_ecma
65 : &_Scanner::_M_eat_escape_posix)
68 template<typename _CharT>
73 if (_M_current == _M_end)
75 _M_token = _S_token_eof;
79 if (_M_state == _S_state_normal)
81 else if (_M_state == _S_state_in_bracket)
83 else if (_M_state == _S_state_in_brace)
87 _GLIBCXX_DEBUG_ASSERT(false);
91 // Differences between styles:
92 // 1) "\(", "\)", "\{" in basic. It's not escaping.
93 // 2) "(?:", "(?=", "(?!" in ECMAScript.
94 template<typename _CharT>
99 auto __c = *_M_current++;
104 if (_M_current == _M_end)
105 __throw_regex_error(regex_constants::error_escape);
108 || (*_M_current != '('
109 && *_M_current != ')'
110 && *_M_current != '{'))
112 (this->*_M_eat_escape)();
119 if (_M_is_ecma() && *_M_current == '?')
121 if (++_M_current == _M_end)
122 __throw_regex_error(regex_constants::error_paren);
124 if (*_M_current == ':')
127 _M_token = _S_token_subexpr_no_group_begin;
129 else if (*_M_current == '=')
132 _M_token = _S_token_subexpr_lookahead_begin;
133 _M_value.assign(1, 'p');
135 else if (*_M_current == '!')
138 _M_token = _S_token_subexpr_lookahead_begin;
139 _M_value.assign(1, 'n');
142 __throw_regex_error(regex_constants::error_paren);
144 else if (_M_flags & regex_constants::nosubs)
145 _M_token = _S_token_subexpr_no_group_begin;
147 _M_token = _S_token_subexpr_begin;
150 _M_token = _S_token_subexpr_end;
153 _M_state = _S_state_in_bracket;
154 _M_at_bracket_start = true;
155 if (_M_current != _M_end && *_M_current == '^')
157 _M_token = _S_token_bracket_neg_begin;
161 _M_token = _S_token_bracket_begin;
165 _M_state = _S_state_in_brace;
166 _M_token = _S_token_interval_begin;
168 else if (((__pos = std::strchr(_M_spec_char, _M_ctype.narrow(__c, '\0')))
173 || (_M_is_grep() && __c == '\n'))
175 auto __it = _M_token_tbl;
176 auto __narrowc = _M_ctype.narrow(__c, '\0');
177 for (; __it->first != '\0'; ++__it)
178 if (__it->first == __narrowc)
180 _M_token = __it->second;
183 _GLIBCXX_DEBUG_ASSERT(false);
187 _M_token = _S_token_ord_char;
188 _M_value.assign(1, __c);
192 // Differences between styles:
193 // 1) different semantics of "[]" and "[^]".
194 // 2) Escaping in bracket expr.
195 template<typename _CharT>
200 if (_M_current == _M_end)
201 __throw_regex_error(regex_constants::error_brack);
203 auto __c = *_M_current++;
207 if (_M_current == _M_end)
208 __throw_regex_error(regex_constants::error_brack);
210 if (*_M_current == '.')
212 _M_token = _S_token_collsymbol;
213 _M_eat_class(*_M_current++);
215 else if (*_M_current == ':')
217 _M_token = _S_token_char_class_name;
218 _M_eat_class(*_M_current++);
220 else if (*_M_current == '=')
222 _M_token = _S_token_equiv_class_name;
223 _M_eat_class(*_M_current++);
227 _M_token = _S_token_ord_char;
228 _M_value.assign(1, __c);
231 // In POSIX, when encountering "[]" or "[^]", the ']' is interpreted
232 // literally. So "[]]" and "[^]]" are valid regexes. See the testcases
233 // `*/empty_range.cc`.
234 else if (__c == ']' && (_M_is_ecma() || !_M_at_bracket_start))
236 _M_token = _S_token_bracket_end;
237 _M_state = _S_state_normal;
239 // ECMAScript and awk permits escaping in bracket.
240 else if (__c == '\\' && (_M_is_ecma() || _M_is_awk()))
241 (this->*_M_eat_escape)();
244 _M_token = _S_token_ord_char;
245 _M_value.assign(1, __c);
247 _M_at_bracket_start = false;
250 // Differences between styles:
251 // 1) "\}" in basic style.
252 template<typename _CharT>
257 if (_M_current == _M_end)
258 __throw_regex_error(regex_constants::error_brace);
260 auto __c = *_M_current++;
262 if (_M_ctype.is(_CtypeT::digit, __c))
264 _M_token = _S_token_dup_count;
265 _M_value.assign(1, __c);
266 while (_M_current != _M_end
267 && _M_ctype.is(_CtypeT::digit, *_M_current))
268 _M_value += *_M_current++;
271 _M_token = _S_token_comma;
273 else if (_M_is_basic())
275 if (__c == '\\' && _M_current != _M_end && *_M_current == '}')
277 _M_state = _S_state_normal;
278 _M_token = _S_token_interval_end;
282 __throw_regex_error(regex_constants::error_badbrace);
286 _M_state = _S_state_normal;
287 _M_token = _S_token_interval_end;
290 __throw_regex_error(regex_constants::error_badbrace);
293 template<typename _CharT>
298 if (_M_current == _M_end)
299 __throw_regex_error(regex_constants::error_escape);
301 auto __c = *_M_current++;
302 auto __pos = _M_find_escape(_M_ctype.narrow(__c, '\0'));
304 if (__pos != nullptr && (__c != 'b' || _M_state == _S_state_in_bracket))
306 _M_token = _S_token_ord_char;
307 _M_value.assign(1, *__pos);
311 _M_token = _S_token_word_bound;
312 _M_value.assign(1, 'p');
316 _M_token = _S_token_word_bound;
317 _M_value.assign(1, 'n');
327 _M_token = _S_token_quoted_class;
328 _M_value.assign(1, __c);
332 if (_M_current == _M_end)
333 __throw_regex_error(regex_constants::error_escape);
334 _M_token = _S_token_ord_char;
335 _M_value.assign(1, *_M_current++);
337 else if (__c == 'x' || __c == 'u')
340 for (int __i = 0; __i < (__c == 'x' ? 2 : 4); __i++)
342 if (_M_current == _M_end
343 || !_M_ctype.is(_CtypeT::xdigit, *_M_current))
344 __throw_regex_error(regex_constants::error_escape);
345 _M_value += *_M_current++;
347 _M_token = _S_token_hex_num;
349 // ECMAScript recognizes multi-digit back-references.
350 else if (_M_ctype.is(_CtypeT::digit, __c))
352 _M_value.assign(1, __c);
353 while (_M_current != _M_end
354 && _M_ctype.is(_CtypeT::digit, *_M_current))
355 _M_value += *_M_current++;
356 _M_token = _S_token_backref;
360 _M_token = _S_token_ord_char;
361 _M_value.assign(1, __c);
365 // Differences between styles:
366 // 1) Extended doesn't support backref, but basic does.
367 template<typename _CharT>
370 _M_eat_escape_posix()
372 if (_M_current == _M_end)
373 __throw_regex_error(regex_constants::error_escape);
375 auto __c = *_M_current;
376 auto __pos = std::strchr(_M_spec_char, _M_ctype.narrow(__c, '\0'));
378 if (__pos != nullptr && *__pos != '\0')
380 _M_token = _S_token_ord_char;
381 _M_value.assign(1, __c);
383 // We MUST judge awk before handling backrefs. There's no backref in awk.
384 else if (_M_is_awk())
389 else if (_M_is_basic() && _M_ctype.is(_CtypeT::digit, __c) && __c != '0')
391 _M_token = _S_token_backref;
392 _M_value.assign(1, __c);
396 #ifdef __STRICT_ANSI__
397 // POSIX says it is undefined to escape ordinary characters
398 __throw_regex_error(regex_constants::error_escape);
400 _M_token = _S_token_ord_char;
401 _M_value.assign(1, __c);
407 template<typename _CharT>
412 auto __c = *_M_current++;
413 auto __pos = _M_find_escape(_M_ctype.narrow(__c, '\0'));
415 if (__pos != nullptr)
417 _M_token = _S_token_ord_char;
418 _M_value.assign(1, *__pos);
420 // \ddd for oct representation
421 else if (_M_ctype.is(_CtypeT::digit, __c)
425 _M_value.assign(1, __c);
428 && _M_current != _M_end
429 && _M_ctype.is(_CtypeT::digit, *_M_current)
430 && *_M_current != '8'
431 && *_M_current != '9';
433 _M_value += *_M_current++;
434 _M_token = _S_token_oct_num;
438 __throw_regex_error(regex_constants::error_escape);
441 // Eats a character class or throws an exception.
442 // __ch could be ':', '.' or '=', _M_current is the char after ']' when
444 template<typename _CharT>
447 _M_eat_class(char __ch)
449 for (_M_value.clear(); _M_current != _M_end && *_M_current != __ch;)
450 _M_value += *_M_current++;
451 if (_M_current == _M_end
452 || *_M_current++ != __ch
453 || _M_current == _M_end // skip __ch
454 || *_M_current++ != ']') // skip ']'
457 __throw_regex_error(regex_constants::error_ctype);
459 __throw_regex_error(regex_constants::error_collate);
463 #ifdef _GLIBCXX_DEBUG
464 template<typename _CharT>
467 _M_print(std::ostream& ostr)
471 case _S_token_anychar:
472 ostr << "any-character\n";
474 case _S_token_backref:
477 case _S_token_bracket_begin:
478 ostr << "bracket-begin\n";
480 case _S_token_bracket_neg_begin:
481 ostr << "bracket-neg-begin\n";
483 case _S_token_bracket_end:
484 ostr << "bracket-end\n";
486 case _S_token_char_class_name:
487 ostr << "char-class-name \"" << _M_value << "\"\n";
489 case _S_token_closure0:
490 ostr << "closure0\n";
492 case _S_token_closure1:
493 ostr << "closure1\n";
495 case _S_token_collsymbol:
496 ostr << "collsymbol \"" << _M_value << "\"\n";
501 case _S_token_dup_count:
502 ostr << "dup count: " << _M_value << "\n";
507 case _S_token_equiv_class_name:
508 ostr << "equiv-class-name \"" << _M_value << "\"\n";
510 case _S_token_interval_begin:
511 ostr << "interval begin\n";
513 case _S_token_interval_end:
514 ostr << "interval end\n";
516 case _S_token_line_begin:
517 ostr << "line begin\n";
519 case _S_token_line_end:
520 ostr << "line end\n";
528 case _S_token_ord_char:
529 ostr << "ordinary character: \"" << _M_value << "\"\n";
531 case _S_token_subexpr_begin:
532 ostr << "subexpr begin\n";
534 case _S_token_subexpr_no_group_begin:
535 ostr << "no grouping subexpr begin\n";
537 case _S_token_subexpr_lookahead_begin:
538 ostr << "lookahead subexpr begin\n";
540 case _S_token_subexpr_end:
541 ostr << "subexpr end\n";
543 case _S_token_unknown:
544 ostr << "-- unknown token --\n";
546 case _S_token_oct_num:
547 ostr << "oct number " << _M_value << "\n";
549 case _S_token_hex_num:
550 ostr << "hex number " << _M_value << "\n";
552 case _S_token_quoted_class:
553 ostr << "quoted class " << "\\" << _M_value << "\n";
556 _GLIBCXX_DEBUG_ASSERT(false);
562 _GLIBCXX_END_NAMESPACE_VERSION
563 } // namespace __detail