libstdc++-v3/include/bits/regex_scanner.tcc

// class template regex -*- C++ -*-

// Copyright (C) 2013-2026 Free Software Foundation, Inc.
//
// This file is part of the GNU ISO C++ Library.  This library is free
// software; you can redistribute it and/or modify it under the
// terms of the GNU General Public License as published by the
// Free Software Foundation; either version 3, or (at your option)
// any later version.

// This library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
// GNU General Public License for more details.

// Under Section 7 of GPL version 3, you are granted additional
// permissions described in the GCC Runtime Library Exception, version
// 3.1, as published by the Free Software Foundation.

// You should have received a copy of the GNU General Public License and
// a copy of the GCC Runtime Library Exception along with this program;
// see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see
// <http://www.gnu.org/licenses/>.

/**
 *  @file bits/regex_scanner.tcc
 *  This is an internal header file, included by other library headers.
 *  Do not attempt to use it directly. @headername{regex}
 */

// FIXME make comments doxygen format.

// N3376 specified 6 regex styles: ECMAScript, basic, extended, grep, egrep
// and awk
// 1) grep is basic except '\n' is treated as '|'
// 2) egrep is extended except '\n' is treated as '|'
// 3) awk is extended except special escaping rules, and there's no
//    back-reference.
//
// References:
//
// ECMAScript: ECMA-262 15.10
//
// basic, extended:
// http://pubs.opengroup.org/onlinepubs/009695399/basedefs/xbd_chap09.html
//
// awk: http://pubs.opengroup.org/onlinepubs/000095399/utilities/awk.html

namespace std _GLIBCXX_VISIBILITY(default)
{
_GLIBCXX_BEGIN_NAMESPACE_VERSION

namespace __detail
{
  template<typename _CharT>
    _Scanner<_CharT>::
    _Scanner(const _CharT* __begin, const _CharT* __end,
	     _FlagT __flags, std::locale __loc)
    : _ScannerBase(__flags),
      _M_current(__begin), _M_end(__end),
      _M_ctype(std::use_facet<_CtypeT>(__loc)),
      _M_eat_escape(_M_is_ecma()
		    ? &_Scanner::_M_eat_escape_ecma
		    : &_Scanner::_M_eat_escape_posix)
    { _M_advance(); }

  template<typename _CharT>
    void
    _Scanner<_CharT>::
    _M_advance()
    {
      if (_M_current == _M_end)
	{
	  _M_token = _S_token_eof;
	  return;
	}

      if (_M_state == _S_state_normal)
	_M_scan_normal();
      else if (_M_state == _S_state_in_bracket)
	_M_scan_in_bracket();
      else if (_M_state == _S_state_in_brace)
	_M_scan_in_brace();
      else
	{
	  __glibcxx_assert(!"unexpected state while processing regex");
	}
    }

  // Differences between styles:
  // 1) "\(", "\)", "\{" in basic. It's not escaping.
  // 2) "(?:", "(?=", "(?!" in ECMAScript.
  template<typename _CharT>
    void
    _Scanner<_CharT>::
    _M_scan_normal()
    {
      auto __c = *_M_current++;

      if (__builtin_strchr(_M_spec_char, _M_ctype.narrow(__c, ' ')) == nullptr)
	{
	  _M_token = _S_token_ord_char;
	  _M_value.assign(1, __c);
	  return;
	}
      if (__c == '\\')
	{
	  if (_M_current == _M_end)
	    __throw_regex_error(
	      regex_constants::error_escape,
	      "Invalid escape at end of regular expression");

	  if (!_M_is_basic()
	      || (*_M_current != '('
		  && *_M_current != ')'
		  && *_M_current != '{'))
	    {
	      (this->*_M_eat_escape)();
	      return;
	    }
	  __c = *_M_current++;
	}
      if (__c == '(')
	{
	  if (_M_is_ecma() && *_M_current == '?')
	    {
	      if (++_M_current == _M_end)
		__throw_regex_error(regex_constants::error_paren);

	      if (*_M_current == ':')
		{
		  ++_M_current;
		  _M_token = _S_token_subexpr_no_group_begin;
		}
	      else if (*_M_current == '=')
		{
		  ++_M_current;
		  _M_token = _S_token_subexpr_lookahead_begin;
		  _M_value.assign(1, 'p');
		}
	      else if (*_M_current == '!')
		{
		  ++_M_current;
		  _M_token = _S_token_subexpr_lookahead_begin;
		  _M_value.assign(1, 'n');
		}
	      else
		__throw_regex_error(regex_constants::error_paren,
				    "Invalid '(?...)' zero-width assertion "
				    "in regular expression");
	    }
	  else if (_M_flags & regex_constants::nosubs)
	    _M_token = _S_token_subexpr_no_group_begin;
	  else
	    _M_token = _S_token_subexpr_begin;
	}
      else if (__c == ')')
	_M_token = _S_token_subexpr_end;
      else if (__c == '[')
	{
	  _M_state = _S_state_in_bracket;
	  _M_at_bracket_start = true;
	  if (_M_current != _M_end && *_M_current == '^')
	    {
	      _M_token = _S_token_bracket_neg_begin;
	      ++_M_current;
	    }
	  else
	    _M_token = _S_token_bracket_begin;
	}
      else if (__c == '{')
	{
	  _M_state = _S_state_in_brace;
	  _M_token = _S_token_interval_begin;
	}
      else if (__builtin_expect(__c == _CharT(0), false))
	{
	  if (!_M_is_ecma())
	    __throw_regex_error(regex_constants::_S_null);
	  _M_token = _S_token_ord_char;
	  _M_value.assign(1, __c);
	}
      else if (__c != ']' && __c != '}')
	{
	  auto __it = _M_token_tbl;
	  auto __narrowc = _M_ctype.narrow(__c, '\0');
	  for (; __it->first != '\0'; ++__it)
	    if (__it->first == __narrowc)
	      {
		_M_token = __it->second;
		return;
	      }
	  __glibcxx_assert(!"unexpected special character in regex");
	}
      else
	{
	  _M_token = _S_token_ord_char;
	  _M_value.assign(1, __c);
	}
    }

  // Differences between styles:
  // 1) different semantics of "[]" and "[^]".
  // 2) Escaping in bracket expr.
  template<typename _CharT>
    void
    _Scanner<_CharT>::
    _M_scan_in_bracket()
    {
      if (_M_current == _M_end)
	__throw_regex_error(regex_constants::error_brack);

      auto __c = *_M_current++;

      if (__c == '-')
	_M_token = _S_token_bracket_dash;
      else if (__c == '[')
	{
	  if (_M_current == _M_end)
	    __throw_regex_error(regex_constants::error_brack,
				"Incomplete '[[' character class in "
				"regular expression");

	  if (*_M_current == '.')
	    {
	      _M_token = _S_token_collsymbol;
	      _M_eat_class(*_M_current++);
	    }
	  else if (*_M_current == ':')
	    {
	      _M_token = _S_token_char_class_name;
	      _M_eat_class(*_M_current++);
	    }
	  else if (*_M_current == '=')
	    {
	      _M_token = _S_token_equiv_class_name;
	      _M_eat_class(*_M_current++);
	    }
	  else
	    {
	      _M_token = _S_token_ord_char;
	      _M_value.assign(1, __c);
	    }
	}
      // In POSIX, when encountering "[]" or "[^]", the ']' is interpreted
      // literally. So "[]]" and "[^]]" are valid regexes. See the testcases
      // `.../empty_range.cc`.
      else if (__c == ']' && (_M_is_ecma() || !_M_at_bracket_start))
	{
	  _M_token = _S_token_bracket_end;
	  _M_state = _S_state_normal;
	}
      // ECMAScript and awk permits escaping in bracket.
      else if (__c == '\\' && (_M_is_ecma() || _M_is_awk()))
	(this->*_M_eat_escape)();
      else
	{
	  _M_token = _S_token_ord_char;
	  _M_value.assign(1, __c);
	}
      _M_at_bracket_start = false;
    }

  // Differences between styles:
  // 1) "\}" in basic style.
  template<typename _CharT>
    void
    _Scanner<_CharT>::
    _M_scan_in_brace()
    {
      if (_M_current == _M_end)
	__throw_regex_error(regex_constants::error_brace);

      auto __c = *_M_current++;

      if (_M_ctype.is(_CtypeT::digit, __c))
	{
	  _M_token = _S_token_dup_count;
	  _M_value.assign(1, __c);
	  while (_M_current != _M_end
		 && _M_ctype.is(_CtypeT::digit, *_M_current))
	    _M_value += *_M_current++;
	}
      else if (__c == ',')
	_M_token = _S_token_comma;
      // basic use \}.
      else if (_M_is_basic())
	{
	  if (__c == '\\' && _M_current != _M_end && *_M_current == '}')
	    {
	      _M_state = _S_state_normal;
	      _M_token = _S_token_interval_end;
	      ++_M_current;
	    }
	  else
	    __throw_regex_error(regex_constants::error_badbrace);
	}
      else if (__c == '}')
	{
	  _M_state = _S_state_normal;
	  _M_token = _S_token_interval_end;
	}
      else
	__throw_regex_error(regex_constants::error_badbrace);
    }

  template<typename _CharT>
    void
    _Scanner<_CharT>::
    _M_eat_escape_ecma()
    {
      if (_M_current == _M_end)
	__throw_regex_error(regex_constants::error_escape);

      auto __c = *_M_current++;
      auto __pos = _M_find_escape(_M_ctype.narrow(__c, '\0'));

      if (__pos != nullptr && (__c != 'b' || _M_state == _S_state_in_bracket))
	{
	  _M_token = _S_token_ord_char;
	  _M_value.assign(1, *__pos);
	}
      else if (__c == 'b')
	{
	  _M_token = _S_token_word_bound;
	  _M_value.assign(1, 'p');
	}
      else if (__c == 'B')
	{
	  _M_token = _S_token_word_bound;
	  _M_value.assign(1, 'n');
	}
      // N3376 28.13
      else if (__c == 'd'
	       || __c == 'D'
	       || __c == 's'
	       || __c == 'S'
	       || __c == 'w'
	       || __c == 'W')
	{
	  _M_token = _S_token_quoted_class;
	  _M_value.assign(1, __c);
	}
      else if (__c == 'c')
	{
	  if (_M_current == _M_end)
	    __throw_regex_error(regex_constants::error_escape,
				"invalid '\\cX' control character in "
				"regular expression");
	  _M_token = _S_token_ord_char;
	  _M_value.assign(1, *_M_current++);
	}
      else if (__c == 'x' || __c == 'u')
	{
	  _M_value.clear();
	  const int __n = __c == 'x' ? 2 : 4;
	  for (int __i = 0; __i < __n; __i++)
	    {
	      if (_M_current == _M_end
		  || !_M_ctype.is(_CtypeT::xdigit, *_M_current))
		__throw_regex_error(regex_constants::error_escape,
				    __n == 2
				    ? "Invalid '\\xNN' control character in "
				      "regular expression"
				    : "Invalid '\\uNNNN' control character in "
				      "regular expression");
	      _M_value += *_M_current++;
	    }
	  _M_token = _S_token_hex_num;
	}
      // ECMAScript recognizes multi-digit back-references.
      else if (_M_ctype.is(_CtypeT::digit, __c))
	{
	  _M_value.assign(1, __c);
	  while (_M_current != _M_end
		 && _M_ctype.is(_CtypeT::digit, *_M_current))
	    _M_value += *_M_current++;
	  _M_token = _S_token_backref;
	}
      else
	{
	  _M_token = _S_token_ord_char;
	  _M_value.assign(1, __c);
	}
    }

  // Differences between styles:
  // 1) Extended doesn't support backref, but basic does.
  template<typename _CharT>
    void
    _Scanner<_CharT>::
    _M_eat_escape_posix()
    {
      if (_M_current == _M_end)
	__throw_regex_error(regex_constants::error_escape);

      auto __c = *_M_current;
      auto __pos = __builtin_strchr(_M_spec_char, _M_ctype.narrow(__c, '\0'));

      if (__pos != nullptr && *__pos != '\0')
	{
	  _M_token = _S_token_ord_char;
	  _M_value.assign(1, __c);
	}
      // We MUST judge awk before handling backrefs. There's no backref in awk.
      else if (_M_is_awk())
	{
	  _M_eat_escape_awk();
	  return;
	}
      else if (_M_is_basic() && _M_ctype.is(_CtypeT::digit, __c) && __c != '0')
	{
	  _M_token = _S_token_backref;
	  _M_value.assign(1, __c);
	}
      else
	{
#ifdef __STRICT_ANSI__
	  // POSIX says it is undefined to escape ordinary characters
	  __throw_regex_error(regex_constants::error_escape);
#else
	  _M_token = _S_token_ord_char;
	  _M_value.assign(1, __c);
#endif
	}
      ++_M_current;
    }

  template<typename _CharT>
    void
    _Scanner<_CharT>::
    _M_eat_escape_awk()
    {
      auto __c = *_M_current++;
      auto __pos = _M_find_escape(_M_ctype.narrow(__c, '\0'));

      if (__pos != nullptr)
	{
	  _M_token = _S_token_ord_char;
	  _M_value.assign(1, *__pos);
	}
      // \ddd for oct representation
      else if (_M_ctype.is(_CtypeT::digit, __c)
	       && __c != '8'
	       && __c != '9')
	{
	  _M_value.assign(1,  __c);
	  for (int __i = 0;
	       __i < 2
	       && _M_current != _M_end
	       && _M_ctype.is(_CtypeT::digit, *_M_current)
	       && *_M_current != '8'
	       && *_M_current != '9';
	       __i++)
	    _M_value += *_M_current++;
	  _M_token = _S_token_oct_num;
	  return;
	}
      else
	__throw_regex_error(regex_constants::error_escape);
    }

  // Eats a character class or throws an exception.
  // __ch could be ':', '.' or '=', _M_current is the char after ']' when
  // returning.
  template<typename _CharT>
    void
    _Scanner<_CharT>::
    _M_eat_class(char __ch)
    {
      for (_M_value.clear(); _M_current != _M_end && *_M_current != __ch;)
	_M_value += *_M_current++;
      if (_M_current == _M_end
	  || *_M_current++ != __ch
	  || _M_current == _M_end // skip __ch
	  || *_M_current++ != ']') // skip ']'
	{
	  __throw_regex_error(__ch == ':' ? regex_constants::error_ctype
					  : regex_constants::error_collate);
	}
    }

#ifdef _GLIBCXX_DEBUG
  template<typename _CharT>
    std::ostream&
    _Scanner<_CharT>::
    _M_print(std::ostream& __ostr)
    {
      switch (_M_token)
      {
      case _S_token_anychar:
	__ostr << "any-character\n";
	break;
      case _S_token_backref:
	__ostr << "backref\n";
	break;
      case _S_token_bracket_begin:
	__ostr << "bracket-begin\n";
	break;
      case _S_token_bracket_neg_begin:
	__ostr << "bracket-neg-begin\n";
	break;
      case _S_token_bracket_end:
	__ostr << "bracket-end\n";
	break;
      case _S_token_char_class_name:
	__ostr << "char-class-name \"" << _M_value << "\"\n";
	break;
      case _S_token_closure0:
	__ostr << "closure0\n";
	break;
      case _S_token_closure1:
	__ostr << "closure1\n";
	break;
      case _S_token_collsymbol:
	__ostr << "collsymbol \"" << _M_value << "\"\n";
	break;
      case _S_token_comma:
	__ostr << "comma\n";
	break;
      case _S_token_dup_count:
	__ostr << "dup count: " << _M_value << "\n";
	break;
      case _S_token_eof:
	__ostr << "EOF\n";
	break;
      case _S_token_equiv_class_name:
	__ostr << "equiv-class-name \"" << _M_value << "\"\n";
	break;
      case _S_token_interval_begin:
	__ostr << "interval begin\n";
	break;
      case _S_token_interval_end:
	__ostr << "interval end\n";
	break;
      case _S_token_line_begin:
	__ostr << "line begin\n";
	break;
      case _S_token_line_end:
	__ostr << "line end\n";
	break;
      case _S_token_opt:
	__ostr << "opt\n";
	break;
      case _S_token_or:
	__ostr << "or\n";
	break;
      case _S_token_ord_char:
	__ostr << "ordinary character: \"" << _M_value << "\"\n";
	break;
      case _S_token_subexpr_begin:
	__ostr << "subexpr begin\n";
	break;
      case _S_token_subexpr_no_group_begin:
	__ostr << "no grouping subexpr begin\n";
	break;
      case _S_token_subexpr_lookahead_begin:
	__ostr << "lookahead subexpr begin\n";
	break;
      case _S_token_subexpr_end:
	__ostr << "subexpr end\n";
	break;
      case _S_token_unknown:
	__ostr << "-- unknown token --\n";
	break;
      case _S_token_oct_num:
	__ostr << "oct number " << _M_value << "\n";
	break;
      case _S_token_hex_num:
	__ostr << "hex number " << _M_value << "\n";
	break;
      case _S_token_quoted_class:
	__ostr << "quoted class " << "\\" << _M_value << "\n";
	break;
      default:
	_GLIBCXX_DEBUG_ASSERT(false);
      }
      return __ostr;
    }
#endif

} // namespace __detail
_GLIBCXX_END_NAMESPACE_VERSION
} // namespace