Chris@16: ///////////////////////////////////////////////////////////////////////////////
Chris@16: // parse_charset.hpp
Chris@16: //
Chris@16: //  Copyright 2008 Eric Niebler. Distributed under the Boost
Chris@16: //  Software License, Version 1.0. (See accompanying file
Chris@16: //  LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
Chris@16: 
Chris@16: #ifndef BOOST_XPRESSIVE_DETAIL_DYNAMIC_PARSE_CHARSET_HPP_EAN_10_04_2005
Chris@16: #define BOOST_XPRESSIVE_DETAIL_DYNAMIC_PARSE_CHARSET_HPP_EAN_10_04_2005
Chris@16: 
Chris@16: // MS compatible compilers support #pragma once
Chris@101: #if defined(_MSC_VER)
Chris@16: # pragma once
Chris@16: #endif
Chris@16: 
Chris@16: #include <boost/config.hpp>
Chris@16: #include <boost/integer.hpp>
Chris@16: #include <boost/mpl/bool.hpp>
Chris@16: #include <boost/throw_exception.hpp>
Chris@16: #include <boost/numeric/conversion/converter.hpp>
Chris@16: #include <boost/xpressive/detail/detail_fwd.hpp>
Chris@16: #include <boost/xpressive/detail/dynamic/parser_enum.hpp>
Chris@16: #include <boost/xpressive/detail/utility/literals.hpp>
Chris@16: #include <boost/xpressive/detail/utility/chset/chset.hpp>
Chris@16: #include <boost/xpressive/regex_constants.hpp>
Chris@16: 
Chris@16: namespace boost { namespace xpressive { namespace detail
Chris@16: {
Chris@16: 
Chris@16: enum escape_type
Chris@16: {
Chris@16:     escape_char
Chris@16:   , escape_mark
Chris@16:   , escape_class
Chris@16: };
Chris@16: 
Chris@16: ///////////////////////////////////////////////////////////////////////////////
Chris@16: // escape_value
Chris@16: //
Chris@16: template<typename Char, typename Class>
Chris@16: struct escape_value
Chris@16: {
Chris@16:     Char ch_;
Chris@16:     int mark_nbr_;
Chris@16:     Class class_;
Chris@16:     escape_type type_;
Chris@16: };
Chris@16: 
Chris@16: ///////////////////////////////////////////////////////////////////////////////
Chris@16: // char_overflow_handler
Chris@16: //
Chris@16: struct char_overflow_handler
Chris@16: {
Chris@16:     void operator ()(numeric::range_check_result result) const // throw(regex_error)
Chris@16:     {
Chris@16:         if(numeric::cInRange != result)
Chris@16:         {
Chris@16:             BOOST_THROW_EXCEPTION(
Chris@16:                 regex_error(
Chris@16:                     regex_constants::error_escape
Chris@16:                   , "character escape too large to fit in target character type"
Chris@16:                 )
Chris@16:             );
Chris@16:         }
Chris@16:     }
Chris@16: };
Chris@16: 
Chris@16: ///////////////////////////////////////////////////////////////////////////////
Chris@16: // parse_escape
Chris@16: //
Chris@16: template<typename FwdIter, typename CompilerTraits>
Chris@16: escape_value<typename iterator_value<FwdIter>::type, typename CompilerTraits::regex_traits::char_class_type>
Chris@16: parse_escape(FwdIter &begin, FwdIter end, CompilerTraits &tr)
Chris@16: {
Chris@16:     using namespace regex_constants;
Chris@16:     typedef typename iterator_value<FwdIter>::type char_type;
Chris@16:     typedef typename CompilerTraits::regex_traits regex_traits;
Chris@16:     typedef typename regex_traits::char_class_type char_class_type;
Chris@16: 
Chris@16:     // define an unsigned type the same size as char_type
Chris@16:     typedef typename boost::uint_t<CHAR_BIT * sizeof(char_type)>::least uchar_t;
Chris@16:     BOOST_MPL_ASSERT_RELATION(sizeof(uchar_t), ==, sizeof(char_type));
Chris@16:     typedef numeric::conversion_traits<uchar_t, int> converstion_traits;
Chris@16: 
Chris@16:     BOOST_XPR_ENSURE_(begin != end, error_escape, "unexpected end of pattern found");
Chris@16:     numeric::converter<int, uchar_t, converstion_traits, char_overflow_handler> converter;
Chris@16:     escape_value<char_type,char_class_type> esc = { 0, 0, 0, escape_char };
Chris@16:     bool const icase = (0 != (regex_constants::icase_ & tr.flags()));
Chris@16:     regex_traits const &rxtraits = tr.traits();
Chris@16:     FwdIter tmp;
Chris@16: 
Chris@16:     esc.class_ = rxtraits.lookup_classname(begin, begin + 1, icase);
Chris@16:     if(0 != esc.class_)
Chris@16:     {
Chris@16:         esc.type_ = escape_class;
Chris@16:         return esc;
Chris@16:     }
Chris@16: 
Chris@16:     if(-1 != rxtraits.value(*begin, 8))
Chris@16:     {
Chris@16:         esc.ch_ = converter(toi(begin, end, rxtraits, 8, 0777));
Chris@16:         return esc;
Chris@16:     }
Chris@16: 
Chris@16:     switch(*begin)
Chris@16:     {
Chris@16:     // bell character
Chris@16:     case BOOST_XPR_CHAR_(char_type, 'a'):
Chris@16:         esc.ch_ = BOOST_XPR_CHAR_(char_type, '\a');
Chris@16:         ++begin;
Chris@16:         break;
Chris@16:     // escape character
Chris@16:     case BOOST_XPR_CHAR_(char_type, 'e'):
Chris@16:         esc.ch_ = converter(27);
Chris@16:         ++begin;
Chris@16:         break;
Chris@16:     // control character
Chris@16:     case BOOST_XPR_CHAR_(char_type, 'c'):
Chris@16:         BOOST_XPR_ENSURE_(++begin != end, error_escape, "unexpected end of pattern found");
Chris@16:         BOOST_XPR_ENSURE_
Chris@16:         (
Chris@16:             rxtraits.in_range(BOOST_XPR_CHAR_(char_type, 'a'), BOOST_XPR_CHAR_(char_type, 'z'), *begin)
Chris@16:          || rxtraits.in_range(BOOST_XPR_CHAR_(char_type, 'A'), BOOST_XPR_CHAR_(char_type, 'Z'), *begin)
Chris@16:           , error_escape
Chris@16:           , "invalid escape control letter; must be one of a-z or A-Z"
Chris@16:         );
Chris@16:         // Convert to character according to ECMA-262, section 15.10.2.10:
Chris@16:         esc.ch_ = converter(*begin % 32);
Chris@16:         ++begin;
Chris@16:         break;
Chris@16:     // formfeed character
Chris@16:     case BOOST_XPR_CHAR_(char_type, 'f'):
Chris@16:         esc.ch_ = BOOST_XPR_CHAR_(char_type, '\f');
Chris@16:         ++begin;
Chris@16:         break;
Chris@16:     // newline
Chris@16:     case BOOST_XPR_CHAR_(char_type, 'n'):
Chris@16:         esc.ch_ = BOOST_XPR_CHAR_(char_type, '\n');
Chris@16:         ++begin;
Chris@16:         break;
Chris@16:     // return
Chris@16:     case BOOST_XPR_CHAR_(char_type, 'r'):
Chris@16:         esc.ch_ = BOOST_XPR_CHAR_(char_type, '\r');
Chris@16:         ++begin;
Chris@16:         break;
Chris@16:     // horizontal tab
Chris@16:     case BOOST_XPR_CHAR_(char_type, 't'):
Chris@16:         esc.ch_ = BOOST_XPR_CHAR_(char_type, '\t');
Chris@16:         ++begin;
Chris@16:         break;
Chris@16:     // vertical tab
Chris@16:     case BOOST_XPR_CHAR_(char_type, 'v'):
Chris@16:         esc.ch_ = BOOST_XPR_CHAR_(char_type, '\v');
Chris@16:         ++begin;
Chris@16:         break;
Chris@16:     // hex escape sequence
Chris@16:     case BOOST_XPR_CHAR_(char_type, 'x'):
Chris@16:         BOOST_XPR_ENSURE_(++begin != end, error_escape, "unexpected end of pattern found");
Chris@16:         tmp = begin;
Chris@16:         esc.ch_ = converter(toi(begin, end, rxtraits, 16, 0xff));
Chris@16:         BOOST_XPR_ENSURE_(2 == std::distance(tmp, begin), error_escape, "invalid hex escape : "
Chris@16:             "must be \\x HexDigit HexDigit");
Chris@16:         break;
Chris@16:     // Unicode escape sequence
Chris@16:     case BOOST_XPR_CHAR_(char_type, 'u'):
Chris@16:         BOOST_XPR_ENSURE_(++begin != end, error_escape, "unexpected end of pattern found");
Chris@16:         tmp = begin;
Chris@16:         esc.ch_ = converter(toi(begin, end, rxtraits, 16, 0xffff));
Chris@16:         BOOST_XPR_ENSURE_(4 == std::distance(tmp, begin), error_escape, "invalid Unicode escape : "
Chris@16:             "must be \\u HexDigit HexDigit HexDigit HexDigit");
Chris@16:         break;
Chris@16:     // backslash
Chris@16:     case BOOST_XPR_CHAR_(char_type, '\\'):
Chris@16:         //esc.ch_ = BOOST_XPR_CHAR_(char_type, '\\');
Chris@16:         //++begin;
Chris@16:         //break;
Chris@16:     // all other escaped characters represent themselves
Chris@16:     default:
Chris@16:         esc.ch_ = *begin;
Chris@16:         ++begin;
Chris@16:         break;
Chris@16:     }
Chris@16: 
Chris@16:     return esc;
Chris@16: }
Chris@16: 
Chris@16: //////////////////////////////////////////////////////////////////////////
Chris@16: // parse_charset
Chris@16: //
Chris@16: template<typename FwdIter, typename RegexTraits, typename CompilerTraits>
Chris@16: inline void parse_charset
Chris@16: (
Chris@16:     FwdIter &begin
Chris@16:   , FwdIter end
Chris@16:   , compound_charset<RegexTraits> &chset
Chris@16:   , CompilerTraits &tr
Chris@16: )
Chris@16: {
Chris@16:     using namespace regex_constants;
Chris@16:     typedef typename RegexTraits::char_type char_type;
Chris@16:     typedef typename RegexTraits::char_class_type char_class_type;
Chris@16:     BOOST_XPR_ENSURE_(begin != end, error_brack, "unexpected end of pattern found");
Chris@16:     RegexTraits const &rxtraits = tr.traits();
Chris@16:     bool const icase = (0 != (regex_constants::icase_ & tr.flags()));
Chris@16:     FwdIter iprev = FwdIter();
Chris@16:     escape_value<char_type, char_class_type> esc = {0, 0, 0, escape_char};
Chris@16:     bool invert = false;
Chris@16: 
Chris@16:     // check to see if we have an inverse charset
Chris@16:     if(begin != end && token_charset_invert == tr.get_charset_token(iprev = begin, end))
Chris@16:     {
Chris@16:         begin = iprev;
Chris@16:         invert = true;
Chris@16:     }
Chris@16: 
Chris@16:     // skip the end token if-and-only-if it is the first token in the charset
Chris@16:     if(begin != end && token_charset_end == tr.get_charset_token(iprev = begin, end))
Chris@16:     {
Chris@16:         for(; begin != iprev; ++begin)
Chris@16:         {
Chris@16:             chset.set_char(*begin, rxtraits, icase);
Chris@16:         }
Chris@16:     }
Chris@16: 
Chris@16:     compiler_token_type tok;
Chris@16:     char_type ch_prev = char_type(), ch_next = char_type();
Chris@16:     bool have_prev = false;
Chris@16: 
Chris@16:     BOOST_XPR_ENSURE_(begin != end, error_brack, "unexpected end of pattern found");
Chris@16: 
Chris@16:     // remember the current position and grab the next token
Chris@16:     iprev = begin;
Chris@16:     tok = tr.get_charset_token(begin, end);
Chris@16:     do
Chris@16:     {
Chris@16:         BOOST_XPR_ENSURE_(begin != end, error_brack, "unexpected end of pattern found");
Chris@16: 
Chris@16:         if(token_charset_hyphen == tok && have_prev)
Chris@16:         {
Chris@16:             // remember the current position
Chris@16:             FwdIter iprev2 = begin;
Chris@16:             have_prev = false;
Chris@16: 
Chris@16:             // ch_prev is lower bound of a range
Chris@16:             switch(tr.get_charset_token(begin, end))
Chris@16:             {
Chris@16:             case token_charset_hyphen:
Chris@16:             case token_charset_invert:
Chris@16:                 begin = iprev2; // un-get these tokens and fall through
Chris@16:                 BOOST_FALLTHROUGH;
Chris@16:             case token_literal:
Chris@16:                 ch_next = *begin++;
Chris@16:                 BOOST_XPR_ENSURE_(ch_prev <= ch_next, error_range, "invalid charset range");
Chris@16:                 chset.set_range(ch_prev, ch_next, rxtraits, icase);
Chris@16:                 continue;
Chris@16:             case token_charset_backspace:
Chris@16:                 ch_next = char_type(8); // backspace
Chris@16:                 BOOST_XPR_ENSURE_(ch_prev <= ch_next, error_range, "invalid charset range");
Chris@16:                 chset.set_range(ch_prev, ch_next, rxtraits, icase);
Chris@16:                 continue;
Chris@16:             case token_escape:
Chris@16:                 esc = parse_escape(begin, end, tr);
Chris@16:                 if(escape_char == esc.type_)
Chris@16:                 {
Chris@16:                     BOOST_XPR_ENSURE_(ch_prev <= esc.ch_, error_range, "invalid charset range");
Chris@16:                     chset.set_range(ch_prev, esc.ch_, rxtraits, icase);
Chris@16:                     continue;
Chris@16:                 }
Chris@16:                 BOOST_FALLTHROUGH;
Chris@16:             case token_charset_end:
Chris@16:             default:                // not a range.
Chris@16:                 begin = iprev;      // backup to hyphen token
Chris@16:                 chset.set_char(ch_prev, rxtraits, icase);
Chris@16:                 chset.set_char(*begin++, rxtraits, icase);
Chris@16:                 continue;
Chris@16:             }
Chris@16:         }
Chris@16: 
Chris@16:         if(have_prev)
Chris@16:         {
Chris@16:             chset.set_char(ch_prev, rxtraits, icase);
Chris@16:             have_prev = false;
Chris@16:         }
Chris@16: 
Chris@16:         switch(tok)
Chris@16:         {
Chris@16:         case token_charset_hyphen:
Chris@16:         case token_charset_invert:
Chris@16:         case token_charset_end:
Chris@16:         case token_posix_charset_end:
Chris@16:             begin = iprev; // un-get these tokens
Chris@16:             ch_prev = *begin++;
Chris@16:             have_prev = true;
Chris@16:             continue;
Chris@16: 
Chris@16:         case token_charset_backspace:
Chris@16:             ch_prev = char_type(8); // backspace
Chris@16:             have_prev = true;
Chris@16:             continue;
Chris@16: 
Chris@16:         case token_posix_charset_begin:
Chris@16:             {
Chris@16:                 FwdIter tmp = begin, start = begin;
Chris@16:                 bool invert = (token_charset_invert == tr.get_charset_token(tmp, end));
Chris@16:                 if(invert)
Chris@16:                 {
Chris@16:                     begin = start = tmp;
Chris@16:                 }
Chris@16:                 while(token_literal == (tok = tr.get_charset_token(begin, end)))
Chris@16:                 {
Chris@16:                     tmp = ++begin;
Chris@16:                     BOOST_XPR_ENSURE_(begin != end, error_brack, "unexpected end of pattern found");
Chris@16:                 }
Chris@16:                 if(token_posix_charset_end == tok)
Chris@16:                 {
Chris@16:                     char_class_type chclass = rxtraits.lookup_classname(start, tmp, icase);
Chris@16:                     BOOST_XPR_ENSURE_(0 != chclass, error_ctype, "unknown class name");
Chris@16:                     chset.set_class(chclass, invert);
Chris@16:                     continue;
Chris@16:                 }
Chris@16:                 begin = iprev; // un-get this token
Chris@16:                 ch_prev = *begin++;
Chris@16:                 have_prev = true;
Chris@16:             }
Chris@16:             continue;
Chris@16: 
Chris@16:         case token_escape:
Chris@16:             esc = parse_escape(begin, end, tr);
Chris@16:             if(escape_char == esc.type_)
Chris@16:             {
Chris@16:                 ch_prev = esc.ch_;
Chris@16:                 have_prev = true;
Chris@16:             }
Chris@16:             else if(escape_class == esc.type_)
Chris@16:             {
Chris@16:                 char_class_type upper_ = lookup_classname(rxtraits, "upper");
Chris@16:                 BOOST_ASSERT(0 != upper_);
Chris@16:                 chset.set_class(esc.class_, rxtraits.isctype(*begin++, upper_));
Chris@16:             }
Chris@16:             else
Chris@16:             {
Chris@16:                 BOOST_ASSERT(false);
Chris@16:             }
Chris@16:             continue;
Chris@16: 
Chris@16:         default:
Chris@16:             ch_prev = *begin++;
Chris@16:             have_prev = true;
Chris@16:             continue;
Chris@16:         }
Chris@16:     }
Chris@16:     while(BOOST_XPR_ENSURE_((iprev = begin) != end, error_brack, "unexpected end of pattern found"),
Chris@16:           token_charset_end != (tok = tr.get_charset_token(begin, end)));
Chris@16: 
Chris@16:     if(have_prev)
Chris@16:     {
Chris@16:         chset.set_char(ch_prev, rxtraits, icase);
Chris@16:     }
Chris@16: 
Chris@16:     if(invert)
Chris@16:     {
Chris@16:         chset.inverse();
Chris@16:     }
Chris@16: }
Chris@16: 
Chris@16: }}} // namespace boost::xpressive::detail
Chris@16: 
Chris@16: #endif