Chris@16: ///////////////////////////////////////////////////////////////////////////////
Chris@16: /// \file cpp_regex_traits.hpp
Chris@16: /// Contains the definition of the cpp_regex_traits\<\> template, which is a
Chris@16: /// wrapper for std::locale that can be used to customize the behavior of
Chris@16: /// static and dynamic regexes.
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_TRAITS_CPP_REGEX_TRAITS_HPP_EAN_10_04_2005
Chris@16: #define BOOST_XPRESSIVE_TRAITS_CPP_REGEX_TRAITS_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 <ios>
Chris@16: #include <string>
Chris@16: #include <locale>
Chris@16: #include <sstream>
Chris@16: #include <climits>
Chris@16: #include <boost/config.hpp>
Chris@16: #include <boost/assert.hpp>
Chris@16: #include <boost/integer.hpp>
Chris@16: #include <boost/mpl/assert.hpp>
Chris@16: #include <boost/static_assert.hpp>
Chris@16: #include <boost/detail/workaround.hpp>
Chris@16: #include <boost/type_traits/is_same.hpp>
Chris@16: #include <boost/xpressive/detail/detail_fwd.hpp>
Chris@16: #include <boost/xpressive/detail/utility/literals.hpp>
Chris@16: 
Chris@16: // From John Maddock:
Chris@16: // Fix for gcc prior to 3.4: std::ctype<wchar_t> doesn't allow masks to be combined, for example:
Chris@16: // std::use_facet<std::ctype<wchar_t> >(locale()).is(std::ctype_base::lower|std::ctype_base::upper, L'a');
Chris@16: // incorrectly returns false.
Chris@16: // NOTE: later version of the gcc define __GLIBCXX__, not __GLIBCPP__
Chris@16: #if BOOST_WORKAROUND(__GLIBCPP__, != 0)
Chris@16: # define BOOST_XPRESSIVE_BUGGY_CTYPE_FACET
Chris@16: #endif
Chris@16: 
Chris@16: namespace boost { namespace xpressive
Chris@16: {
Chris@16: 
Chris@16: namespace detail
Chris@16: {
Chris@16:     // define an unsigned integral typedef of the same size as std::ctype_base::mask
Chris@16:     typedef boost::uint_t<sizeof(std::ctype_base::mask) * CHAR_BIT>::least umask_t;
Chris@16:     BOOST_MPL_ASSERT_RELATION(sizeof(std::ctype_base::mask), ==, sizeof(umask_t));
Chris@16: 
Chris@16:     // Calculate what the size of the umaskex_t type should be to fix the 3 extra bitmasks
Chris@16:     //   11 char categories in ctype_base
Chris@16:     // +  3 extra categories for xpressive
Chris@16:     // = 14 total bits needed
Chris@16:     int const umaskex_bits = (14 > (sizeof(umask_t) * CHAR_BIT)) ? 14 : sizeof(umask_t) * CHAR_BIT;
Chris@16: 
Chris@16:     // define an unsigned integral type with at least umaskex_bits
Chris@16:     typedef boost::uint_t<umaskex_bits>::fast umaskex_t;
Chris@16:     BOOST_MPL_ASSERT_RELATION(sizeof(umask_t), <=, sizeof(umaskex_t));
Chris@16: 
Chris@16:     // cast a ctype mask to a umaskex_t
Chris@16:     template<std::ctype_base::mask Mask>
Chris@16:     struct mask_cast
Chris@16:     {
Chris@16:         BOOST_STATIC_CONSTANT(umaskex_t, value = static_cast<umask_t>(Mask));
Chris@16:     };
Chris@16: 
Chris@16:     #ifdef __CYGWIN__
Chris@16:     // Work around a gcc warning on cygwin
Chris@16:     template<>
Chris@16:     struct mask_cast<std::ctype_base::print>
Chris@16:     {
Chris@16:         BOOST_MPL_ASSERT_RELATION('\227', ==, std::ctype_base::print);
Chris@16:         BOOST_STATIC_CONSTANT(umaskex_t, value = 0227);
Chris@16:     };
Chris@16:     #endif
Chris@16: 
Chris@16:     #ifndef BOOST_NO_INCLASS_MEMBER_INITIALIZATION
Chris@16:     template<std::ctype_base::mask Mask>
Chris@16:     umaskex_t const mask_cast<Mask>::value;
Chris@16:     #endif
Chris@16: 
Chris@16:     #ifndef BOOST_XPRESSIVE_BUGGY_CTYPE_FACET
Chris@16:     // an unsigned integer with the highest bit set
Chris@16:     umaskex_t const highest_bit = static_cast<umaskex_t>(1) << (sizeof(umaskex_t) * CHAR_BIT - 1);
Chris@16: 
Chris@16:     ///////////////////////////////////////////////////////////////////////////////
Chris@16:     // unused_mask
Chris@16:     //   find a bit in an int that isn't set
Chris@16:     template<umaskex_t In, umaskex_t Out = highest_bit, bool Done = (0 == (Out & In))>
Chris@16:     struct unused_mask
Chris@16:     {
Chris@16:         BOOST_STATIC_ASSERT(1 != Out);
Chris@16:         BOOST_STATIC_CONSTANT(umaskex_t, value = (unused_mask<In, (Out >> 1)>::value));
Chris@16:     };
Chris@16: 
Chris@16:     template<umaskex_t In, umaskex_t Out>
Chris@16:     struct unused_mask<In, Out, true>
Chris@16:     {
Chris@16:         BOOST_STATIC_CONSTANT(umaskex_t, value = Out);
Chris@16:     };
Chris@16: 
Chris@16:     #ifndef BOOST_NO_INCLASS_MEMBER_INITIALIZATION
Chris@16:     template<umaskex_t In, umaskex_t Out, bool Done>
Chris@16:     umaskex_t const unused_mask<In, Out, Done>::value;
Chris@16:     #endif
Chris@16: 
Chris@16:     umaskex_t const std_ctype_alnum = mask_cast<std::ctype_base::alnum>::value;
Chris@16:     umaskex_t const std_ctype_alpha = mask_cast<std::ctype_base::alpha>::value;
Chris@16:     umaskex_t const std_ctype_cntrl = mask_cast<std::ctype_base::cntrl>::value;
Chris@16:     umaskex_t const std_ctype_digit = mask_cast<std::ctype_base::digit>::value;
Chris@16:     umaskex_t const std_ctype_graph = mask_cast<std::ctype_base::graph>::value;
Chris@16:     umaskex_t const std_ctype_lower = mask_cast<std::ctype_base::lower>::value;
Chris@16:     umaskex_t const std_ctype_print = mask_cast<std::ctype_base::print>::value;
Chris@16:     umaskex_t const std_ctype_punct = mask_cast<std::ctype_base::punct>::value;
Chris@16:     umaskex_t const std_ctype_space = mask_cast<std::ctype_base::space>::value;
Chris@16:     umaskex_t const std_ctype_upper = mask_cast<std::ctype_base::upper>::value;
Chris@16:     umaskex_t const std_ctype_xdigit = mask_cast<std::ctype_base::xdigit>::value;
Chris@16: 
Chris@16:     // Reserve some bits for the implementation
Chris@16:     #if defined(__GLIBCXX__)
Chris@16:     umaskex_t const std_ctype_reserved = 0x8000;
Chris@16:     #elif defined(_CPPLIB_VER) && defined(BOOST_WINDOWS)
Chris@16:     umaskex_t const std_ctype_reserved = 0x8200;
Chris@16:     #elif defined(_LIBCPP_VERSION)
Chris@16:     umaskex_t const std_ctype_reserved = 0x8000;
Chris@16:     #else
Chris@16:     umaskex_t const std_ctype_reserved = 0;
Chris@16:     #endif
Chris@16: 
Chris@16:     // Bitwise-or all the ctype masks together
Chris@16:     umaskex_t const all_ctype_masks = std_ctype_reserved
Chris@16:       | std_ctype_alnum | std_ctype_alpha | std_ctype_cntrl | std_ctype_digit
Chris@16:       | std_ctype_graph | std_ctype_lower | std_ctype_print | std_ctype_punct
Chris@16:       | std_ctype_space | std_ctype_upper | std_ctype_xdigit;
Chris@16: 
Chris@16:     // define a new mask for "underscore" ("word" == alnum | underscore)
Chris@16:     umaskex_t const non_std_ctype_underscore = unused_mask<all_ctype_masks>::value;
Chris@16: 
Chris@16:     // define a new mask for "blank"
Chris@16:     umaskex_t const non_std_ctype_blank = unused_mask<all_ctype_masks | non_std_ctype_underscore>::value;
Chris@16: 
Chris@16:     // define a new mask for "newline"
Chris@16:     umaskex_t const non_std_ctype_newline = unused_mask<all_ctype_masks | non_std_ctype_underscore | non_std_ctype_blank>::value;
Chris@16: 
Chris@16:     #else
Chris@16:     ///////////////////////////////////////////////////////////////////////////////
Chris@16:     // Ugly work-around for buggy ctype facets.
Chris@16:     umaskex_t const std_ctype_alnum = 1 << 0;
Chris@16:     umaskex_t const std_ctype_alpha = 1 << 1;
Chris@16:     umaskex_t const std_ctype_cntrl = 1 << 2;
Chris@16:     umaskex_t const std_ctype_digit = 1 << 3;
Chris@16:     umaskex_t const std_ctype_graph = 1 << 4;
Chris@16:     umaskex_t const std_ctype_lower = 1 << 5;
Chris@16:     umaskex_t const std_ctype_print = 1 << 6;
Chris@16:     umaskex_t const std_ctype_punct = 1 << 7;
Chris@16:     umaskex_t const std_ctype_space = 1 << 8;
Chris@16:     umaskex_t const std_ctype_upper = 1 << 9;
Chris@16:     umaskex_t const std_ctype_xdigit = 1 << 10;
Chris@16:     umaskex_t const non_std_ctype_underscore = 1 << 11;
Chris@16:     umaskex_t const non_std_ctype_blank = 1 << 12;
Chris@16:     umaskex_t const non_std_ctype_newline = 1 << 13;
Chris@16: 
Chris@16:     static umaskex_t const std_masks[] =
Chris@16:     {
Chris@16:         mask_cast<std::ctype_base::alnum>::value
Chris@16:       , mask_cast<std::ctype_base::alpha>::value
Chris@16:       , mask_cast<std::ctype_base::cntrl>::value
Chris@16:       , mask_cast<std::ctype_base::digit>::value
Chris@16:       , mask_cast<std::ctype_base::graph>::value
Chris@16:       , mask_cast<std::ctype_base::lower>::value
Chris@16:       , mask_cast<std::ctype_base::print>::value
Chris@16:       , mask_cast<std::ctype_base::punct>::value
Chris@16:       , mask_cast<std::ctype_base::space>::value
Chris@16:       , mask_cast<std::ctype_base::upper>::value
Chris@16:       , mask_cast<std::ctype_base::xdigit>::value
Chris@16:     };
Chris@16: 
Chris@16:     inline int mylog2(umaskex_t i)
Chris@16:     {
Chris@16:         return "\0\0\1\0\2\0\0\0\3"[i & 0xf]
Chris@16:              + "\0\4\5\0\6\0\0\0\7"[(i & 0xf0) >> 04]
Chris@16:              + "\0\10\11\0\12\0\0\0\13"[(i & 0xf00) >> 010];
Chris@16:     }
Chris@16:     #endif
Chris@16: 
Chris@16:     // convenient constant for the extra masks
Chris@16:     umaskex_t const non_std_ctype_masks = non_std_ctype_underscore | non_std_ctype_blank | non_std_ctype_newline;
Chris@16: 
Chris@16:     ///////////////////////////////////////////////////////////////////////////////
Chris@16:     // cpp_regex_traits_base
Chris@16:     //   BUGBUG this should be replaced with a regex facet that lets you query for
Chris@16:     //   an array of underscore characters and an array of line separator characters.
Chris@16:     template<typename Char, std::size_t SizeOfChar = sizeof(Char)>
Chris@16:     struct cpp_regex_traits_base
Chris@16:     {
Chris@16:     protected:
Chris@16:         void imbue(std::locale const &)
Chris@16:         {
Chris@16:         }
Chris@16: 
Chris@16:         static bool is(std::ctype<Char> const &ct, Char ch, umaskex_t mask)
Chris@16:         {
Chris@16:             #ifndef BOOST_XPRESSIVE_BUGGY_CTYPE_FACET
Chris@16: 
Chris@16:             if(ct.is((std::ctype_base::mask)(umask_t)mask, ch))
Chris@16:             {
Chris@16:                 return true;
Chris@16:             }
Chris@16: 
Chris@16:             // HACKHACK Cygwin and mingw have buggy ctype facets for wchar_t
Chris@16:             #if defined(__CYGWIN__) || defined(__MINGW32_VERSION)
Chris@16:             if (std::ctype_base::xdigit == ((std::ctype_base::mask)(umask_t)mask & std::ctype_base::xdigit))
Chris@16:             {
Chris@16:                 typename std::char_traits<Char>::int_type i = std::char_traits<Char>::to_int_type(ch);
Chris@16:                 if(UCHAR_MAX >= i && std::isxdigit(static_cast<int>(i)))
Chris@16:                     return true;
Chris@16:             }
Chris@16:             #endif
Chris@16: 
Chris@16:             #else
Chris@16: 
Chris@16:             umaskex_t tmp = mask & ~non_std_ctype_masks;
Chris@16:             for(umaskex_t i; 0 != (i = (tmp & (~tmp+1))); tmp &= ~i)
Chris@16:             {
Chris@16:                 std::ctype_base::mask m = (std::ctype_base::mask)(umask_t)std_masks[mylog2(i)];
Chris@16:                 if(ct.is(m, ch))
Chris@16:                 {
Chris@16:                     return true;
Chris@16:                 }
Chris@16:             }
Chris@16: 
Chris@16:             #endif
Chris@16: 
Chris@16:             return ((mask & non_std_ctype_blank) && cpp_regex_traits_base::is_blank(ch))
Chris@16:                 || ((mask & non_std_ctype_underscore) && cpp_regex_traits_base::is_underscore(ch))
Chris@16:                 || ((mask & non_std_ctype_newline) && cpp_regex_traits_base::is_newline(ch));
Chris@16:         }
Chris@16: 
Chris@16:     private:
Chris@16:         static bool is_blank(Char ch)
Chris@16:         {
Chris@16:             BOOST_MPL_ASSERT_RELATION('\t', ==, L'\t');
Chris@16:             BOOST_MPL_ASSERT_RELATION(' ', ==, L' ');
Chris@16:             return L' ' == ch || L'\t' == ch;
Chris@16:         }
Chris@16: 
Chris@16:         static bool is_underscore(Char ch)
Chris@16:         {
Chris@16:             BOOST_MPL_ASSERT_RELATION('_', ==, L'_');
Chris@16:             return L'_' == ch;
Chris@16:         }
Chris@16: 
Chris@16:         static bool is_newline(Char ch)
Chris@16:         {
Chris@16:             BOOST_MPL_ASSERT_RELATION('\r', ==, L'\r');
Chris@16:             BOOST_MPL_ASSERT_RELATION('\n', ==, L'\n');
Chris@16:             BOOST_MPL_ASSERT_RELATION('\f', ==, L'\f');
Chris@16:             return L'\r' == ch || L'\n' == ch || L'\f' == ch
Chris@16:                 || (1 < SizeOfChar && (0x2028u == ch || 0x2029u == ch || 0x85u == ch));
Chris@16:         }
Chris@16:     };
Chris@16: 
Chris@16:     #ifndef BOOST_XPRESSIVE_BUGGY_CTYPE_FACET
Chris@16: 
Chris@16:     template<typename Char>
Chris@16:     struct cpp_regex_traits_base<Char, 1>
Chris@16:     {
Chris@16:     protected:
Chris@16:         void imbue(std::locale const &loc)
Chris@16:         {
Chris@16:             int i = 0;
Chris@16:             Char allchars[UCHAR_MAX + 1];
Chris@16:             for(i = 0; i <= static_cast<int>(UCHAR_MAX); ++i)
Chris@16:             {
Chris@16:                 allchars[i] = static_cast<Char>(i);
Chris@16:             }
Chris@16: 
Chris@16:             std::ctype<Char> const &ct = BOOST_USE_FACET(std::ctype<Char>, loc);
Chris@16:             std::ctype_base::mask tmp[UCHAR_MAX + 1];
Chris@16:             ct.is(allchars, allchars + UCHAR_MAX + 1, tmp);
Chris@16:             for(i = 0; i <= static_cast<int>(UCHAR_MAX); ++i)
Chris@16:             {
Chris@16:                 this->masks_[i] = static_cast<umask_t>(tmp[i]);
Chris@16:                 BOOST_ASSERT(0 == (this->masks_[i] & non_std_ctype_masks));
Chris@16:             }
Chris@16: 
Chris@16:             this->masks_[static_cast<unsigned char>('_')] |= non_std_ctype_underscore;
Chris@16:             this->masks_[static_cast<unsigned char>(' ')] |= non_std_ctype_blank;
Chris@16:             this->masks_[static_cast<unsigned char>('\t')] |= non_std_ctype_blank;
Chris@16:             this->masks_[static_cast<unsigned char>('\n')] |= non_std_ctype_newline;
Chris@16:             this->masks_[static_cast<unsigned char>('\r')] |= non_std_ctype_newline;
Chris@16:             this->masks_[static_cast<unsigned char>('\f')] |= non_std_ctype_newline;
Chris@16:         }
Chris@16: 
Chris@16:         bool is(std::ctype<Char> const &, Char ch, umaskex_t mask) const
Chris@16:         {
Chris@16:             return 0 != (this->masks_[static_cast<unsigned char>(ch)] & mask);
Chris@16:         }
Chris@16: 
Chris@16:     private:
Chris@16:         umaskex_t masks_[UCHAR_MAX + 1];
Chris@16:     };
Chris@16: 
Chris@16:     #endif
Chris@16: 
Chris@16: } // namespace detail
Chris@16: 
Chris@16: 
Chris@16: ///////////////////////////////////////////////////////////////////////////////
Chris@16: // cpp_regex_traits
Chris@16: //
Chris@16: /// \brief Encapsaulates a \c std::locale for use by the
Chris@16: /// \c basic_regex\<\> class template.
Chris@16: template<typename Char>
Chris@16: struct cpp_regex_traits
Chris@16:   : detail::cpp_regex_traits_base<Char>
Chris@16: {
Chris@16:     typedef Char char_type;
Chris@16:     typedef std::basic_string<char_type> string_type;
Chris@16:     typedef std::locale locale_type;
Chris@16:     typedef detail::umaskex_t char_class_type;
Chris@16:     typedef regex_traits_version_2_tag version_tag;
Chris@16:     typedef detail::cpp_regex_traits_base<Char> base_type;
Chris@16: 
Chris@16:     /// Initialize a cpp_regex_traits object to use the specified std::locale,
Chris@16:     /// or the global std::locale if none is specified.
Chris@16:     ///
Chris@16:     cpp_regex_traits(locale_type const &loc = locale_type())
Chris@16:       : base_type()
Chris@16:       , loc_()
Chris@16:     {
Chris@16:         this->imbue(loc);
Chris@16:     }
Chris@16: 
Chris@16:     /// Checks two cpp_regex_traits objects for equality
Chris@16:     ///
Chris@16:     /// \return this->getloc() == that.getloc().
Chris@16:     bool operator ==(cpp_regex_traits<char_type> const &that) const
Chris@16:     {
Chris@16:         return this->loc_ == that.loc_;
Chris@16:     }
Chris@16: 
Chris@16:     /// Checks two cpp_regex_traits objects for inequality
Chris@16:     ///
Chris@16:     /// \return this->getloc() != that.getloc().
Chris@16:     bool operator !=(cpp_regex_traits<char_type> const &that) const
Chris@16:     {
Chris@16:         return this->loc_ != that.loc_;
Chris@16:     }
Chris@16: 
Chris@16:     /// Convert a char to a Char
Chris@16:     ///
Chris@16:     /// \param ch The source character.
Chris@16:     /// \return std::use_facet\<std::ctype\<char_type\> \>(this->getloc()).widen(ch).
Chris@16:     char_type widen(char ch) const
Chris@16:     {
Chris@16:         return this->ctype_->widen(ch);
Chris@16:     }
Chris@16: 
Chris@16:     /// Returns a hash value for a Char in the range [0, UCHAR_MAX]
Chris@16:     ///
Chris@16:     /// \param ch The source character.
Chris@16:     /// \return a value between 0 and UCHAR_MAX, inclusive.
Chris@16:     static unsigned char hash(char_type ch)
Chris@16:     {
Chris@16:         return static_cast<unsigned char>(std::char_traits<Char>::to_int_type(ch));
Chris@16:     }
Chris@16: 
Chris@16:     /// No-op
Chris@16:     ///
Chris@16:     /// \param ch The source character.
Chris@16:     /// \return ch
Chris@16:     static char_type translate(char_type ch)
Chris@16:     {
Chris@16:         return ch;
Chris@16:     }
Chris@16: 
Chris@16:     /// Converts a character to lower-case using the internally-stored std::locale.
Chris@16:     ///
Chris@16:     /// \param ch The source character.
Chris@16:     /// \return std::tolower(ch, this->getloc()).
Chris@16:     char_type translate_nocase(char_type ch) const
Chris@16:     {
Chris@16:         return this->ctype_->tolower(ch);
Chris@16:     }
Chris@16: 
Chris@16:     /// Converts a character to lower-case using the internally-stored std::locale.
Chris@16:     ///
Chris@16:     /// \param ch The source character.
Chris@16:     /// \return std::tolower(ch, this->getloc()).
Chris@16:     char_type tolower(char_type ch) const
Chris@16:     {
Chris@16:         return this->ctype_->tolower(ch);
Chris@16:     }
Chris@16: 
Chris@16:     /// Converts a character to upper-case using the internally-stored std::locale.
Chris@16:     ///
Chris@16:     /// \param ch The source character.
Chris@16:     /// \return std::toupper(ch, this->getloc()).
Chris@16:     char_type toupper(char_type ch) const
Chris@16:     {
Chris@16:         return this->ctype_->toupper(ch);
Chris@16:     }
Chris@16: 
Chris@16:     /// Returns a \c string_type containing all the characters that compare equal
Chris@16:     /// disregrarding case to the one passed in. This function can only be called
Chris@16:     /// if <tt>has_fold_case\<cpp_regex_traits\<Char\> \>::value</tt> is \c true.
Chris@16:     ///
Chris@16:     /// \param ch The source character.
Chris@16:     /// \return \c string_type containing all chars which are equal to \c ch when disregarding
Chris@16:     ///     case
Chris@16:     string_type fold_case(char_type ch) const
Chris@16:     {
Chris@16:         BOOST_MPL_ASSERT((is_same<char_type, char>));
Chris@16:         char_type ntcs[] = {
Chris@16:             this->ctype_->tolower(ch)
Chris@16:           , this->ctype_->toupper(ch)
Chris@16:           , 0
Chris@16:         };
Chris@16:         if(ntcs[1] == ntcs[0])
Chris@16:             ntcs[1] = 0;
Chris@16:         return string_type(ntcs);
Chris@16:     }
Chris@16: 
Chris@16:     /// Checks to see if a character is within a character range.
Chris@16:     ///
Chris@16:     /// \param first The bottom of the range, inclusive.
Chris@16:     /// \param last The top of the range, inclusive.
Chris@16:     /// \param ch The source character.
Chris@16:     /// \return first <= ch && ch <= last.
Chris@16:     static bool in_range(char_type first, char_type last, char_type ch)
Chris@16:     {
Chris@16:         return first <= ch && ch <= last;
Chris@16:     }
Chris@16: 
Chris@16:     /// Checks to see if a character is within a character range, irregardless of case.
Chris@16:     ///
Chris@16:     /// \param first The bottom of the range, inclusive.
Chris@16:     /// \param last The top of the range, inclusive.
Chris@16:     /// \param ch The source character.
Chris@16:     /// \return in_range(first, last, ch) || in_range(first, last, tolower(ch, this->getloc())) ||
Chris@16:     ///     in_range(first, last, toupper(ch, this->getloc()))
Chris@16:     /// \attention The default implementation doesn't do proper Unicode
Chris@16:     ///     case folding, but this is the best we can do with the standard
Chris@16:     ///     ctype facet.
Chris@16:     bool in_range_nocase(char_type first, char_type last, char_type ch) const
Chris@16:     {
Chris@16:         // NOTE: this default implementation doesn't do proper Unicode
Chris@16:         // case folding, but this is the best we can do with the standard
Chris@16:         // std::ctype facet.
Chris@16:         return this->in_range(first, last, ch)
Chris@16:             || this->in_range(first, last, this->ctype_->toupper(ch))
Chris@16:             || this->in_range(first, last, this->ctype_->tolower(ch));
Chris@16:     }
Chris@16: 
Chris@16:     /// INTERNAL ONLY
Chris@16:     //string_type transform(char_type const *begin, char_type const *end) const
Chris@16:     //{
Chris@16:     //    return this->collate_->transform(begin, end);
Chris@16:     //}
Chris@16: 
Chris@16:     /// Returns a sort key for the character sequence designated by the iterator range [F1, F2)
Chris@16:     /// such that if the character sequence [G1, G2) sorts before the character sequence [H1, H2)
Chris@16:     /// then v.transform(G1, G2) \< v.transform(H1, H2).
Chris@16:     ///
Chris@16:     /// \attention Not currently used
Chris@16:     template<typename FwdIter>
Chris@16:     string_type transform(FwdIter, FwdIter) const
Chris@16:     {
Chris@16:         //string_type str(begin, end);
Chris@16:         //return this->transform(str.data(), str.data() + str.size());
Chris@16: 
Chris@16:         BOOST_ASSERT(false);
Chris@16:         return string_type();
Chris@16:     }
Chris@16: 
Chris@16:     /// Returns a sort key for the character sequence designated by the iterator range [F1, F2)
Chris@16:     /// such that if the character sequence [G1, G2) sorts before the character sequence [H1, H2)
Chris@16:     /// when character case is not considered then
Chris@16:     /// v.transform_primary(G1, G2) \< v.transform_primary(H1, H2).
Chris@16:     ///
Chris@16:     /// \attention Not currently used
Chris@16:     template<typename FwdIter>
Chris@16:     string_type transform_primary(FwdIter, FwdIter ) const
Chris@16:     {
Chris@16:         BOOST_ASSERT(false); // TODO implement me
Chris@16:         return string_type();
Chris@16:     }
Chris@16: 
Chris@16:     /// Returns a sequence of characters that represents the collating element
Chris@16:     /// consisting of the character sequence designated by the iterator range [F1, F2).
Chris@16:     /// Returns an empty string if the character sequence is not a valid collating element.
Chris@16:     ///
Chris@16:     /// \attention Not currently used
Chris@16:     template<typename FwdIter>
Chris@16:     string_type lookup_collatename(FwdIter, FwdIter) const
Chris@16:     {
Chris@16:         BOOST_ASSERT(false); // TODO implement me
Chris@16:         return string_type();
Chris@16:     }
Chris@16: 
Chris@16:     /// For the character class name represented by the specified character sequence,
Chris@16:     /// return the corresponding bitmask representation.
Chris@16:     ///
Chris@16:     /// \param begin A forward iterator to the start of the character sequence representing
Chris@16:     ///     the name of the character class.
Chris@16:     /// \param end The end of the character sequence.
Chris@16:     /// \param icase Specifies whether the returned bitmask should represent the case-insensitive
Chris@16:     ///     version of the character class.
Chris@16:     /// \return A bitmask representing the character class.
Chris@16:     template<typename FwdIter>
Chris@16:     char_class_type lookup_classname(FwdIter begin, FwdIter end, bool icase) const
Chris@16:     {
Chris@16:         static detail::umaskex_t const icase_masks =
Chris@16:             detail::std_ctype_lower | detail::std_ctype_upper;
Chris@16: 
Chris@16:         BOOST_ASSERT(begin != end);
Chris@16:         char_class_type char_class = this->lookup_classname_impl_(begin, end);
Chris@16:         if(0 == char_class)
Chris@16:         {
Chris@16:             // convert the string to lowercase
Chris@16:             string_type classname(begin, end);
Chris@16:             for(typename string_type::size_type i = 0, len = classname.size(); i < len; ++i)
Chris@16:             {
Chris@16:                 classname[i] = this->translate_nocase(classname[i]);
Chris@16:             }
Chris@16:             char_class = this->lookup_classname_impl_(classname.begin(), classname.end());
Chris@16:         }
Chris@16:         // erase case-sensitivity if icase==true
Chris@16:         if(icase && 0 != (char_class & icase_masks))
Chris@16:         {
Chris@16:             char_class |= icase_masks;
Chris@16:         }
Chris@16:         return char_class;
Chris@16:     }
Chris@16: 
Chris@16:     /// Tests a character against a character class bitmask.
Chris@16:     ///
Chris@16:     /// \param ch The character to test.
Chris@16:     /// \param mask The character class bitmask against which to test.
Chris@16:     /// \pre mask is a bitmask returned by lookup_classname, or is several such masks bit-or'ed
Chris@16:     ///     together.
Chris@16:     /// \return true if the character is a member of any of the specified character classes, false
Chris@16:     ///     otherwise.
Chris@16:     bool isctype(char_type ch, char_class_type mask) const
Chris@16:     {
Chris@16:         return this->base_type::is(*this->ctype_, ch, mask);
Chris@16:     }
Chris@16: 
Chris@16:     /// Convert a digit character into the integer it represents.
Chris@16:     ///
Chris@16:     /// \param ch The digit character.
Chris@16:     /// \param radix The radix to use for the conversion.
Chris@16:     /// \pre radix is one of 8, 10, or 16.
Chris@16:     /// \return -1 if ch is not a digit character, the integer value of the character otherwise.
Chris@16:     ///     The conversion is performed by imbueing a std::stringstream with this-\>getloc();
Chris@16:     ///     setting the radix to one of oct, hex or dec; inserting ch into the stream; and
Chris@16:     ///     extracting an int.
Chris@16:     int value(char_type ch, int radix) const
Chris@16:     {
Chris@16:         BOOST_ASSERT(8 == radix || 10 == radix || 16 == radix);
Chris@16:         int val = -1;
Chris@16:         std::basic_stringstream<char_type> str;
Chris@16:         str.imbue(this->getloc());
Chris@16:         str << (8 == radix ? std::oct : (16 == radix ? std::hex : std::dec));
Chris@16:         str.put(ch);
Chris@16:         str >> val;
Chris@16:         return str.fail() ? -1 : val;
Chris@16:     }
Chris@16: 
Chris@16:     /// Imbues *this with loc
Chris@16:     ///
Chris@16:     /// \param loc A std::locale.
Chris@16:     /// \return the previous std::locale used by *this.
Chris@16:     locale_type imbue(locale_type loc)
Chris@16:     {
Chris@16:         locale_type old_loc = this->loc_;
Chris@16:         this->loc_ = loc;
Chris@16:         this->ctype_ = &BOOST_USE_FACET(std::ctype<char_type>, this->loc_);
Chris@16:         //this->collate_ = &BOOST_USE_FACET(std::collate<char_type>, this->loc_);
Chris@16:         this->base_type::imbue(this->loc_);
Chris@16:         return old_loc;
Chris@16:     }
Chris@16: 
Chris@16:     /// Returns the current std::locale used by *this.
Chris@16:     ///
Chris@16:     locale_type getloc() const
Chris@16:     {
Chris@16:         return this->loc_;
Chris@16:     }
Chris@16: 
Chris@16: private:
Chris@16: 
Chris@16:     ///////////////////////////////////////////////////////////////////////////////
Chris@16:     // char_class_pair
Chris@16:     /// INTERNAL ONLY
Chris@16:     struct char_class_pair
Chris@16:     {
Chris@16:         char_type const *class_name_;
Chris@16:         char_class_type class_type_;
Chris@16:     };
Chris@16: 
Chris@16:     ///////////////////////////////////////////////////////////////////////////////
Chris@16:     // char_class
Chris@16:     /// INTERNAL ONLY
Chris@16:     static char_class_pair const &char_class(std::size_t j)
Chris@16:     {
Chris@16:         static BOOST_CONSTEXPR_OR_CONST char_class_pair s_char_class_map[] =
Chris@16:         {
Chris@16:             { BOOST_XPR_CSTR_(char_type, "alnum"),  detail::std_ctype_alnum }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "alpha"),  detail::std_ctype_alpha }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "blank"),  detail::non_std_ctype_blank }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "cntrl"),  detail::std_ctype_cntrl }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "d"),      detail::std_ctype_digit }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "digit"),  detail::std_ctype_digit }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "graph"),  detail::std_ctype_graph }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "lower"),  detail::std_ctype_lower }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "newline"),detail::non_std_ctype_newline }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "print"),  detail::std_ctype_print }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "punct"),  detail::std_ctype_punct }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "s"),      detail::std_ctype_space }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "space"),  detail::std_ctype_space }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "upper"),  detail::std_ctype_upper }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "w"),      detail::std_ctype_alnum | detail::non_std_ctype_underscore }
Chris@16:           , { BOOST_XPR_CSTR_(char_type, "xdigit"), detail::std_ctype_xdigit }
Chris@16:           , { 0, 0 }
Chris@16:         };
Chris@16:         return s_char_class_map[j];
Chris@16:     }
Chris@16: 
Chris@16:     ///////////////////////////////////////////////////////////////////////////////
Chris@16:     // lookup_classname_impl
Chris@16:     /// INTERNAL ONLY
Chris@16:     template<typename FwdIter>
Chris@16:     static char_class_type lookup_classname_impl_(FwdIter begin, FwdIter end)
Chris@16:     {
Chris@16:         // find the classname
Chris@16:         typedef cpp_regex_traits<Char> this_t;
Chris@16:         for(std::size_t j = 0; 0 != this_t::char_class(j).class_name_; ++j)
Chris@16:         {
Chris@16:             if(this_t::compare_(this_t::char_class(j).class_name_, begin, end))
Chris@16:             {
Chris@16:                 return this_t::char_class(j).class_type_;
Chris@16:             }
Chris@16:         }
Chris@16:         return 0;
Chris@16:     }
Chris@16: 
Chris@16:     /// INTERNAL ONLY
Chris@16:     template<typename FwdIter>
Chris@16:     static bool compare_(char_type const *name, FwdIter begin, FwdIter end)
Chris@16:     {
Chris@16:         for(; *name && begin != end; ++name, ++begin)
Chris@16:         {
Chris@16:             if(*name != *begin)
Chris@16:             {
Chris@16:                 return false;
Chris@16:             }
Chris@16:         }
Chris@16:         return !*name && begin == end;
Chris@16:     }
Chris@16: 
Chris@16:     locale_type loc_;
Chris@16:     std::ctype<char_type> const *ctype_;
Chris@16:     //std::collate<char_type> const *collate_;
Chris@16: };
Chris@16: 
Chris@16: ///////////////////////////////////////////////////////////////////////////////
Chris@16: // cpp_regex_traits<>::hash specializations
Chris@16: template<>
Chris@16: inline unsigned char cpp_regex_traits<unsigned char>::hash(unsigned char ch)
Chris@16: {
Chris@16:     return ch;
Chris@16: }
Chris@16: 
Chris@16: template<>
Chris@16: inline unsigned char cpp_regex_traits<char>::hash(char ch)
Chris@16: {
Chris@16:     return static_cast<unsigned char>(ch);
Chris@16: }
Chris@16: 
Chris@16: template<>
Chris@16: inline unsigned char cpp_regex_traits<signed char>::hash(signed char ch)
Chris@16: {
Chris@16:     return static_cast<unsigned char>(ch);
Chris@16: }
Chris@16: 
Chris@16: #ifndef BOOST_XPRESSIVE_NO_WREGEX
Chris@16: template<>
Chris@16: inline unsigned char cpp_regex_traits<wchar_t>::hash(wchar_t ch)
Chris@16: {
Chris@16:     return static_cast<unsigned char>(ch);
Chris@16: }
Chris@16: #endif
Chris@16: 
Chris@16: // Narrow C++ traits has fold_case() member function.
Chris@16: template<>
Chris@16: struct has_fold_case<cpp_regex_traits<char> >
Chris@16:   : mpl::true_
Chris@16: {
Chris@16: };
Chris@16: 
Chris@16: 
Chris@16: }}
Chris@16: 
Chris@16: #endif