cpp_regex_traits.hpp

来自「support vector clustering for vc++」· HPP 代码 · 共 677 行 · 第 1/2 页

///////////////////////////////////////////////////////////////////////////////
/// \file cpp_regex_traits.hpp
/// Contains the definition of the cpp_regex_traits\<\> template, which is a 
/// wrapper for std::locale that can be used to customize the behavior of
/// static and dynamic regexes.
//
//  Copyright 2004 Eric Niebler. Distributed under the Boost
//  Software License, Version 1.0. (See accompanying file
//  LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)

#ifndef BOOST_XPRESSIVE_TRAITS_CPP_REGEX_TRAITS_HPP_EAN_10_04_2005
#define BOOST_XPRESSIVE_TRAITS_CPP_REGEX_TRAITS_HPP_EAN_10_04_2005

// MS compatible compilers support #pragma once
#if defined(_MSC_VER) && (_MSC_VER >= 1020)
# pragma once
#endif

#include <string>
#include <locale>
#include <sstream>
#include <boost/config.hpp>
#include <boost/assert.hpp>
#include <boost/integer.hpp>
#include <boost/mpl/assert.hpp>
#include <boost/type_traits/is_same.hpp>
#include <boost/xpressive/detail/detail_fwd.hpp>
#include <boost/xpressive/detail/utility/literals.hpp>

// From John Maddock:
// Fix for gcc prior to 3.4: std::ctype<wchar_t> doesn't allow masks to be combined, for example:
// std::use_facet<std::ctype<wchar_t> >(locale()).is(std::ctype_base::lower|std::ctype_base::upper, L'a');
// incorrectly returns false.
// NOTE: later version of the gcc define __GLIBCXX__, not __GLIBCPP__
#if BOOST_WORKAROUND(__GLIBCPP__, != 0)
# define BOOST_XPRESSIVE_BUGGY_CTYPE_FACET
#endif

namespace boost { namespace xpressive
{

namespace detail
{
    // define an unsigned integral typedef of the same size as std::ctype_base::mask
    typedef boost::uint_t<sizeof(std::ctype_base::mask) * CHAR_BIT>::least umask_t;
    BOOST_MPL_ASSERT_RELATION(sizeof(std::ctype_base::mask), ==, sizeof(umask_t));

    // Calculate what the size of the umaskex_t type should be to fix the 3 extra bitmasks
    //   11 char categories in ctype_base
    // +  3 extra categories for xpressive
    // = 14 total bits needed
    int const umaskex_bits = (14 > (sizeof(umask_t) * CHAR_BIT)) ? 14 : sizeof(umask_t) * CHAR_BIT;

    // define an unsigned integral type with at least umaskex_bits
    typedef boost::uint_t<umaskex_bits>::fast umaskex_t;
    BOOST_MPL_ASSERT_RELATION(sizeof(umask_t), <=, sizeof(umaskex_t));

    // cast a ctype mask to a umaskex_t
    template<std::ctype_base::mask Mask>
    struct mask_cast
    {
        BOOST_STATIC_CONSTANT(umaskex_t, value = static_cast<umask_t>(Mask));
    };

    #ifdef __CYGWIN__
    // Work around a gcc warning on cygwin
    template<>
    struct mask_cast<std::ctype_base::print>
    {
        BOOST_MPL_ASSERT_RELATION('\227', ==, std::ctype_base::print);
        BOOST_STATIC_CONSTANT(umaskex_t, value = 0227);
    };
    #endif

    #ifndef BOOST_NO_INCLASS_MEMBER_INITIALIZATION
    template<std::ctype_base::mask Mask>
    umaskex_t const mask_cast<Mask>::value;
    #endif

    #ifndef BOOST_XPRESSIVE_BUGGY_CTYPE_FACET
    // an unsigned integer with the highest bit set
    umaskex_t const highest_bit = 1 << (sizeof(umaskex_t) * CHAR_BIT - 1);

    ///////////////////////////////////////////////////////////////////////////////
    // unused_mask
    //   find a bit in an int that isn't set
    template<umaskex_t In, umaskex_t Out = highest_bit, bool Done = (0 == (Out & In))>
    struct unused_mask
    {
        BOOST_MPL_ASSERT_RELATION(1, !=, Out);
        BOOST_STATIC_CONSTANT(umaskex_t, value = (unused_mask<In, (Out >> 1)>::value));
    };

    template<umaskex_t In, umaskex_t Out>
    struct unused_mask<In, Out, true>
    {
        BOOST_STATIC_CONSTANT(umaskex_t, value = Out);
    };

    #ifndef BOOST_NO_INCLASS_MEMBER_INITIALIZATION
    template<umaskex_t In, umaskex_t Out, bool Done>
    umaskex_t const unused_mask<In, Out, Done>::value;
    #endif

    umaskex_t const std_ctype_alnum = mask_cast<std::ctype_base::alnum>::value;
    umaskex_t const std_ctype_alpha = mask_cast<std::ctype_base::alpha>::value;
    umaskex_t const std_ctype_cntrl = mask_cast<std::ctype_base::cntrl>::value;
    umaskex_t const std_ctype_digit = mask_cast<std::ctype_base::digit>::value;
    umaskex_t const std_ctype_graph = mask_cast<std::ctype_base::graph>::value;
    umaskex_t const std_ctype_lower = mask_cast<std::ctype_base::lower>::value;
    umaskex_t const std_ctype_print = mask_cast<std::ctype_base::print>::value;
    umaskex_t const std_ctype_punct = mask_cast<std::ctype_base::punct>::value;
    umaskex_t const std_ctype_space = mask_cast<std::ctype_base::space>::value;
    umaskex_t const std_ctype_upper = mask_cast<std::ctype_base::upper>::value;
    umaskex_t const std_ctype_xdigit = mask_cast<std::ctype_base::xdigit>::value;

    // Reserve some bits for the implementation
    #if defined(__GLIBCXX__) && __GLIBCXX__ >= 20050209
    umaskex_t const std_ctype_reserved = 0x8000;
    #else
    umaskex_t const std_ctype_reserved = 0;
    #endif

    // Bitwise-or all the ctype masks together
    umaskex_t const all_ctype_masks = std_ctype_reserved
      | std_ctype_alnum | std_ctype_alpha | std_ctype_cntrl | std_ctype_digit
      | std_ctype_graph | std_ctype_lower | std_ctype_print | std_ctype_punct
      | std_ctype_space | std_ctype_upper | std_ctype_xdigit;

    // define a new mask for "underscore" ("word" == alnum | underscore)
    umaskex_t const non_std_ctype_underscore = unused_mask<all_ctype_masks>::value;

    // define a new mask for "blank"
    umaskex_t const non_std_ctype_blank = unused_mask<all_ctype_masks | non_std_ctype_underscore>::value;

    // define a new mask for "newline"
    umaskex_t const non_std_ctype_newline = unused_mask<all_ctype_masks | non_std_ctype_underscore | non_std_ctype_blank>::value;

    #else
    ///////////////////////////////////////////////////////////////////////////////
    // Ugly work-around for buggy ctype facets.
    umaskex_t const std_ctype_alnum = 1 << 0;
    umaskex_t const std_ctype_alpha = 1 << 1;
    umaskex_t const std_ctype_cntrl = 1 << 2;
    umaskex_t const std_ctype_digit = 1 << 3;
    umaskex_t const std_ctype_graph = 1 << 4;
    umaskex_t const std_ctype_lower = 1 << 5;
    umaskex_t const std_ctype_print = 1 << 6;
    umaskex_t const std_ctype_punct = 1 << 7;
    umaskex_t const std_ctype_space = 1 << 8;
    umaskex_t const std_ctype_upper = 1 << 9;
    umaskex_t const std_ctype_xdigit = 1 << 10;
    umaskex_t const non_std_ctype_underscore = 1 << 11;
    umaskex_t const non_std_ctype_blank = 1 << 12;
    umaskex_t const non_std_ctype_newline = 1 << 13;

    static umaskex_t const std_masks[] = 
    {
        mask_cast<std::ctype_base::alnum>::value
      , mask_cast<std::ctype_base::alpha>::value
      , mask_cast<std::ctype_base::cntrl>::value
      , mask_cast<std::ctype_base::digit>::value
      , mask_cast<std::ctype_base::graph>::value
      , mask_cast<std::ctype_base::lower>::value
      , mask_cast<std::ctype_base::print>::value
      , mask_cast<std::ctype_base::punct>::value
      , mask_cast<std::ctype_base::space>::value
      , mask_cast<std::ctype_base::upper>::value
      , mask_cast<std::ctype_base::xdigit>::value
    };

    inline int mylog2(umaskex_t i)
    {
        return "\0\0\1\0\2\0\0\0\3"[i & 0xf]
             + "\0\4\5\0\6\0\0\0\7"[(i & 0xf0) >> 04]
             + "\0\10\11\0\12\0\0\0\13"[(i & 0xf00) >> 010];
    }
    #endif

    // convenient constant for the extra masks
    umaskex_t const non_std_ctype_masks = non_std_ctype_underscore | non_std_ctype_blank | non_std_ctype_newline;

    ///////////////////////////////////////////////////////////////////////////////
    // cpp_regex_traits_base
    //   BUGBUG this should be replaced with a regex facet that lets you query for
    //   an array of underscore characters and an array of line separator characters.
    template<typename Char, std::size_t SizeOfChar = sizeof(Char)>
    struct cpp_regex_traits_base
    {
    protected:
        void imbue(std::locale const &)
        {
        }

        static bool is(std::ctype<Char> const &ct, Char ch, umaskex_t mask)
        {
            #ifndef BOOST_XPRESSIVE_BUGGY_CTYPE_FACET

            if(ct.is((std::ctype_base::mask)(umask_t)mask, ch))
            {
                return true;
            }

            #else

            umaskex_t tmp = mask & ~non_std_ctype_masks;
            for(umaskex_t i; 0 != (i = (tmp & (~tmp+1))); tmp &= ~i)
            {
                std::ctype_base::mask m = (std::ctype_base::mask)(umask_t)std_masks[mylog2(i)];
                if(ct.is(m, ch))
                {
                    return true;
                }
            }

            #endif

            return ((mask & non_std_ctype_blank) && cpp_regex_traits_base::is_blank(ch))
                || ((mask & non_std_ctype_underscore) && cpp_regex_traits_base::is_underscore(ch))
                || ((mask & non_std_ctype_newline) && cpp_regex_traits_base::is_newline(ch));
        }

    private:
        static bool is_blank(Char ch)
        {
            BOOST_MPL_ASSERT_RELATION('\t', ==, L'\t');
            return L'\t' == ch;
        }

        static bool is_underscore(Char ch)
        {
            BOOST_MPL_ASSERT_RELATION('_', ==, L'_');
            return L'_' == ch;
        }

        static bool is_newline(Char ch)
        {
            BOOST_MPL_ASSERT_RELATION('\r', ==, L'\r');
            BOOST_MPL_ASSERT_RELATION('\n', ==, L'\n');
            BOOST_MPL_ASSERT_RELATION('\f', ==, L'\f');
            return L'\r' == ch || L'\n' == ch || L'\f' == ch
                || (1 < SizeOfChar && (0x2028u == ch || 0x2029u == ch || 0x85u == ch));
        }
    };

    #ifndef BOOST_XPRESSIVE_BUGGY_CTYPE_FACET

    template<typename Char>
    struct cpp_regex_traits_base<Char, 1>
    {
    protected:
        void imbue(std::locale const &loc)
        {
            int i = 0;
            Char allchars[UCHAR_MAX + 1];
            for(i = 0; i <= UCHAR_MAX; ++i)
            {
                allchars[i] = static_cast<Char>(i);
            }

            std::ctype<Char> const &ct = BOOST_USE_FACET(std::ctype<Char>, loc);
            std::ctype_base::mask tmp[UCHAR_MAX + 1];
            ct.is(allchars, allchars + UCHAR_MAX + 1, tmp);
            for(i = 0; i <= UCHAR_MAX; ++i)
            {
                this->masks_[i] = static_cast<umask_t>(tmp[i]);
                BOOST_ASSERT(0 == (this->masks_[i] & non_std_ctype_masks));
            }

            this->masks_[static_cast<unsigned char>('_')] |= non_std_ctype_underscore;
            this->masks_[static_cast<unsigned char>(' ')] |= non_std_ctype_blank;
            this->masks_[static_cast<unsigned char>('\t')] |= non_std_ctype_blank;
            this->masks_[static_cast<unsigned char>('\n')] |= non_std_ctype_newline;
            this->masks_[static_cast<unsigned char>('\r')] |= non_std_ctype_newline;
            this->masks_[static_cast<unsigned char>('\f')] |= non_std_ctype_newline;
        }
 
        bool is(std::ctype<Char> const &, Char ch, umaskex_t mask) const
        {
            return 0 != (this->masks_[static_cast<unsigned char>(ch)] & mask);
        }

    private:
        umaskex_t masks_[UCHAR_MAX + 1];
    };

    #endif

    template<typename Char>
    struct version_tag
    {
        typedef regex_traits_version_1_tag type;
    };

    template<>
    struct version_tag<char>
    {
        typedef regex_traits_version_1_case_fold_tag type;
    };

} // namespace detail

///////////////////////////////////////////////////////////////////////////////
// cpp_regex_traits
//
/// \brief Encapsaulates a std::locale for use by the 
/// basic_regex\<\> class template.
template<typename Char>
struct cpp_regex_traits
  : detail::cpp_regex_traits_base<Char>
{
    typedef Char char_type;
    typedef std::basic_string<char_type> string_type;
    typedef std::locale locale_type;
    typedef detail::umaskex_t char_class_type;
    typedef typename detail::version_tag<Char>::type version_tag;
    typedef detail::cpp_regex_traits_base<Char> base_type;

    /// Initialize a cpp_regex_traits object to use the specified std::locale,
    /// or the global std::locale if none is specified.
    ///
    cpp_regex_traits(locale_type const &loc = locale_type())
      : base_type()
      , loc_()
    {
        this->imbue(loc);
    }

    /// Checks two cpp_regex_traits objects for equality
    ///
    /// \return this->getloc() == that.getloc().
    bool operator ==(cpp_regex_traits<char_type> const &that) const
    {
        return this->loc_ == that.loc_;
    }

    /// Checks two cpp_regex_traits objects for inequality
    ///
    /// \return this->getloc() != that.getloc().