support/char_encoding/standard_wide.hpp

0001 /*=============================================================================
0002     Copyright (c) 2001-2011 Hartmut Kaiser
0003     Copyright (c) 2001-2011 Joel de Guzman
0004
0005     Distributed under the Boost Software License, Version 1.0. (See accompanying
0006     file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
0007 =============================================================================*/
0008 #if !defined(BOOST_SPIRIT_STANDARD_WIDE_NOVEMBER_10_2006_0913AM)
0009 #define BOOST_SPIRIT_STANDARD_WIDE_NOVEMBER_10_2006_0913AM
0010
0011 #if defined(_MSC_VER)
0012 #pragma once
0013 #endif
0014
0015 #include <cwctype>
0016 #include <string>
0017
0018 #include <boost/assert.hpp>
0019 #include <boost/cstdint.hpp>
0020 #include <boost/spirit/home/support/assert_msg.hpp>
0021
0022 #include <boost/type_traits/make_unsigned.hpp>
0023
0024 namespace boost { namespace spirit { namespace traits
0025 {
0026     template <std::size_t N>
0027     struct wchar_t_size
0028     {
0029         BOOST_SPIRIT_ASSERT_MSG(N == 1 || N == 2 || N == 4,
0030             not_supported_size_of_wchar_t, ());
0031     };
0032
0033     template <> struct wchar_t_size<1> { enum { mask = 0xff }; };
0034     template <> struct wchar_t_size<2> { enum { mask = 0xffff }; };
0035     template <> struct wchar_t_size<4> { enum { mask = 0xffffffff }; };
0036
0037 }}}
0038
0039 namespace boost { namespace spirit { namespace char_encoding
0040 {
0041     ///////////////////////////////////////////////////////////////////////////
0042     //  Test characters for specified conditions (using std wchar_t functions)
0043     ///////////////////////////////////////////////////////////////////////////
0044
0045     struct standard_wide
0046     {
0047         typedef wchar_t char_type;
0048         typedef wchar_t classify_type;
0049
0050         template <typename Char>
0051         static typename std::char_traits<Char>::int_type
0052         to_int_type(Char ch)
0053         {
0054             return std::char_traits<Char>::to_int_type(ch);
0055         }
0056
0057         template <typename Char>
0058         static Char
0059         to_char_type(typename std::char_traits<Char>::int_type ch)
0060         {
0061             return std::char_traits<Char>::to_char_type(ch);
0062         }
0063
0064         static bool
0065         ischar(int ch)
0066         {
0067             // we have to watch out for sign extensions (casting is there to
0068             // silence certain compilers complaining about signed/unsigned
0069             // mismatch)
0070             return (
0071                 std::size_t(0) ==
0072                     std::size_t(ch & ~traits::wchar_t_size<sizeof(wchar_t)>::mask) ||
0073                 std::size_t(~0) ==
0074                     std::size_t(ch | traits::wchar_t_size<sizeof(wchar_t)>::mask)
0075             ) != 0;     // any wchar_t, but no other bits set
0076         }
0077
0078         static bool
0079         isalnum(wchar_t ch)
0080         {
0081             using namespace std;
0082             return iswalnum(to_int_type(ch)) != 0;
0083         }
0084
0085         static bool
0086         isalpha(wchar_t ch)
0087         {
0088             using namespace std;
0089             return iswalpha(to_int_type(ch)) != 0;
0090         }
0091
0092         static bool
0093         iscntrl(wchar_t ch)
0094         {
0095             using namespace std;
0096             return iswcntrl(to_int_type(ch)) != 0;
0097         }
0098
0099         static bool
0100         isdigit(wchar_t ch)
0101         {
0102             using namespace std;
0103             return iswdigit(to_int_type(ch)) != 0;
0104         }
0105
0106         static bool
0107         isgraph(wchar_t ch)
0108         {
0109             using namespace std;
0110             return iswgraph(to_int_type(ch)) != 0;
0111         }
0112
0113         static bool
0114         islower(wchar_t ch)
0115         {
0116             using namespace std;
0117             return iswlower(to_int_type(ch)) != 0;
0118         }
0119
0120         static bool
0121         isprint(wchar_t ch)
0122         {
0123             using namespace std;
0124             return iswprint(to_int_type(ch)) != 0;
0125         }
0126
0127         static bool
0128         ispunct(wchar_t ch)
0129         {
0130             using namespace std;
0131             return iswpunct(to_int_type(ch)) != 0;
0132         }
0133
0134         static bool
0135         isspace(wchar_t ch)
0136         {
0137             using namespace std;
0138             return iswspace(to_int_type(ch)) != 0;
0139         }
0140
0141         static bool
0142         isupper(wchar_t ch)
0143         {
0144             using namespace std;
0145             return iswupper(to_int_type(ch)) != 0;
0146         }
0147
0148         static bool
0149         isxdigit(wchar_t ch)
0150         {
0151             using namespace std;
0152             return iswxdigit(to_int_type(ch)) != 0;
0153         }
0154
0155         static bool
0156         isblank BOOST_PREVENT_MACRO_SUBSTITUTION (wchar_t ch)
0157         {
0158             return (ch == L' ' || ch == L'\t');
0159         }
0160
0161         ///////////////////////////////////////////////////////////////////////
0162         //  Simple character conversions
0163         ///////////////////////////////////////////////////////////////////////
0164
0165         static wchar_t
0166         tolower(wchar_t ch)
0167         {
0168             using namespace std;
0169             return isupper(ch) ?
0170                 to_char_type<wchar_t>(towlower(to_int_type(ch))) : ch;
0171         }
0172
0173         static wchar_t
0174         toupper(wchar_t ch)
0175         {
0176             using namespace std;
0177             return islower(ch) ?
0178                 to_char_type<wchar_t>(towupper(to_int_type(ch))) : ch;
0179         }
0180
0181         static ::boost::uint32_t
0182         toucs4(wchar_t ch)
0183         {
0184             return static_cast<make_unsigned<wchar_t>::type>(ch);
0185         }
0186     };
0187 }}}
0188
0189 #endif