support/char_encoding/unicode.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_UNICODE_1_JANUARY_12_2010_0728PM)
0009 #define BOOST_SPIRIT_UNICODE_1_JANUARY_12_2010_0728PM
0010
0011 #if defined(_MSC_VER)
0012 #pragma once
0013 #endif
0014
0015 #include <boost/cstdint.hpp>
0016 #include <boost/spirit/home/support/char_encoding/unicode/query.hpp>
0017
0018 namespace boost { namespace spirit { namespace char_encoding
0019 {
0020     ///////////////////////////////////////////////////////////////////////////
0021     //  Test characters for specified conditions (using iso8859-1)
0022     ///////////////////////////////////////////////////////////////////////////
0023     struct unicode
0024     {
0025 #ifdef BOOST_NO_CXX11_CHAR32_T
0026         typedef ::boost::uint32_t char_type;
0027 #else
0028         typedef char32_t char_type;
0029 #endif
0030         typedef ::boost::uint32_t classify_type;
0031
0032     ///////////////////////////////////////////////////////////////////////////
0033     //  Posix stuff
0034     ///////////////////////////////////////////////////////////////////////////
0035         static bool
0036         isascii_(char_type ch)
0037         {
0038             return 0 == (ch & ~0x7f);
0039         }
0040
0041         static bool
0042         ischar(char_type ch)
0043         {
0044             // unicode code points in the range 0x00 to 0x10FFFF
0045             return ch <= 0x10FFFF;
0046         }
0047
0048         static bool
0049         isalnum(char_type ch)
0050         {
0051             return ucd::is_alphanumeric(ch);
0052         }
0053
0054         static bool
0055         isalpha(char_type ch)
0056         {
0057             return ucd::is_alphabetic(ch);
0058         }
0059
0060         static bool
0061         isdigit(char_type ch)
0062         {
0063             return ucd::is_decimal_number(ch);
0064         }
0065
0066         static bool
0067         isxdigit(char_type ch)
0068         {
0069             return ucd::is_hex_digit(ch);
0070         }
0071
0072         static bool
0073         iscntrl(char_type ch)
0074         {
0075             return ucd::is_control(ch);
0076         }
0077
0078         static bool
0079         isgraph(char_type ch)
0080         {
0081             return ucd::is_graph(ch);
0082         }
0083
0084         static bool
0085         islower(char_type ch)
0086         {
0087             return ucd::is_lowercase(ch);
0088         }
0089
0090         static bool
0091         isprint(char_type ch)
0092         {
0093             return ucd::is_print(ch);
0094         }
0095
0096         static bool
0097         ispunct(char_type ch)
0098         {
0099             return ucd::is_punctuation(ch);
0100         }
0101
0102         static bool
0103         isspace(char_type ch)
0104         {
0105             return ucd::is_white_space(ch);
0106         }
0107
0108         static bool
0109         isblank BOOST_PREVENT_MACRO_SUBSTITUTION (char_type ch)
0110         {
0111             return ucd::is_blank(ch);
0112         }
0113
0114         static bool
0115         isupper(char_type ch)
0116         {
0117             return ucd::is_uppercase(ch);
0118         }
0119
0120     ///////////////////////////////////////////////////////////////////////////
0121     //  Simple character conversions
0122     ///////////////////////////////////////////////////////////////////////////
0123
0124         static char_type
0125         tolower(char_type ch)
0126         {
0127             return ucd::to_lowercase(ch);
0128         }
0129
0130         static char_type
0131         toupper(char_type ch)
0132         {
0133             return ucd::to_uppercase(ch);
0134         }
0135
0136         static ::boost::uint32_t
0137         toucs4(char_type ch)
0138         {
0139             return ch;
0140         }
0141
0142     ///////////////////////////////////////////////////////////////////////////
0143     //  Major Categories
0144     ///////////////////////////////////////////////////////////////////////////
0145 #define BOOST_SPIRIT_MAJOR_CATEGORY(name)                                       \
0146         static bool                                                             \
0147         is_##name(char_type ch)                                                 \
0148         {                                                                       \
0149             return ucd::get_major_category(ch) == ucd::properties::name;        \
0150         }                                                                       \
0151         /***/
0152
0153         BOOST_SPIRIT_MAJOR_CATEGORY(letter)
0154         BOOST_SPIRIT_MAJOR_CATEGORY(mark)
0155         BOOST_SPIRIT_MAJOR_CATEGORY(number)
0156         BOOST_SPIRIT_MAJOR_CATEGORY(separator)
0157         BOOST_SPIRIT_MAJOR_CATEGORY(other)
0158         BOOST_SPIRIT_MAJOR_CATEGORY(punctuation)
0159         BOOST_SPIRIT_MAJOR_CATEGORY(symbol)
0160
0161     ///////////////////////////////////////////////////////////////////////////
0162     //  General Categories
0163     ///////////////////////////////////////////////////////////////////////////
0164 #define BOOST_SPIRIT_CATEGORY(name)                                             \
0165         static bool                                                             \
0166         is_##name(char_type ch)                                                 \
0167         {                                                                       \
0168             return ucd::get_category(ch) == ucd::properties::name;              \
0169         }                                                                       \
0170         /***/
0171
0172         BOOST_SPIRIT_CATEGORY(uppercase_letter)
0173         BOOST_SPIRIT_CATEGORY(lowercase_letter)
0174         BOOST_SPIRIT_CATEGORY(titlecase_letter)
0175         BOOST_SPIRIT_CATEGORY(modifier_letter)
0176         BOOST_SPIRIT_CATEGORY(other_letter)
0177
0178         BOOST_SPIRIT_CATEGORY(nonspacing_mark)
0179         BOOST_SPIRIT_CATEGORY(enclosing_mark)
0180         BOOST_SPIRIT_CATEGORY(spacing_mark)
0181
0182         BOOST_SPIRIT_CATEGORY(decimal_number)
0183         BOOST_SPIRIT_CATEGORY(letter_number)
0184         BOOST_SPIRIT_CATEGORY(other_number)
0185
0186         BOOST_SPIRIT_CATEGORY(space_separator)
0187         BOOST_SPIRIT_CATEGORY(line_separator)
0188         BOOST_SPIRIT_CATEGORY(paragraph_separator)
0189
0190         BOOST_SPIRIT_CATEGORY(control)
0191         BOOST_SPIRIT_CATEGORY(format)
0192         BOOST_SPIRIT_CATEGORY(private_use)
0193         BOOST_SPIRIT_CATEGORY(surrogate)
0194         BOOST_SPIRIT_CATEGORY(unassigned)
0195
0196         BOOST_SPIRIT_CATEGORY(dash_punctuation)
0197         BOOST_SPIRIT_CATEGORY(open_punctuation)
0198         BOOST_SPIRIT_CATEGORY(close_punctuation)
0199         BOOST_SPIRIT_CATEGORY(connector_punctuation)
0200         BOOST_SPIRIT_CATEGORY(other_punctuation)
0201         BOOST_SPIRIT_CATEGORY(initial_punctuation)
0202         BOOST_SPIRIT_CATEGORY(final_punctuation)
0203
0204         BOOST_SPIRIT_CATEGORY(math_symbol)
0205         BOOST_SPIRIT_CATEGORY(currency_symbol)
0206         BOOST_SPIRIT_CATEGORY(modifier_symbol)
0207         BOOST_SPIRIT_CATEGORY(other_symbol)
0208
0209     ///////////////////////////////////////////////////////////////////////////
0210     //  Derived Categories
0211     ///////////////////////////////////////////////////////////////////////////
0212 #define BOOST_SPIRIT_DERIVED_CATEGORY(name)                                     \
0213         static bool                                                             \
0214         is_##name(char_type ch)                                                 \
0215         {                                                                       \
0216             return ucd::is_##name(ch);                                          \
0217         }                                                                       \
0218         /***/
0219
0220         BOOST_SPIRIT_DERIVED_CATEGORY(alphabetic)
0221         BOOST_SPIRIT_DERIVED_CATEGORY(uppercase)
0222         BOOST_SPIRIT_DERIVED_CATEGORY(lowercase)
0223         BOOST_SPIRIT_DERIVED_CATEGORY(white_space)
0224         BOOST_SPIRIT_DERIVED_CATEGORY(hex_digit)
0225         BOOST_SPIRIT_DERIVED_CATEGORY(noncharacter_code_point)
0226         BOOST_SPIRIT_DERIVED_CATEGORY(default_ignorable_code_point)
0227
0228     ///////////////////////////////////////////////////////////////////////////
0229     //  Scripts
0230     ///////////////////////////////////////////////////////////////////////////
0231 #define BOOST_SPIRIT_SCRIPT(name)                                               \
0232         static bool                                                             \
0233         is_##name(char_type ch)                                                 \
0234         {                                                                       \
0235             return ucd::get_script(ch) == ucd::properties::name;                \
0236         }                                                                       \
0237         /***/
0238
0239         BOOST_SPIRIT_SCRIPT(adlam)
0240         BOOST_SPIRIT_SCRIPT(caucasian_albanian)
0241         BOOST_SPIRIT_SCRIPT(ahom)
0242         BOOST_SPIRIT_SCRIPT(arabic)
0243         BOOST_SPIRIT_SCRIPT(imperial_aramaic)
0244         BOOST_SPIRIT_SCRIPT(armenian)
0245         BOOST_SPIRIT_SCRIPT(avestan)
0246         BOOST_SPIRIT_SCRIPT(balinese)
0247         BOOST_SPIRIT_SCRIPT(bamum)
0248         BOOST_SPIRIT_SCRIPT(bassa_vah)
0249         BOOST_SPIRIT_SCRIPT(batak)
0250         BOOST_SPIRIT_SCRIPT(bengali)
0251         BOOST_SPIRIT_SCRIPT(bhaiksuki)
0252         BOOST_SPIRIT_SCRIPT(bopomofo)
0253         BOOST_SPIRIT_SCRIPT(brahmi)
0254         BOOST_SPIRIT_SCRIPT(braille)
0255         BOOST_SPIRIT_SCRIPT(buginese)
0256         BOOST_SPIRIT_SCRIPT(buhid)
0257         BOOST_SPIRIT_SCRIPT(chakma)
0258         BOOST_SPIRIT_SCRIPT(canadian_aboriginal)
0259         BOOST_SPIRIT_SCRIPT(carian)
0260         BOOST_SPIRIT_SCRIPT(cham)
0261         BOOST_SPIRIT_SCRIPT(cherokee)
0262         BOOST_SPIRIT_SCRIPT(chorasmian)
0263         BOOST_SPIRIT_SCRIPT(coptic)
0264         BOOST_SPIRIT_SCRIPT(cypro_minoan)
0265         BOOST_SPIRIT_SCRIPT(cypriot)
0266         BOOST_SPIRIT_SCRIPT(cyrillic)
0267         BOOST_SPIRIT_SCRIPT(devanagari)
0268         BOOST_SPIRIT_SCRIPT(dives_akuru)
0269         BOOST_SPIRIT_SCRIPT(dogra)
0270         BOOST_SPIRIT_SCRIPT(deseret)
0271         BOOST_SPIRIT_SCRIPT(duployan)
0272         BOOST_SPIRIT_SCRIPT(egyptian_hieroglyphs)
0273         BOOST_SPIRIT_SCRIPT(elbasan)
0274         BOOST_SPIRIT_SCRIPT(elymaic)
0275         BOOST_SPIRIT_SCRIPT(ethiopic)
0276         BOOST_SPIRIT_SCRIPT(georgian)
0277         BOOST_SPIRIT_SCRIPT(glagolitic)
0278         BOOST_SPIRIT_SCRIPT(gunjala_gondi)
0279         BOOST_SPIRIT_SCRIPT(masaram_gondi)
0280         BOOST_SPIRIT_SCRIPT(gothic)
0281         BOOST_SPIRIT_SCRIPT(grantha)
0282         BOOST_SPIRIT_SCRIPT(greek)
0283         BOOST_SPIRIT_SCRIPT(gujarati)
0284         BOOST_SPIRIT_SCRIPT(gurmukhi)
0285         BOOST_SPIRIT_SCRIPT(hangul)
0286         BOOST_SPIRIT_SCRIPT(han)
0287         BOOST_SPIRIT_SCRIPT(hanunoo)
0288         BOOST_SPIRIT_SCRIPT(hatran)
0289         BOOST_SPIRIT_SCRIPT(hebrew)
0290         BOOST_SPIRIT_SCRIPT(hiragana)
0291         BOOST_SPIRIT_SCRIPT(anatolian_hieroglyphs)
0292         BOOST_SPIRIT_SCRIPT(pahawh_hmong)
0293         BOOST_SPIRIT_SCRIPT(nyiakeng_puachue_hmong)
0294         BOOST_SPIRIT_SCRIPT(katakana_or_hiragana)
0295         BOOST_SPIRIT_SCRIPT(old_hungarian)
0296         BOOST_SPIRIT_SCRIPT(old_italic)
0297         BOOST_SPIRIT_SCRIPT(javanese)
0298         BOOST_SPIRIT_SCRIPT(kayah_li)
0299         BOOST_SPIRIT_SCRIPT(katakana)
0300         BOOST_SPIRIT_SCRIPT(kawi)
0301         BOOST_SPIRIT_SCRIPT(kharoshthi)
0302         BOOST_SPIRIT_SCRIPT(khmer)
0303         BOOST_SPIRIT_SCRIPT(khojki)
0304         BOOST_SPIRIT_SCRIPT(khitan_small_script)
0305         BOOST_SPIRIT_SCRIPT(kannada)
0306         BOOST_SPIRIT_SCRIPT(kaithi)
0307         BOOST_SPIRIT_SCRIPT(tai_tham)
0308         BOOST_SPIRIT_SCRIPT(lao)
0309         BOOST_SPIRIT_SCRIPT(latin)
0310         BOOST_SPIRIT_SCRIPT(lepcha)
0311         BOOST_SPIRIT_SCRIPT(limbu)
0312         BOOST_SPIRIT_SCRIPT(linear_a)
0313         BOOST_SPIRIT_SCRIPT(linear_b)
0314         BOOST_SPIRIT_SCRIPT(lisu)
0315         BOOST_SPIRIT_SCRIPT(lycian)
0316         BOOST_SPIRIT_SCRIPT(lydian)
0317         BOOST_SPIRIT_SCRIPT(mahajani)
0318         BOOST_SPIRIT_SCRIPT(makasar)
0319         BOOST_SPIRIT_SCRIPT(mandaic)
0320         BOOST_SPIRIT_SCRIPT(manichaean)
0321         BOOST_SPIRIT_SCRIPT(marchen)
0322         BOOST_SPIRIT_SCRIPT(medefaidrin)
0323         BOOST_SPIRIT_SCRIPT(mende_kikakui)
0324         BOOST_SPIRIT_SCRIPT(meroitic_cursive)
0325         BOOST_SPIRIT_SCRIPT(meroitic_hieroglyphs)
0326         BOOST_SPIRIT_SCRIPT(malayalam)
0327         BOOST_SPIRIT_SCRIPT(modi)
0328         BOOST_SPIRIT_SCRIPT(mongolian)
0329         BOOST_SPIRIT_SCRIPT(mro)
0330         BOOST_SPIRIT_SCRIPT(meetei_mayek)
0331         BOOST_SPIRIT_SCRIPT(multani)
0332         BOOST_SPIRIT_SCRIPT(myanmar)
0333         BOOST_SPIRIT_SCRIPT(nag_mundari)
0334         BOOST_SPIRIT_SCRIPT(nandinagari)
0335         BOOST_SPIRIT_SCRIPT(old_north_arabian)
0336         BOOST_SPIRIT_SCRIPT(nabataean)
0337         BOOST_SPIRIT_SCRIPT(newa)
0338         BOOST_SPIRIT_SCRIPT(nko)
0339         BOOST_SPIRIT_SCRIPT(nushu)
0340         BOOST_SPIRIT_SCRIPT(ogham)
0341         BOOST_SPIRIT_SCRIPT(ol_chiki)
0342         BOOST_SPIRIT_SCRIPT(old_turkic)
0343         BOOST_SPIRIT_SCRIPT(oriya)
0344         BOOST_SPIRIT_SCRIPT(osage)
0345         BOOST_SPIRIT_SCRIPT(osmanya)
0346         BOOST_SPIRIT_SCRIPT(old_uyghur)
0347         BOOST_SPIRIT_SCRIPT(palmyrene)
0348         BOOST_SPIRIT_SCRIPT(pau_cin_hau)
0349         BOOST_SPIRIT_SCRIPT(old_permic)
0350         BOOST_SPIRIT_SCRIPT(phags_pa)
0351         BOOST_SPIRIT_SCRIPT(inscriptional_pahlavi)
0352         BOOST_SPIRIT_SCRIPT(psalter_pahlavi)
0353         BOOST_SPIRIT_SCRIPT(phoenician)
0354         BOOST_SPIRIT_SCRIPT(miao)
0355         BOOST_SPIRIT_SCRIPT(inscriptional_parthian)
0356         BOOST_SPIRIT_SCRIPT(rejang)
0357         BOOST_SPIRIT_SCRIPT(hanifi_rohingya)
0358         BOOST_SPIRIT_SCRIPT(runic)
0359         BOOST_SPIRIT_SCRIPT(samaritan)
0360         BOOST_SPIRIT_SCRIPT(old_south_arabian)
0361         BOOST_SPIRIT_SCRIPT(saurashtra)
0362         BOOST_SPIRIT_SCRIPT(signwriting)
0363         BOOST_SPIRIT_SCRIPT(shavian)
0364         BOOST_SPIRIT_SCRIPT(sharada)
0365         BOOST_SPIRIT_SCRIPT(siddham)
0366         BOOST_SPIRIT_SCRIPT(khudawadi)
0367         BOOST_SPIRIT_SCRIPT(sinhala)
0368         BOOST_SPIRIT_SCRIPT(sogdian)
0369         BOOST_SPIRIT_SCRIPT(old_sogdian)
0370         BOOST_SPIRIT_SCRIPT(sora_sompeng)
0371         BOOST_SPIRIT_SCRIPT(soyombo)
0372         BOOST_SPIRIT_SCRIPT(sundanese)
0373         BOOST_SPIRIT_SCRIPT(syloti_nagri)
0374         BOOST_SPIRIT_SCRIPT(syriac)
0375         BOOST_SPIRIT_SCRIPT(tagbanwa)
0376         BOOST_SPIRIT_SCRIPT(takri)
0377         BOOST_SPIRIT_SCRIPT(tai_le)
0378         BOOST_SPIRIT_SCRIPT(new_tai_lue)
0379         BOOST_SPIRIT_SCRIPT(tamil)
0380         BOOST_SPIRIT_SCRIPT(tangut)
0381         BOOST_SPIRIT_SCRIPT(tai_viet)
0382         BOOST_SPIRIT_SCRIPT(telugu)
0383         BOOST_SPIRIT_SCRIPT(tifinagh)
0384         BOOST_SPIRIT_SCRIPT(tagalog)
0385         BOOST_SPIRIT_SCRIPT(thaana)
0386         BOOST_SPIRIT_SCRIPT(thai)
0387         BOOST_SPIRIT_SCRIPT(tibetan)
0388         BOOST_SPIRIT_SCRIPT(tirhuta)
0389         BOOST_SPIRIT_SCRIPT(tangsa)
0390         BOOST_SPIRIT_SCRIPT(toto)
0391         BOOST_SPIRIT_SCRIPT(ugaritic)
0392         BOOST_SPIRIT_SCRIPT(vai)
0393         BOOST_SPIRIT_SCRIPT(vithkuqi)
0394         BOOST_SPIRIT_SCRIPT(warang_citi)
0395         BOOST_SPIRIT_SCRIPT(wancho)
0396         BOOST_SPIRIT_SCRIPT(old_persian)
0397         BOOST_SPIRIT_SCRIPT(cuneiform)
0398         BOOST_SPIRIT_SCRIPT(yezidi)
0399         BOOST_SPIRIT_SCRIPT(yi)
0400         BOOST_SPIRIT_SCRIPT(zanabazar_square)
0401         BOOST_SPIRIT_SCRIPT(inherited)
0402         BOOST_SPIRIT_SCRIPT(common)
0403         BOOST_SPIRIT_SCRIPT(unknown)
0404
0405 #undef BOOST_SPIRIT_MAJOR_CATEGORY
0406 #undef BOOST_SPIRIT_CATEGORY
0407 #undef BOOST_SPIRIT_DERIVED_CATEGORY
0408 #undef BOOST_SPIRIT_SCRIPT
0409
0410     };
0411
0412 }}}
0413
0414 #endif
0415