char_encoding/unicode/query.hpp

0001 /*=============================================================================
0002     Copyright (c) 2001-2011 Joel de Guzman
0003
0004     Distributed under the Boost Software License, Version 1.0. (See accompanying
0005     file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
0006
0007     Autogenerated by MultiStageTable.py (Unicode multi-stage
0008     table builder) (c) Peter Kankowski, 2008
0009 ==============================================================================*/
0010 #if !defined(BOOST_SPIRIT_UNICODE_QUERY_FEBRUARY_2_2010)
0011 #define BOOST_SPIRIT_UNICODE_QUERY_FEBRUARY_2_2010
0012
0013 #include <boost/cstdint.hpp>
0014
0015 # include "category_table.hpp"
0016 # include "script_table.hpp"
0017 # include "lowercase_table.hpp"
0018 # include "uppercase_table.hpp"
0019
0020 namespace boost { namespace spirit { namespace ucd
0021 {
0022     // This header provides Basic (Level 1) Unicode Support
0023     // See http://unicode.org/reports/tr18/ for details
0024
0025     struct properties
0026     {
0027         // bit pattern: xxMMMCCC
0028         // MMM: major_category
0029         // CCC: category
0030
0031         enum major_category
0032         {
0033             letter,
0034             mark,
0035             number,
0036             separator,
0037             other,
0038             punctuation,
0039             symbol
0040         };
0041
0042         enum category
0043         {
0044             uppercase_letter = 0,   // [Lu] an uppercase letter
0045             lowercase_letter,       // [Ll] a lowercase letter
0046             titlecase_letter,       // [Lt] a digraphic character, with first part uppercase
0047             modifier_letter,        // [Lm] a modifier letter
0048             other_letter,           // [Lo] other letters, including syllables and ideographs
0049
0050             nonspacing_mark = 8,    // [Mn] a nonspacing combining mark (zero advance width)
0051             enclosing_mark,         // [Me] an enclosing combining mark
0052             spacing_mark,           // [Mc] a spacing combining mark (positive advance width)
0053
0054             decimal_number = 16,    // [Nd] a decimal digit
0055             letter_number,          // [Nl] a letterlike numeric character
0056             other_number,           // [No] a numeric character of other type
0057
0058             space_separator = 24,   // [Zs] a space character (of various non-zero widths)
0059             line_separator,         // [Zl] U+2028 LINE SEPARATOR only
0060             paragraph_separator,    // [Zp] U+2029 PARAGRAPH SEPARATOR only
0061
0062             control = 32,           // [Cc] a C0 or C1 control code
0063             format,                 // [Cf] a format control character
0064             private_use,            // [Co] a private-use character
0065             surrogate,              // [Cs] a surrogate code point
0066             unassigned,             // [Cn] a reserved unassigned code point or a noncharacter
0067
0068             dash_punctuation = 40,  // [Pd] a dash or hyphen punctuation mark
0069             open_punctuation,       // [Ps] an opening punctuation mark (of a pair)
0070             close_punctuation,      // [Pe] a closing punctuation mark (of a pair)
0071             connector_punctuation,  // [Pc] a connecting punctuation mark, like a tie
0072             other_punctuation,      // [Po] a punctuation mark of other type
0073             initial_punctuation,    // [Pi] an initial quotation mark
0074             final_punctuation,      // [Pf] a final quotation mark
0075
0076             math_symbol = 48,       // [Sm] a symbol of primarily mathematical use
0077             currency_symbol,        // [Sc] a currency sign
0078             modifier_symbol,        // [Sk] a non-letterlike modifier symbol
0079             other_symbol            // [So] a symbol of other type
0080         };
0081
0082         enum derived_properties
0083         {
0084             alphabetic = 64,
0085             uppercase = 128,
0086             lowercase = 256,
0087             white_space = 512,
0088             hex_digit = 1024,
0089             noncharacter_code_point = 2048,
0090             default_ignorable_code_point = 4096
0091         };
0092
0093         enum script
0094         {
0095             adlam,
0096             caucasian_albanian,
0097             ahom,
0098             arabic,
0099             imperial_aramaic,
0100             armenian,
0101             avestan,
0102             balinese,
0103             bamum,
0104             bassa_vah,
0105             batak,
0106             bengali,
0107             bhaiksuki,
0108             bopomofo,
0109             brahmi,
0110             braille,
0111             buginese,
0112             buhid,
0113             chakma,
0114             canadian_aboriginal,
0115             carian,
0116             cham,
0117             cherokee,
0118             chorasmian,
0119             coptic,
0120             cypro_minoan,
0121             cypriot,
0122             cyrillic,
0123             devanagari,
0124             dives_akuru,
0125             dogra,
0126             deseret,
0127             duployan,
0128             egyptian_hieroglyphs,
0129             elbasan,
0130             elymaic,
0131             ethiopic,
0132             georgian,
0133             glagolitic,
0134             gunjala_gondi,
0135             masaram_gondi,
0136             gothic,
0137             grantha,
0138             greek,
0139             gujarati,
0140             gurmukhi,
0141             hangul,
0142             han,
0143             hanunoo,
0144             hatran,
0145             hebrew,
0146             hiragana,
0147             anatolian_hieroglyphs,
0148             pahawh_hmong,
0149             nyiakeng_puachue_hmong,
0150             katakana_or_hiragana,
0151             old_hungarian,
0152             old_italic,
0153             javanese,
0154             kayah_li,
0155             katakana,
0156             kawi,
0157             kharoshthi,
0158             khmer,
0159             khojki,
0160             khitan_small_script,
0161             kannada,
0162             kaithi,
0163             tai_tham,
0164             lao,
0165             latin,
0166             lepcha,
0167             limbu,
0168             linear_a,
0169             linear_b,
0170             lisu,
0171             lycian,
0172             lydian,
0173             mahajani,
0174             makasar,
0175             mandaic,
0176             manichaean,
0177             marchen,
0178             medefaidrin,
0179             mende_kikakui,
0180             meroitic_cursive,
0181             meroitic_hieroglyphs,
0182             malayalam,
0183             modi,
0184             mongolian,
0185             mro,
0186             meetei_mayek,
0187             multani,
0188             myanmar,
0189             nag_mundari,
0190             nandinagari,
0191             old_north_arabian,
0192             nabataean,
0193             newa,
0194             nko,
0195             nushu,
0196             ogham,
0197             ol_chiki,
0198             old_turkic,
0199             oriya,
0200             osage,
0201             osmanya,
0202             old_uyghur,
0203             palmyrene,
0204             pau_cin_hau,
0205             old_permic,
0206             phags_pa,
0207             inscriptional_pahlavi,
0208             psalter_pahlavi,
0209             phoenician,
0210             miao,
0211             inscriptional_parthian,
0212             rejang,
0213             hanifi_rohingya,
0214             runic,
0215             samaritan,
0216             old_south_arabian,
0217             saurashtra,
0218             signwriting,
0219             shavian,
0220             sharada,
0221             siddham,
0222             khudawadi,
0223             sinhala,
0224             sogdian,
0225             old_sogdian,
0226             sora_sompeng,
0227             soyombo,
0228             sundanese,
0229             syloti_nagri,
0230             syriac,
0231             tagbanwa,
0232             takri,
0233             tai_le,
0234             new_tai_lue,
0235             tamil,
0236             tangut,
0237             tai_viet,
0238             telugu,
0239             tifinagh,
0240             tagalog,
0241             thaana,
0242             thai,
0243             tibetan,
0244             tirhuta,
0245             tangsa,
0246             toto,
0247             ugaritic,
0248             vai,
0249             vithkuqi,
0250             warang_citi,
0251             wancho,
0252             old_persian,
0253             cuneiform,
0254             yezidi,
0255             yi,
0256             zanabazar_square,
0257             inherited,
0258             common,
0259             unknown
0260         };
0261     };
0262
0263     inline properties::category get_category(::boost::uint32_t ch)
0264     {
0265         return static_cast<properties::category>(detail::category_lookup(ch) & 0x3F);
0266     }
0267
0268     inline properties::major_category get_major_category(::boost::uint32_t ch)
0269     {
0270         return static_cast<properties::major_category>(get_category(ch) >> 3);
0271     }
0272
0273     inline bool is_punctuation(::boost::uint32_t ch)
0274     {
0275         return get_major_category(ch) == properties::punctuation;
0276     }
0277
0278     inline bool is_decimal_number(::boost::uint32_t ch)
0279     {
0280         return get_category(ch) == properties::decimal_number;
0281     }
0282
0283     inline bool is_hex_digit(::boost::uint32_t ch)
0284     {
0285         return (detail::category_lookup(ch) & properties::hex_digit) != 0;
0286     }
0287
0288     inline bool is_control(::boost::uint32_t ch)
0289     {
0290         return get_category(ch) == properties::control;
0291     }
0292
0293     inline bool is_alphabetic(::boost::uint32_t ch)
0294     {
0295         return (detail::category_lookup(ch) & properties::alphabetic) != 0;
0296     }
0297
0298     inline bool is_alphanumeric(::boost::uint32_t ch)
0299     {
0300         return is_decimal_number(ch) || is_alphabetic(ch);
0301     }
0302
0303     inline bool is_uppercase(::boost::uint32_t ch)
0304     {
0305         return (detail::category_lookup(ch) & properties::uppercase) != 0;
0306     }
0307
0308     inline bool is_lowercase(::boost::uint32_t ch)
0309     {
0310         return (detail::category_lookup(ch) & properties::lowercase) != 0;
0311     }
0312
0313     inline bool is_white_space(::boost::uint32_t ch)
0314     {
0315         return (detail::category_lookup(ch) & properties::white_space) != 0;
0316     }
0317
0318     inline bool is_blank(::boost::uint32_t ch)
0319     {
0320         switch (ch)
0321         {
0322             case '\n': case '\v': case '\f': case '\r':
0323                 return false;
0324             default:
0325                 return is_white_space(ch)
0326                 && !(   get_category(ch) == properties::line_separator
0327                     ||  get_category(ch) == properties::paragraph_separator
0328                     );
0329         }
0330     }
0331
0332     inline bool is_graph(::boost::uint32_t ch)
0333     {
0334         return !(   is_white_space(ch)
0335                 ||  get_category(ch) == properties::control
0336                 ||  get_category(ch) == properties::surrogate
0337                 ||  get_category(ch) == properties::unassigned
0338                 );
0339     }
0340
0341     inline bool is_print(::boost::uint32_t ch)
0342     {
0343         return (is_graph(ch) || is_blank(ch)) && !is_control(ch);
0344     }
0345
0346     inline bool is_noncharacter_code_point(::boost::uint32_t ch)
0347     {
0348         return (detail::category_lookup(ch) & properties::noncharacter_code_point) != 0;
0349     }
0350
0351     inline bool is_default_ignorable_code_point(::boost::uint32_t ch)
0352     {
0353         return (detail::category_lookup(ch) & properties::default_ignorable_code_point) != 0;
0354     }
0355
0356     inline properties::script get_script(::boost::uint32_t ch)
0357     {
0358         return static_cast<properties::script>(detail::script_lookup(ch));
0359     }
0360
0361     inline ::boost::uint32_t to_lowercase(::boost::uint32_t ch)
0362     {
0363         // The table returns 0 to signal that this code maps to itself
0364         ::boost::uint32_t r = detail::lowercase_lookup(ch);
0365         return (r == 0)? ch : r;
0366     }
0367
0368     inline ::boost::uint32_t to_uppercase(::boost::uint32_t ch)
0369     {
0370         // The table returns 0 to signal that this code maps to itself
0371         ::boost::uint32_t r = detail::uppercase_lookup(ch);
0372         return (r == 0)? ch : r;
0373     }
0374 }}}
0375
0376 #endif