src/hb-unicode.h

   1 /*
   2  * Copyright (C) 2009  Red Hat, Inc.
   3  *
   4  *  This is part of HarfBuzz, a text shaping library.
   5  *
   6  * Permission is hereby granted, without written agreement and without
   7  * license or royalty fees, to use, copy, modify, and distribute this
   8  * software and its documentation for any purpose, provided that the
   9  * above copyright notice and the following two paragraphs appear in
  10  * all copies of this software.
  11  *
  12  * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
  13  * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
  14  * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
  15  * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
  16  * DAMAGE.
  17  *
  18  * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
  19  * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
  20  * FITNESS FOR A PARTICULAR PURPOSE.  THE SOFTWARE PROVIDED HEREUNDER IS
  21  * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
  22  * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
  23  *
  24  * Red Hat Author(s): Behdad Esfahbod
  25  */
  26
  27 #ifndef HB_UNICODE_H
  28 #define HB_UNICODE_H
  29
  30 #include "hb-common.h"
  31
  32 HB_BEGIN_DECLS
  33
  34 /* Unicode General Category property */
  35 typedef enum
  36 {
  37   HB_CATEGORY_CONTROL,
  38   HB_CATEGORY_FORMAT,
  39   HB_CATEGORY_UNASSIGNED,
  40   HB_CATEGORY_PRIVATE_USE,
  41   HB_CATEGORY_SURROGATE,
  42   HB_CATEGORY_LOWERCASE_LETTER,
  43   HB_CATEGORY_MODIFIER_LETTER,
  44   HB_CATEGORY_OTHER_LETTER,
  45   HB_CATEGORY_TITLECASE_LETTER,
  46   HB_CATEGORY_UPPERCASE_LETTER,
  47   HB_CATEGORY_COMBINING_MARK,
  48   HB_CATEGORY_ENCLOSING_MARK,
  49   HB_CATEGORY_NON_SPACING_MARK,
  50   HB_CATEGORY_DECIMAL_NUMBER,
  51   HB_CATEGORY_LETTER_NUMBER,
  52   HB_CATEGORY_OTHER_NUMBER,
  53   HB_CATEGORY_CONNECT_PUNCTUATION,
  54   HB_CATEGORY_DASH_PUNCTUATION,
  55   HB_CATEGORY_CLOSE_PUNCTUATION,
  56   HB_CATEGORY_FINAL_PUNCTUATION,
  57   HB_CATEGORY_INITIAL_PUNCTUATION,
  58   HB_CATEGORY_OTHER_PUNCTUATION,
  59   HB_CATEGORY_OPEN_PUNCTUATION,
  60   HB_CATEGORY_CURRENCY_SYMBOL,
  61   HB_CATEGORY_MODIFIER_SYMBOL,
  62   HB_CATEGORY_MATH_SYMBOL,
  63   HB_CATEGORY_OTHER_SYMBOL,
  64   HB_CATEGORY_LINE_SEPARATOR,
  65   HB_CATEGORY_PARAGRAPH_SEPARATOR,
  66   HB_CATEGORY_SPACE_SEPARATOR
  67 } hb_category_t;
  68
  69 /* Unicode Script property */
  70 typedef enum
  71 {                               /* ISO 15924 code */
  72   HB_SCRIPT_INVALID_CODE = -1,
  73   HB_SCRIPT_COMMON       = 0,   /* Zyyy */
  74   HB_SCRIPT_INHERITED,          /* Qaai */
  75   HB_SCRIPT_ARABIC,             /* Arab */
  76   HB_SCRIPT_ARMENIAN,           /* Armn */
  77   HB_SCRIPT_BENGALI,            /* Beng */
  78   HB_SCRIPT_BOPOMOFO,           /* Bopo */
  79   HB_SCRIPT_CHEROKEE,           /* Cher */
  80   HB_SCRIPT_COPTIC,             /* Qaac */
  81   HB_SCRIPT_CYRILLIC,           /* Cyrl (Cyrs) */
  82   HB_SCRIPT_DESERET,            /* Dsrt */
  83   HB_SCRIPT_DEVANAGARI,         /* Deva */
  84   HB_SCRIPT_ETHIOPIC,           /* Ethi */
  85   HB_SCRIPT_GEORGIAN,           /* Geor (Geon, Geoa) */
  86   HB_SCRIPT_GOTHIC,             /* Goth */
  87   HB_SCRIPT_GREEK,              /* Grek */
  88   HB_SCRIPT_GUJARATI,           /* Gujr */
  89   HB_SCRIPT_GURMUKHI,           /* Guru */
  90   HB_SCRIPT_HAN,                /* Hani */
  91   HB_SCRIPT_HANGUL,             /* Hang */
  92   HB_SCRIPT_HEBREW,             /* Hebr */
  93   HB_SCRIPT_HIRAGANA,           /* Hira */
  94   HB_SCRIPT_KANNADA,            /* Knda */
  95   HB_SCRIPT_KATAKANA,           /* Kana */
  96   HB_SCRIPT_KHMER,              /* Khmr */
  97   HB_SCRIPT_LAO,                /* Laoo */
  98   HB_SCRIPT_LATIN,              /* Latn (Latf, Latg) */
  99   HB_SCRIPT_MALAYALAM,          /* Mlym */
 100   HB_SCRIPT_MONGOLIAN,          /* Mong */
 101   HB_SCRIPT_MYANMAR,            /* Mymr */
 102   HB_SCRIPT_OGHAM,              /* Ogam */
 103   HB_SCRIPT_OLD_ITALIC,         /* Ital */
 104   HB_SCRIPT_ORIYA,              /* Orya */
 105   HB_SCRIPT_RUNIC,              /* Runr */
 106   HB_SCRIPT_SINHALA,            /* Sinh */
 107   HB_SCRIPT_SYRIAC,             /* Syrc (Syrj, Syrn, Syre) */
 108   HB_SCRIPT_TAMIL,              /* Taml */
 109   HB_SCRIPT_TELUGU,             /* Telu */
 110   HB_SCRIPT_THAANA,             /* Thaa */
 111   HB_SCRIPT_THAI,               /* Thai */
 112   HB_SCRIPT_TIBETAN,            /* Tibt */
 113   HB_SCRIPT_CANADIAN_ABORIGINAL, /* Cans */
 114   HB_SCRIPT_YI,                 /* Yiii */
 115   HB_SCRIPT_TAGALOG,            /* Tglg */
 116   HB_SCRIPT_HANUNOO,            /* Hano */
 117   HB_SCRIPT_BUHID,              /* Buhd */
 118   HB_SCRIPT_TAGBANWA,           /* Tagb */
 119
 120   /* Unicode-4.0 additions */
 121   HB_SCRIPT_BRAILLE,            /* Brai */
 122   HB_SCRIPT_CYPRIOT,            /* Cprt */
 123   HB_SCRIPT_LIMBU,              /* Limb */
 124   HB_SCRIPT_OSMANYA,            /* Osma */
 125   HB_SCRIPT_SHAVIAN,            /* Shaw */
 126   HB_SCRIPT_LINEAR_B,           /* Linb */
 127   HB_SCRIPT_TAI_LE,             /* Tale */
 128   HB_SCRIPT_UGARITIC,           /* Ugar */
 129
 130   /* Unicode-4.1 additions */
 131   HB_SCRIPT_NEW_TAI_LUE,        /* Talu */
 132   HB_SCRIPT_BUGINESE,           /* Bugi */
 133   HB_SCRIPT_GLAGOLITIC,         /* Glag */
 134   HB_SCRIPT_TIFINAGH,           /* Tfng */
 135   HB_SCRIPT_SYLOTI_NAGRI,       /* Sylo */
 136   HB_SCRIPT_OLD_PERSIAN,        /* Xpeo */
 137   HB_SCRIPT_KHAROSHTHI,         /* Khar */
 138
 139   /* Unicode-5.0 additions */
 140   HB_SCRIPT_UNKNOWN,            /* Zzzz */
 141   HB_SCRIPT_BALINESE,           /* Bali */
 142   HB_SCRIPT_CUNEIFORM,          /* Xsux */
 143   HB_SCRIPT_PHOENICIAN,         /* Phnx */
 144   HB_SCRIPT_PHAGS_PA,           /* Phag */
 145   HB_SCRIPT_NKO,                /* Nkoo */
 146
 147   /* Unicode-5.1 additions */
 148   HB_SCRIPT_KAYAH_LI,           /* Kali */
 149   HB_SCRIPT_LEPCHA,             /* Lepc */
 150   HB_SCRIPT_REJANG,             /* Rjng */
 151   HB_SCRIPT_SUNDANESE,          /* Sund */
 152   HB_SCRIPT_SAURASHTRA,         /* Saur */
 153   HB_SCRIPT_CHAM,               /* Cham */
 154   HB_SCRIPT_OL_CHIKI,           /* Olck */
 155   HB_SCRIPT_VAI,                /* Vaii */
 156   HB_SCRIPT_CARIAN,             /* Cari */
 157   HB_SCRIPT_LYCIAN,             /* Lyci */
 158   HB_SCRIPT_LYDIAN              /* Lydi */
 159 } hb_script_t;
 160
 161
 162 /*
 163  * hb_unicode_funcs_t
 164  */
 165
 166 typedef struct _hb_unicode_funcs_t hb_unicode_funcs_t;
 167
 168 hb_unicode_funcs_t *
 169 hb_unicode_funcs_create (void);
 170
 171 hb_unicode_funcs_t *
 172 hb_unicode_funcs_reference (hb_unicode_funcs_t *ufuncs);
 173
 174 unsigned int
 175 hb_unicode_funcs_get_reference_count (hb_unicode_funcs_t *ufuncs);
 176
 177 void
 178 hb_unicode_funcs_destroy (hb_unicode_funcs_t *ufuncs);
 179
 180 hb_unicode_funcs_t *
 181 hb_unicode_funcs_copy (hb_unicode_funcs_t *ufuncs);
 182
 183 void
 184 hb_unicode_funcs_make_immutable (hb_unicode_funcs_t *ufuncs);
 185
 186
 187 /* funcs */
 188
 189 typedef hb_codepoint_t (*hb_unicode_get_mirroring_func_t) (hb_codepoint_t unicode);
 190 typedef hb_category_t (*hb_unicode_get_general_category_func_t) (hb_codepoint_t unicode);
 191 typedef hb_script_t (*hb_unicode_get_script_func_t) (hb_codepoint_t unicode);
 192 typedef unsigned int (*hb_unicode_get_combining_class_func_t) (hb_codepoint_t unicode);
 193 typedef unsigned int (*hb_unicode_get_eastasian_width_func_t) (hb_codepoint_t unicode);
 194
 195
 196 void
 197 hb_unicode_funcs_set_mirroring_func (hb_unicode_funcs_t *ufuncs,
 198                                      hb_unicode_get_mirroring_func_t mirroring_func);
 199
 200 void
 201 hb_unicode_funcs_set_general_category_func (hb_unicode_funcs_t *ufuncs,
 202                                             hb_unicode_get_general_category_func_t general_category_func);
 203
 204 void
 205 hb_unicode_funcs_set_script_func (hb_unicode_funcs_t *ufuncs,
 206                                   hb_unicode_get_script_func_t script_func);
 207
 208 void
 209 hb_unicode_funcs_set_combining_class_func (hb_unicode_funcs_t *ufuncs,
 210                                            hb_unicode_get_combining_class_func_t combining_class_func);
 211
 212 void
 213 hb_unicode_funcs_set_eastasian_width_func (hb_unicode_funcs_t *ufuncs,
 214                                            hb_unicode_get_eastasian_width_func_t eastasian_width_func);
 215
 216
 217 hb_codepoint_t
 218 hb_unicode_get_mirroring (hb_unicode_funcs_t *ufuncs,
 219                           hb_codepoint_t unicode);
 220
 221 hb_category_t
 222 hb_unicode_get_general_category (hb_unicode_funcs_t *ufuncs,
 223                                  hb_codepoint_t unicode);
 224
 225 hb_script_t
 226 hb_unicode_get_script (hb_unicode_funcs_t *ufuncs,
 227                        hb_codepoint_t unicode);
 228
 229 unsigned int
 230 hb_unicode_get_combining_class (hb_unicode_funcs_t *ufuncs,
 231                                 hb_codepoint_t unicode);
 232
 233 unsigned int
 234 hb_unicode_get_eastasian_width (hb_unicode_funcs_t *ufuncs,
 235                                 hb_codepoint_t unicode);
 236
 237
 238 HB_END_DECLS
 239
 240 #endif /* HB_UNICODE_H */