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
  35 /* Unicode General Category property */
  36 typedef enum
  37 {
  38   HB_CATEGORY_CONTROL,
  39   HB_CATEGORY_FORMAT,
  40   HB_CATEGORY_UNASSIGNED,
  41   HB_CATEGORY_PRIVATE_USE,
  42   HB_CATEGORY_SURROGATE,
  43   HB_CATEGORY_LOWERCASE_LETTER,
  44   HB_CATEGORY_MODIFIER_LETTER,
  45   HB_CATEGORY_OTHER_LETTER,
  46   HB_CATEGORY_TITLECASE_LETTER,
  47   HB_CATEGORY_UPPERCASE_LETTER,
  48   HB_CATEGORY_COMBINING_MARK,
  49   HB_CATEGORY_ENCLOSING_MARK,
  50   HB_CATEGORY_NON_SPACING_MARK,
  51   HB_CATEGORY_DECIMAL_NUMBER,
  52   HB_CATEGORY_LETTER_NUMBER,
  53   HB_CATEGORY_OTHER_NUMBER,
  54   HB_CATEGORY_CONNECT_PUNCTUATION,
  55   HB_CATEGORY_DASH_PUNCTUATION,
  56   HB_CATEGORY_CLOSE_PUNCTUATION,
  57   HB_CATEGORY_FINAL_PUNCTUATION,
  58   HB_CATEGORY_INITIAL_PUNCTUATION,
  59   HB_CATEGORY_OTHER_PUNCTUATION,
  60   HB_CATEGORY_OPEN_PUNCTUATION,
  61   HB_CATEGORY_CURRENCY_SYMBOL,
  62   HB_CATEGORY_MODIFIER_SYMBOL,
  63   HB_CATEGORY_MATH_SYMBOL,
  64   HB_CATEGORY_OTHER_SYMBOL,
  65   HB_CATEGORY_LINE_SEPARATOR,
  66   HB_CATEGORY_PARAGRAPH_SEPARATOR,
  67   HB_CATEGORY_SPACE_SEPARATOR
  68 } hb_category_t;
  69
  70 /* Unicode Script property */
  71 typedef enum
  72 {                               /* ISO 15924 code */
  73   HB_SCRIPT_INVALID_CODE = -1,
  74   HB_SCRIPT_COMMON       = 0,   /* Zyyy */
  75   HB_SCRIPT_INHERITED,          /* Qaai */
  76   HB_SCRIPT_ARABIC,             /* Arab */
  77   HB_SCRIPT_ARMENIAN,           /* Armn */
  78   HB_SCRIPT_BENGALI,            /* Beng */
  79   HB_SCRIPT_BOPOMOFO,           /* Bopo */
  80   HB_SCRIPT_CHEROKEE,           /* Cher */
  81   HB_SCRIPT_COPTIC,             /* Qaac */
  82   HB_SCRIPT_CYRILLIC,           /* Cyrl (Cyrs) */
  83   HB_SCRIPT_DESERET,            /* Dsrt */
  84   HB_SCRIPT_DEVANAGARI,         /* Deva */
  85   HB_SCRIPT_ETHIOPIC,           /* Ethi */
  86   HB_SCRIPT_GEORGIAN,           /* Geor (Geon, Geoa) */
  87   HB_SCRIPT_GOTHIC,             /* Goth */
  88   HB_SCRIPT_GREEK,              /* Grek */
  89   HB_SCRIPT_GUJARATI,           /* Gujr */
  90   HB_SCRIPT_GURMUKHI,           /* Guru */
  91   HB_SCRIPT_HAN,                /* Hani */
  92   HB_SCRIPT_HANGUL,             /* Hang */
  93   HB_SCRIPT_HEBREW,             /* Hebr */
  94   HB_SCRIPT_HIRAGANA,           /* Hira */
  95   HB_SCRIPT_KANNADA,            /* Knda */
  96   HB_SCRIPT_KATAKANA,           /* Kana */
  97   HB_SCRIPT_KHMER,              /* Khmr */
  98   HB_SCRIPT_LAO,                /* Laoo */
  99   HB_SCRIPT_LATIN,              /* Latn (Latf, Latg) */
 100   HB_SCRIPT_MALAYALAM,          /* Mlym */
 101   HB_SCRIPT_MONGOLIAN,          /* Mong */
 102   HB_SCRIPT_MYANMAR,            /* Mymr */
 103   HB_SCRIPT_OGHAM,              /* Ogam */
 104   HB_SCRIPT_OLD_ITALIC,         /* Ital */
 105   HB_SCRIPT_ORIYA,              /* Orya */
 106   HB_SCRIPT_RUNIC,              /* Runr */
 107   HB_SCRIPT_SINHALA,            /* Sinh */
 108   HB_SCRIPT_SYRIAC,             /* Syrc (Syrj, Syrn, Syre) */
 109   HB_SCRIPT_TAMIL,              /* Taml */
 110   HB_SCRIPT_TELUGU,             /* Telu */
 111   HB_SCRIPT_THAANA,             /* Thaa */
 112   HB_SCRIPT_THAI,               /* Thai */
 113   HB_SCRIPT_TIBETAN,            /* Tibt */
 114   HB_SCRIPT_CANADIAN_ABORIGINAL, /* Cans */
 115   HB_SCRIPT_YI,                 /* Yiii */
 116   HB_SCRIPT_TAGALOG,            /* Tglg */
 117   HB_SCRIPT_HANUNOO,            /* Hano */
 118   HB_SCRIPT_BUHID,              /* Buhd */
 119   HB_SCRIPT_TAGBANWA,           /* Tagb */
 120
 121   /* Unicode-4.0 additions */
 122   HB_SCRIPT_BRAILLE,            /* Brai */
 123   HB_SCRIPT_CYPRIOT,            /* Cprt */
 124   HB_SCRIPT_LIMBU,              /* Limb */
 125   HB_SCRIPT_OSMANYA,            /* Osma */
 126   HB_SCRIPT_SHAVIAN,            /* Shaw */
 127   HB_SCRIPT_LINEAR_B,           /* Linb */
 128   HB_SCRIPT_TAI_LE,             /* Tale */
 129   HB_SCRIPT_UGARITIC,           /* Ugar */
 130
 131   /* Unicode-4.1 additions */
 132   HB_SCRIPT_NEW_TAI_LUE,        /* Talu */
 133   HB_SCRIPT_BUGINESE,           /* Bugi */
 134   HB_SCRIPT_GLAGOLITIC,         /* Glag */
 135   HB_SCRIPT_TIFINAGH,           /* Tfng */
 136   HB_SCRIPT_SYLOTI_NAGRI,       /* Sylo */
 137   HB_SCRIPT_OLD_PERSIAN,        /* Xpeo */
 138   HB_SCRIPT_KHAROSHTHI,         /* Khar */
 139
 140   /* Unicode-5.0 additions */
 141   HB_SCRIPT_UNKNOWN,            /* Zzzz */
 142   HB_SCRIPT_BALINESE,           /* Bali */
 143   HB_SCRIPT_CUNEIFORM,          /* Xsux */
 144   HB_SCRIPT_PHOENICIAN,         /* Phnx */
 145   HB_SCRIPT_PHAGS_PA,           /* Phag */
 146   HB_SCRIPT_NKO,                /* Nkoo */
 147
 148   /* Unicode-5.1 additions */
 149   HB_SCRIPT_KAYAH_LI,           /* Kali */
 150   HB_SCRIPT_LEPCHA,             /* Lepc */
 151   HB_SCRIPT_REJANG,             /* Rjng */
 152   HB_SCRIPT_SUNDANESE,          /* Sund */
 153   HB_SCRIPT_SAURASHTRA,         /* Saur */
 154   HB_SCRIPT_CHAM,               /* Cham */
 155   HB_SCRIPT_OL_CHIKI,           /* Olck */
 156   HB_SCRIPT_VAI,                /* Vaii */
 157   HB_SCRIPT_CARIAN,             /* Cari */
 158   HB_SCRIPT_LYCIAN,             /* Lyci */
 159   HB_SCRIPT_LYDIAN,             /* Lydi */
 160
 161   /* Unicode-5.2 additions */
 162   HB_SCRIPT_AVESTAN,                /* Avst */
 163   HB_SCRIPT_BAMUM,                  /* Bamu */
 164   HB_SCRIPT_EGYPTIAN_HIEROGLYPHS,   /* Egyp */
 165   HB_SCRIPT_IMPERIAL_ARAMAIC,       /* Armi */
 166   HB_SCRIPT_INSCRIPTIONAL_PAHLAVI,  /* Phli */
 167   HB_SCRIPT_INSCRIPTIONAL_PARTHIAN, /* Prti */
 168   HB_SCRIPT_JAVANESE,               /* Java */
 169   HB_SCRIPT_KAITHI,                 /* Kthi */
 170   HB_SCRIPT_LISU,                   /* Lisu */
 171   HB_SCRIPT_MEITEI_MAYEK,           /* Mtei */
 172   HB_SCRIPT_OLD_SOUTH_ARABIAN,      /* Sarb */
 173   HB_SCRIPT_OLD_TURKIC,             /* Orkh */
 174   HB_SCRIPT_SAMARITAN,              /* Samr */
 175   HB_SCRIPT_TAI_THAM,               /* Lana */
 176   HB_SCRIPT_TAI_VIET                /* Tavt */
 177 } hb_script_t;
 178
 179
 180 /*
 181  * hb_unicode_funcs_t
 182  */
 183
 184 typedef struct _hb_unicode_funcs_t hb_unicode_funcs_t;
 185
 186 hb_unicode_funcs_t *
 187 hb_unicode_funcs_create (void);
 188
 189 hb_unicode_funcs_t *
 190 hb_unicode_funcs_reference (hb_unicode_funcs_t *ufuncs);
 191
 192 unsigned int
 193 hb_unicode_funcs_get_reference_count (hb_unicode_funcs_t *ufuncs);
 194
 195 void
 196 hb_unicode_funcs_destroy (hb_unicode_funcs_t *ufuncs);
 197
 198 hb_unicode_funcs_t *
 199 hb_unicode_funcs_copy (hb_unicode_funcs_t *ufuncs);
 200
 201 void
 202 hb_unicode_funcs_make_immutable (hb_unicode_funcs_t *ufuncs);
 203
 204 hb_bool_t
 205 hb_unicode_funcs_is_immutable (hb_unicode_funcs_t *ufuncs);
 206
 207 /*
 208  * funcs
 209  */
 210
 211
 212 /* typedefs */
 213
 214 typedef hb_codepoint_t (*hb_unicode_get_mirroring_func_t) (hb_codepoint_t unicode);
 215 typedef hb_category_t (*hb_unicode_get_general_category_func_t) (hb_codepoint_t unicode);
 216 typedef hb_script_t (*hb_unicode_get_script_func_t) (hb_codepoint_t unicode);
 217 typedef unsigned int (*hb_unicode_get_combining_class_func_t) (hb_codepoint_t unicode);
 218 typedef unsigned int (*hb_unicode_get_eastasian_width_func_t) (hb_codepoint_t unicode);
 219
 220
 221 /* setters */
 222
 223 void
 224 hb_unicode_funcs_set_mirroring_func (hb_unicode_funcs_t *ufuncs,
 225                                      hb_unicode_get_mirroring_func_t mirroring_func);
 226
 227 void
 228 hb_unicode_funcs_set_general_category_func (hb_unicode_funcs_t *ufuncs,
 229                                             hb_unicode_get_general_category_func_t general_category_func);
 230
 231 void
 232 hb_unicode_funcs_set_script_func (hb_unicode_funcs_t *ufuncs,
 233                                   hb_unicode_get_script_func_t script_func);
 234
 235 void
 236 hb_unicode_funcs_set_combining_class_func (hb_unicode_funcs_t *ufuncs,
 237                                            hb_unicode_get_combining_class_func_t combining_class_func);
 238
 239 void
 240 hb_unicode_funcs_set_eastasian_width_func (hb_unicode_funcs_t *ufuncs,
 241                                            hb_unicode_get_eastasian_width_func_t eastasian_width_func);
 242
 243
 244 /* getters */
 245
 246 /* These never return NULL.  Return fallback defaults instead. */
 247
 248 hb_unicode_get_mirroring_func_t
 249 hb_unicode_funcs_get_mirroring_func (hb_unicode_funcs_t *ufuncs);
 250
 251 hb_unicode_get_general_category_func_t
 252 hb_unicode_funcs_get_general_category_func (hb_unicode_funcs_t *ufuncs);
 253
 254 hb_unicode_get_script_func_t
 255 hb_unicode_funcs_get_script_func (hb_unicode_funcs_t *ufuncs);
 256
 257 hb_unicode_get_combining_class_func_t
 258 hb_unicode_funcs_get_combining_class_func (hb_unicode_funcs_t *ufuncs);
 259
 260 hb_unicode_get_eastasian_width_func_t
 261 hb_unicode_funcs_get_eastasian_width_func (hb_unicode_funcs_t *ufuncs);
 262
 263
 264 /* accessors */
 265
 266 hb_codepoint_t
 267 hb_unicode_get_mirroring (hb_unicode_funcs_t *ufuncs,
 268                           hb_codepoint_t unicode);
 269
 270 hb_category_t
 271 hb_unicode_get_general_category (hb_unicode_funcs_t *ufuncs,
 272                                  hb_codepoint_t unicode);
 273
 274 hb_script_t
 275 hb_unicode_get_script (hb_unicode_funcs_t *ufuncs,
 276                        hb_codepoint_t unicode);
 277
 278 unsigned int
 279 hb_unicode_get_combining_class (hb_unicode_funcs_t *ufuncs,
 280                                 hb_codepoint_t unicode);
 281
 282 unsigned int
 283 hb_unicode_get_eastasian_width (hb_unicode_funcs_t *ufuncs,
 284                                 hb_codepoint_t unicode);
 285
 286
 287 HB_END_DECLS
 288
 289 #endif /* HB_UNICODE_H */