src/hb-ot-shape-complex-default.cc

   1 /*
   2  * Copyright © 2010,2012  Google, 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  * Google Author(s): Behdad Esfahbod
  25  */
  26
  27 #include "hb-ot-shape-complex-private.hh"
  28
  29
  30 /* TODO Add kana, and other small shapers here */
  31
  32
  33 /* The default shaper *only* adds additional per-script features.*/
  34
  35 static const hb_tag_t hangul_features[] =
  36 {
  37   HB_TAG('l','j','m','o'),
  38   HB_TAG('v','j','m','o'),
  39   HB_TAG('t','j','m','o'),
  40   HB_TAG_NONE
  41 };
  42
  43 static const hb_tag_t tibetan_features[] =
  44 {
  45   HB_TAG('a','b','v','s'),
  46   HB_TAG('b','l','w','s'),
  47   HB_TAG('a','b','v','m'),
  48   HB_TAG('b','l','w','m'),
  49   HB_TAG_NONE
  50 };
  51
  52 static void
  53 collect_features_default (hb_ot_shape_planner_t *plan)
  54 {
  55   const hb_tag_t *script_features = NULL;
  56
  57   switch ((hb_tag_t) plan->props.script)
  58   {
  59     /* Unicode-1.1 additions */
  60     case HB_SCRIPT_HANGUL:
  61       script_features = hangul_features;
  62       break;
  63
  64     /* Unicode-2.0 additions */
  65     case HB_SCRIPT_TIBETAN:
  66       script_features = tibetan_features;
  67       break;
  68   }
  69
  70   for (; script_features && *script_features; script_features++)
  71     plan->map.add_bool_feature (*script_features);
  72 }
  73
  74 static hb_ot_shape_normalization_mode_t
  75 normalization_preference_default (const hb_segment_properties_t *props)
  76 {
  77   switch ((hb_tag_t) props->script)
  78   {
  79     /* Unicode-1.1 additions */
  80     case HB_SCRIPT_HANGUL:
  81       return HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_FULL;
  82   }
  83   return HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS;
  84 }
  85
  86 static bool
  87 compose_default (const hb_ot_shape_normalize_context_t *c,
  88                  hb_codepoint_t  a,
  89                  hb_codepoint_t  b,
  90                  hb_codepoint_t *ab)
  91 {
  92   /* Hebrew presentation-form shaping.
  93    * https://bugzilla.mozilla.org/show_bug.cgi?id=728866
  94    * Hebrew presentation forms with dagesh, for characters 0x05D0..0x05EA;
  95    * Note that some letters do not have a dagesh presForm encoded.
  96    */
  97   static const hb_codepoint_t sDageshForms[0x05EA - 0x05D0 + 1] = {
  98     0xFB30, /* ALEF */
  99     0xFB31, /* BET */
 100     0xFB32, /* GIMEL */
 101     0xFB33, /* DALET */
 102     0xFB34, /* HE */
 103     0xFB35, /* VAV */
 104     0xFB36, /* ZAYIN */
 105     0x0000, /* HET */
 106     0xFB38, /* TET */
 107     0xFB39, /* YOD */
 108     0xFB3A, /* FINAL KAF */
 109     0xFB3B, /* KAF */
 110     0xFB3C, /* LAMED */
 111     0x0000, /* FINAL MEM */
 112     0xFB3E, /* MEM */
 113     0x0000, /* FINAL NUN */
 114     0xFB40, /* NUN */
 115     0xFB41, /* SAMEKH */
 116     0x0000, /* AYIN */
 117     0xFB43, /* FINAL PE */
 118     0xFB44, /* PE */
 119     0x0000, /* FINAL TSADI */
 120     0xFB46, /* TSADI */
 121     0xFB47, /* QOF */
 122     0xFB48, /* RESH */
 123     0xFB49, /* SHIN */
 124     0xFB4A /* TAV */
 125   };
 126
 127   bool found = c->unicode->compose (a, b, ab);
 128
 129   if (!found && (b & ~0x7F) == 0x0580) {
 130       /* Special-case Hebrew presentation forms that are excluded from
 131        * standard normalization, but wanted for old fonts. */
 132       switch (b) {
 133       case 0x05B4: /* HIRIQ */
 134           if (a == 0x05D9) { /* YOD */
 135               *ab = 0xFB1D;
 136               found = true;
 137           }
 138           break;
 139       case 0x05B7: /* patah */
 140           if (a == 0x05F2) { /* YIDDISH YOD YOD */
 141               *ab = 0xFB1F;
 142               found = true;
 143           } else if (a == 0x05D0) { /* ALEF */
 144               *ab = 0xFB2E;
 145               found = true;
 146           }
 147           break;
 148       case 0x05B8: /* QAMATS */
 149           if (a == 0x05D0) { /* ALEF */
 150               *ab = 0xFB2F;
 151               found = true;
 152           }
 153           break;
 154       case 0x05B9: /* HOLAM */
 155           if (a == 0x05D5) { /* VAV */
 156               *ab = 0xFB4B;
 157               found = true;
 158           }
 159           break;
 160       case 0x05BC: /* DAGESH */
 161           if (a >= 0x05D0 && a <= 0x05EA) {
 162               *ab = sDageshForms[a - 0x05D0];
 163               found = (*ab != 0);
 164           } else if (a == 0xFB2A) { /* SHIN WITH SHIN DOT */
 165               *ab = 0xFB2C;
 166               found = true;
 167           } else if (a == 0xFB2B) { /* SHIN WITH SIN DOT */
 168               *ab = 0xFB2D;
 169               found = true;
 170           }
 171           break;
 172       case 0x05BF: /* RAFE */
 173           switch (a) {
 174           case 0x05D1: /* BET */
 175               *ab = 0xFB4C;
 176               found = true;
 177               break;
 178           case 0x05DB: /* KAF */
 179               *ab = 0xFB4D;
 180               found = true;
 181               break;
 182           case 0x05E4: /* PE */
 183               *ab = 0xFB4E;
 184               found = true;
 185               break;
 186           }
 187           break;
 188       case 0x05C1: /* SHIN DOT */
 189           if (a == 0x05E9) { /* SHIN */
 190               *ab = 0xFB2A;
 191               found = true;
 192           } else if (a == 0xFB49) { /* SHIN WITH DAGESH */
 193               *ab = 0xFB2C;
 194               found = true;
 195           }
 196           break;
 197       case 0x05C2: /* SIN DOT */
 198           if (a == 0x05E9) { /* SHIN */
 199               *ab = 0xFB2B;
 200               found = true;
 201           } else if (a == 0xFB49) { /* SHIN WITH DAGESH */
 202               *ab = 0xFB2D;
 203               found = true;
 204           }
 205           break;
 206       }
 207   }
 208
 209   return found;
 210 }
 211
 212 const hb_ot_complex_shaper_t _hb_ot_complex_shaper_default =
 213 {
 214   "default",
 215   collect_features_default,
 216   NULL, /* override_features */
 217   NULL, /* data_create */
 218   NULL, /* data_destroy */
 219   NULL, /* preprocess_text */
 220   normalization_preference_default,
 221   NULL, /* decompose */
 222   compose_default,
 223   NULL, /* setup_masks */
 224   true, /* zero_width_attached_marks */
 225   true, /* fallback_position */
 226 };