Home | History | Annotate | Download | only in src
      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_global_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   return HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS;
     78 }
     79 
     80 static bool
     81 compose_default (const hb_ot_shape_normalize_context_t *c,
     82 		 hb_codepoint_t  a,
     83 		 hb_codepoint_t  b,
     84 		 hb_codepoint_t *ab)
     85 {
     86   /* Hebrew presentation-form shaping.
     87    * https://bugzilla.mozilla.org/show_bug.cgi?id=728866
     88    * Hebrew presentation forms with dagesh, for characters 0x05D0..0x05EA;
     89    * Note that some letters do not have a dagesh presForm encoded.
     90    */
     91   static const hb_codepoint_t sDageshForms[0x05EA - 0x05D0 + 1] = {
     92     0xFB30, /* ALEF */
     93     0xFB31, /* BET */
     94     0xFB32, /* GIMEL */
     95     0xFB33, /* DALET */
     96     0xFB34, /* HE */
     97     0xFB35, /* VAV */
     98     0xFB36, /* ZAYIN */
     99     0x0000, /* HET */
    100     0xFB38, /* TET */
    101     0xFB39, /* YOD */
    102     0xFB3A, /* FINAL KAF */
    103     0xFB3B, /* KAF */
    104     0xFB3C, /* LAMED */
    105     0x0000, /* FINAL MEM */
    106     0xFB3E, /* MEM */
    107     0x0000, /* FINAL NUN */
    108     0xFB40, /* NUN */
    109     0xFB41, /* SAMEKH */
    110     0x0000, /* AYIN */
    111     0xFB43, /* FINAL PE */
    112     0xFB44, /* PE */
    113     0x0000, /* FINAL TSADI */
    114     0xFB46, /* TSADI */
    115     0xFB47, /* QOF */
    116     0xFB48, /* RESH */
    117     0xFB49, /* SHIN */
    118     0xFB4A /* TAV */
    119   };
    120 
    121   bool found = c->unicode->compose (a, b, ab);
    122 
    123   if (!found && (b & ~0x7F) == 0x0580) {
    124       /* Special-case Hebrew presentation forms that are excluded from
    125        * standard normalization, but wanted for old fonts. */
    126       switch (b) {
    127       case 0x05B4: /* HIRIQ */
    128 	  if (a == 0x05D9) { /* YOD */
    129 	      *ab = 0xFB1D;
    130 	      found = true;
    131 	  }
    132 	  break;
    133       case 0x05B7: /* patah */
    134 	  if (a == 0x05F2) { /* YIDDISH YOD YOD */
    135 	      *ab = 0xFB1F;
    136 	      found = true;
    137 	  } else if (a == 0x05D0) { /* ALEF */
    138 	      *ab = 0xFB2E;
    139 	      found = true;
    140 	  }
    141 	  break;
    142       case 0x05B8: /* QAMATS */
    143 	  if (a == 0x05D0) { /* ALEF */
    144 	      *ab = 0xFB2F;
    145 	      found = true;
    146 	  }
    147 	  break;
    148       case 0x05B9: /* HOLAM */
    149 	  if (a == 0x05D5) { /* VAV */
    150 	      *ab = 0xFB4B;
    151 	      found = true;
    152 	  }
    153 	  break;
    154       case 0x05BC: /* DAGESH */
    155 	  if (a >= 0x05D0 && a <= 0x05EA) {
    156 	      *ab = sDageshForms[a - 0x05D0];
    157 	      found = (*ab != 0);
    158 	  } else if (a == 0xFB2A) { /* SHIN WITH SHIN DOT */
    159 	      *ab = 0xFB2C;
    160 	      found = true;
    161 	  } else if (a == 0xFB2B) { /* SHIN WITH SIN DOT */
    162 	      *ab = 0xFB2D;
    163 	      found = true;
    164 	  }
    165 	  break;
    166       case 0x05BF: /* RAFE */
    167 	  switch (a) {
    168 	  case 0x05D1: /* BET */
    169 	      *ab = 0xFB4C;
    170 	      found = true;
    171 	      break;
    172 	  case 0x05DB: /* KAF */
    173 	      *ab = 0xFB4D;
    174 	      found = true;
    175 	      break;
    176 	  case 0x05E4: /* PE */
    177 	      *ab = 0xFB4E;
    178 	      found = true;
    179 	      break;
    180 	  }
    181 	  break;
    182       case 0x05C1: /* SHIN DOT */
    183 	  if (a == 0x05E9) { /* SHIN */
    184 	      *ab = 0xFB2A;
    185 	      found = true;
    186 	  } else if (a == 0xFB49) { /* SHIN WITH DAGESH */
    187 	      *ab = 0xFB2C;
    188 	      found = true;
    189 	  }
    190 	  break;
    191       case 0x05C2: /* SIN DOT */
    192 	  if (a == 0x05E9) { /* SHIN */
    193 	      *ab = 0xFB2B;
    194 	      found = true;
    195 	  } else if (a == 0xFB49) { /* SHIN WITH DAGESH */
    196 	      *ab = 0xFB2D;
    197 	      found = true;
    198 	  }
    199 	  break;
    200       }
    201   }
    202 
    203   return found;
    204 }
    205 
    206 const hb_ot_complex_shaper_t _hb_ot_complex_shaper_default =
    207 {
    208   "default",
    209   collect_features_default,
    210   NULL, /* override_features */
    211   NULL, /* data_create */
    212   NULL, /* data_destroy */
    213   NULL, /* preprocess_text */
    214   normalization_preference_default,
    215   NULL, /* decompose */
    216   compose_default,
    217   NULL, /* setup_masks */
    218   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_UNICODE_LATE,
    219   true, /* fallback_position */
    220 };
    221