Home | History | Annotate | Download | only in src
      1 /*
      2  * Copyright  2010,2011,2012  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 #ifndef HB_OT_SHAPE_COMPLEX_PRIVATE_HH
     28 #define HB_OT_SHAPE_COMPLEX_PRIVATE_HH
     29 
     30 #include "hb-private.hh"
     31 
     32 #include "hb-ot-shape-private.hh"
     33 #include "hb-ot-shape-normalize-private.hh"
     34 
     35 
     36 
     37 /* buffer var allocations, used by complex shapers */
     38 #define complex_var_u8_0()	var2.u8[2]
     39 #define complex_var_u8_1()	var2.u8[3]
     40 
     41 
     42 enum hb_ot_shape_zero_width_marks_type_t {
     43   HB_OT_SHAPE_ZERO_WIDTH_MARKS_NONE,
     44 //  HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_UNICODE_EARLY,
     45   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_UNICODE_LATE,
     46   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_EARLY,
     47   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_LATE,
     48 
     49   HB_OT_SHAPE_ZERO_WIDTH_MARKS_DEFAULT = HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_UNICODE_LATE
     50 };
     51 
     52 
     53 /* Master OT shaper list */
     54 #define HB_COMPLEX_SHAPERS_IMPLEMENT_SHAPERS \
     55   HB_COMPLEX_SHAPER_IMPLEMENT (default) /* should be first */ \
     56   HB_COMPLEX_SHAPER_IMPLEMENT (arabic) \
     57   HB_COMPLEX_SHAPER_IMPLEMENT (hangul) \
     58   HB_COMPLEX_SHAPER_IMPLEMENT (hebrew) \
     59   HB_COMPLEX_SHAPER_IMPLEMENT (indic) \
     60   HB_COMPLEX_SHAPER_IMPLEMENT (myanmar) \
     61   HB_COMPLEX_SHAPER_IMPLEMENT (sea) \
     62   HB_COMPLEX_SHAPER_IMPLEMENT (thai) \
     63   HB_COMPLEX_SHAPER_IMPLEMENT (tibetan) \
     64   /* ^--- Add new shapers here */
     65 
     66 
     67 struct hb_ot_complex_shaper_t
     68 {
     69   char name[8];
     70 
     71   /* collect_features()
     72    * Called during shape_plan().
     73    * Shapers should use plan->map to add their features and callbacks.
     74    * May be NULL.
     75    */
     76   void (*collect_features) (hb_ot_shape_planner_t *plan);
     77 
     78   /* override_features()
     79    * Called during shape_plan().
     80    * Shapers should use plan->map to override features and add callbacks after
     81    * common features are added.
     82    * May be NULL.
     83    */
     84   void (*override_features) (hb_ot_shape_planner_t *plan);
     85 
     86 
     87   /* data_create()
     88    * Called at the end of shape_plan().
     89    * Whatever shapers return will be accessible through plan->data later.
     90    * If NULL is returned, means a plan failure.
     91    */
     92   void *(*data_create) (const hb_ot_shape_plan_t *plan);
     93 
     94   /* data_destroy()
     95    * Called when the shape_plan is being destroyed.
     96    * plan->data is passed here for destruction.
     97    * If NULL is returned, means a plan failure.
     98    * May be NULL.
     99    */
    100   void (*data_destroy) (void *data);
    101 
    102 
    103   /* preprocess_text()
    104    * Called during shape().
    105    * Shapers can use to modify text before shaping starts.
    106    * May be NULL.
    107    */
    108   void (*preprocess_text) (const hb_ot_shape_plan_t *plan,
    109 			   hb_buffer_t              *buffer,
    110 			   hb_font_t                *font);
    111 
    112 
    113   hb_ot_shape_normalization_mode_t normalization_preference;
    114 
    115   /* decompose()
    116    * Called during shape()'s normalization.
    117    * May be NULL.
    118    */
    119   bool (*decompose) (const hb_ot_shape_normalize_context_t *c,
    120 		     hb_codepoint_t  ab,
    121 		     hb_codepoint_t *a,
    122 		     hb_codepoint_t *b);
    123 
    124   /* compose()
    125    * Called during shape()'s normalization.
    126    * May be NULL.
    127    */
    128   bool (*compose) (const hb_ot_shape_normalize_context_t *c,
    129 		   hb_codepoint_t  a,
    130 		   hb_codepoint_t  b,
    131 		   hb_codepoint_t *ab);
    132 
    133   /* setup_masks()
    134    * Called during shape().
    135    * Shapers should use map to get feature masks and set on buffer.
    136    * Shapers may NOT modify characters.
    137    * May be NULL.
    138    */
    139   void (*setup_masks) (const hb_ot_shape_plan_t *plan,
    140 		       hb_buffer_t              *buffer,
    141 		       hb_font_t                *font);
    142 
    143   hb_ot_shape_zero_width_marks_type_t zero_width_marks;
    144 
    145   bool fallback_position;
    146 };
    147 
    148 #define HB_COMPLEX_SHAPER_IMPLEMENT(name) extern HB_INTERNAL const hb_ot_complex_shaper_t _hb_ot_complex_shaper_##name;
    149 HB_COMPLEX_SHAPERS_IMPLEMENT_SHAPERS
    150 #undef HB_COMPLEX_SHAPER_IMPLEMENT
    151 
    152 
    153 static inline const hb_ot_complex_shaper_t *
    154 hb_ot_shape_complex_categorize (const hb_ot_shape_planner_t *planner)
    155 {
    156   switch ((hb_tag_t) planner->props.script)
    157   {
    158     default:
    159       return &_hb_ot_complex_shaper_default;
    160 
    161 
    162     /* Unicode-1.1 additions */
    163     case HB_SCRIPT_ARABIC:
    164 
    165     /* Unicode-3.0 additions */
    166     case HB_SCRIPT_MONGOLIAN:
    167     case HB_SCRIPT_SYRIAC:
    168 
    169     /* Unicode-5.0 additions */
    170     case HB_SCRIPT_NKO:
    171     case HB_SCRIPT_PHAGS_PA:
    172 
    173     /* Unicode-6.0 additions */
    174     case HB_SCRIPT_MANDAIC:
    175 
    176       /* For Arabic script, use the Arabic shaper even if no OT script tag was found.
    177        * This is because we do fallback shaping for Arabic script (and not others). */
    178       if (planner->map.chosen_script[0] != HB_OT_TAG_DEFAULT_SCRIPT ||
    179 	  planner->props.script == HB_SCRIPT_ARABIC)
    180 	return &_hb_ot_complex_shaper_arabic;
    181       else
    182 	return &_hb_ot_complex_shaper_default;
    183 
    184 
    185     /* Unicode-1.1 additions */
    186     case HB_SCRIPT_THAI:
    187     case HB_SCRIPT_LAO:
    188 
    189       return &_hb_ot_complex_shaper_thai;
    190 
    191 
    192     /* Unicode-1.1 additions */
    193     case HB_SCRIPT_HANGUL:
    194 
    195       return &_hb_ot_complex_shaper_hangul;
    196 
    197 
    198     /* Unicode-2.0 additions */
    199     case HB_SCRIPT_TIBETAN:
    200 
    201       return &_hb_ot_complex_shaper_tibetan;
    202 
    203 
    204     /* Unicode-1.1 additions */
    205     case HB_SCRIPT_HEBREW:
    206 
    207       return &_hb_ot_complex_shaper_hebrew;
    208 
    209 
    210     /* ^--- Add new shapers here */
    211 
    212 
    213 #if 0
    214     /* Note:
    215      *
    216      * These disabled scripts are listed in ucd/IndicSyllabicCategory.txt, but according
    217      * to Martin Hosken and Jonathan Kew do not require complex shaping.
    218      *
    219      * TODO We should automate figuring out which scripts do not need complex shaping
    220      *
    221      * TODO We currently keep data for these scripts in our indic table.  Need to fix the
    222      * generator to not do that.
    223      */
    224 
    225 
    226     /* Simple? */
    227 
    228     /* Unicode-3.2 additions */
    229     case HB_SCRIPT_BUHID:
    230     case HB_SCRIPT_HANUNOO:
    231 
    232     /* Unicode-5.1 additions */
    233     case HB_SCRIPT_SAURASHTRA:
    234 
    235     /* Unicode-6.0 additions */
    236     case HB_SCRIPT_BATAK:
    237     case HB_SCRIPT_BRAHMI:
    238 
    239 
    240     /* Simple */
    241 
    242     /* Unicode-1.1 additions */
    243     /* These have their own shaper now. */
    244     case HB_SCRIPT_LAO:
    245     case HB_SCRIPT_THAI:
    246 
    247     /* Unicode-3.2 additions */
    248     case HB_SCRIPT_TAGALOG:
    249     case HB_SCRIPT_TAGBANWA:
    250 
    251     /* Unicode-4.0 additions */
    252     case HB_SCRIPT_LIMBU:
    253     case HB_SCRIPT_TAI_LE:
    254 
    255     /* Unicode-4.1 additions */
    256     case HB_SCRIPT_KHAROSHTHI:
    257     case HB_SCRIPT_SYLOTI_NAGRI:
    258 
    259     /* Unicode-5.1 additions */
    260     case HB_SCRIPT_KAYAH_LI:
    261 
    262     /* Unicode-5.2 additions */
    263     case HB_SCRIPT_TAI_VIET:
    264 
    265 
    266 #endif
    267 
    268     /* Unicode-1.1 additions */
    269     case HB_SCRIPT_BENGALI:
    270     case HB_SCRIPT_DEVANAGARI:
    271     case HB_SCRIPT_GUJARATI:
    272     case HB_SCRIPT_GURMUKHI:
    273     case HB_SCRIPT_KANNADA:
    274     case HB_SCRIPT_MALAYALAM:
    275     case HB_SCRIPT_ORIYA:
    276     case HB_SCRIPT_TAMIL:
    277     case HB_SCRIPT_TELUGU:
    278 
    279     /* Unicode-3.0 additions */
    280     case HB_SCRIPT_SINHALA:
    281 
    282     /* Unicode-5.0 additions */
    283     case HB_SCRIPT_BALINESE:
    284 
    285     /* Unicode-5.1 additions */
    286     case HB_SCRIPT_LEPCHA:
    287     case HB_SCRIPT_REJANG:
    288     case HB_SCRIPT_SUNDANESE:
    289 
    290     /* Unicode-5.2 additions */
    291     case HB_SCRIPT_JAVANESE:
    292     case HB_SCRIPT_KAITHI:
    293     case HB_SCRIPT_MEETEI_MAYEK:
    294 
    295     /* Unicode-6.0 additions */
    296 
    297     /* Unicode-6.1 additions */
    298     case HB_SCRIPT_CHAKMA:
    299     case HB_SCRIPT_SHARADA:
    300     case HB_SCRIPT_TAKRI:
    301 
    302       /* If the designer designed the font for the 'DFLT' script,
    303        * use the default shaper.  Otherwise, use the Indic shaper.
    304        * Note that for some simple scripts, there may not be *any*
    305        * GSUB/GPOS needed, so there may be no scripts found! */
    306       if (planner->map.chosen_script[0] == HB_TAG ('D','F','L','T'))
    307 	return &_hb_ot_complex_shaper_default;
    308       else
    309 	return &_hb_ot_complex_shaper_indic;
    310 
    311     case HB_SCRIPT_KHMER:
    312       /* A number of Khmer fonts in the wild don't have a 'pref' feature,
    313        * and as such won't shape properly via the Indic shaper;
    314        * however, they typically have 'liga' / 'clig' features that implement
    315        * the necessary "reordering" by means of ligature substitutions.
    316        * So we send such pref-less fonts through the generic shaper instead. */
    317       if (planner->map.found_script[0] &&
    318 	  hb_ot_layout_language_find_feature (planner->face, HB_OT_TAG_GSUB,
    319 					      planner->map.script_index[0],
    320 					      planner->map.language_index[0],
    321 					      HB_TAG ('p','r','e','f'),
    322 					      NULL))
    323 	return &_hb_ot_complex_shaper_indic;
    324       else
    325 	return &_hb_ot_complex_shaper_default;
    326 
    327     case HB_SCRIPT_MYANMAR:
    328       /* For Myanmar, we only want to use the Myanmar shaper if the "new" script
    329        * tag is found.  For "old" script tag we want to use the default shaper. */
    330       if (planner->map.chosen_script[0] == HB_TAG ('m','y','m','2'))
    331 	return &_hb_ot_complex_shaper_myanmar;
    332       else
    333 	return &_hb_ot_complex_shaper_default;
    334 
    335     /* Unicode-4.1 additions */
    336     case HB_SCRIPT_BUGINESE:
    337     case HB_SCRIPT_NEW_TAI_LUE:
    338 
    339     /* Unicode-5.1 additions */
    340     case HB_SCRIPT_CHAM:
    341 
    342     /* Unicode-5.2 additions */
    343     case HB_SCRIPT_TAI_THAM:
    344 
    345       /* If the designer designed the font for the 'DFLT' script,
    346        * use the default shaper.  Otherwise, use the Indic shaper.
    347        * Note that for some simple scripts, there may not be *any*
    348        * GSUB/GPOS needed, so there may be no scripts found! */
    349       if (planner->map.chosen_script[0] == HB_TAG ('D','F','L','T'))
    350 	return &_hb_ot_complex_shaper_default;
    351       else
    352 	return &_hb_ot_complex_shaper_sea;
    353   }
    354 }
    355 
    356 
    357 #endif /* HB_OT_SHAPE_COMPLEX_PRIVATE_HH */
    358