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 
     50 
     51 /* Master OT shaper list */
     52 #define HB_COMPLEX_SHAPERS_IMPLEMENT_SHAPERS \
     53   HB_COMPLEX_SHAPER_IMPLEMENT (default) /* should be first */ \
     54   HB_COMPLEX_SHAPER_IMPLEMENT (arabic) \
     55   HB_COMPLEX_SHAPER_IMPLEMENT (indic) \
     56   HB_COMPLEX_SHAPER_IMPLEMENT (myanmar) \
     57   HB_COMPLEX_SHAPER_IMPLEMENT (sea) \
     58   HB_COMPLEX_SHAPER_IMPLEMENT (thai) \
     59   /* ^--- Add new shapers here */
     60 
     61 
     62 struct hb_ot_complex_shaper_t
     63 {
     64   char name[8];
     65 
     66   /* collect_features()
     67    * Called during shape_plan().
     68    * Shapers should use plan->map to add their features and callbacks.
     69    * May be NULL.
     70    */
     71   void (*collect_features) (hb_ot_shape_planner_t *plan);
     72 
     73   /* override_features()
     74    * Called during shape_plan().
     75    * Shapers should use plan->map to override features and add callbacks after
     76    * common features are added.
     77    * May be NULL.
     78    */
     79   void (*override_features) (hb_ot_shape_planner_t *plan);
     80 
     81 
     82   /* data_create()
     83    * Called at the end of shape_plan().
     84    * Whatever shapers return will be accessible through plan->data later.
     85    * If NULL is returned, means a plan failure.
     86    */
     87   void *(*data_create) (const hb_ot_shape_plan_t *plan);
     88 
     89   /* data_destroy()
     90    * Called when the shape_plan is being destroyed.
     91    * plan->data is passed here for destruction.
     92    * If NULL is returned, means a plan failure.
     93    * May be NULL.
     94    */
     95   void (*data_destroy) (void *data);
     96 
     97 
     98   /* preprocess_text()
     99    * Called during shape().
    100    * Shapers can use to modify text before shaping starts.
    101    * May be NULL.
    102    */
    103   void (*preprocess_text) (const hb_ot_shape_plan_t *plan,
    104 			   hb_buffer_t              *buffer,
    105 			   hb_font_t                *font);
    106 
    107 
    108   /* normalization_preference()
    109    * Called during shape().
    110    * May be NULL.
    111    */
    112   hb_ot_shape_normalization_mode_t
    113   (*normalization_preference) (const hb_segment_properties_t *props);
    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     case HB_SCRIPT_MONGOLIAN:
    165     case HB_SCRIPT_SYRIAC:
    166 
    167     /* Unicode-5.0 additions */
    168     case HB_SCRIPT_NKO:
    169     case HB_SCRIPT_PHAGS_PA:
    170 
    171     /* Unicode-6.0 additions */
    172     case HB_SCRIPT_MANDAIC:
    173 
    174       /* For Arabic script, use the Arabic shaper even if no OT script tag was found.
    175        * This is because we do fallback shaping for Arabic script (and not others). */
    176       if (planner->map.chosen_script[0] != HB_OT_TAG_DEFAULT_SCRIPT ||
    177 	  planner->props.script == HB_SCRIPT_ARABIC)
    178 	return &_hb_ot_complex_shaper_arabic;
    179       else
    180 	return &_hb_ot_complex_shaper_default;
    181 
    182 
    183     /* Unicode-1.1 additions */
    184     case HB_SCRIPT_THAI:
    185     case HB_SCRIPT_LAO:
    186 
    187       return &_hb_ot_complex_shaper_thai;
    188 
    189 
    190 #if 0
    191     /* Note:
    192      * Currently we don't have a separate Hangul shaper.  The default shaper handles
    193      * Hangul by enabling jamo features.  We may want to implement a separate shaper
    194      * in the future.  See this thread for details of what such a shaper would do:
    195      *
    196      *   http://lists.freedesktop.org/archives/harfbuzz/2013-April/003070.html
    197      */
    198     /* Unicode-1.1 additions */
    199     case HB_SCRIPT_HANGUL:
    200 
    201       return &_hb_ot_complex_shaper_hangul;
    202 #endif
    203 
    204 
    205     /* ^--- Add new shapers here */
    206 
    207 
    208 #if 0
    209     /* Note:
    210      *
    211      * These disabled scripts are listed in ucd/IndicSyllabicCategory.txt, but according
    212      * to Martin Hosken and Jonathan Kew do not require complex shaping.
    213      *
    214      * TODO We should automate figuring out which scripts do not need complex shaping
    215      *
    216      * TODO We currently keep data for these scripts in our indic table.  Need to fix the
    217      * generator to not do that.
    218      */
    219 
    220 
    221     /* Simple? */
    222 
    223     /* Unicode-3.2 additions */
    224     case HB_SCRIPT_BUHID:
    225     case HB_SCRIPT_HANUNOO:
    226 
    227     /* Unicode-5.1 additions */
    228     case HB_SCRIPT_SAURASHTRA:
    229 
    230     /* Unicode-6.0 additions */
    231     case HB_SCRIPT_BATAK:
    232     case HB_SCRIPT_BRAHMI:
    233 
    234 
    235     /* Simple */
    236 
    237     /* Unicode-1.1 additions */
    238     /* These have their own shaper now. */
    239     case HB_SCRIPT_LAO:
    240     case HB_SCRIPT_THAI:
    241 
    242     /* Unicode-2.0 additions */
    243     case HB_SCRIPT_TIBETAN:
    244 
    245     /* Unicode-3.2 additions */
    246     case HB_SCRIPT_TAGALOG:
    247     case HB_SCRIPT_TAGBANWA:
    248 
    249     /* Unicode-4.0 additions */
    250     case HB_SCRIPT_LIMBU:
    251     case HB_SCRIPT_TAI_LE:
    252 
    253     /* Unicode-4.1 additions */
    254     case HB_SCRIPT_KHAROSHTHI:
    255     case HB_SCRIPT_SYLOTI_NAGRI:
    256 
    257     /* Unicode-5.1 additions */
    258     case HB_SCRIPT_KAYAH_LI:
    259 
    260     /* Unicode-5.2 additions */
    261     case HB_SCRIPT_TAI_VIET:
    262 
    263 
    264 #endif
    265 
    266     /* Unicode-1.1 additions */
    267     case HB_SCRIPT_BENGALI:
    268     case HB_SCRIPT_DEVANAGARI:
    269     case HB_SCRIPT_GUJARATI:
    270     case HB_SCRIPT_GURMUKHI:
    271     case HB_SCRIPT_KANNADA:
    272     case HB_SCRIPT_MALAYALAM:
    273     case HB_SCRIPT_ORIYA:
    274     case HB_SCRIPT_TAMIL:
    275     case HB_SCRIPT_TELUGU:
    276 
    277     /* Unicode-3.0 additions */
    278     case HB_SCRIPT_SINHALA:
    279 
    280     /* Unicode-4.1 additions */
    281     case HB_SCRIPT_BUGINESE:
    282 
    283     /* Unicode-5.0 additions */
    284     case HB_SCRIPT_BALINESE:
    285 
    286     /* Unicode-5.1 additions */
    287     case HB_SCRIPT_LEPCHA:
    288     case HB_SCRIPT_REJANG:
    289     case HB_SCRIPT_SUNDANESE:
    290 
    291     /* Unicode-5.2 additions */
    292     case HB_SCRIPT_JAVANESE:
    293     case HB_SCRIPT_KAITHI:
    294     case HB_SCRIPT_MEETEI_MAYEK:
    295 
    296     /* Unicode-6.0 additions */
    297 
    298     /* Unicode-6.1 additions */
    299     case HB_SCRIPT_CHAKMA:
    300     case HB_SCRIPT_SHARADA:
    301     case HB_SCRIPT_TAKRI:
    302 
    303       /* If the designer designed the font for the 'DFLT' script,
    304        * use the default shaper.  Otherwise, use the Indic shaper.
    305        * Note that for some simple scripts, there may not be *any*
    306        * GSUB/GPOS needed, so there may be no scripts found! */
    307       if (planner->map.chosen_script[0] == HB_TAG ('D','F','L','T'))
    308 	return &_hb_ot_complex_shaper_default;
    309       else
    310 	return &_hb_ot_complex_shaper_indic;
    311 
    312     case HB_SCRIPT_KHMER:
    313       /* A number of Khmer fonts in the wild don't have a 'pref' feature,
    314        * and as such won't shape properly via the Indic shaper;
    315        * however, they typically have 'liga' / 'clig' features that implement
    316        * the necessary "reordering" by means of ligature substitutions.
    317        * So we send such pref-less fonts through the generic shaper instead. */
    318       if (planner->map.found_script[0] &&
    319 	  hb_ot_layout_language_find_feature (planner->face, HB_OT_TAG_GSUB,
    320 					      planner->map.script_index[0],
    321 					      planner->map.language_index[0],
    322 					      HB_TAG ('p','r','e','f'),
    323 					      NULL))
    324 	return &_hb_ot_complex_shaper_indic;
    325       else
    326 	return &_hb_ot_complex_shaper_default;
    327 
    328     case HB_SCRIPT_MYANMAR:
    329       /* For Myanmar, we only want to use the Myanmar shaper if the "new" script
    330        * tag is found.  For "old" script tag we want to use the default shaper. */
    331       if (planner->map.chosen_script[0] == HB_TAG ('m','y','m','2'))
    332 	return &_hb_ot_complex_shaper_myanmar;
    333       else
    334 	return &_hb_ot_complex_shaper_default;
    335 
    336     /* Unicode-4.1 additions */
    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