michael@0: /*
michael@0:  * Copyright © 2010,2011,2012  Google, Inc.
michael@0:  *
michael@0:  *  This is part of HarfBuzz, a text shaping library.
michael@0:  *
michael@0:  * Permission is hereby granted, without written agreement and without
michael@0:  * license or royalty fees, to use, copy, modify, and distribute this
michael@0:  * software and its documentation for any purpose, provided that the
michael@0:  * above copyright notice and the following two paragraphs appear in
michael@0:  * all copies of this software.
michael@0:  *
michael@0:  * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
michael@0:  * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
michael@0:  * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
michael@0:  * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
michael@0:  * DAMAGE.
michael@0:  *
michael@0:  * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
michael@0:  * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
michael@0:  * FITNESS FOR A PARTICULAR PURPOSE.  THE SOFTWARE PROVIDED HEREUNDER IS
michael@0:  * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
michael@0:  * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
michael@0:  *
michael@0:  * Google Author(s): Behdad Esfahbod
michael@0:  */
michael@0: 
michael@0: #ifndef HB_OT_SHAPE_COMPLEX_PRIVATE_HH
michael@0: #define HB_OT_SHAPE_COMPLEX_PRIVATE_HH
michael@0: 
michael@0: #include "hb-private.hh"
michael@0: 
michael@0: #include "hb-ot-shape-private.hh"
michael@0: #include "hb-ot-shape-normalize-private.hh"
michael@0: 
michael@0: 
michael@0: 
michael@0: /* buffer var allocations, used by complex shapers */
michael@0: #define complex_var_u8_0()	var2.u8[2]
michael@0: #define complex_var_u8_1()	var2.u8[3]
michael@0: 
michael@0: 
michael@0: enum hb_ot_shape_zero_width_marks_type_t {
michael@0:   HB_OT_SHAPE_ZERO_WIDTH_MARKS_NONE,
michael@0: //  HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_UNICODE_EARLY,
michael@0:   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_UNICODE_LATE,
michael@0:   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_EARLY,
michael@0:   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_LATE,
michael@0: 
michael@0:   HB_OT_SHAPE_ZERO_WIDTH_MARKS_DEFAULT = HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_UNICODE_LATE
michael@0: };
michael@0: 
michael@0: 
michael@0: /* Master OT shaper list */
michael@0: #define HB_COMPLEX_SHAPERS_IMPLEMENT_SHAPERS \
michael@0:   HB_COMPLEX_SHAPER_IMPLEMENT (default) /* should be first */ \
michael@0:   HB_COMPLEX_SHAPER_IMPLEMENT (arabic) \
michael@0:   HB_COMPLEX_SHAPER_IMPLEMENT (hangul) \
michael@0:   HB_COMPLEX_SHAPER_IMPLEMENT (hebrew) \
michael@0:   HB_COMPLEX_SHAPER_IMPLEMENT (indic) \
michael@0:   HB_COMPLEX_SHAPER_IMPLEMENT (myanmar) \
michael@0:   HB_COMPLEX_SHAPER_IMPLEMENT (sea) \
michael@0:   HB_COMPLEX_SHAPER_IMPLEMENT (thai) \
michael@0:   HB_COMPLEX_SHAPER_IMPLEMENT (tibetan) \
michael@0:   /* ^--- Add new shapers here */
michael@0: 
michael@0: 
michael@0: struct hb_ot_complex_shaper_t
michael@0: {
michael@0:   char name[8];
michael@0: 
michael@0:   /* collect_features()
michael@0:    * Called during shape_plan().
michael@0:    * Shapers should use plan->map to add their features and callbacks.
michael@0:    * May be NULL.
michael@0:    */
michael@0:   void (*collect_features) (hb_ot_shape_planner_t *plan);
michael@0: 
michael@0:   /* override_features()
michael@0:    * Called during shape_plan().
michael@0:    * Shapers should use plan->map to override features and add callbacks after
michael@0:    * common features are added.
michael@0:    * May be NULL.
michael@0:    */
michael@0:   void (*override_features) (hb_ot_shape_planner_t *plan);
michael@0: 
michael@0: 
michael@0:   /* data_create()
michael@0:    * Called at the end of shape_plan().
michael@0:    * Whatever shapers return will be accessible through plan->data later.
michael@0:    * If NULL is returned, means a plan failure.
michael@0:    */
michael@0:   void *(*data_create) (const hb_ot_shape_plan_t *plan);
michael@0: 
michael@0:   /* data_destroy()
michael@0:    * Called when the shape_plan is being destroyed.
michael@0:    * plan->data is passed here for destruction.
michael@0:    * If NULL is returned, means a plan failure.
michael@0:    * May be NULL.
michael@0:    */
michael@0:   void (*data_destroy) (void *data);
michael@0: 
michael@0: 
michael@0:   /* preprocess_text()
michael@0:    * Called during shape().
michael@0:    * Shapers can use to modify text before shaping starts.
michael@0:    * May be NULL.
michael@0:    */
michael@0:   void (*preprocess_text) (const hb_ot_shape_plan_t *plan,
michael@0: 			   hb_buffer_t              *buffer,
michael@0: 			   hb_font_t                *font);
michael@0: 
michael@0: 
michael@0:   hb_ot_shape_normalization_mode_t normalization_preference;
michael@0: 
michael@0:   /* decompose()
michael@0:    * Called during shape()'s normalization.
michael@0:    * May be NULL.
michael@0:    */
michael@0:   bool (*decompose) (const hb_ot_shape_normalize_context_t *c,
michael@0: 		     hb_codepoint_t  ab,
michael@0: 		     hb_codepoint_t *a,
michael@0: 		     hb_codepoint_t *b);
michael@0: 
michael@0:   /* compose()
michael@0:    * Called during shape()'s normalization.
michael@0:    * May be NULL.
michael@0:    */
michael@0:   bool (*compose) (const hb_ot_shape_normalize_context_t *c,
michael@0: 		   hb_codepoint_t  a,
michael@0: 		   hb_codepoint_t  b,
michael@0: 		   hb_codepoint_t *ab);
michael@0: 
michael@0:   /* setup_masks()
michael@0:    * Called during shape().
michael@0:    * Shapers should use map to get feature masks and set on buffer.
michael@0:    * Shapers may NOT modify characters.
michael@0:    * May be NULL.
michael@0:    */
michael@0:   void (*setup_masks) (const hb_ot_shape_plan_t *plan,
michael@0: 		       hb_buffer_t              *buffer,
michael@0: 		       hb_font_t                *font);
michael@0: 
michael@0:   hb_ot_shape_zero_width_marks_type_t zero_width_marks;
michael@0: 
michael@0:   bool fallback_position;
michael@0: };
michael@0: 
michael@0: #define HB_COMPLEX_SHAPER_IMPLEMENT(name) extern HB_INTERNAL const hb_ot_complex_shaper_t _hb_ot_complex_shaper_##name;
michael@0: HB_COMPLEX_SHAPERS_IMPLEMENT_SHAPERS
michael@0: #undef HB_COMPLEX_SHAPER_IMPLEMENT
michael@0: 
michael@0: 
michael@0: static inline const hb_ot_complex_shaper_t *
michael@0: hb_ot_shape_complex_categorize (const hb_ot_shape_planner_t *planner)
michael@0: {
michael@0:   switch ((hb_tag_t) planner->props.script)
michael@0:   {
michael@0:     default:
michael@0:       return &_hb_ot_complex_shaper_default;
michael@0: 
michael@0: 
michael@0:     /* Unicode-1.1 additions */
michael@0:     case HB_SCRIPT_ARABIC:
michael@0: 
michael@0:     /* Unicode-3.0 additions */
michael@0:     case HB_SCRIPT_MONGOLIAN:
michael@0:     case HB_SCRIPT_SYRIAC:
michael@0: 
michael@0:     /* Unicode-5.0 additions */
michael@0:     case HB_SCRIPT_NKO:
michael@0:     case HB_SCRIPT_PHAGS_PA:
michael@0: 
michael@0:     /* Unicode-6.0 additions */
michael@0:     case HB_SCRIPT_MANDAIC:
michael@0: 
michael@0:       /* For Arabic script, use the Arabic shaper even if no OT script tag was found.
michael@0:        * This is because we do fallback shaping for Arabic script (and not others). */
michael@0:       if (planner->map.chosen_script[0] != HB_OT_TAG_DEFAULT_SCRIPT ||
michael@0: 	  planner->props.script == HB_SCRIPT_ARABIC)
michael@0: 	return &_hb_ot_complex_shaper_arabic;
michael@0:       else
michael@0: 	return &_hb_ot_complex_shaper_default;
michael@0: 
michael@0: 
michael@0:     /* Unicode-1.1 additions */
michael@0:     case HB_SCRIPT_THAI:
michael@0:     case HB_SCRIPT_LAO:
michael@0: 
michael@0:       return &_hb_ot_complex_shaper_thai;
michael@0: 
michael@0: 
michael@0:     /* Unicode-1.1 additions */
michael@0:     case HB_SCRIPT_HANGUL:
michael@0: 
michael@0:       return &_hb_ot_complex_shaper_hangul;
michael@0: 
michael@0: 
michael@0:     /* Unicode-2.0 additions */
michael@0:     case HB_SCRIPT_TIBETAN:
michael@0: 
michael@0:       return &_hb_ot_complex_shaper_tibetan;
michael@0: 
michael@0: 
michael@0:     /* Unicode-1.1 additions */
michael@0:     case HB_SCRIPT_HEBREW:
michael@0: 
michael@0:       return &_hb_ot_complex_shaper_hebrew;
michael@0: 
michael@0: 
michael@0:     /* ^--- Add new shapers here */
michael@0: 
michael@0: 
michael@0: #if 0
michael@0:     /* Note:
michael@0:      *
michael@0:      * These disabled scripts are listed in ucd/IndicSyllabicCategory.txt, but according
michael@0:      * to Martin Hosken and Jonathan Kew do not require complex shaping.
michael@0:      *
michael@0:      * TODO We should automate figuring out which scripts do not need complex shaping
michael@0:      *
michael@0:      * TODO We currently keep data for these scripts in our indic table.  Need to fix the
michael@0:      * generator to not do that.
michael@0:      */
michael@0: 
michael@0: 
michael@0:     /* Simple? */
michael@0: 
michael@0:     /* Unicode-3.2 additions */
michael@0:     case HB_SCRIPT_BUHID:
michael@0:     case HB_SCRIPT_HANUNOO:
michael@0: 
michael@0:     /* Unicode-5.1 additions */
michael@0:     case HB_SCRIPT_SAURASHTRA:
michael@0: 
michael@0:     /* Unicode-6.0 additions */
michael@0:     case HB_SCRIPT_BATAK:
michael@0:     case HB_SCRIPT_BRAHMI:
michael@0: 
michael@0: 
michael@0:     /* Simple */
michael@0: 
michael@0:     /* Unicode-1.1 additions */
michael@0:     /* These have their own shaper now. */
michael@0:     case HB_SCRIPT_LAO:
michael@0:     case HB_SCRIPT_THAI:
michael@0: 
michael@0:     /* Unicode-3.2 additions */
michael@0:     case HB_SCRIPT_TAGALOG:
michael@0:     case HB_SCRIPT_TAGBANWA:
michael@0: 
michael@0:     /* Unicode-4.0 additions */
michael@0:     case HB_SCRIPT_LIMBU:
michael@0:     case HB_SCRIPT_TAI_LE:
michael@0: 
michael@0:     /* Unicode-4.1 additions */
michael@0:     case HB_SCRIPT_KHAROSHTHI:
michael@0:     case HB_SCRIPT_SYLOTI_NAGRI:
michael@0: 
michael@0:     /* Unicode-5.1 additions */
michael@0:     case HB_SCRIPT_KAYAH_LI:
michael@0: 
michael@0:     /* Unicode-5.2 additions */
michael@0:     case HB_SCRIPT_TAI_VIET:
michael@0: 
michael@0: 
michael@0: #endif
michael@0: 
michael@0:     /* Unicode-1.1 additions */
michael@0:     case HB_SCRIPT_BENGALI:
michael@0:     case HB_SCRIPT_DEVANAGARI:
michael@0:     case HB_SCRIPT_GUJARATI:
michael@0:     case HB_SCRIPT_GURMUKHI:
michael@0:     case HB_SCRIPT_KANNADA:
michael@0:     case HB_SCRIPT_MALAYALAM:
michael@0:     case HB_SCRIPT_ORIYA:
michael@0:     case HB_SCRIPT_TAMIL:
michael@0:     case HB_SCRIPT_TELUGU:
michael@0: 
michael@0:     /* Unicode-3.0 additions */
michael@0:     case HB_SCRIPT_SINHALA:
michael@0: 
michael@0:     /* Unicode-5.0 additions */
michael@0:     case HB_SCRIPT_BALINESE:
michael@0: 
michael@0:     /* Unicode-5.1 additions */
michael@0:     case HB_SCRIPT_LEPCHA:
michael@0:     case HB_SCRIPT_REJANG:
michael@0:     case HB_SCRIPT_SUNDANESE:
michael@0: 
michael@0:     /* Unicode-5.2 additions */
michael@0:     case HB_SCRIPT_JAVANESE:
michael@0:     case HB_SCRIPT_KAITHI:
michael@0:     case HB_SCRIPT_MEETEI_MAYEK:
michael@0: 
michael@0:     /* Unicode-6.0 additions */
michael@0: 
michael@0:     /* Unicode-6.1 additions */
michael@0:     case HB_SCRIPT_CHAKMA:
michael@0:     case HB_SCRIPT_SHARADA:
michael@0:     case HB_SCRIPT_TAKRI:
michael@0: 
michael@0:       /* If the designer designed the font for the 'DFLT' script,
michael@0:        * use the default shaper.  Otherwise, use the Indic shaper.
michael@0:        * Note that for some simple scripts, there may not be *any*
michael@0:        * GSUB/GPOS needed, so there may be no scripts found! */
michael@0:       if (planner->map.chosen_script[0] == HB_TAG ('D','F','L','T'))
michael@0: 	return &_hb_ot_complex_shaper_default;
michael@0:       else
michael@0: 	return &_hb_ot_complex_shaper_indic;
michael@0: 
michael@0:     case HB_SCRIPT_KHMER:
michael@0:       /* A number of Khmer fonts in the wild don't have a 'pref' feature,
michael@0:        * and as such won't shape properly via the Indic shaper;
michael@0:        * however, they typically have 'liga' / 'clig' features that implement
michael@0:        * the necessary "reordering" by means of ligature substitutions.
michael@0:        * So we send such pref-less fonts through the generic shaper instead. */
michael@0:       if (planner->map.found_script[0] &&
michael@0: 	  hb_ot_layout_language_find_feature (planner->face, HB_OT_TAG_GSUB,
michael@0: 					      planner->map.script_index[0],
michael@0: 					      planner->map.language_index[0],
michael@0: 					      HB_TAG ('p','r','e','f'),
michael@0: 					      NULL))
michael@0: 	return &_hb_ot_complex_shaper_indic;
michael@0:       else
michael@0: 	return &_hb_ot_complex_shaper_default;
michael@0: 
michael@0:     case HB_SCRIPT_MYANMAR:
michael@0:       /* For Myanmar, we only want to use the Myanmar shaper if the "new" script
michael@0:        * tag is found.  For "old" script tag we want to use the default shaper. */
michael@0:       if (planner->map.chosen_script[0] == HB_TAG ('m','y','m','2'))
michael@0: 	return &_hb_ot_complex_shaper_myanmar;
michael@0:       else
michael@0: 	return &_hb_ot_complex_shaper_default;
michael@0: 
michael@0:     /* Unicode-4.1 additions */
michael@0:     case HB_SCRIPT_BUGINESE:
michael@0:     case HB_SCRIPT_NEW_TAI_LUE:
michael@0: 
michael@0:     /* Unicode-5.1 additions */
michael@0:     case HB_SCRIPT_CHAM:
michael@0: 
michael@0:     /* Unicode-5.2 additions */
michael@0:     case HB_SCRIPT_TAI_THAM:
michael@0: 
michael@0:       /* If the designer designed the font for the 'DFLT' script,
michael@0:        * use the default shaper.  Otherwise, use the Indic shaper.
michael@0:        * Note that for some simple scripts, there may not be *any*
michael@0:        * GSUB/GPOS needed, so there may be no scripts found! */
michael@0:       if (planner->map.chosen_script[0] == HB_TAG ('D','F','L','T'))
michael@0: 	return &_hb_ot_complex_shaper_default;
michael@0:       else
michael@0: 	return &_hb_ot_complex_shaper_sea;
michael@0:   }
michael@0: }
michael@0: 
michael@0: 
michael@0: #endif /* HB_OT_SHAPE_COMPLEX_PRIVATE_HH */