michael@0: /*
michael@0:  * Copyright © 2011,2012,2013  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: #include "hb-ot-shape-complex-indic-private.hh"
michael@0: 
michael@0: /* buffer var allocations */
michael@0: #define sea_category() complex_var_u8_0() /* indic_category_t */
michael@0: #define sea_position() complex_var_u8_1() /* indic_position_t */
michael@0: 
michael@0: 
michael@0: /*
michael@0:  * South-East Asian shaper.
michael@0:  * Loosely based on the Myanmar spec / shaper.
michael@0:  * There is no OpenType spec for this.
michael@0:  */
michael@0: 
michael@0: static const hb_tag_t
michael@0: basic_features[] =
michael@0: {
michael@0:   /*
michael@0:    * Basic features.
michael@0:    * These features are applied in order, one at a time, after initial_reordering.
michael@0:    */
michael@0:   HB_TAG('p','r','e','f'),
michael@0:   HB_TAG('a','b','v','f'),
michael@0:   HB_TAG('b','l','w','f'),
michael@0:   HB_TAG('p','s','t','f'),
michael@0: };
michael@0: static const hb_tag_t
michael@0: other_features[] =
michael@0: {
michael@0:   /*
michael@0:    * Other features.
michael@0:    * These features are applied all at once, after final_reordering.
michael@0:    */
michael@0:   HB_TAG('p','r','e','s'),
michael@0:   HB_TAG('a','b','v','s'),
michael@0:   HB_TAG('b','l','w','s'),
michael@0:   HB_TAG('p','s','t','s'),
michael@0:   /* Positioning features, though we don't care about the types. */
michael@0:   HB_TAG('d','i','s','t'),
michael@0: };
michael@0: 
michael@0: static void
michael@0: setup_syllables (const hb_ot_shape_plan_t *plan,
michael@0: 		 hb_font_t *font,
michael@0: 		 hb_buffer_t *buffer);
michael@0: static void
michael@0: initial_reordering (const hb_ot_shape_plan_t *plan,
michael@0: 		    hb_font_t *font,
michael@0: 		    hb_buffer_t *buffer);
michael@0: static void
michael@0: final_reordering (const hb_ot_shape_plan_t *plan,
michael@0: 		  hb_font_t *font,
michael@0: 		  hb_buffer_t *buffer);
michael@0: 
michael@0: static void
michael@0: collect_features_sea (hb_ot_shape_planner_t *plan)
michael@0: {
michael@0:   hb_ot_map_builder_t *map = &plan->map;
michael@0: 
michael@0:   /* Do this before any lookups have been applied. */
michael@0:   map->add_gsub_pause (setup_syllables);
michael@0: 
michael@0:   map->add_global_bool_feature (HB_TAG('l','o','c','l'));
michael@0:   /* The Indic specs do not require ccmp, but we apply it here since if
michael@0:    * there is a use of it, it's typically at the beginning. */
michael@0:   map->add_global_bool_feature (HB_TAG('c','c','m','p'));
michael@0: 
michael@0:   map->add_gsub_pause (initial_reordering);
michael@0:   for (unsigned int i = 0; i < ARRAY_LENGTH (basic_features); i++)
michael@0:   {
michael@0:     map->add_feature (basic_features[i], 1, F_GLOBAL | F_MANUAL_ZWJ);
michael@0:     map->add_gsub_pause (NULL);
michael@0:   }
michael@0:   map->add_gsub_pause (final_reordering);
michael@0:   for (unsigned int i = 0; i < ARRAY_LENGTH (other_features); i++)
michael@0:     map->add_feature (other_features[i], 1, F_GLOBAL | F_MANUAL_ZWJ);
michael@0: }
michael@0: 
michael@0: static void
michael@0: override_features_sea (hb_ot_shape_planner_t *plan)
michael@0: {
michael@0:   plan->map.add_feature (HB_TAG('l','i','g','a'), 0, F_GLOBAL);
michael@0: }
michael@0: 
michael@0: 
michael@0: enum syllable_type_t {
michael@0:   consonant_syllable,
michael@0:   broken_cluster,
michael@0:   non_sea_cluster,
michael@0: };
michael@0: 
michael@0: #include "hb-ot-shape-complex-sea-machine.hh"
michael@0: 
michael@0: 
michael@0: /* Note: This enum is duplicated in the -machine.rl source file.
michael@0:  * Not sure how to avoid duplication. */
michael@0: enum sea_category_t {
michael@0: //  OT_C    = 1,
michael@0:   OT_GB   = 12, /* Generic Base XXX DOTTED CIRCLE only for now */
michael@0: //  OT_H    = 4,  /* Halant */
michael@0:   OT_IV   = 2,  /* Independent Vowel */
michael@0:   OT_MR   = 22, /* Medial Ra */
michael@0: //  OT_CM   = 17, /* Consonant Medial */
michael@0:   OT_VAbv = 26,
michael@0:   OT_VBlw = 27,
michael@0:   OT_VPre = 28,
michael@0:   OT_VPst = 29,
michael@0:   OT_T    = 3,  /* Tone Marks */
michael@0: //  OT_A    = 10, /* Anusvara */
michael@0: };
michael@0: 
michael@0: static inline void
michael@0: set_sea_properties (hb_glyph_info_t &info)
michael@0: {
michael@0:   hb_codepoint_t u = info.codepoint;
michael@0:   unsigned int type = hb_indic_get_categories (u);
michael@0:   indic_category_t cat = (indic_category_t) (type & 0x7F);
michael@0:   indic_position_t pos = (indic_position_t) (type >> 8);
michael@0: 
michael@0:   /* Medial Ra */
michael@0:   if (u == 0x1A55 || u == 0xAA34)
michael@0:     cat = (indic_category_t) OT_MR;
michael@0: 
michael@0:   if (cat == OT_M)
michael@0:   {
michael@0:     switch ((int) pos)
michael@0:     {
michael@0:       case POS_PRE_C:	cat = (indic_category_t) OT_VPre; break;
michael@0:       case POS_ABOVE_C:	cat = (indic_category_t) OT_VAbv; break;
michael@0:       case POS_BELOW_C:	cat = (indic_category_t) OT_VBlw; break;
michael@0:       case POS_POST_C:	cat = (indic_category_t) OT_VPst; break;
michael@0:     }
michael@0:   }
michael@0: 
michael@0:   info.sea_category() = (sea_category_t) cat;
michael@0:   info.sea_position() = pos;
michael@0: }
michael@0: 
michael@0: 
michael@0: static void
michael@0: setup_masks_sea (const hb_ot_shape_plan_t *plan HB_UNUSED,
michael@0: 		 hb_buffer_t              *buffer,
michael@0: 		 hb_font_t                *font HB_UNUSED)
michael@0: {
michael@0:   HB_BUFFER_ALLOCATE_VAR (buffer, sea_category);
michael@0:   HB_BUFFER_ALLOCATE_VAR (buffer, sea_position);
michael@0: 
michael@0:   /* We cannot setup masks here.  We save information about characters
michael@0:    * and setup masks later on in a pause-callback. */
michael@0: 
michael@0:   unsigned int count = buffer->len;
michael@0:   for (unsigned int i = 0; i < count; i++)
michael@0:     set_sea_properties (buffer->info[i]);
michael@0: }
michael@0: 
michael@0: static void
michael@0: setup_syllables (const hb_ot_shape_plan_t *plan HB_UNUSED,
michael@0: 		 hb_font_t *font HB_UNUSED,
michael@0: 		 hb_buffer_t *buffer)
michael@0: {
michael@0:   find_syllables (buffer);
michael@0: }
michael@0: 
michael@0: static int
michael@0: compare_sea_order (const hb_glyph_info_t *pa, const hb_glyph_info_t *pb)
michael@0: {
michael@0:   int a = pa->sea_position();
michael@0:   int b = pb->sea_position();
michael@0: 
michael@0:   return a < b ? -1 : a == b ? 0 : +1;
michael@0: }
michael@0: 
michael@0: 
michael@0: static void
michael@0: initial_reordering_consonant_syllable (const hb_ot_shape_plan_t *plan,
michael@0: 				       hb_face_t *face,
michael@0: 				       hb_buffer_t *buffer,
michael@0: 				       unsigned int start, unsigned int end)
michael@0: {
michael@0:   hb_glyph_info_t *info = buffer->info;
michael@0:   unsigned int base = start;
michael@0: 
michael@0:   /* Reorder! */
michael@0:   unsigned int i = start;
michael@0:   for (; i < base; i++)
michael@0:     info[i].sea_position() = POS_PRE_C;
michael@0:   if (i < end)
michael@0:   {
michael@0:     info[i].sea_position() = POS_BASE_C;
michael@0:     i++;
michael@0:   }
michael@0:   for (; i < end; i++)
michael@0:   {
michael@0:     if (info[i].sea_category() == OT_MR) /* Pre-base reordering */
michael@0:     {
michael@0:       info[i].sea_position() = POS_PRE_C;
michael@0:       continue;
michael@0:     }
michael@0:     if (info[i].sea_category() == OT_VPre) /* Left matra */
michael@0:     {
michael@0:       info[i].sea_position() = POS_PRE_M;
michael@0:       continue;
michael@0:     }
michael@0: 
michael@0:     info[i].sea_position() = POS_AFTER_MAIN;
michael@0:   }
michael@0: 
michael@0:   buffer->merge_clusters (start, end);
michael@0:   /* Sit tight, rock 'n roll! */
michael@0:   hb_bubble_sort (info + start, end - start, compare_sea_order);
michael@0: }
michael@0: 
michael@0: static void
michael@0: initial_reordering_broken_cluster (const hb_ot_shape_plan_t *plan,
michael@0: 				   hb_face_t *face,
michael@0: 				   hb_buffer_t *buffer,
michael@0: 				   unsigned int start, unsigned int end)
michael@0: {
michael@0:   /* We already inserted dotted-circles, so just call the consonant_syllable. */
michael@0:   initial_reordering_consonant_syllable (plan, face, buffer, start, end);
michael@0: }
michael@0: 
michael@0: static void
michael@0: initial_reordering_non_sea_cluster (const hb_ot_shape_plan_t *plan HB_UNUSED,
michael@0: 				    hb_face_t *face HB_UNUSED,
michael@0: 				    hb_buffer_t *buffer HB_UNUSED,
michael@0: 				    unsigned int start HB_UNUSED, unsigned int end HB_UNUSED)
michael@0: {
michael@0:   /* Nothing to do right now.  If we ever switch to using the output
michael@0:    * buffer in the reordering process, we'd need to next_glyph() here. */
michael@0: }
michael@0: 
michael@0: 
michael@0: static void
michael@0: initial_reordering_syllable (const hb_ot_shape_plan_t *plan,
michael@0: 			     hb_face_t *face,
michael@0: 			     hb_buffer_t *buffer,
michael@0: 			     unsigned int start, unsigned int end)
michael@0: {
michael@0:   syllable_type_t syllable_type = (syllable_type_t) (buffer->info[start].syllable() & 0x0F);
michael@0:   switch (syllable_type) {
michael@0:   case consonant_syllable:	initial_reordering_consonant_syllable  (plan, face, buffer, start, end); return;
michael@0:   case broken_cluster:		initial_reordering_broken_cluster      (plan, face, buffer, start, end); return;
michael@0:   case non_sea_cluster:		initial_reordering_non_sea_cluster     (plan, face, buffer, start, end); return;
michael@0:   }
michael@0: }
michael@0: 
michael@0: static inline void
michael@0: insert_dotted_circles (const hb_ot_shape_plan_t *plan HB_UNUSED,
michael@0: 		       hb_font_t *font,
michael@0: 		       hb_buffer_t *buffer)
michael@0: {
michael@0:   /* Note: This loop is extra overhead, but should not be measurable. */
michael@0:   bool has_broken_syllables = false;
michael@0:   unsigned int count = buffer->len;
michael@0:   for (unsigned int i = 0; i < count; i++)
michael@0:     if ((buffer->info[i].syllable() & 0x0F) == broken_cluster) {
michael@0:       has_broken_syllables = true;
michael@0:       break;
michael@0:     }
michael@0:   if (likely (!has_broken_syllables))
michael@0:     return;
michael@0: 
michael@0: 
michael@0:   hb_codepoint_t dottedcircle_glyph;
michael@0:   if (!font->get_glyph (0x25CC, 0, &dottedcircle_glyph))
michael@0:     return;
michael@0: 
michael@0:   hb_glyph_info_t dottedcircle = {0};
michael@0:   dottedcircle.codepoint = 0x25CC;
michael@0:   set_sea_properties (dottedcircle);
michael@0:   dottedcircle.codepoint = dottedcircle_glyph;
michael@0: 
michael@0:   buffer->clear_output ();
michael@0: 
michael@0:   buffer->idx = 0;
michael@0:   unsigned int last_syllable = 0;
michael@0:   while (buffer->idx < buffer->len)
michael@0:   {
michael@0:     unsigned int syllable = buffer->cur().syllable();
michael@0:     syllable_type_t syllable_type = (syllable_type_t) (syllable & 0x0F);
michael@0:     if (unlikely (last_syllable != syllable && syllable_type == broken_cluster))
michael@0:     {
michael@0:       last_syllable = syllable;
michael@0: 
michael@0:       hb_glyph_info_t info = dottedcircle;
michael@0:       info.cluster = buffer->cur().cluster;
michael@0:       info.mask = buffer->cur().mask;
michael@0:       info.syllable() = buffer->cur().syllable();
michael@0: 
michael@0:       buffer->output_info (info);
michael@0:     }
michael@0:     else
michael@0:       buffer->next_glyph ();
michael@0:   }
michael@0: 
michael@0:   buffer->swap_buffers ();
michael@0: }
michael@0: 
michael@0: static void
michael@0: initial_reordering (const hb_ot_shape_plan_t *plan,
michael@0: 		    hb_font_t *font,
michael@0: 		    hb_buffer_t *buffer)
michael@0: {
michael@0:   insert_dotted_circles (plan, font, buffer);
michael@0: 
michael@0:   hb_glyph_info_t *info = buffer->info;
michael@0:   unsigned int count = buffer->len;
michael@0:   if (unlikely (!count)) return;
michael@0:   unsigned int last = 0;
michael@0:   unsigned int last_syllable = info[0].syllable();
michael@0:   for (unsigned int i = 1; i < count; i++)
michael@0:     if (last_syllable != info[i].syllable()) {
michael@0:       initial_reordering_syllable (plan, font->face, buffer, last, i);
michael@0:       last = i;
michael@0:       last_syllable = info[last].syllable();
michael@0:     }
michael@0:   initial_reordering_syllable (plan, font->face, buffer, last, count);
michael@0: }
michael@0: 
michael@0: static void
michael@0: final_reordering (const hb_ot_shape_plan_t *plan,
michael@0: 		  hb_font_t *font HB_UNUSED,
michael@0: 		  hb_buffer_t *buffer)
michael@0: {
michael@0:   hb_glyph_info_t *info = buffer->info;
michael@0:   unsigned int count = buffer->len;
michael@0: 
michael@0:   /* Zero syllables now... */
michael@0:   for (unsigned int i = 0; i < count; i++)
michael@0:     info[i].syllable() = 0;
michael@0: 
michael@0:   HB_BUFFER_DEALLOCATE_VAR (buffer, sea_category);
michael@0:   HB_BUFFER_DEALLOCATE_VAR (buffer, sea_position);
michael@0: }
michael@0: 
michael@0: 
michael@0: const hb_ot_complex_shaper_t _hb_ot_complex_shaper_sea =
michael@0: {
michael@0:   "sea",
michael@0:   collect_features_sea,
michael@0:   override_features_sea,
michael@0:   NULL, /* data_create */
michael@0:   NULL, /* data_destroy */
michael@0:   NULL, /* preprocess_text */
michael@0:   HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS_NO_SHORT_CIRCUIT,
michael@0:   NULL, /* decompose */
michael@0:   NULL, /* compose */
michael@0:   setup_masks_sea,
michael@0:   HB_OT_SHAPE_ZERO_WIDTH_MARKS_NONE,
michael@0:   false, /* fallback_position */
michael@0: };