michael@0: /*
michael@0:  * Copyright © 2010,2012  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: #include "hb-ot-shape-complex-private.hh"
michael@0: #include "hb-ot-shape-private.hh"
michael@0: 
michael@0: 
michael@0: /* buffer var allocations */
michael@0: #define arabic_shaping_action() complex_var_u8_0() /* arabic shaping action */
michael@0: 
michael@0: 
michael@0: /*
michael@0:  * Bits used in the joining tables
michael@0:  */
michael@0: enum {
michael@0:   JOINING_TYPE_U		= 0,
michael@0:   JOINING_TYPE_L		= 1,
michael@0:   JOINING_TYPE_R		= 2,
michael@0:   JOINING_TYPE_D		= 3,
michael@0:   JOINING_TYPE_C		= JOINING_TYPE_D,
michael@0:   JOINING_GROUP_ALAPH		= 4,
michael@0:   JOINING_GROUP_DALATH_RISH	= 5,
michael@0:   NUM_STATE_MACHINE_COLS	= 6,
michael@0: 
michael@0:   JOINING_TYPE_T = 7,
michael@0:   JOINING_TYPE_X = 8  /* means: use general-category to choose between U or T. */
michael@0: };
michael@0: 
michael@0: /*
michael@0:  * Joining types:
michael@0:  */
michael@0: 
michael@0: #include "hb-ot-shape-complex-arabic-table.hh"
michael@0: 
michael@0: static unsigned int get_joining_type (hb_codepoint_t u, hb_unicode_general_category_t gen_cat)
michael@0: {
michael@0:   if (likely (hb_in_range<hb_codepoint_t> (u, JOINING_TABLE_FIRST, JOINING_TABLE_LAST))) {
michael@0:     unsigned int j_type = joining_table[u - JOINING_TABLE_FIRST];
michael@0:     if (likely (j_type != JOINING_TYPE_X))
michael@0:       return j_type;
michael@0:   }
michael@0: 
michael@0:   /* Mongolian joining data is not in ArabicJoining.txt yet. */
michael@0:   if (unlikely (hb_in_range<hb_codepoint_t> (u, 0x1800, 0x18AF)))
michael@0:   {
michael@0:     if (unlikely (hb_in_range<hb_codepoint_t> (u, 0x1880, 0x1886)))
michael@0:       return JOINING_TYPE_U;
michael@0: 
michael@0:     /* All letters, SIBE SYLLABLE BOUNDARY MARKER, and NIRUGU are D */
michael@0:     if ((FLAG(gen_cat) & (FLAG (HB_UNICODE_GENERAL_CATEGORY_OTHER_LETTER) |
michael@0: 			  FLAG (HB_UNICODE_GENERAL_CATEGORY_MODIFIER_LETTER)))
michael@0: 	|| u == 0x1807 || u == 0x180A)
michael@0:       return JOINING_TYPE_D;
michael@0:   }
michael@0: 
michael@0:   /* 'Phags-pa joining data is not in ArabicJoining.txt yet. */
michael@0:   if (unlikely (hb_in_range<hb_codepoint_t> (u, 0xA840, 0xA872)))
michael@0:   {
michael@0:       if (unlikely (u == 0xA872))
michael@0: 	return JOINING_TYPE_L;
michael@0: 
michael@0:       return JOINING_TYPE_D;
michael@0:   }
michael@0: 
michael@0:   if (unlikely (hb_in_range<hb_codepoint_t> (u, 0x200C, 0x200D)))
michael@0:   {
michael@0:     return u == 0x200C ? JOINING_TYPE_U : JOINING_TYPE_C;
michael@0:   }
michael@0: 
michael@0:   return (FLAG(gen_cat) & (FLAG(HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK) | FLAG(HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK) | FLAG(HB_UNICODE_GENERAL_CATEGORY_FORMAT))) ?
michael@0: 	 JOINING_TYPE_T : JOINING_TYPE_U;
michael@0: }
michael@0: 
michael@0: static const hb_tag_t arabic_features[] =
michael@0: {
michael@0:   HB_TAG('i','n','i','t'),
michael@0:   HB_TAG('m','e','d','i'),
michael@0:   HB_TAG('f','i','n','a'),
michael@0:   HB_TAG('i','s','o','l'),
michael@0:   /* Syriac */
michael@0:   HB_TAG('m','e','d','2'),
michael@0:   HB_TAG('f','i','n','2'),
michael@0:   HB_TAG('f','i','n','3'),
michael@0:   HB_TAG_NONE
michael@0: };
michael@0: 
michael@0: 
michael@0: /* Same order as the feature array */
michael@0: enum {
michael@0:   INIT,
michael@0:   MEDI,
michael@0:   FINA,
michael@0:   ISOL,
michael@0: 
michael@0:   /* Syriac */
michael@0:   MED2,
michael@0:   FIN2,
michael@0:   FIN3,
michael@0: 
michael@0:   NONE,
michael@0: 
michael@0:   ARABIC_NUM_FEATURES = NONE
michael@0: };
michael@0: 
michael@0: static const struct arabic_state_table_entry {
michael@0: 	uint8_t prev_action;
michael@0: 	uint8_t curr_action;
michael@0: 	uint16_t next_state;
michael@0: } arabic_state_table[][NUM_STATE_MACHINE_COLS] =
michael@0: {
michael@0:   /*   jt_U,          jt_L,          jt_R,          jt_D,          jg_ALAPH,      jg_DALATH_RISH */
michael@0: 
michael@0:   /* State 0: prev was U, not willing to join. */
michael@0:   { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,6}, },
michael@0: 
michael@0:   /* State 1: prev was R or ISOL/ALAPH, not willing to join. */
michael@0:   { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN2,5}, {NONE,ISOL,6}, },
michael@0: 
michael@0:   /* State 2: prev was D/L in ISOL form, willing to join. */
michael@0:   { {NONE,NONE,0}, {NONE,ISOL,2}, {INIT,FINA,1}, {INIT,FINA,3}, {INIT,FINA,4}, {INIT,FINA,6}, },
michael@0: 
michael@0:   /* State 3: prev was D in FINA form, willing to join. */
michael@0:   { {NONE,NONE,0}, {NONE,ISOL,2}, {MEDI,FINA,1}, {MEDI,FINA,3}, {MEDI,FINA,4}, {MEDI,FINA,6}, },
michael@0: 
michael@0:   /* State 4: prev was FINA ALAPH, not willing to join. */
michael@0:   { {NONE,NONE,0}, {NONE,ISOL,2}, {MED2,ISOL,1}, {MED2,ISOL,2}, {MED2,FIN2,5}, {MED2,ISOL,6}, },
michael@0: 
michael@0:   /* State 5: prev was FIN2/FIN3 ALAPH, not willing to join. */
michael@0:   { {NONE,NONE,0}, {NONE,ISOL,2}, {ISOL,ISOL,1}, {ISOL,ISOL,2}, {ISOL,FIN2,5}, {ISOL,ISOL,6}, },
michael@0: 
michael@0:   /* State 6: prev was DALATH/RISH, not willing to join. */
michael@0:   { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN3,5}, {NONE,ISOL,6}, }
michael@0: };
michael@0: 
michael@0: 
michael@0: static void
michael@0: nuke_joiners (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: arabic_fallback_shape (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_arabic (hb_ot_shape_planner_t *plan)
michael@0: {
michael@0:   hb_ot_map_builder_t *map = &plan->map;
michael@0: 
michael@0:   /* For Language forms (in ArabicOT speak), we do the iso/fina/medi/init together,
michael@0:    * then rlig and calt each in their own stage.  This makes IranNastaliq's ALLAH
michael@0:    * ligature work correctly. It's unfortunate though...
michael@0:    *
michael@0:    * This also makes Arial Bold in Windows7 work.  See:
michael@0:    * https://bugzilla.mozilla.org/show_bug.cgi?id=644184
michael@0:    *
michael@0:    * TODO: Add test cases for these two.
michael@0:    */
michael@0: 
michael@0:   map->add_gsub_pause (nuke_joiners);
michael@0: 
michael@0:   map->add_global_bool_feature (HB_TAG('c','c','m','p'));
michael@0:   map->add_global_bool_feature (HB_TAG('l','o','c','l'));
michael@0: 
michael@0:   map->add_gsub_pause (NULL);
michael@0: 
michael@0:   for (unsigned int i = 0; i < ARABIC_NUM_FEATURES; i++)
michael@0:     map->add_feature (arabic_features[i], 1, i < 4 ? F_HAS_FALLBACK : F_NONE); /* The first four features have fallback. */
michael@0: 
michael@0:   map->add_gsub_pause (NULL);
michael@0: 
michael@0:   map->add_feature (HB_TAG('r','l','i','g'), 1, F_GLOBAL|F_HAS_FALLBACK);
michael@0:   map->add_gsub_pause (arabic_fallback_shape);
michael@0: 
michael@0:   map->add_global_bool_feature (HB_TAG('c','a','l','t'));
michael@0:   map->add_gsub_pause (NULL);
michael@0: 
michael@0:   map->add_global_bool_feature (HB_TAG('m','s','e','t'));
michael@0: }
michael@0: 
michael@0: #include "hb-ot-shape-complex-arabic-fallback.hh"
michael@0: 
michael@0: struct arabic_shape_plan_t
michael@0: {
michael@0:   ASSERT_POD ();
michael@0: 
michael@0:   /* The "+ 1" in the next array is to accommodate for the "NONE" command,
michael@0:    * which is not an OpenType feature, but this simplifies the code by not
michael@0:    * having to do a "if (... < NONE) ..." and just rely on the fact that
michael@0:    * mask_array[NONE] == 0. */
michael@0:   hb_mask_t mask_array[ARABIC_NUM_FEATURES + 1];
michael@0: 
michael@0:   bool do_fallback;
michael@0:   arabic_fallback_plan_t *fallback_plan;
michael@0: };
michael@0: 
michael@0: static void *
michael@0: data_create_arabic (const hb_ot_shape_plan_t *plan)
michael@0: {
michael@0:   arabic_shape_plan_t *arabic_plan = (arabic_shape_plan_t *) calloc (1, sizeof (arabic_shape_plan_t));
michael@0:   if (unlikely (!arabic_plan))
michael@0:     return NULL;
michael@0: 
michael@0:   arabic_plan->do_fallback = plan->props.script == HB_SCRIPT_ARABIC;
michael@0:   for (unsigned int i = 0; i < ARABIC_NUM_FEATURES; i++) {
michael@0:     arabic_plan->mask_array[i] = plan->map.get_1_mask (arabic_features[i]);
michael@0:     if (i < 4)
michael@0:       arabic_plan->do_fallback = arabic_plan->do_fallback && plan->map.needs_fallback (arabic_features[i]);
michael@0:   }
michael@0: 
michael@0:   return arabic_plan;
michael@0: }
michael@0: 
michael@0: static void
michael@0: data_destroy_arabic (void *data)
michael@0: {
michael@0:   arabic_shape_plan_t *arabic_plan = (arabic_shape_plan_t *) data;
michael@0: 
michael@0:   arabic_fallback_plan_destroy (arabic_plan->fallback_plan);
michael@0: 
michael@0:   free (data);
michael@0: }
michael@0: 
michael@0: static void
michael@0: arabic_joining (hb_buffer_t *buffer)
michael@0: {
michael@0:   unsigned int count = buffer->len;
michael@0:   unsigned int prev = (unsigned int) -1, state = 0;
michael@0: 
michael@0:   HB_BUFFER_ALLOCATE_VAR (buffer, arabic_shaping_action);
michael@0: 
michael@0:   /* Check pre-context */
michael@0:   if (!(buffer->flags & HB_BUFFER_FLAG_BOT))
michael@0:     for (unsigned int i = 0; i < buffer->context_len[0]; i++)
michael@0:     {
michael@0:       unsigned int this_type = get_joining_type (buffer->context[0][i], buffer->unicode->general_category (buffer->context[0][i]));
michael@0: 
michael@0:       if (unlikely (this_type == JOINING_TYPE_T))
michael@0: 	continue;
michael@0: 
michael@0:       const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
michael@0:       state = entry->next_state;
michael@0:       break;
michael@0:     }
michael@0: 
michael@0:   for (unsigned int i = 0; i < count; i++)
michael@0:   {
michael@0:     unsigned int this_type = get_joining_type (buffer->info[i].codepoint, _hb_glyph_info_get_general_category (&buffer->info[i]));
michael@0: 
michael@0:     if (unlikely (this_type == JOINING_TYPE_T)) {
michael@0:       buffer->info[i].arabic_shaping_action() = NONE;
michael@0:       continue;
michael@0:     }
michael@0: 
michael@0:     const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
michael@0: 
michael@0:     if (entry->prev_action != NONE && prev != (unsigned int) -1)
michael@0:       for (; prev < i; prev++)
michael@0: 	buffer->info[prev].arabic_shaping_action() = entry->prev_action;
michael@0: 
michael@0:     buffer->info[i].arabic_shaping_action() = entry->curr_action;
michael@0: 
michael@0:     prev = i;
michael@0:     state = entry->next_state;
michael@0:   }
michael@0: 
michael@0:   if (!(buffer->flags & HB_BUFFER_FLAG_EOT))
michael@0:     for (unsigned int i = 0; i < buffer->context_len[1]; i++)
michael@0:     {
michael@0:       unsigned int this_type = get_joining_type (buffer->context[1][i], buffer->unicode->general_category (buffer->context[1][i]));
michael@0: 
michael@0:       if (unlikely (this_type == JOINING_TYPE_T))
michael@0: 	continue;
michael@0: 
michael@0:       const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
michael@0:       if (entry->prev_action != NONE && prev != (unsigned int) -1)
michael@0: 	buffer->info[prev].arabic_shaping_action() = entry->prev_action;
michael@0:       break;
michael@0:     }
michael@0: 
michael@0: 
michael@0:   HB_BUFFER_DEALLOCATE_VAR (buffer, arabic_shaping_action);
michael@0: }
michael@0: 
michael@0: static void
michael@0: setup_masks_arabic (const hb_ot_shape_plan_t *plan,
michael@0: 		    hb_buffer_t              *buffer,
michael@0: 		    hb_font_t                *font HB_UNUSED)
michael@0: {
michael@0:   const arabic_shape_plan_t *arabic_plan = (const arabic_shape_plan_t *) plan->data;
michael@0: 
michael@0:   arabic_joining (buffer);
michael@0:   unsigned int count = buffer->len;
michael@0:   for (unsigned int i = 0; i < count; i++)
michael@0:     buffer->info[i].mask |= arabic_plan->mask_array[buffer->info[i].arabic_shaping_action()];
michael@0: }
michael@0: 
michael@0: 
michael@0: static void
michael@0: nuke_joiners (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:   unsigned int count = buffer->len;
michael@0:   for (unsigned int i = 0; i < count; i++)
michael@0:     if (_hb_glyph_info_is_zwj (&buffer->info[i]))
michael@0:       _hb_glyph_info_flip_joiners (&buffer->info[i]);
michael@0: }
michael@0: 
michael@0: static void
michael@0: arabic_fallback_shape (const hb_ot_shape_plan_t *plan,
michael@0: 		       hb_font_t *font,
michael@0: 		       hb_buffer_t *buffer)
michael@0: {
michael@0:   const arabic_shape_plan_t *arabic_plan = (const arabic_shape_plan_t *) plan->data;
michael@0: 
michael@0:   if (!arabic_plan->do_fallback)
michael@0:     return;
michael@0: 
michael@0: retry:
michael@0:   arabic_fallback_plan_t *fallback_plan = (arabic_fallback_plan_t *) hb_atomic_ptr_get (&arabic_plan->fallback_plan);
michael@0:   if (unlikely (!fallback_plan))
michael@0:   {
michael@0:     /* This sucks.  We need a font to build the fallback plan... */
michael@0:     fallback_plan = arabic_fallback_plan_create (plan, font);
michael@0:     if (unlikely (!hb_atomic_ptr_cmpexch (&(const_cast<arabic_shape_plan_t *> (arabic_plan))->fallback_plan, NULL, fallback_plan))) {
michael@0:       arabic_fallback_plan_destroy (fallback_plan);
michael@0:       goto retry;
michael@0:     }
michael@0:   }
michael@0: 
michael@0:   arabic_fallback_plan_shape (fallback_plan, font, buffer);
michael@0: }
michael@0: 
michael@0: 
michael@0: const hb_ot_complex_shaper_t _hb_ot_complex_shaper_arabic =
michael@0: {
michael@0:   "arabic",
michael@0:   collect_features_arabic,
michael@0:   NULL, /* override_features */
michael@0:   data_create_arabic,
michael@0:   data_destroy_arabic,
michael@0:   NULL, /* preprocess_text_arabic */
michael@0:   HB_OT_SHAPE_NORMALIZATION_MODE_DEFAULT,
michael@0:   NULL, /* decompose */
michael@0:   NULL, /* compose */
michael@0:   setup_masks_arabic,
michael@0:   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_LATE,
michael@0:   true, /* fallback_position */
michael@0: };