src/hb-ot-shape-complex-arabic.cc

   1 /*
   2  * Copyright © 2010,2012  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 #include "hb-ot-shape-complex-private.hh"
  28 #include "hb-ot-shape-private.hh"
  29
  30
  31 /* buffer var allocations */
  32 #define arabic_shaping_action() complex_var_u8_0() /* arabic shaping action */
  33
  34
  35 /*
  36  * Bits used in the joining tables
  37  */
  38 enum {
  39   JOINING_TYPE_U                = 0,
  40   JOINING_TYPE_R                = 1,
  41   JOINING_TYPE_D                = 2,
  42   JOINING_TYPE_C                = JOINING_TYPE_D,
  43   JOINING_GROUP_ALAPH           = 3,
  44   JOINING_GROUP_DALATH_RISH     = 4,
  45   NUM_STATE_MACHINE_COLS        = 5,
  46
  47   /* We deliberately don't have a JOINING_TYPE_L since that's unused in Unicode. */
  48
  49   JOINING_TYPE_T = 6,
  50   JOINING_TYPE_X = 7  /* means: use general-category to choose between U or T. */
  51 };
  52
  53 /*
  54  * Joining types:
  55  */
  56
  57 #include "hb-ot-shape-complex-arabic-table.hh"
  58
  59 static unsigned int get_joining_type (hb_codepoint_t u, hb_unicode_general_category_t gen_cat)
  60 {
  61   if (likely (hb_in_range<hb_codepoint_t> (u, JOINING_TABLE_FIRST, JOINING_TABLE_LAST))) {
  62     unsigned int j_type = joining_table[u - JOINING_TABLE_FIRST];
  63     if (likely (j_type != JOINING_TYPE_X))
  64       return j_type;
  65   }
  66
  67   /* Mongolian joining data is not in ArabicJoining.txt yet. */
  68   if (unlikely (hb_in_range<hb_codepoint_t> (u, 0x1800, 0x18AF)))
  69   {
  70     if (unlikely (hb_in_range<hb_codepoint_t> (u, 0x1880, 0x1886)))
  71       return JOINING_TYPE_U;
  72
  73     /* All letters, SIBE SYLLABLE BOUNDARY MARKER, and NIRUGU are D */
  74     if ((FLAG(gen_cat) & (FLAG (HB_UNICODE_GENERAL_CATEGORY_OTHER_LETTER) |
  75                           FLAG (HB_UNICODE_GENERAL_CATEGORY_MODIFIER_LETTER)))
  76         || u == 0x1807 || u == 0x180A)
  77       return JOINING_TYPE_D;
  78   }
  79
  80   /* 'Phags-pa joining data is not in ArabicJoining.txt yet. */
  81   if (unlikely (hb_in_range<hb_codepoint_t> (u, 0xA840, 0xA872)))
  82   {
  83       if (unlikely (u == 0xA872))
  84         /* XXX Looks like this should be TYPE_L, but we don't support that yet! */
  85         return JOINING_TYPE_R;
  86
  87       return JOINING_TYPE_D;
  88   }
  89
  90   if (unlikely (hb_in_range<hb_codepoint_t> (u, 0x200C, 0x200D)))
  91   {
  92     return u == 0x200C ? JOINING_TYPE_U : JOINING_TYPE_C;
  93   }
  94
  95   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))) ?
  96          JOINING_TYPE_T : JOINING_TYPE_U;
  97 }
  98
  99 static const hb_tag_t arabic_features[] =
 100 {
 101   HB_TAG('i','n','i','t'),
 102   HB_TAG('m','e','d','i'),
 103   HB_TAG('f','i','n','a'),
 104   HB_TAG('i','s','o','l'),
 105   /* Syriac */
 106   HB_TAG('m','e','d','2'),
 107   HB_TAG('f','i','n','2'),
 108   HB_TAG('f','i','n','3'),
 109   HB_TAG_NONE
 110 };
 111
 112
 113 /* Same order as the feature array */
 114 enum {
 115   INIT,
 116   MEDI,
 117   FINA,
 118   ISOL,
 119
 120   /* Syriac */
 121   MED2,
 122   FIN2,
 123   FIN3,
 124
 125   NONE,
 126
 127   ARABIC_NUM_FEATURES = NONE
 128 };
 129
 130 static const struct arabic_state_table_entry {
 131         uint8_t prev_action;
 132         uint8_t curr_action;
 133         uint16_t next_state;
 134 } arabic_state_table[][NUM_STATE_MACHINE_COLS] =
 135 {
 136   /*   jt_U,          jt_R,          jt_D,          jg_ALAPH,      jg_DALATH_RISH */
 137
 138   /* State 0: prev was U, not willing to join. */
 139   { {NONE,NONE,0}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,6}, },
 140
 141   /* State 1: prev was R or ISOL/ALAPH, not willing to join. */
 142   { {NONE,NONE,0}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN2,5}, {NONE,ISOL,6}, },
 143
 144   /* State 2: prev was D/ISOL, willing to join. */
 145   { {NONE,NONE,0}, {INIT,FINA,1}, {INIT,FINA,3}, {INIT,FINA,4}, {INIT,FINA,6}, },
 146
 147   /* State 3: prev was D/FINA, willing to join. */
 148   { {NONE,NONE,0}, {MEDI,FINA,1}, {MEDI,FINA,3}, {MEDI,FINA,4}, {MEDI,FINA,6}, },
 149
 150   /* State 4: prev was FINA ALAPH, not willing to join. */
 151   { {NONE,NONE,0}, {MED2,ISOL,1}, {MED2,ISOL,2}, {MED2,FIN2,5}, {MED2,ISOL,6}, },
 152
 153   /* State 5: prev was FIN2/FIN3 ALAPH, not willing to join. */
 154   { {NONE,NONE,0}, {ISOL,ISOL,1}, {ISOL,ISOL,2}, {ISOL,FIN2,5}, {ISOL,ISOL,6}, },
 155
 156   /* State 6: prev was DALATH/RISH, not willing to join. */
 157   { {NONE,NONE,0}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN3,5}, {NONE,ISOL,6}, }
 158 };
 159
 160
 161 static void
 162 arabic_fallback_shape (const hb_ot_shape_plan_t *plan,
 163                        hb_font_t *font,
 164                        hb_buffer_t *buffer);
 165
 166 static void
 167 collect_features_arabic (hb_ot_shape_planner_t *plan)
 168 {
 169   hb_ot_map_builder_t *map = &plan->map;
 170
 171   /* For Language forms (in ArabicOT speak), we do the iso/fina/medi/init together,
 172    * then rlig and calt each in their own stage.  This makes IranNastaliq's ALLAH
 173    * ligature work correctly. It's unfortunate though...
 174    *
 175    * This also makes Arial Bold in Windows7 work.  See:
 176    * https://bugzilla.mozilla.org/show_bug.cgi?id=644184
 177    *
 178    * TODO: Add test cases for these two.
 179    */
 180
 181   map->add_bool_feature (HB_TAG('c','c','m','p'));
 182   map->add_bool_feature (HB_TAG('l','o','c','l'));
 183
 184   map->add_gsub_pause (NULL);
 185
 186   for (unsigned int i = 0; i < ARABIC_NUM_FEATURES; i++)
 187     map->add_bool_feature (arabic_features[i], false, i < 4); /* The first four features have fallback. */
 188
 189   map->add_gsub_pause (NULL);
 190
 191   map->add_bool_feature (HB_TAG('r','l','i','g'), true, true);
 192   map->add_gsub_pause (arabic_fallback_shape);
 193
 194   map->add_bool_feature (HB_TAG('c','a','l','t'));
 195   map->add_gsub_pause (NULL);
 196
 197   map->add_bool_feature (HB_TAG('c','s','w','h'));
 198   map->add_bool_feature (HB_TAG('d','l','i','g'));
 199   map->add_bool_feature (HB_TAG('m','s','e','t'));
 200 }
 201
 202 #include "hb-ot-shape-complex-arabic-fallback.hh"
 203
 204 struct arabic_shape_plan_t
 205 {
 206   ASSERT_POD ();
 207
 208   /* The "+ 1" in the next array is to accommodate for the "NONE" command,
 209    * which is not an OpenType feature, but this simplifies the code by not
 210    * having to do a "if (... < NONE) ..." and just rely on the fact that
 211    * mask_array[NONE] == 0. */
 212   hb_mask_t mask_array[ARABIC_NUM_FEATURES + 1];
 213
 214   bool do_fallback;
 215   arabic_fallback_plan_t *fallback_plan;
 216 };
 217
 218 static void *
 219 data_create_arabic (const hb_ot_shape_plan_t *plan)
 220 {
 221   arabic_shape_plan_t *arabic_plan = (arabic_shape_plan_t *) calloc (1, sizeof (arabic_shape_plan_t));
 222   if (unlikely (!arabic_plan))
 223     return NULL;
 224
 225   arabic_plan->do_fallback = plan->props.script == HB_SCRIPT_ARABIC;
 226   for (unsigned int i = 0; i < ARABIC_NUM_FEATURES; i++) {
 227     arabic_plan->mask_array[i] = plan->map.get_1_mask (arabic_features[i]);
 228     if (i < 4)
 229       arabic_plan->do_fallback = arabic_plan->do_fallback && plan->map.needs_fallback (arabic_features[i]);
 230   }
 231
 232   return arabic_plan;
 233 }
 234
 235 static void
 236 data_destroy_arabic (void *data)
 237 {
 238   arabic_shape_plan_t *arabic_plan = (arabic_shape_plan_t *) data;
 239
 240   arabic_fallback_plan_destroy (arabic_plan->fallback_plan);
 241
 242   free (data);
 243 }
 244
 245 static void
 246 arabic_joining (hb_buffer_t *buffer)
 247 {
 248   unsigned int count = buffer->len;
 249   unsigned int prev = (unsigned int) -1, state = 0;
 250
 251   HB_BUFFER_ALLOCATE_VAR (buffer, arabic_shaping_action);
 252
 253   /* Check pre-context */
 254   if (!(buffer->flags & HB_BUFFER_FLAG_BOT))
 255     for (unsigned int i = 0; i < buffer->context_len[0]; i++)
 256     {
 257       unsigned int this_type = get_joining_type (buffer->context[0][i], buffer->unicode->general_category (buffer->context[0][i]));
 258
 259       if (unlikely (this_type == JOINING_TYPE_T))
 260         continue;
 261
 262       const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
 263       state = entry->next_state;
 264       break;
 265     }
 266
 267   for (unsigned int i = 0; i < count; i++)
 268   {
 269     unsigned int this_type = get_joining_type (buffer->info[i].codepoint, _hb_glyph_info_get_general_category (&buffer->info[i]));
 270
 271     if (unlikely (this_type == JOINING_TYPE_T)) {
 272       buffer->info[i].arabic_shaping_action() = NONE;
 273       continue;
 274     }
 275
 276     const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
 277
 278     if (entry->prev_action != NONE && prev != (unsigned int) -1)
 279       buffer->info[prev].arabic_shaping_action() = entry->prev_action;
 280
 281     buffer->info[i].arabic_shaping_action() = entry->curr_action;
 282
 283     prev = i;
 284     state = entry->next_state;
 285   }
 286
 287   if (!(buffer->flags & HB_BUFFER_FLAG_EOT))
 288     for (unsigned int i = 0; i < buffer->context_len[1]; i++)
 289     {
 290       unsigned int this_type = get_joining_type (buffer->context[1][i], buffer->unicode->general_category (buffer->context[1][i]));
 291
 292       if (unlikely (this_type == JOINING_TYPE_T))
 293         continue;
 294
 295       const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
 296       if (entry->prev_action != NONE && prev != (unsigned int) -1)
 297         buffer->info[prev].arabic_shaping_action() = entry->prev_action;
 298       break;
 299     }
 300
 301
 302   HB_BUFFER_DEALLOCATE_VAR (buffer, arabic_shaping_action);
 303 }
 304
 305 static void
 306 setup_masks_arabic (const hb_ot_shape_plan_t *plan,
 307                     hb_buffer_t              *buffer,
 308                     hb_font_t                *font HB_UNUSED)
 309 {
 310   const arabic_shape_plan_t *arabic_plan = (const arabic_shape_plan_t *) plan->data;
 311
 312   arabic_joining (buffer);
 313   unsigned int count = buffer->len;
 314   for (unsigned int i = 0; i < count; i++)
 315     buffer->info[i].mask |= arabic_plan->mask_array[buffer->info[i].arabic_shaping_action()];
 316 }
 317
 318
 319 static void
 320 arabic_fallback_shape (const hb_ot_shape_plan_t *plan,
 321                        hb_font_t *font,
 322                        hb_buffer_t *buffer)
 323 {
 324   const arabic_shape_plan_t *arabic_plan = (const arabic_shape_plan_t *) plan->data;
 325
 326   if (!arabic_plan->do_fallback)
 327     return;
 328
 329 retry:
 330   arabic_fallback_plan_t *fallback_plan = (arabic_fallback_plan_t *) hb_atomic_ptr_get (&arabic_plan->fallback_plan);
 331   if (unlikely (!fallback_plan))
 332   {
 333     /* This sucks.  We need a font to build the fallback plan... */
 334     fallback_plan = arabic_fallback_plan_create (plan, font);
 335     if (unlikely (!hb_atomic_ptr_cmpexch (&(const_cast<arabic_shape_plan_t *> (arabic_plan))->fallback_plan, NULL, fallback_plan))) {
 336       arabic_fallback_plan_destroy (fallback_plan);
 337       goto retry;
 338     }
 339   }
 340
 341   arabic_fallback_plan_shape (fallback_plan, font, buffer);
 342 }
 343
 344
 345 const hb_ot_complex_shaper_t _hb_ot_complex_shaper_arabic =
 346 {
 347   "arabic",
 348   collect_features_arabic,
 349   NULL, /* override_features */
 350   data_create_arabic,
 351   data_destroy_arabic,
 352   NULL, /* preprocess_text_arabic */
 353   NULL, /* normalization_preference */
 354   NULL, /* decompose */
 355   NULL, /* compose */
 356   setup_masks_arabic,
 357   true, /* zero_width_attached_marks */
 358   true, /* fallback_position */
 359 };