third_party/harfbuzz-ng/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_L                = 1,
  41   JOINING_TYPE_R                = 2,
  42   JOINING_TYPE_D                = 3,
  43   JOINING_TYPE_C                = JOINING_TYPE_D,
  44   JOINING_GROUP_ALAPH           = 4,
  45   JOINING_GROUP_DALATH_RISH     = 5,
  46   NUM_STATE_MACHINE_COLS        = 6,
  47
  48   JOINING_TYPE_T = 7,
  49   JOINING_TYPE_X = 8  /* means: use general-category to choose between U or T. */
  50 };
  51
  52 /*
  53  * Joining types:
  54  */
  55
  56 #include "hb-ot-shape-complex-arabic-table.hh"
  57
  58 static unsigned int get_joining_type (hb_codepoint_t u, hb_unicode_general_category_t gen_cat)
  59 {
  60   if (likely (hb_in_range<hb_codepoint_t> (u, JOINING_TABLE_FIRST, JOINING_TABLE_LAST))) {
  61     unsigned int j_type = joining_table[u - JOINING_TABLE_FIRST];
  62     if (likely (j_type != JOINING_TYPE_X))
  63       return j_type;
  64   }
  65
  66   /* Mongolian joining data is not in ArabicJoining.txt yet. */
  67   if (unlikely (hb_in_range<hb_codepoint_t> (u, 0x1800, 0x18AF)))
  68   {
  69     if (unlikely (hb_in_range<hb_codepoint_t> (u, 0x1880, 0x1886)))
  70       return JOINING_TYPE_U;
  71
  72     /* All letters, SIBE SYLLABLE BOUNDARY MARKER, and NIRUGU are D */
  73     if ((FLAG(gen_cat) & (FLAG (HB_UNICODE_GENERAL_CATEGORY_OTHER_LETTER) |
  74                           FLAG (HB_UNICODE_GENERAL_CATEGORY_MODIFIER_LETTER)))
  75         || u == 0x1807 || u == 0x180A)
  76       return JOINING_TYPE_D;
  77   }
  78
  79   /* 'Phags-pa joining data is not in ArabicJoining.txt yet. */
  80   if (unlikely (hb_in_range<hb_codepoint_t> (u, 0xA840, 0xA872)))
  81   {
  82       if (unlikely (u == 0xA872))
  83         return JOINING_TYPE_L;
  84
  85       return JOINING_TYPE_D;
  86   }
  87
  88   if (unlikely (hb_in_range<hb_codepoint_t> (u, 0x200C, 0x200D)))
  89   {
  90     return u == 0x200C ? JOINING_TYPE_U : JOINING_TYPE_C;
  91   }
  92
  93   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))) ?
  94          JOINING_TYPE_T : JOINING_TYPE_U;
  95 }
  96
  97 static const hb_tag_t arabic_features[] =
  98 {
  99   HB_TAG('i','n','i','t'),
 100   HB_TAG('m','e','d','i'),
 101   HB_TAG('f','i','n','a'),
 102   HB_TAG('i','s','o','l'),
 103   /* Syriac */
 104   HB_TAG('m','e','d','2'),
 105   HB_TAG('f','i','n','2'),
 106   HB_TAG('f','i','n','3'),
 107   HB_TAG_NONE
 108 };
 109
 110
 111 /* Same order as the feature array */
 112 enum {
 113   INIT,
 114   MEDI,
 115   FINA,
 116   ISOL,
 117
 118   /* Syriac */
 119   MED2,
 120   FIN2,
 121   FIN3,
 122
 123   NONE,
 124
 125   ARABIC_NUM_FEATURES = NONE
 126 };
 127
 128 static const struct arabic_state_table_entry {
 129         uint8_t prev_action;
 130         uint8_t curr_action;
 131         uint16_t next_state;
 132 } arabic_state_table[][NUM_STATE_MACHINE_COLS] =
 133 {
 134   /*   jt_U,          jt_L,          jt_R,          jt_D,          jg_ALAPH,      jg_DALATH_RISH */
 135
 136   /* State 0: prev was U, not willing to join. */
 137   { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,6}, },
 138
 139   /* State 1: prev was R or ISOL/ALAPH, not willing to join. */
 140   { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN2,5}, {NONE,ISOL,6}, },
 141
 142   /* State 2: prev was D/L in ISOL form, willing to join. */
 143   { {NONE,NONE,0}, {NONE,ISOL,2}, {INIT,FINA,1}, {INIT,FINA,3}, {INIT,FINA,4}, {INIT,FINA,6}, },
 144
 145   /* State 3: prev was D in FINA form, willing to join. */
 146   { {NONE,NONE,0}, {NONE,ISOL,2}, {MEDI,FINA,1}, {MEDI,FINA,3}, {MEDI,FINA,4}, {MEDI,FINA,6}, },
 147
 148   /* State 4: prev was FINA ALAPH, not willing to join. */
 149   { {NONE,NONE,0}, {NONE,ISOL,2}, {MED2,ISOL,1}, {MED2,ISOL,2}, {MED2,FIN2,5}, {MED2,ISOL,6}, },
 150
 151   /* State 5: prev was FIN2/FIN3 ALAPH, not willing to join. */
 152   { {NONE,NONE,0}, {NONE,ISOL,2}, {ISOL,ISOL,1}, {ISOL,ISOL,2}, {ISOL,FIN2,5}, {ISOL,ISOL,6}, },
 153
 154   /* State 6: prev was DALATH/RISH, not willing to join. */
 155   { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN3,5}, {NONE,ISOL,6}, }
 156 };
 157
 158
 159 static void
 160 nuke_joiners (const hb_ot_shape_plan_t *plan,
 161               hb_font_t *font,
 162               hb_buffer_t *buffer);
 163
 164 static void
 165 arabic_fallback_shape (const hb_ot_shape_plan_t *plan,
 166                        hb_font_t *font,
 167                        hb_buffer_t *buffer);
 168
 169 static void
 170 collect_features_arabic (hb_ot_shape_planner_t *plan)
 171 {
 172   hb_ot_map_builder_t *map = &plan->map;
 173
 174   /* For Language forms (in ArabicOT speak), we do the iso/fina/medi/init together,
 175    * then rlig and calt each in their own stage.  This makes IranNastaliq's ALLAH
 176    * ligature work correctly. It's unfortunate though...
 177    *
 178    * This also makes Arial Bold in Windows7 work.  See:
 179    * https://bugzilla.mozilla.org/show_bug.cgi?id=644184
 180    *
 181    * TODO: Add test cases for these two.
 182    */
 183
 184   map->add_gsub_pause (nuke_joiners);
 185
 186   map->add_global_bool_feature (HB_TAG('c','c','m','p'));
 187   map->add_global_bool_feature (HB_TAG('l','o','c','l'));
 188
 189   map->add_gsub_pause (NULL);
 190
 191   for (unsigned int i = 0; i < ARABIC_NUM_FEATURES; i++)
 192     map->add_feature (arabic_features[i], 1, i < 4 ? F_HAS_FALLBACK : F_NONE); /* The first four features have fallback. */
 193
 194   map->add_gsub_pause (NULL);
 195
 196   map->add_feature (HB_TAG('r','l','i','g'), 1, F_GLOBAL|F_HAS_FALLBACK);
 197   map->add_gsub_pause (arabic_fallback_shape);
 198
 199   map->add_global_bool_feature (HB_TAG('c','a','l','t'));
 200   map->add_gsub_pause (NULL);
 201
 202   map->add_global_bool_feature (HB_TAG('c','s','w','h'));
 203   map->add_global_bool_feature (HB_TAG('m','s','e','t'));
 204 }
 205
 206 #include "hb-ot-shape-complex-arabic-fallback.hh"
 207
 208 struct arabic_shape_plan_t
 209 {
 210   ASSERT_POD ();
 211
 212   /* The "+ 1" in the next array is to accommodate for the "NONE" command,
 213    * which is not an OpenType feature, but this simplifies the code by not
 214    * having to do a "if (... < NONE) ..." and just rely on the fact that
 215    * mask_array[NONE] == 0. */
 216   hb_mask_t mask_array[ARABIC_NUM_FEATURES + 1];
 217
 218   bool do_fallback;
 219   arabic_fallback_plan_t *fallback_plan;
 220 };
 221
 222 static void *
 223 data_create_arabic (const hb_ot_shape_plan_t *plan)
 224 {
 225   arabic_shape_plan_t *arabic_plan = (arabic_shape_plan_t *) calloc (1, sizeof (arabic_shape_plan_t));
 226   if (unlikely (!arabic_plan))
 227     return NULL;
 228
 229   arabic_plan->do_fallback = plan->props.script == HB_SCRIPT_ARABIC;
 230   for (unsigned int i = 0; i < ARABIC_NUM_FEATURES; i++) {
 231     arabic_plan->mask_array[i] = plan->map.get_1_mask (arabic_features[i]);
 232     if (i < 4)
 233       arabic_plan->do_fallback = arabic_plan->do_fallback && plan->map.needs_fallback (arabic_features[i]);
 234   }
 235
 236   return arabic_plan;
 237 }
 238
 239 static void
 240 data_destroy_arabic (void *data)
 241 {
 242   arabic_shape_plan_t *arabic_plan = (arabic_shape_plan_t *) data;
 243
 244   arabic_fallback_plan_destroy (arabic_plan->fallback_plan);
 245
 246   free (data);
 247 }
 248
 249 static void
 250 arabic_joining (hb_buffer_t *buffer)
 251 {
 252   unsigned int count = buffer->len;
 253   unsigned int prev = (unsigned int) -1, state = 0;
 254
 255   HB_BUFFER_ALLOCATE_VAR (buffer, arabic_shaping_action);
 256
 257   /* Check pre-context */
 258   if (!(buffer->flags & HB_BUFFER_FLAG_BOT))
 259     for (unsigned int i = 0; i < buffer->context_len[0]; i++)
 260     {
 261       unsigned int this_type = get_joining_type (buffer->context[0][i], buffer->unicode->general_category (buffer->context[0][i]));
 262
 263       if (unlikely (this_type == JOINING_TYPE_T))
 264         continue;
 265
 266       const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
 267       state = entry->next_state;
 268       break;
 269     }
 270
 271   for (unsigned int i = 0; i < count; i++)
 272   {
 273     unsigned int this_type = get_joining_type (buffer->info[i].codepoint, _hb_glyph_info_get_general_category (&buffer->info[i]));
 274
 275     if (unlikely (this_type == JOINING_TYPE_T)) {
 276       buffer->info[i].arabic_shaping_action() = NONE;
 277       continue;
 278     }
 279
 280     const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
 281
 282     if (entry->prev_action != NONE && prev != (unsigned int) -1)
 283       for (; prev < i; prev++)
 284         buffer->info[prev].arabic_shaping_action() = entry->prev_action;
 285
 286     buffer->info[i].arabic_shaping_action() = entry->curr_action;
 287
 288     prev = i;
 289     state = entry->next_state;
 290   }
 291
 292   if (!(buffer->flags & HB_BUFFER_FLAG_EOT))
 293     for (unsigned int i = 0; i < buffer->context_len[1]; i++)
 294     {
 295       unsigned int this_type = get_joining_type (buffer->context[1][i], buffer->unicode->general_category (buffer->context[1][i]));
 296
 297       if (unlikely (this_type == JOINING_TYPE_T))
 298         continue;
 299
 300       const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
 301       if (entry->prev_action != NONE && prev != (unsigned int) -1)
 302         buffer->info[prev].arabic_shaping_action() = entry->prev_action;
 303       break;
 304     }
 305
 306
 307   HB_BUFFER_DEALLOCATE_VAR (buffer, arabic_shaping_action);
 308 }
 309
 310 static void
 311 setup_masks_arabic (const hb_ot_shape_plan_t *plan,
 312                     hb_buffer_t              *buffer,
 313                     hb_font_t                *font HB_UNUSED)
 314 {
 315   const arabic_shape_plan_t *arabic_plan = (const arabic_shape_plan_t *) plan->data;
 316
 317   arabic_joining (buffer);
 318   unsigned int count = buffer->len;
 319   for (unsigned int i = 0; i < count; i++)
 320     buffer->info[i].mask |= arabic_plan->mask_array[buffer->info[i].arabic_shaping_action()];
 321 }
 322
 323
 324 static void
 325 nuke_joiners (const hb_ot_shape_plan_t *plan HB_UNUSED,
 326               hb_font_t *font HB_UNUSED,
 327               hb_buffer_t *buffer)
 328 {
 329   unsigned int count = buffer->len;
 330   for (unsigned int i = 0; i < count; i++)
 331     if (_hb_glyph_info_is_zwj (&buffer->info[i]))
 332       _hb_glyph_info_flip_joiners (&buffer->info[i]);
 333 }
 334
 335 static void
 336 arabic_fallback_shape (const hb_ot_shape_plan_t *plan,
 337                        hb_font_t *font,
 338                        hb_buffer_t *buffer)
 339 {
 340   const arabic_shape_plan_t *arabic_plan = (const arabic_shape_plan_t *) plan->data;
 341
 342   if (!arabic_plan->do_fallback)
 343     return;
 344
 345 retry:
 346   arabic_fallback_plan_t *fallback_plan = (arabic_fallback_plan_t *) hb_atomic_ptr_get (&arabic_plan->fallback_plan);
 347   if (unlikely (!fallback_plan))
 348   {
 349     /* This sucks.  We need a font to build the fallback plan... */
 350     fallback_plan = arabic_fallback_plan_create (plan, font);
 351     if (unlikely (!hb_atomic_ptr_cmpexch (&(const_cast<arabic_shape_plan_t *> (arabic_plan))->fallback_plan, NULL, fallback_plan))) {
 352       arabic_fallback_plan_destroy (fallback_plan);
 353       goto retry;
 354     }
 355   }
 356
 357   arabic_fallback_plan_shape (fallback_plan, font, buffer);
 358 }
 359
 360
 361 const hb_ot_complex_shaper_t _hb_ot_complex_shaper_arabic =
 362 {
 363   "arabic",
 364   collect_features_arabic,
 365   NULL, /* override_features */
 366   data_create_arabic,
 367   data_destroy_arabic,
 368   NULL, /* preprocess_text_arabic */
 369   NULL, /* normalization_preference */
 370   NULL, /* decompose */
 371   NULL, /* compose */
 372   setup_masks_arabic,
 373   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_LATE,
 374   true, /* fallback_position */
 375 };