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