third_party/harfbuzz-ng/src/hb-ot-shape-complex-sea.cc

   1 /*
   2  * Copyright © 2011,2012,2013  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-indic-private.hh"
  28
  29 /* buffer var allocations */
  30 #define sea_category() complex_var_u8_0() /* indic_category_t */
  31 #define sea_position() complex_var_u8_1() /* indic_position_t */
  32
  33
  34 /*
  35  * South-East Asian shaper.
  36  * Loosely based on the Myanmar spec / shaper.
  37  * There is no OpenType spec for this.
  38  */
  39
  40 static const hb_tag_t
  41 basic_features[] =
  42 {
  43   /*
  44    * Basic features.
  45    * These features are applied in order, one at a time, after initial_reordering.
  46    */
  47   HB_TAG('p','r','e','f'),
  48   HB_TAG('a','b','v','f'),
  49   HB_TAG('b','l','w','f'),
  50   HB_TAG('p','s','t','f'),
  51 };
  52 static const hb_tag_t
  53 other_features[] =
  54 {
  55   /*
  56    * Other features.
  57    * These features are applied all at once, after final_reordering.
  58    */
  59   HB_TAG('p','r','e','s'),
  60   HB_TAG('a','b','v','s'),
  61   HB_TAG('b','l','w','s'),
  62   HB_TAG('p','s','t','s'),
  63   /* Positioning features, though we don't care about the types. */
  64   HB_TAG('d','i','s','t'),
  65 };
  66
  67 static void
  68 setup_syllables (const hb_ot_shape_plan_t *plan,
  69                  hb_font_t *font,
  70                  hb_buffer_t *buffer);
  71 static void
  72 initial_reordering (const hb_ot_shape_plan_t *plan,
  73                     hb_font_t *font,
  74                     hb_buffer_t *buffer);
  75 static void
  76 final_reordering (const hb_ot_shape_plan_t *plan,
  77                   hb_font_t *font,
  78                   hb_buffer_t *buffer);
  79
  80 static void
  81 collect_features_sea (hb_ot_shape_planner_t *plan)
  82 {
  83   hb_ot_map_builder_t *map = &plan->map;
  84
  85   /* Do this before any lookups have been applied. */
  86   map->add_gsub_pause (setup_syllables);
  87
  88   map->add_global_bool_feature (HB_TAG('l','o','c','l'));
  89   /* The Indic specs do not require ccmp, but we apply it here since if
  90    * there is a use of it, it's typically at the beginning. */
  91   map->add_global_bool_feature (HB_TAG('c','c','m','p'));
  92
  93   map->add_gsub_pause (initial_reordering);
  94   for (unsigned int i = 0; i < ARRAY_LENGTH (basic_features); i++)
  95   {
  96     map->add_feature (basic_features[i], 1, F_GLOBAL | F_MANUAL_ZWJ);
  97     map->add_gsub_pause (NULL);
  98   }
  99   map->add_gsub_pause (final_reordering);
 100   for (unsigned int i = 0; i < ARRAY_LENGTH (other_features); i++)
 101     map->add_feature (other_features[i], 1, F_GLOBAL | F_MANUAL_ZWJ);
 102 }
 103
 104 static void
 105 override_features_sea (hb_ot_shape_planner_t *plan)
 106 {
 107   plan->map.add_feature (HB_TAG('l','i','g','a'), 0, F_GLOBAL);
 108 }
 109
 110
 111 enum syllable_type_t {
 112   consonant_syllable,
 113   broken_cluster,
 114   non_sea_cluster,
 115 };
 116
 117 #include "hb-ot-shape-complex-sea-machine.hh"
 118
 119
 120 /* Note: This enum is duplicated in the -machine.rl source file.
 121  * Not sure how to avoid duplication. */
 122 enum sea_category_t {
 123 //  OT_C    = 1,
 124   OT_GB   = 12, /* Generic Base XXX DOTTED CIRCLE only for now */
 125 //  OT_H    = 4,  /* Halant */
 126   OT_IV   = 2,  /* Independent Vowel */
 127   OT_MR   = 22, /* Medial Ra */
 128 //  OT_CM   = 17, /* Consonant Medial */
 129   OT_VAbv = 26,
 130   OT_VBlw = 27,
 131   OT_VPre = 28,
 132   OT_VPst = 29,
 133   OT_T    = 3,  /* Tone Marks */
 134 //  OT_A    = 10, /* Anusvara */
 135 };
 136
 137 static inline void
 138 set_sea_properties (hb_glyph_info_t &info)
 139 {
 140   hb_codepoint_t u = info.codepoint;
 141   unsigned int type = hb_indic_get_categories (u);
 142   indic_category_t cat = (indic_category_t) (type & 0x7F);
 143   indic_position_t pos = (indic_position_t) (type >> 8);
 144
 145   /* Medial Ra */
 146   if (u == 0x1A55 || u == 0xAA34)
 147     cat = (indic_category_t) OT_MR;
 148
 149   if (cat == OT_M)
 150   {
 151     switch ((int) pos)
 152     {
 153       case POS_PRE_C:   cat = (indic_category_t) OT_VPre; break;
 154       case POS_ABOVE_C: cat = (indic_category_t) OT_VAbv; break;
 155       case POS_BELOW_C: cat = (indic_category_t) OT_VBlw; break;
 156       case POS_POST_C:  cat = (indic_category_t) OT_VPst; break;
 157     }
 158   }
 159
 160   info.sea_category() = (sea_category_t) cat;
 161   info.sea_position() = pos;
 162 }
 163
 164
 165 static void
 166 setup_masks_sea (const hb_ot_shape_plan_t *plan HB_UNUSED,
 167                  hb_buffer_t              *buffer,
 168                  hb_font_t                *font HB_UNUSED)
 169 {
 170   HB_BUFFER_ALLOCATE_VAR (buffer, sea_category);
 171   HB_BUFFER_ALLOCATE_VAR (buffer, sea_position);
 172
 173   /* We cannot setup masks here.  We save information about characters
 174    * and setup masks later on in a pause-callback. */
 175
 176   unsigned int count = buffer->len;
 177   for (unsigned int i = 0; i < count; i++)
 178     set_sea_properties (buffer->info[i]);
 179 }
 180
 181 static void
 182 setup_syllables (const hb_ot_shape_plan_t *plan HB_UNUSED,
 183                  hb_font_t *font HB_UNUSED,
 184                  hb_buffer_t *buffer)
 185 {
 186   find_syllables (buffer);
 187 }
 188
 189 static int
 190 compare_sea_order (const hb_glyph_info_t *pa, const hb_glyph_info_t *pb)
 191 {
 192   int a = pa->sea_position();
 193   int b = pb->sea_position();
 194
 195   return a < b ? -1 : a == b ? 0 : +1;
 196 }
 197
 198
 199 static void
 200 initial_reordering_consonant_syllable (const hb_ot_shape_plan_t *plan,
 201                                        hb_face_t *face,
 202                                        hb_buffer_t *buffer,
 203                                        unsigned int start, unsigned int end)
 204 {
 205   hb_glyph_info_t *info = buffer->info;
 206   unsigned int base = start;
 207
 208   /* Reorder! */
 209   unsigned int i = start;
 210   for (; i < base; i++)
 211     info[i].sea_position() = POS_PRE_C;
 212   if (i < end)
 213   {
 214     info[i].sea_position() = POS_BASE_C;
 215     i++;
 216   }
 217   for (; i < end; i++)
 218   {
 219     if (info[i].sea_category() == OT_MR) /* Pre-base reordering */
 220     {
 221       info[i].sea_position() = POS_PRE_C;
 222       continue;
 223     }
 224     if (info[i].sea_category() == OT_VPre) /* Left matra */
 225     {
 226       info[i].sea_position() = POS_PRE_M;
 227       continue;
 228     }
 229
 230     info[i].sea_position() = POS_AFTER_MAIN;
 231   }
 232
 233   buffer->merge_clusters (start, end);
 234   /* Sit tight, rock 'n roll! */
 235   hb_bubble_sort (info + start, end - start, compare_sea_order);
 236 }
 237
 238 static void
 239 initial_reordering_broken_cluster (const hb_ot_shape_plan_t *plan,
 240                                    hb_face_t *face,
 241                                    hb_buffer_t *buffer,
 242                                    unsigned int start, unsigned int end)
 243 {
 244   /* We already inserted dotted-circles, so just call the consonant_syllable. */
 245   initial_reordering_consonant_syllable (plan, face, buffer, start, end);
 246 }
 247
 248 static void
 249 initial_reordering_non_sea_cluster (const hb_ot_shape_plan_t *plan HB_UNUSED,
 250                                     hb_face_t *face HB_UNUSED,
 251                                     hb_buffer_t *buffer HB_UNUSED,
 252                                     unsigned int start HB_UNUSED, unsigned int end HB_UNUSED)
 253 {
 254   /* Nothing to do right now.  If we ever switch to using the output
 255    * buffer in the reordering process, we'd need to next_glyph() here. */
 256 }
 257
 258
 259 static void
 260 initial_reordering_syllable (const hb_ot_shape_plan_t *plan,
 261                              hb_face_t *face,
 262                              hb_buffer_t *buffer,
 263                              unsigned int start, unsigned int end)
 264 {
 265   syllable_type_t syllable_type = (syllable_type_t) (buffer->info[start].syllable() & 0x0F);
 266   switch (syllable_type) {
 267   case consonant_syllable:      initial_reordering_consonant_syllable  (plan, face, buffer, start, end); return;
 268   case broken_cluster:          initial_reordering_broken_cluster      (plan, face, buffer, start, end); return;
 269   case non_sea_cluster: initial_reordering_non_sea_cluster (plan, face, buffer, start, end); return;
 270   }
 271 }
 272
 273 static inline void
 274 insert_dotted_circles (const hb_ot_shape_plan_t *plan HB_UNUSED,
 275                        hb_font_t *font,
 276                        hb_buffer_t *buffer)
 277 {
 278   /* Note: This loop is extra overhead, but should not be measurable. */
 279   bool has_broken_syllables = false;
 280   unsigned int count = buffer->len;
 281   for (unsigned int i = 0; i < count; i++)
 282     if ((buffer->info[i].syllable() & 0x0F) == broken_cluster) {
 283       has_broken_syllables = true;
 284       break;
 285     }
 286   if (likely (!has_broken_syllables))
 287     return;
 288
 289
 290   hb_codepoint_t dottedcircle_glyph;
 291   if (!font->get_glyph (0x25CC, 0, &dottedcircle_glyph))
 292     return;
 293
 294   hb_glyph_info_t dottedcircle = {0};
 295   dottedcircle.codepoint = 0x25CC;
 296   set_sea_properties (dottedcircle);
 297   dottedcircle.codepoint = dottedcircle_glyph;
 298
 299   buffer->clear_output ();
 300
 301   buffer->idx = 0;
 302   unsigned int last_syllable = 0;
 303   while (buffer->idx < buffer->len)
 304   {
 305     unsigned int syllable = buffer->cur().syllable();
 306     syllable_type_t syllable_type = (syllable_type_t) (syllable & 0x0F);
 307     if (unlikely (last_syllable != syllable && syllable_type == broken_cluster))
 308     {
 309       last_syllable = syllable;
 310
 311       hb_glyph_info_t info = dottedcircle;
 312       info.cluster = buffer->cur().cluster;
 313       info.mask = buffer->cur().mask;
 314       info.syllable() = buffer->cur().syllable();
 315
 316       buffer->output_info (info);
 317     }
 318     else
 319       buffer->next_glyph ();
 320   }
 321
 322   buffer->swap_buffers ();
 323 }
 324
 325 static void
 326 initial_reordering (const hb_ot_shape_plan_t *plan,
 327                     hb_font_t *font,
 328                     hb_buffer_t *buffer)
 329 {
 330   insert_dotted_circles (plan, font, buffer);
 331
 332   hb_glyph_info_t *info = buffer->info;
 333   unsigned int count = buffer->len;
 334   if (unlikely (!count)) return;
 335   unsigned int last = 0;
 336   unsigned int last_syllable = info[0].syllable();
 337   for (unsigned int i = 1; i < count; i++)
 338     if (last_syllable != info[i].syllable()) {
 339       initial_reordering_syllable (plan, font->face, buffer, last, i);
 340       last = i;
 341       last_syllable = info[last].syllable();
 342     }
 343   initial_reordering_syllable (plan, font->face, buffer, last, count);
 344 }
 345
 346 static void
 347 final_reordering (const hb_ot_shape_plan_t *plan,
 348                   hb_font_t *font HB_UNUSED,
 349                   hb_buffer_t *buffer)
 350 {
 351   hb_glyph_info_t *info = buffer->info;
 352   unsigned int count = buffer->len;
 353
 354   /* Zero syllables now... */
 355   for (unsigned int i = 0; i < count; i++)
 356     info[i].syllable() = 0;
 357
 358   HB_BUFFER_DEALLOCATE_VAR (buffer, sea_category);
 359   HB_BUFFER_DEALLOCATE_VAR (buffer, sea_position);
 360 }
 361
 362
 363 static hb_ot_shape_normalization_mode_t
 364 normalization_preference_sea (const hb_segment_properties_t *props HB_UNUSED)
 365 {
 366   return HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS_NO_SHORT_CIRCUIT;
 367 }
 368
 369
 370 const hb_ot_complex_shaper_t _hb_ot_complex_shaper_sea =
 371 {
 372   "sea",
 373   collect_features_sea,
 374   override_features_sea,
 375   NULL, /* data_create */
 376   NULL, /* data_destroy */
 377   NULL, /* preprocess_text */
 378   normalization_preference_sea,
 379   NULL, /* decompose */
 380   NULL, /* compose */
 381   setup_masks_sea,
 382   HB_OT_SHAPE_ZERO_WIDTH_MARKS_NONE,
 383   false, /* fallback_position */
 384 };