third_party/harfbuzz-ng/src/hb-ot-shape-complex-myanmar.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 myanmar_category() complex_var_u8_0() /* myanmar_category_t */
  31 #define myanmar_position() complex_var_u8_1() /* myanmar_position_t */
  32
  33
  34 /*
  35  * Myanmar shaper.
  36  */
  37
  38 static const hb_tag_t
  39 basic_features[] =
  40 {
  41   /*
  42    * Basic features.
  43    * These features are applied in order, one at a time, after initial_reordering.
  44    */
  45   HB_TAG('r','p','h','f'),
  46   HB_TAG('p','r','e','f'),
  47   HB_TAG('b','l','w','f'),
  48   HB_TAG('p','s','t','f'),
  49 };
  50 static const hb_tag_t
  51 other_features[] =
  52 {
  53   /*
  54    * Other features.
  55    * These features are applied all at once, after final_reordering.
  56    */
  57   HB_TAG('p','r','e','s'),
  58   HB_TAG('a','b','v','s'),
  59   HB_TAG('b','l','w','s'),
  60   HB_TAG('p','s','t','s'),
  61   /* Positioning features, though we don't care about the types. */
  62   HB_TAG('d','i','s','t'),
  63   /* Pre-release version of Windows 8 Myanmar font had abvm,blwm
  64    * features.  The released Windows 8 version of the font (as well
  65    * as the released spec) used 'mark' instead.  The Windows 8
  66    * shaper however didn't apply 'mark' but did apply 'mkmk'.
  67    * Perhaps it applied abvm/blwm.  This was fixed in a Windows 8
  68    * update, so now it applies mark/mkmk.  We are guessing that
  69    * it still applies abvm/blwm too.
  70    */
  71   HB_TAG('a','b','v','m'),
  72   HB_TAG('b','l','w','m'),
  73 };
  74
  75 static void
  76 setup_syllables (const hb_ot_shape_plan_t *plan,
  77                  hb_font_t *font,
  78                  hb_buffer_t *buffer);
  79 static void
  80 initial_reordering (const hb_ot_shape_plan_t *plan,
  81                     hb_font_t *font,
  82                     hb_buffer_t *buffer);
  83 static void
  84 final_reordering (const hb_ot_shape_plan_t *plan,
  85                   hb_font_t *font,
  86                   hb_buffer_t *buffer);
  87
  88 static void
  89 collect_features_myanmar (hb_ot_shape_planner_t *plan)
  90 {
  91   hb_ot_map_builder_t *map = &plan->map;
  92
  93   /* Do this before any lookups have been applied. */
  94   map->add_gsub_pause (setup_syllables);
  95
  96   map->add_global_bool_feature (HB_TAG('l','o','c','l'));
  97   /* The Indic specs do not require ccmp, but we apply it here since if
  98    * there is a use of it, it's typically at the beginning. */
  99   map->add_global_bool_feature (HB_TAG('c','c','m','p'));
 100
 101
 102   map->add_gsub_pause (initial_reordering);
 103   for (unsigned int i = 0; i < ARRAY_LENGTH (basic_features); i++)
 104   {
 105     map->add_feature (basic_features[i], 1, F_GLOBAL | F_MANUAL_ZWJ);
 106     map->add_gsub_pause (NULL);
 107   }
 108   map->add_gsub_pause (final_reordering);
 109   for (unsigned int i = 0; i < ARRAY_LENGTH (other_features); i++)
 110     map->add_feature (other_features[i], 1, F_GLOBAL | F_MANUAL_ZWJ);
 111 }
 112
 113 static void
 114 override_features_myanmar (hb_ot_shape_planner_t *plan)
 115 {
 116   plan->map.add_feature (HB_TAG('l','i','g','a'), 0, F_GLOBAL);
 117 }
 118
 119
 120 enum syllable_type_t {
 121   consonant_syllable,
 122   punctuation_cluster,
 123   broken_cluster,
 124   non_myanmar_cluster,
 125 };
 126
 127 #include "hb-ot-shape-complex-myanmar-machine.hh"
 128
 129
 130 /* Note: This enum is duplicated in the -machine.rl source file.
 131  * Not sure how to avoid duplication. */
 132 enum myanmar_category_t {
 133   OT_As  = 18, /* Asat */
 134   OT_D   = 19, /* Digits except zero */
 135   OT_D0  = 20, /* Digit zero */
 136   OT_DB  = OT_N, /* Dot below */
 137   OT_GB  = OT_PLACEHOLDER,
 138   OT_MH  = 21, /* Various consonant medial types */
 139   OT_MR  = 22, /* Various consonant medial types */
 140   OT_MW  = 23, /* Various consonant medial types */
 141   OT_MY  = 24, /* Various consonant medial types */
 142   OT_PT  = 25, /* Pwo and other tones */
 143   OT_VAbv = 26,
 144   OT_VBlw = 27,
 145   OT_VPre = 28,
 146   OT_VPst = 29,
 147   OT_VS   = 30, /* Variation selectors */
 148   OT_P    = 31  /* Punctuation */
 149 };
 150
 151
 152 static inline bool
 153 is_one_of (const hb_glyph_info_t &info, unsigned int flags)
 154 {
 155   /* If it ligated, all bets are off. */
 156   if (_hb_glyph_info_ligated (&info)) return false;
 157   return !!(FLAG (info.myanmar_category()) & flags);
 158 }
 159
 160 static inline bool
 161 is_consonant (const hb_glyph_info_t &info)
 162 {
 163   return is_one_of (info, CONSONANT_FLAGS);
 164 }
 165
 166
 167 static inline void
 168 set_myanmar_properties (hb_glyph_info_t &info)
 169 {
 170   hb_codepoint_t u = info.codepoint;
 171   unsigned int type = hb_indic_get_categories (u);
 172   indic_category_t cat = (indic_category_t) (type & 0x7Fu);
 173   indic_position_t pos = (indic_position_t) (type >> 8);
 174
 175   /* Myanmar
 176    * http://www.microsoft.com/typography/OpenTypeDev/myanmar/intro.htm#analyze
 177    */
 178   if (unlikely (hb_in_range (u, 0xFE00u, 0xFE0Fu)))
 179     cat = (indic_category_t) OT_VS;
 180
 181   switch (u)
 182   {
 183     case 0x104Eu:
 184       cat = (indic_category_t) OT_C; /* The spec says C, IndicSyllableCategory doesn't have. */
 185       break;
 186
 187     case 0x002Du: case 0x00A0u: case 0x00D7u: case 0x2012u:
 188     case 0x2013u: case 0x2014u: case 0x2015u: case 0x2022u:
 189     case 0x25CCu: case 0x25FBu: case 0x25FCu: case 0x25FDu:
 190     case 0x25FEu:
 191       cat = (indic_category_t) OT_GB;
 192       break;
 193
 194     case 0x1004u: case 0x101Bu: case 0x105Au:
 195       cat = (indic_category_t) OT_Ra;
 196       break;
 197
 198     case 0x1032u: case 0x1036u:
 199       cat = (indic_category_t) OT_A;
 200       break;
 201
 202     case 0x103Au:
 203       cat = (indic_category_t) OT_As;
 204       break;
 205
 206     case 0x1041u: case 0x1042u: case 0x1043u: case 0x1044u:
 207     case 0x1045u: case 0x1046u: case 0x1047u: case 0x1048u:
 208     case 0x1049u: case 0x1090u: case 0x1091u: case 0x1092u:
 209     case 0x1093u: case 0x1094u: case 0x1095u: case 0x1096u:
 210     case 0x1097u: case 0x1098u: case 0x1099u:
 211       cat = (indic_category_t) OT_D;
 212       break;
 213
 214     case 0x1040u:
 215       cat = (indic_category_t) OT_D; /* XXX The spec says D0, but Uniscribe doesn't seem to do. */
 216       break;
 217
 218     case 0x103Eu: case 0x1060u:
 219       cat = (indic_category_t) OT_MH;
 220       break;
 221
 222     case 0x103Cu:
 223       cat = (indic_category_t) OT_MR;
 224       break;
 225
 226     case 0x103Du: case 0x1082u:
 227       cat = (indic_category_t) OT_MW;
 228       break;
 229
 230     case 0x103Bu: case 0x105Eu: case 0x105Fu:
 231       cat = (indic_category_t) OT_MY;
 232       break;
 233
 234     case 0x1063u: case 0x1064u: case 0x1069u: case 0x106Au:
 235     case 0x106Bu: case 0x106Cu: case 0x106Du: case 0xAA7Bu:
 236       cat = (indic_category_t) OT_PT;
 237       break;
 238
 239     case 0x1038u: case 0x1087u: case 0x1088u: case 0x1089u:
 240     case 0x108Au: case 0x108Bu: case 0x108Cu: case 0x108Du:
 241     case 0x108Fu: case 0x109Au: case 0x109Bu: case 0x109Cu:
 242       cat = (indic_category_t) OT_SM;
 243       break;
 244
 245     case 0x104Au: case 0x104Bu:
 246       cat = (indic_category_t) OT_P;
 247       break;
 248   }
 249
 250   if (cat == OT_M)
 251   {
 252     switch ((int) pos)
 253     {
 254       case POS_PRE_C:   cat = (indic_category_t) OT_VPre;
 255                         pos = POS_PRE_M;                  break;
 256       case POS_ABOVE_C: cat = (indic_category_t) OT_VAbv; break;
 257       case POS_BELOW_C: cat = (indic_category_t) OT_VBlw; break;
 258       case POS_POST_C:  cat = (indic_category_t) OT_VPst; break;
 259     }
 260   }
 261
 262   info.myanmar_category() = (myanmar_category_t) cat;
 263   info.myanmar_position() = pos;
 264 }
 265
 266
 267
 268 static void
 269 setup_masks_myanmar (const hb_ot_shape_plan_t *plan HB_UNUSED,
 270                    hb_buffer_t              *buffer,
 271                    hb_font_t                *font HB_UNUSED)
 272 {
 273   HB_BUFFER_ALLOCATE_VAR (buffer, myanmar_category);
 274   HB_BUFFER_ALLOCATE_VAR (buffer, myanmar_position);
 275
 276   /* We cannot setup masks here.  We save information about characters
 277    * and setup masks later on in a pause-callback. */
 278
 279   unsigned int count = buffer->len;
 280   hb_glyph_info_t *info = buffer->info;
 281   for (unsigned int i = 0; i < count; i++)
 282     set_myanmar_properties (info[i]);
 283 }
 284
 285 static void
 286 setup_syllables (const hb_ot_shape_plan_t *plan HB_UNUSED,
 287                  hb_font_t *font HB_UNUSED,
 288                  hb_buffer_t *buffer)
 289 {
 290   find_syllables (buffer);
 291 }
 292
 293 static int
 294 compare_myanmar_order (const hb_glyph_info_t *pa, const hb_glyph_info_t *pb)
 295 {
 296   int a = pa->myanmar_position();
 297   int b = pb->myanmar_position();
 298
 299   return a < b ? -1 : a == b ? 0 : +1;
 300 }
 301
 302
 303 /* Rules from:
 304  * http://www.microsoft.com/typography/OpenTypeDev/myanmar/intro.htm */
 305
 306 static void
 307 initial_reordering_consonant_syllable (const hb_ot_shape_plan_t *plan,
 308                                        hb_face_t *face,
 309                                        hb_buffer_t *buffer,
 310                                        unsigned int start, unsigned int end)
 311 {
 312   hb_glyph_info_t *info = buffer->info;
 313
 314   unsigned int base = end;
 315   bool has_reph = false;
 316
 317   {
 318     unsigned int limit = start;
 319     if (start + 3 <= end &&
 320         info[start  ].myanmar_category() == OT_Ra &&
 321         info[start+1].myanmar_category() == OT_As &&
 322         info[start+2].myanmar_category() == OT_H)
 323     {
 324       limit += 3;
 325       base = start;
 326       has_reph = true;
 327     }
 328
 329     {
 330       if (!has_reph)
 331         base = limit;
 332
 333       for (unsigned int i = limit; i < end; i++)
 334         if (is_consonant (info[i]))
 335         {
 336           base = i;
 337           break;
 338         }
 339     }
 340   }
 341
 342   /* Reorder! */
 343   {
 344     unsigned int i = start;
 345     for (; i < start + (has_reph ? 3 : 0); i++)
 346       info[i].myanmar_position() = POS_AFTER_MAIN;
 347     for (; i < base; i++)
 348       info[i].myanmar_position() = POS_PRE_C;
 349     if (i < end)
 350     {
 351       info[i].myanmar_position() = POS_BASE_C;
 352       i++;
 353     }
 354     indic_position_t pos = POS_AFTER_MAIN;
 355     /* The following loop may be ugly, but it implements all of
 356      * Myanmar reordering! */
 357     for (; i < end; i++)
 358     {
 359       if (info[i].myanmar_category() == OT_MR) /* Pre-base reordering */
 360       {
 361         info[i].myanmar_position() = POS_PRE_C;
 362         continue;
 363       }
 364       if (info[i].myanmar_position() < POS_BASE_C) /* Left matra */
 365       {
 366         continue;
 367       }
 368
 369       if (pos == POS_AFTER_MAIN && info[i].myanmar_category() == OT_VBlw)
 370       {
 371         pos = POS_BELOW_C;
 372         info[i].myanmar_position() = pos;
 373         continue;
 374       }
 375
 376       if (pos == POS_BELOW_C && info[i].myanmar_category() == OT_A)
 377       {
 378         info[i].myanmar_position() = POS_BEFORE_SUB;
 379         continue;
 380       }
 381       if (pos == POS_BELOW_C && info[i].myanmar_category() == OT_VBlw)
 382       {
 383         info[i].myanmar_position() = pos;
 384         continue;
 385       }
 386       if (pos == POS_BELOW_C && info[i].myanmar_category() != OT_A)
 387       {
 388         pos = POS_AFTER_SUB;
 389         info[i].myanmar_position() = pos;
 390         continue;
 391       }
 392       info[i].myanmar_position() = pos;
 393     }
 394   }
 395
 396   buffer->merge_clusters (start, end);
 397   /* Sit tight, rock 'n roll! */
 398   hb_bubble_sort (info + start, end - start, compare_myanmar_order);
 399 }
 400
 401 static void
 402 initial_reordering_broken_cluster (const hb_ot_shape_plan_t *plan,
 403                                    hb_face_t *face,
 404                                    hb_buffer_t *buffer,
 405                                    unsigned int start, unsigned int end)
 406 {
 407   /* We already inserted dotted-circles, so just call the consonant_syllable. */
 408   initial_reordering_consonant_syllable (plan, face, buffer, start, end);
 409 }
 410
 411 static void
 412 initial_reordering_punctuation_cluster (const hb_ot_shape_plan_t *plan HB_UNUSED,
 413                                         hb_face_t *face HB_UNUSED,
 414                                         hb_buffer_t *buffer HB_UNUSED,
 415                                         unsigned int start HB_UNUSED, unsigned int end HB_UNUSED)
 416 {
 417   /* Nothing to do right now.  If we ever switch to using the output
 418    * buffer in the reordering process, we'd need to next_glyph() here. */
 419 }
 420
 421 static void
 422 initial_reordering_non_myanmar_cluster (const hb_ot_shape_plan_t *plan HB_UNUSED,
 423                                         hb_face_t *face HB_UNUSED,
 424                                         hb_buffer_t *buffer HB_UNUSED,
 425                                         unsigned int start HB_UNUSED, unsigned int end HB_UNUSED)
 426 {
 427   /* Nothing to do right now.  If we ever switch to using the output
 428    * buffer in the reordering process, we'd need to next_glyph() here. */
 429 }
 430
 431
 432 static void
 433 initial_reordering_syllable (const hb_ot_shape_plan_t *plan,
 434                              hb_face_t *face,
 435                              hb_buffer_t *buffer,
 436                              unsigned int start, unsigned int end)
 437 {
 438   syllable_type_t syllable_type = (syllable_type_t) (buffer->info[start].syllable() & 0x0F);
 439   switch (syllable_type) {
 440   case consonant_syllable:      initial_reordering_consonant_syllable  (plan, face, buffer, start, end); return;
 441   case punctuation_cluster:     initial_reordering_punctuation_cluster (plan, face, buffer, start, end); return;
 442   case broken_cluster:          initial_reordering_broken_cluster      (plan, face, buffer, start, end); return;
 443   case non_myanmar_cluster:     initial_reordering_non_myanmar_cluster (plan, face, buffer, start, end); return;
 444   }
 445 }
 446
 447 static inline void
 448 insert_dotted_circles (const hb_ot_shape_plan_t *plan HB_UNUSED,
 449                        hb_font_t *font,
 450                        hb_buffer_t *buffer)
 451 {
 452   /* Note: This loop is extra overhead, but should not be measurable. */
 453   bool has_broken_syllables = false;
 454   unsigned int count = buffer->len;
 455   hb_glyph_info_t *info = buffer->info;
 456   for (unsigned int i = 0; i < count; i++)
 457     if ((info[i].syllable() & 0x0F) == broken_cluster)
 458     {
 459       has_broken_syllables = true;
 460       break;
 461     }
 462   if (likely (!has_broken_syllables))
 463     return;
 464
 465
 466   hb_codepoint_t dottedcircle_glyph;
 467   if (!font->get_glyph (0x25CCu, 0, &dottedcircle_glyph))
 468     return;
 469
 470   hb_glyph_info_t dottedcircle = {0};
 471   dottedcircle.codepoint = 0x25CCu;
 472   set_myanmar_properties (dottedcircle);
 473   dottedcircle.codepoint = dottedcircle_glyph;
 474
 475   buffer->clear_output ();
 476
 477   buffer->idx = 0;
 478   unsigned int last_syllable = 0;
 479   while (buffer->idx < buffer->len)
 480   {
 481     unsigned int syllable = buffer->cur().syllable();
 482     syllable_type_t syllable_type = (syllable_type_t) (syllable & 0x0F);
 483     if (unlikely (last_syllable != syllable && syllable_type == broken_cluster))
 484     {
 485       last_syllable = syllable;
 486
 487       hb_glyph_info_t info = dottedcircle;
 488       info.cluster = buffer->cur().cluster;
 489       info.mask = buffer->cur().mask;
 490       info.syllable() = buffer->cur().syllable();
 491
 492       buffer->output_info (info);
 493     }
 494     else
 495       buffer->next_glyph ();
 496   }
 497
 498   buffer->swap_buffers ();
 499 }
 500
 501 static void
 502 initial_reordering (const hb_ot_shape_plan_t *plan,
 503                     hb_font_t *font,
 504                     hb_buffer_t *buffer)
 505 {
 506   insert_dotted_circles (plan, font, buffer);
 507
 508   hb_glyph_info_t *info = buffer->info;
 509   unsigned int count = buffer->len;
 510   if (unlikely (!count)) return;
 511   unsigned int last = 0;
 512   unsigned int last_syllable = info[0].syllable();
 513   for (unsigned int i = 1; i < count; i++)
 514     if (last_syllable != info[i].syllable()) {
 515       initial_reordering_syllable (plan, font->face, buffer, last, i);
 516       last = i;
 517       last_syllable = info[last].syllable();
 518     }
 519   initial_reordering_syllable (plan, font->face, buffer, last, count);
 520 }
 521
 522 static void
 523 final_reordering (const hb_ot_shape_plan_t *plan,
 524                   hb_font_t *font HB_UNUSED,
 525                   hb_buffer_t *buffer)
 526 {
 527   hb_glyph_info_t *info = buffer->info;
 528   unsigned int count = buffer->len;
 529
 530   /* Zero syllables now... */
 531   for (unsigned int i = 0; i < count; i++)
 532     info[i].syllable() = 0;
 533
 534   HB_BUFFER_DEALLOCATE_VAR (buffer, myanmar_category);
 535   HB_BUFFER_DEALLOCATE_VAR (buffer, myanmar_position);
 536 }
 537
 538
 539 /* Uniscribe seems to have a shaper for 'mymr' that is like the
 540  * generic shaper, except that it zeros mark advances GDEF_LATE. */
 541 const hb_ot_complex_shaper_t _hb_ot_complex_shaper_myanmar_old =
 542 {
 543   "default",
 544   NULL, /* collect_features */
 545   NULL, /* override_features */
 546   NULL, /* data_create */
 547   NULL, /* data_destroy */
 548   NULL, /* preprocess_text */
 549   HB_OT_SHAPE_NORMALIZATION_MODE_DEFAULT,
 550   NULL, /* decompose */
 551   NULL, /* compose */
 552   NULL, /* setup_masks */
 553   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_LATE,
 554   true, /* fallback_position */
 555 };
 556
 557 const hb_ot_complex_shaper_t _hb_ot_complex_shaper_myanmar =
 558 {
 559   "myanmar",
 560   collect_features_myanmar,
 561   override_features_myanmar,
 562   NULL, /* data_create */
 563   NULL, /* data_destroy */
 564   NULL, /* preprocess_text */
 565   HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS_NO_SHORT_CIRCUIT,
 566   NULL, /* decompose */
 567   NULL, /* compose */
 568   setup_masks_myanmar,
 569   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_EARLY,
 570   false, /* fallback_position */
 571 };