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 };
  64
  65 static void
  66 setup_syllables (const hb_ot_shape_plan_t *plan,
  67                  hb_font_t *font,
  68                  hb_buffer_t *buffer);
  69 static void
  70 initial_reordering (const hb_ot_shape_plan_t *plan,
  71                     hb_font_t *font,
  72                     hb_buffer_t *buffer);
  73 static void
  74 final_reordering (const hb_ot_shape_plan_t *plan,
  75                   hb_font_t *font,
  76                   hb_buffer_t *buffer);
  77
  78 static void
  79 collect_features_myanmar (hb_ot_shape_planner_t *plan)
  80 {
  81   hb_ot_map_builder_t *map = &plan->map;
  82
  83   /* Do this before any lookups have been applied. */
  84   map->add_gsub_pause (setup_syllables);
  85
  86   map->add_global_bool_feature (HB_TAG('l','o','c','l'));
  87   /* The Indic specs do not require ccmp, but we apply it here since if
  88    * there is a use of it, it's typically at the beginning. */
  89   map->add_global_bool_feature (HB_TAG('c','c','m','p'));
  90
  91
  92   map->add_gsub_pause (initial_reordering);
  93   for (unsigned int i = 0; i < ARRAY_LENGTH (basic_features); i++)
  94   {
  95     map->add_feature (basic_features[i], 1, F_GLOBAL | F_MANUAL_ZWJ);
  96     map->add_gsub_pause (NULL);
  97   }
  98   map->add_gsub_pause (final_reordering);
  99   for (unsigned int i = 0; i < ARRAY_LENGTH (other_features); i++)
 100     map->add_feature (other_features[i], 1, F_GLOBAL | F_MANUAL_ZWJ);
 101 }
 102
 103 static void
 104 override_features_myanmar (hb_ot_shape_planner_t *plan)
 105 {
 106   plan->map.add_feature (HB_TAG('l','i','g','a'), 0, F_GLOBAL);
 107
 108   /*
 109    * Note:
 110    *
 111    * Spec says 'mark' is used, and the mmrtext.ttf font from
 112    * Windows 8 has lookups for it.  But testing suggests that
 113    * Windows 8 Uniscribe is NOT applying it.  It *is* applying
 114    * 'mkmk' however.
 115    */
 116   if (hb_options ().uniscribe_bug_compatible)
 117     plan->map.add_feature (HB_TAG('m','a','r','k'), 0, F_GLOBAL);
 118 }
 119
 120
 121 enum syllable_type_t {
 122   consonant_syllable,
 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_DOTTEDCIRCLE,
 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 };
 149
 150
 151 static inline bool
 152 is_one_of (const hb_glyph_info_t &info, unsigned int flags)
 153 {
 154   /* If it ligated, all bets are off. */
 155   if (is_a_ligature (info)) return false;
 156   return !!(FLAG (info.myanmar_category()) & flags);
 157 }
 158
 159 /* Note:
 160  *
 161  * We treat Vowels and placeholders as if they were consonants.  This is safe because Vowels
 162  * cannot happen in a consonant syllable.  The plus side however is, we can call the
 163  * consonant syllable logic from the vowel syllable function and get it all right! */
 164 #define CONSONANT_FLAGS (FLAG (OT_C) | FLAG (OT_CM) | FLAG (OT_Ra) | FLAG (OT_V) | FLAG (OT_NBSP) | FLAG (OT_GB))
 165 static inline bool
 166 is_consonant (const hb_glyph_info_t &info)
 167 {
 168   return is_one_of (info, CONSONANT_FLAGS);
 169 }
 170
 171
 172 static inline void
 173 set_myanmar_properties (hb_glyph_info_t &info)
 174 {
 175   hb_codepoint_t u = info.codepoint;
 176   unsigned int type = hb_indic_get_categories (u);
 177   indic_category_t cat = (indic_category_t) (type & 0x7F);
 178   indic_position_t pos = (indic_position_t) (type >> 8);
 179
 180   /* Myanmar
 181    * http://www.microsoft.com/typography/OpenTypeDev/myanmar/intro.htm#analyze
 182    */
 183   if (unlikely (hb_in_range<hb_codepoint_t> (u, 0xFE00, 0xFE0F)))
 184     cat = (indic_category_t) OT_VS;
 185   else if (unlikely (u == 0x200C)) cat = (indic_category_t) OT_ZWNJ;
 186   else if (unlikely (u == 0x200D)) cat = (indic_category_t) OT_ZWJ;
 187
 188   switch (u)
 189   {
 190     case 0x002D: case 0x00A0: case 0x00D7: case 0x2012:
 191     case 0x2013: case 0x2014: case 0x2015: case 0x2022:
 192     case 0x25CC: case 0x25FB: case 0x25FC: case 0x25FD:
 193     case 0x25FE:
 194       cat = (indic_category_t) OT_GB;
 195       break;
 196
 197     case 0x1004: case 0x101B: case 0x105A:
 198       cat = (indic_category_t) OT_Ra;
 199       break;
 200
 201     case 0x1032: case 0x1036:
 202       cat = (indic_category_t) OT_A;
 203       break;
 204
 205     case 0x103A:
 206       cat = (indic_category_t) OT_As;
 207       break;
 208
 209     case 0x1041: case 0x1042: case 0x1043: case 0x1044:
 210     case 0x1045: case 0x1046: case 0x1047: case 0x1048:
 211     case 0x1049: case 0x1090: case 0x1091: case 0x1092:
 212     case 0x1093: case 0x1094: case 0x1095: case 0x1096:
 213     case 0x1097: case 0x1098: case 0x1099:
 214       cat = (indic_category_t) OT_D;
 215       break;
 216
 217     case 0x1040:
 218       cat = (indic_category_t) OT_D; /* XXX The spec says D0, but Uniscribe doesn't seem to do. */
 219       break;
 220
 221     case 0x103E: case 0x1060:
 222       cat = (indic_category_t) OT_MH;
 223       break;
 224
 225     case 0x103C:
 226       cat = (indic_category_t) OT_MR;
 227       break;
 228
 229     case 0x103D: case 0x1082:
 230       cat = (indic_category_t) OT_MW;
 231       break;
 232
 233     case 0x103B: case 0x105E: case 0x105F:
 234       cat = (indic_category_t) OT_MY;
 235       break;
 236
 237     case 0x1063: case 0x1064: case 0x1069: case 0x106A:
 238     case 0x106B: case 0x106C: case 0x106D: case 0xAA7B:
 239       cat = (indic_category_t) OT_PT;
 240       break;
 241
 242     case 0x1038: case 0x1087: case 0x1088: case 0x1089:
 243     case 0x108A: case 0x108B: case 0x108C: case 0x108D:
 244     case 0x108F: case 0x109A: case 0x109B: case 0x109C:
 245       cat = (indic_category_t) OT_SM;
 246       break;
 247   }
 248
 249   if (cat == OT_M)
 250   {
 251     switch ((int) pos)
 252     {
 253       case POS_PRE_C:   cat = (indic_category_t) OT_VPre;
 254                         pos = POS_PRE_M;                  break;
 255       case POS_ABOVE_C: cat = (indic_category_t) OT_VAbv; break;
 256       case POS_BELOW_C: cat = (indic_category_t) OT_VBlw; break;
 257       case POS_POST_C:  cat = (indic_category_t) OT_VPst; break;
 258     }
 259   }
 260
 261   info.myanmar_category() = (myanmar_category_t) cat;
 262   info.myanmar_position() = pos;
 263 }
 264
 265
 266
 267 static void
 268 setup_masks_myanmar (const hb_ot_shape_plan_t *plan HB_UNUSED,
 269                    hb_buffer_t              *buffer,
 270                    hb_font_t                *font HB_UNUSED)
 271 {
 272   HB_BUFFER_ALLOCATE_VAR (buffer, myanmar_category);
 273   HB_BUFFER_ALLOCATE_VAR (buffer, myanmar_position);
 274
 275   /* We cannot setup masks here.  We save information about characters
 276    * and setup masks later on in a pause-callback. */
 277
 278   unsigned int count = buffer->len;
 279   for (unsigned int i = 0; i < count; i++)
 280     set_myanmar_properties (buffer->info[i]);
 281 }
 282
 283 static void
 284 setup_syllables (const hb_ot_shape_plan_t *plan HB_UNUSED,
 285                  hb_font_t *font HB_UNUSED,
 286                  hb_buffer_t *buffer)
 287 {
 288   find_syllables (buffer);
 289 }
 290
 291 static int
 292 compare_myanmar_order (const hb_glyph_info_t *pa, const hb_glyph_info_t *pb)
 293 {
 294   int a = pa->myanmar_position();
 295   int b = pb->myanmar_position();
 296
 297   return a < b ? -1 : a == b ? 0 : +1;
 298 }
 299
 300
 301 /* Rules from:
 302  * http://www.microsoft.com/typography/OpenTypeDev/myanmar/intro.htm */
 303
 304 static void
 305 initial_reordering_consonant_syllable (const hb_ot_shape_plan_t *plan,
 306                                        hb_face_t *face,
 307                                        hb_buffer_t *buffer,
 308                                        unsigned int start, unsigned int end)
 309 {
 310   hb_glyph_info_t *info = buffer->info;
 311
 312   unsigned int base = end;
 313   bool has_reph = false;
 314
 315   {
 316     unsigned int limit = start;
 317     if (start + 3 <= end &&
 318         info[start  ].myanmar_category() == OT_Ra &&
 319         info[start+1].myanmar_category() == OT_As &&
 320         info[start+2].myanmar_category() == OT_H)
 321     {
 322       limit += 3;
 323       base = start;
 324       has_reph = true;
 325     }
 326
 327     {
 328       if (!has_reph)
 329         base = limit;
 330
 331       for (unsigned int i = limit; i < end; i++)
 332         if (is_consonant (info[i]))
 333         {
 334           base = i;
 335           break;
 336         }
 337     }
 338   }
 339
 340   /* Reorder! */
 341   {
 342     unsigned int i = start;
 343     for (; i < start + (has_reph ? 3 : 0); i++)
 344       info[i].myanmar_position() = POS_AFTER_MAIN;
 345     for (; i < base; i++)
 346       info[i].myanmar_position() = POS_PRE_C;
 347     if (i < end)
 348     {
 349       info[i].myanmar_position() = POS_BASE_C;
 350       i++;
 351     }
 352     indic_position_t pos = POS_AFTER_MAIN;
 353     /* The following loop may be ugly, but it implements all of
 354      * Myanmar reordering! */
 355     for (; i < end; i++)
 356     {
 357       if (info[i].myanmar_category() == OT_MR) /* Pre-base reordering */
 358       {
 359         info[i].myanmar_position() = POS_PRE_C;
 360         continue;
 361       }
 362       if (info[i].myanmar_position() < POS_BASE_C) /* Left matra */
 363       {
 364         continue;
 365       }
 366
 367       if (pos == POS_AFTER_MAIN && info[i].myanmar_category() == OT_VBlw)
 368       {
 369         pos = POS_BELOW_C;
 370         info[i].myanmar_position() = pos;
 371         continue;
 372       }
 373
 374       if (pos == POS_BELOW_C && info[i].myanmar_category() == OT_A)
 375       {
 376         info[i].myanmar_position() = POS_BEFORE_SUB;
 377         continue;
 378       }
 379       if (pos == POS_BELOW_C && info[i].myanmar_category() == OT_VBlw)
 380       {
 381         info[i].myanmar_position() = pos;
 382         continue;
 383       }
 384       if (pos == POS_BELOW_C && info[i].myanmar_category() != OT_A)
 385       {
 386         pos = POS_AFTER_SUB;
 387         info[i].myanmar_position() = pos;
 388         continue;
 389       }
 390       info[i].myanmar_position() = pos;
 391     }
 392   }
 393
 394   buffer->merge_clusters (start, end);
 395   /* Sit tight, rock 'n roll! */
 396   hb_bubble_sort (info + start, end - start, compare_myanmar_order);
 397 }
 398
 399 static void
 400 initial_reordering_broken_cluster (const hb_ot_shape_plan_t *plan,
 401                                    hb_face_t *face,
 402                                    hb_buffer_t *buffer,
 403                                    unsigned int start, unsigned int end)
 404 {
 405   /* We already inserted dotted-circles, so just call the consonant_syllable. */
 406   initial_reordering_consonant_syllable (plan, face, buffer, start, end);
 407 }
 408
 409 static void
 410 initial_reordering_non_myanmar_cluster (const hb_ot_shape_plan_t *plan HB_UNUSED,
 411                                         hb_face_t *face HB_UNUSED,
 412                                         hb_buffer_t *buffer HB_UNUSED,
 413                                         unsigned int start HB_UNUSED, unsigned int end HB_UNUSED)
 414 {
 415   /* Nothing to do right now.  If we ever switch to using the output
 416    * buffer in the reordering process, we'd need to next_glyph() here. */
 417 }
 418
 419
 420 static void
 421 initial_reordering_syllable (const hb_ot_shape_plan_t *plan,
 422                              hb_face_t *face,
 423                              hb_buffer_t *buffer,
 424                              unsigned int start, unsigned int end)
 425 {
 426   syllable_type_t syllable_type = (syllable_type_t) (buffer->info[start].syllable() & 0x0F);
 427   switch (syllable_type) {
 428   case consonant_syllable:      initial_reordering_consonant_syllable  (plan, face, buffer, start, end); return;
 429   case broken_cluster:          initial_reordering_broken_cluster      (plan, face, buffer, start, end); return;
 430   case non_myanmar_cluster:     initial_reordering_non_myanmar_cluster (plan, face, buffer, start, end); return;
 431   }
 432 }
 433
 434 static inline void
 435 insert_dotted_circles (const hb_ot_shape_plan_t *plan HB_UNUSED,
 436                        hb_font_t *font,
 437                        hb_buffer_t *buffer)
 438 {
 439   /* Note: This loop is extra overhead, but should not be measurable. */
 440   bool has_broken_syllables = false;
 441   unsigned int count = buffer->len;
 442   for (unsigned int i = 0; i < count; i++)
 443     if ((buffer->info[i].syllable() & 0x0F) == broken_cluster) {
 444       has_broken_syllables = true;
 445       break;
 446     }
 447   if (likely (!has_broken_syllables))
 448     return;
 449
 450
 451   hb_codepoint_t dottedcircle_glyph;
 452   if (!font->get_glyph (0x25CC, 0, &dottedcircle_glyph))
 453     return;
 454
 455   hb_glyph_info_t dottedcircle = {0};
 456   dottedcircle.codepoint = 0x25CC;
 457   set_myanmar_properties (dottedcircle);
 458   dottedcircle.codepoint = dottedcircle_glyph;
 459
 460   buffer->clear_output ();
 461
 462   buffer->idx = 0;
 463   unsigned int last_syllable = 0;
 464   while (buffer->idx < buffer->len)
 465   {
 466     unsigned int syllable = buffer->cur().syllable();
 467     syllable_type_t syllable_type = (syllable_type_t) (syllable & 0x0F);
 468     if (unlikely (last_syllable != syllable && syllable_type == broken_cluster))
 469     {
 470       last_syllable = syllable;
 471
 472       hb_glyph_info_t info = dottedcircle;
 473       info.cluster = buffer->cur().cluster;
 474       info.mask = buffer->cur().mask;
 475       info.syllable() = buffer->cur().syllable();
 476
 477       buffer->output_info (info);
 478     }
 479     else
 480       buffer->next_glyph ();
 481   }
 482
 483   buffer->swap_buffers ();
 484 }
 485
 486 static void
 487 initial_reordering (const hb_ot_shape_plan_t *plan,
 488                     hb_font_t *font,
 489                     hb_buffer_t *buffer)
 490 {
 491   insert_dotted_circles (plan, font, buffer);
 492
 493   hb_glyph_info_t *info = buffer->info;
 494   unsigned int count = buffer->len;
 495   if (unlikely (!count)) return;
 496   unsigned int last = 0;
 497   unsigned int last_syllable = info[0].syllable();
 498   for (unsigned int i = 1; i < count; i++)
 499     if (last_syllable != info[i].syllable()) {
 500       initial_reordering_syllable (plan, font->face, buffer, last, i);
 501       last = i;
 502       last_syllable = info[last].syllable();
 503     }
 504   initial_reordering_syllable (plan, font->face, buffer, last, count);
 505 }
 506
 507 static void
 508 final_reordering (const hb_ot_shape_plan_t *plan,
 509                   hb_font_t *font HB_UNUSED,
 510                   hb_buffer_t *buffer)
 511 {
 512   hb_glyph_info_t *info = buffer->info;
 513   unsigned int count = buffer->len;
 514
 515   /* Zero syllables now... */
 516   for (unsigned int i = 0; i < count; i++)
 517     info[i].syllable() = 0;
 518
 519   HB_BUFFER_DEALLOCATE_VAR (buffer, myanmar_category);
 520   HB_BUFFER_DEALLOCATE_VAR (buffer, myanmar_position);
 521 }
 522
 523
 524 static hb_ot_shape_normalization_mode_t
 525 normalization_preference_myanmar (const hb_segment_properties_t *props HB_UNUSED)
 526 {
 527   return HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS_NO_SHORT_CIRCUIT;
 528 }
 529
 530
 531 const hb_ot_complex_shaper_t _hb_ot_complex_shaper_myanmar =
 532 {
 533   "myanmar",
 534   collect_features_myanmar,
 535   override_features_myanmar,
 536   NULL, /* data_create */
 537   NULL, /* data_destroy */
 538   NULL, /* preprocess_text */
 539   normalization_preference_myanmar,
 540   NULL, /* decompose */
 541   NULL, /* compose */
 542   setup_masks_myanmar,
 543   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_EARLY,
 544   false, /* fallback_position */
 545 };