gfx/harfbuzz/src/hb-ot-shaper-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.hh"
  28
  29 #ifndef HB_NO_OT_SHAPE
  30
  31 #include "hb-ot-shaper-myanmar-machine.hh"
  32 #include "hb-ot-shaper-indic.hh"
  33 #include "hb-ot-layout.hh"
  34
  35
  36 /*
  37  * Myanmar shaper.
  38  */
  39
  40
  41 static const hb_tag_t
  42 myanmar_basic_features[] =
  43 {
  44   /*
  45    * Basic features.
  46    * These features are applied in order, one at a time, after reordering,
  47    * constrained to the syllable.
  48    */
  49   HB_TAG('r','p','h','f'),
  50   HB_TAG('p','r','e','f'),
  51   HB_TAG('b','l','w','f'),
  52   HB_TAG('p','s','t','f'),
  53 };
  54 static const hb_tag_t
  55 myanmar_other_features[] =
  56 {
  57   /*
  58    * Other features.
  59    * These features are applied all at once, after clearing syllables.
  60    */
  61   HB_TAG('p','r','e','s'),
  62   HB_TAG('a','b','v','s'),
  63   HB_TAG('b','l','w','s'),
  64   HB_TAG('p','s','t','s'),
  65 };
  66
  67 static inline void
  68 set_myanmar_properties (hb_glyph_info_t &info)
  69 {
  70   hb_codepoint_t u = info.codepoint;
  71   unsigned int type = hb_indic_get_categories (u);
  72
  73   info.myanmar_category() = (myanmar_category_t) (type & 0xFFu);
  74 }
  75
  76
  77 static inline bool
  78 is_one_of_myanmar (const hb_glyph_info_t &info, unsigned int flags)
  79 {
  80   /* If it ligated, all bets are off. */
  81   if (_hb_glyph_info_ligated (&info)) return false;
  82   return !!(FLAG_UNSAFE (info.myanmar_category()) & flags);
  83 }
  84
  85 /* Note:
  86  *
  87  * We treat Vowels and placeholders as if they were consonants.  This is safe because Vowels
  88  * cannot happen in a consonant syllable.  The plus side however is, we can call the
  89  * consonant syllable logic from the vowel syllable function and get it all right!
  90  *
  91  * Keep in sync with consonant_categories in the generator. */
  92 #define CONSONANT_FLAGS_MYANMAR (FLAG (M_Cat(C)) | FLAG (M_Cat(CS)) | FLAG (M_Cat(Ra)) | /* FLAG (M_Cat(CM)) | */ FLAG (M_Cat(IV)) | FLAG (M_Cat(GB)) | FLAG (M_Cat(DOTTEDCIRCLE)))
  93
  94 static inline bool
  95 is_consonant_myanmar (const hb_glyph_info_t &info)
  96 {
  97   return is_one_of_myanmar (info, CONSONANT_FLAGS_MYANMAR);
  98 }
  99
 100
 101 static bool
 102 setup_syllables_myanmar (const hb_ot_shape_plan_t *plan,
 103                          hb_font_t *font,
 104                          hb_buffer_t *buffer);
 105 static bool
 106 reorder_myanmar (const hb_ot_shape_plan_t *plan,
 107                  hb_font_t *font,
 108                  hb_buffer_t *buffer);
 109
 110 static void
 111 collect_features_myanmar (hb_ot_shape_planner_t *plan)
 112 {
 113   hb_ot_map_builder_t *map = &plan->map;
 114
 115   /* Do this before any lookups have been applied. */
 116   map->add_gsub_pause (setup_syllables_myanmar);
 117
 118   map->enable_feature (HB_TAG('l','o','c','l'), F_PER_SYLLABLE);
 119   /* The Indic specs do not require ccmp, but we apply it here since if
 120    * there is a use of it, it's typically at the beginning. */
 121   map->enable_feature (HB_TAG('c','c','m','p'), F_PER_SYLLABLE);
 122
 123
 124   map->add_gsub_pause (reorder_myanmar);
 125
 126   for (unsigned int i = 0; i < ARRAY_LENGTH (myanmar_basic_features); i++)
 127   {
 128     map->enable_feature (myanmar_basic_features[i], F_MANUAL_ZWJ | F_PER_SYLLABLE);
 129     map->add_gsub_pause (nullptr);
 130   }
 131   map->add_gsub_pause (hb_syllabic_clear_var); // Don't need syllables anymore, use stop to free buffer var
 132
 133   for (unsigned int i = 0; i < ARRAY_LENGTH (myanmar_other_features); i++)
 134     map->enable_feature (myanmar_other_features[i], F_MANUAL_ZWJ);
 135 }
 136
 137 static void
 138 setup_masks_myanmar (const hb_ot_shape_plan_t *plan HB_UNUSED,
 139                      hb_buffer_t              *buffer,
 140                      hb_font_t                *font HB_UNUSED)
 141 {
 142   HB_BUFFER_ALLOCATE_VAR (buffer, myanmar_category);
 143   HB_BUFFER_ALLOCATE_VAR (buffer, myanmar_position);
 144
 145   /* No masks, we just save information about characters. */
 146
 147   unsigned int count = buffer->len;
 148   hb_glyph_info_t *info = buffer->info;
 149   for (unsigned int i = 0; i < count; i++)
 150     set_myanmar_properties (info[i]);
 151 }
 152
 153 static bool
 154 setup_syllables_myanmar (const hb_ot_shape_plan_t *plan HB_UNUSED,
 155                          hb_font_t *font HB_UNUSED,
 156                          hb_buffer_t *buffer)
 157 {
 158   HB_BUFFER_ALLOCATE_VAR (buffer, syllable);
 159   find_syllables_myanmar (buffer);
 160   foreach_syllable (buffer, start, end)
 161     buffer->unsafe_to_break (start, end);
 162   return false;
 163 }
 164
 165 static int
 166 compare_myanmar_order (const hb_glyph_info_t *pa, const hb_glyph_info_t *pb)
 167 {
 168   int a = pa->myanmar_position();
 169   int b = pb->myanmar_position();
 170
 171   return (int) a - (int) b;
 172 }
 173
 174
 175 /* Rules from:
 176  * https://docs.microsoft.com/en-us/typography/script-development/myanmar */
 177
 178 static void
 179 initial_reordering_consonant_syllable (hb_buffer_t *buffer,
 180                                        unsigned int start, unsigned int end)
 181 {
 182   hb_glyph_info_t *info = buffer->info;
 183
 184   unsigned int base = end;
 185   bool has_reph = false;
 186
 187   {
 188     unsigned int limit = start;
 189     if (start + 3 <= end &&
 190         info[start  ].myanmar_category() == M_Cat(Ra) &&
 191         info[start+1].myanmar_category() == M_Cat(As) &&
 192         info[start+2].myanmar_category() == M_Cat(H))
 193     {
 194       limit += 3;
 195       base = start;
 196       has_reph = true;
 197     }
 198
 199     {
 200       if (!has_reph)
 201         base = limit;
 202
 203       for (unsigned int i = limit; i < end; i++)
 204         if (is_consonant_myanmar (info[i]))
 205         {
 206           base = i;
 207           break;
 208         }
 209     }
 210   }
 211
 212   /* Reorder! */
 213   {
 214     unsigned int i = start;
 215     for (; i < start + (has_reph ? 3 : 0); i++)
 216       info[i].myanmar_position() = POS_AFTER_MAIN;
 217     for (; i < base; i++)
 218       info[i].myanmar_position() = POS_PRE_C;
 219     if (i < end)
 220     {
 221       info[i].myanmar_position() = POS_BASE_C;
 222       i++;
 223     }
 224     myanmar_position_t pos = POS_AFTER_MAIN;
 225     /* The following loop may be ugly, but it implements all of
 226      * Myanmar reordering! */
 227     for (; i < end; i++)
 228     {
 229       if (info[i].myanmar_category() == M_Cat(MR)) /* Pre-base reordering */
 230       {
 231         info[i].myanmar_position() = POS_PRE_C;
 232         continue;
 233       }
 234       if (info[i].myanmar_category() == M_Cat(VPre)) /* Left matra */
 235       {
 236         info[i].myanmar_position() = POS_PRE_M;
 237         continue;
 238       }
 239       if (info[i].myanmar_category() == M_Cat(VS))
 240       {
 241         info[i].myanmar_position() = info[i - 1].myanmar_position();
 242         continue;
 243       }
 244
 245       if (pos == POS_AFTER_MAIN && info[i].myanmar_category() == M_Cat(VBlw))
 246       {
 247         pos = POS_BELOW_C;
 248         info[i].myanmar_position() = pos;
 249         continue;
 250       }
 251
 252       if (pos == POS_BELOW_C && info[i].myanmar_category() == M_Cat(A))
 253       {
 254         info[i].myanmar_position() = POS_BEFORE_SUB;
 255         continue;
 256       }
 257       if (pos == POS_BELOW_C && info[i].myanmar_category() == M_Cat(VBlw))
 258       {
 259         info[i].myanmar_position() = pos;
 260         continue;
 261       }
 262       if (pos == POS_BELOW_C && info[i].myanmar_category() != M_Cat(A))
 263       {
 264         pos = POS_AFTER_SUB;
 265         info[i].myanmar_position() = pos;
 266         continue;
 267       }
 268       info[i].myanmar_position() = pos;
 269     }
 270   }
 271
 272   /* Sit tight, rock 'n roll! */
 273   buffer->sort (start, end, compare_myanmar_order);
 274
 275   /* Flip left-matra sequence. */
 276   unsigned first_left_matra = end;
 277   unsigned last_left_matra = end;
 278   for (unsigned int i = start; i < end; i++)
 279   {
 280     if (info[i].myanmar_position() == POS_PRE_M)
 281     {
 282       if (first_left_matra == end)
 283         first_left_matra = i;
 284       last_left_matra = i;
 285     }
 286   }
 287   /* https://github.com/harfbuzz/harfbuzz/issues/3863 */
 288   if (first_left_matra < last_left_matra)
 289   {
 290     /* No need to merge clusters, done already? */
 291     buffer->reverse_range (first_left_matra, last_left_matra + 1);
 292     /* Reverse back VS, etc. */
 293     unsigned i = first_left_matra;
 294     for (unsigned j = i; j <= last_left_matra; j++)
 295       if (info[j].myanmar_category() == M_Cat(VPre))
 296       {
 297         buffer->reverse_range (i, j + 1);
 298         i = j + 1;
 299       }
 300   }
 301 }
 302
 303 static void
 304 reorder_syllable_myanmar (const hb_ot_shape_plan_t *plan HB_UNUSED,
 305                           hb_face_t *face HB_UNUSED,
 306                           hb_buffer_t *buffer,
 307                           unsigned int start, unsigned int end)
 308 {
 309   myanmar_syllable_type_t syllable_type = (myanmar_syllable_type_t) (buffer->info[start].syllable() & 0x0F);
 310   switch (syllable_type) {
 311
 312     case myanmar_broken_cluster: /* We already inserted dotted-circles, so just call the consonant_syllable. */
 313     case myanmar_consonant_syllable:
 314       initial_reordering_consonant_syllable  (buffer, start, end);
 315       break;
 316
 317     case myanmar_non_myanmar_cluster:
 318       break;
 319   }
 320 }
 321
 322 static bool
 323 reorder_myanmar (const hb_ot_shape_plan_t *plan,
 324                  hb_font_t *font,
 325                  hb_buffer_t *buffer)
 326 {
 327   bool ret = false;
 328   if (buffer->message (font, "start reordering myanmar"))
 329   {
 330     if (hb_syllabic_insert_dotted_circles (font, buffer,
 331                                            myanmar_broken_cluster,
 332                                            M_Cat(DOTTEDCIRCLE)))
 333       ret = true;
 334
 335     foreach_syllable (buffer, start, end)
 336       reorder_syllable_myanmar (plan, font->face, buffer, start, end);
 337     (void) buffer->message (font, "end reordering myanmar");
 338   }
 339
 340   HB_BUFFER_DEALLOCATE_VAR (buffer, myanmar_category);
 341   HB_BUFFER_DEALLOCATE_VAR (buffer, myanmar_position);
 342
 343   return ret;
 344 }
 345
 346
 347 const hb_ot_shaper_t _hb_ot_shaper_myanmar =
 348 {
 349   collect_features_myanmar,
 350   nullptr, /* override_features */
 351   nullptr, /* data_create */
 352   nullptr, /* data_destroy */
 353   nullptr, /* preprocess_text */
 354   nullptr, /* postprocess_glyphs */
 355   nullptr, /* decompose */
 356   nullptr, /* compose */
 357   setup_masks_myanmar,
 358   nullptr, /* reorder_marks */
 359   HB_TAG_NONE, /* gpos_tag */
 360   HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS_NO_SHORT_CIRCUIT,
 361   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_EARLY,
 362   false, /* fallback_position */
 363 };
 364
 365
 366 #ifndef HB_NO_OT_SHAPER_MYANMAR_ZAWGYI
 367 /* Ugly Zawgyi encoding.
 368  * Disable all auto processing.
 369  * https://github.com/harfbuzz/harfbuzz/issues/1162 */
 370 const hb_ot_shaper_t _hb_ot_shaper_myanmar_zawgyi =
 371 {
 372   nullptr, /* collect_features */
 373   nullptr, /* override_features */
 374   nullptr, /* data_create */
 375   nullptr, /* data_destroy */
 376   nullptr, /* preprocess_text */
 377   nullptr, /* postprocess_glyphs */
 378   nullptr, /* decompose */
 379   nullptr, /* compose */
 380   nullptr, /* setup_masks */
 381   nullptr, /* reorder_marks */
 382   HB_TAG_NONE, /* gpos_tag */
 383   HB_OT_SHAPE_NORMALIZATION_MODE_NONE,
 384   HB_OT_SHAPE_ZERO_WIDTH_MARKS_NONE,
 385   false, /* fallback_position */
 386 };
 387 #endif
 388
 389
 390 #endif