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_SAFE (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 0x1039u:
203 cat = (indic_category_t) OT_H;
204 break;
205
206 case 0x103Au:
207 cat = (indic_category_t) OT_As;
208 break;
209
210 case 0x1041u: case 0x1042u: case 0x1043u: case 0x1044u:
211 case 0x1045u: case 0x1046u: case 0x1047u: case 0x1048u:
212 case 0x1049u: case 0x1090u: case 0x1091u: case 0x1092u:
213 case 0x1093u: case 0x1094u: case 0x1095u: case 0x1096u:
214 case 0x1097u: case 0x1098u: case 0x1099u:
215 cat = (indic_category_t) OT_D;
216 break;
217
218 case 0x1040u:
219 cat = (indic_category_t) OT_D; /* XXX The spec says D0, but Uniscribe doesn't seem to do. */
220 break;
221
222 case 0x103Eu: case 0x1060u:
223 cat = (indic_category_t) OT_MH;
224 break;
225
226 case 0x103Cu:
227 cat = (indic_category_t) OT_MR;
228 break;
229
230 case 0x103Du: case 0x1082u:
231 cat = (indic_category_t) OT_MW;
232 break;
233
234 case 0x103Bu: case 0x105Eu: case 0x105Fu:
235 cat = (indic_category_t) OT_MY;
236 break;
237
238 case 0x1063u: case 0x1064u: case 0x1069u: case 0x106Au:
239 case 0x106Bu: case 0x106Cu: case 0x106Du: case 0xAA7Bu:
240 cat = (indic_category_t) OT_PT;
241 break;
242
243 case 0x1038u: case 0x1087u: case 0x1088u: case 0x1089u:
244 case 0x108Au: case 0x108Bu: case 0x108Cu: case 0x108Du:
245 case 0x108Fu: case 0x109Au: case 0x109Bu: case 0x109Cu:
246 cat = (indic_category_t) OT_SM;
247 break;
248
249 case 0x104Au: case 0x104Bu:
250 cat = (indic_category_t) OT_P;
251 break;
252
253 case 0xAA74u: case 0xAA75u: case 0xAA76u:
254 /* https://github.com/roozbehp/unicode-data/issues/3 */
255 cat = (indic_category_t) OT_C;
256 break;
257 }
258
259 if (cat == OT_M)
260 {
261 switch ((int) pos)
262 {
263 case POS_PRE_C: cat = (indic_category_t) OT_VPre;
264 pos = POS_PRE_M; break;
265 case POS_ABOVE_C: cat = (indic_category_t) OT_VAbv; break;
266 case POS_BELOW_C: cat = (indic_category_t) OT_VBlw; break;
267 case POS_POST_C: cat = (indic_category_t) OT_VPst; break;
268 }
269 }
270
271 info.myanmar_category() = (myanmar_category_t) cat;
272 info.myanmar_position() = pos;
273 }
274
275
276
277 static void
278 setup_masks_myanmar (const hb_ot_shape_plan_t *plan HB_UNUSED,
279 hb_buffer_t *buffer,
280 hb_font_t *font HB_UNUSED)
281 {
282 HB_BUFFER_ALLOCATE_VAR (buffer, myanmar_category);
283 HB_BUFFER_ALLOCATE_VAR (buffer, myanmar_position);
284
285 /* We cannot setup masks here. We save information about characters
286 * and setup masks later on in a pause-callback. */
287
288 unsigned int count = buffer->len;
289 hb_glyph_info_t *info = buffer->info;
290 for (unsigned int i = 0; i < count; i++)
291 set_myanmar_properties (info[i]);
292 }
293
294 static void
295 setup_syllables (const hb_ot_shape_plan_t *plan HB_UNUSED,
296 hb_font_t *font HB_UNUSED,
297 hb_buffer_t *buffer)
298 {
299 find_syllables (buffer);
300 }
301
302 static int
303 compare_myanmar_order (const hb_glyph_info_t *pa, const hb_glyph_info_t *pb)
304 {
305 int a = pa->myanmar_position();
306 int b = pb->myanmar_position();
307
308 return a < b ? -1 : a == b ? 0 : +1;
309 }
310
311
312 /* Rules from:
313 * http://www.microsoft.com/typography/OpenTypeDev/myanmar/intro.htm */
314
315 static void
316 initial_reordering_consonant_syllable (hb_buffer_t *buffer,
317 unsigned int start, unsigned int end)
318 {
319 hb_glyph_info_t *info = buffer->info;
320
321 unsigned int base = end;
322 bool has_reph = false;
323
324 {
325 unsigned int limit = start;
326 if (start + 3 <= end &&
327 info[start ].myanmar_category() == OT_Ra &&
328 info[start+1].myanmar_category() == OT_As &&
329 info[start+2].myanmar_category() == OT_H)
330 {
331 limit += 3;
332 base = start;
333 has_reph = true;
334 }
335
336 {
337 if (!has_reph)
338 base = limit;
339
340 for (unsigned int i = limit; i < end; i++)
341 if (is_consonant (info[i]))
342 {
343 base = i;
344 break;
345 }
346 }
347 }
348
349 /* Reorder! */
350 {
351 unsigned int i = start;
352 for (; i < start + (has_reph ? 3 : 0); i++)
353 info[i].myanmar_position() = POS_AFTER_MAIN;
354 for (; i < base; i++)
355 info[i].myanmar_position() = POS_PRE_C;
356 if (i < end)
357 {
358 info[i].myanmar_position() = POS_BASE_C;
359 i++;
360 }
361 indic_position_t pos = POS_AFTER_MAIN;
362 /* The following loop may be ugly, but it implements all of
363 * Myanmar reordering! */
364 for (; i < end; i++)
365 {
366 if (info[i].myanmar_category() == OT_MR) /* Pre-base reordering */
367 {
368 info[i].myanmar_position() = POS_PRE_C;
369 continue;
370 }
371 if (info[i].myanmar_position() < POS_BASE_C) /* Left matra */
372 {
373 continue;
374 }
375
376 if (pos == POS_AFTER_MAIN && info[i].myanmar_category() == OT_VBlw)
377 {
378 pos = POS_BELOW_C;
379 info[i].myanmar_position() = pos;
380 continue;
381 }
382
383 if (pos == POS_BELOW_C && info[i].myanmar_category() == OT_A)
384 {
385 info[i].myanmar_position() = POS_BEFORE_SUB;
386 continue;
387 }
388 if (pos == POS_BELOW_C && info[i].myanmar_category() == OT_VBlw)
389 {
390 info[i].myanmar_position() = pos;
391 continue;
392 }
393 if (pos == POS_BELOW_C && info[i].myanmar_category() != OT_A)
394 {
395 pos = POS_AFTER_SUB;
396 info[i].myanmar_position() = pos;
397 continue;
398 }
399 info[i].myanmar_position() = pos;
400 }
401 }
402
403 /* Sit tight, rock 'n roll! */
404 buffer->sort (start, end, compare_myanmar_order);
405 }
406
407 static void
408 initial_reordering_syllable (const hb_ot_shape_plan_t *plan,
409 hb_face_t *face,
410 hb_buffer_t *buffer,
411 unsigned int start, unsigned int end)
412 {
413 syllable_type_t syllable_type = (syllable_type_t) (buffer->info[start].syllable() & 0x0F);
414 switch (syllable_type) {
415
416 case broken_cluster: /* We already inserted dotted-circles, so just call the consonant_syllable. */
417 case consonant_syllable:
418 initial_reordering_consonant_syllable (buffer, start, end);
419 break;
420
421 case punctuation_cluster:
422 case non_myanmar_cluster:
423 break;
424 }
425 }
426
427 static inline void
428 insert_dotted_circles (const hb_ot_shape_plan_t *plan HB_UNUSED,
429 hb_font_t *font,
430 hb_buffer_t *buffer)
431 {
432 /* Note: This loop is extra overhead, but should not be measurable. */
433 bool has_broken_syllables = false;
434 unsigned int count = buffer->len;
435 hb_glyph_info_t *info = buffer->info;
436 for (unsigned int i = 0; i < count; i++)
437 if ((info[i].syllable() & 0x0F) == broken_cluster)
438 {
439 has_broken_syllables = true;
440 break;
441 }
442 if (likely (!has_broken_syllables))
443 return;
444
445
446 hb_codepoint_t dottedcircle_glyph;
447 if (!font->get_nominal_glyph (0x25CCu, &dottedcircle_glyph))
448 return;
449
450 hb_glyph_info_t dottedcircle = {0};
451 dottedcircle.codepoint = 0x25CCu;
452 set_myanmar_properties (dottedcircle);
453 dottedcircle.codepoint = dottedcircle_glyph;
454
455 buffer->clear_output ();
456
457 buffer->idx = 0;
458 unsigned int last_syllable = 0;
459 while (buffer->idx < buffer->len && !buffer->in_error)
460 {
461 unsigned int syllable = buffer->cur().syllable();
462 syllable_type_t syllable_type = (syllable_type_t) (syllable & 0x0F);
463 if (unlikely (last_syllable != syllable && syllable_type == broken_cluster))
464 {
465 last_syllable = syllable;
466
467 hb_glyph_info_t ginfo = dottedcircle;
468 ginfo.cluster = buffer->cur().cluster;
469 ginfo.mask = buffer->cur().mask;
470 ginfo.syllable() = buffer->cur().syllable();
471
472 buffer->output_info (ginfo);
473 }
474 else
475 buffer->next_glyph ();
476 }
477
478 buffer->swap_buffers ();
479 }
480
481 static void
482 initial_reordering (const hb_ot_shape_plan_t *plan,
483 hb_font_t *font,
484 hb_buffer_t *buffer)
485 {
486 insert_dotted_circles (plan, font, buffer);
487
488 foreach_syllable (buffer, start, end)
489 initial_reordering_syllable (plan, font->face, buffer, start, end);
490 }
491
492 static void
493 final_reordering (const hb_ot_shape_plan_t *plan,
494 hb_font_t *font HB_UNUSED,
495 hb_buffer_t *buffer)
496 {
497 hb_glyph_info_t *info = buffer->info;
498 unsigned int count = buffer->len;
499
500 /* Zero syllables now... */
501 for (unsigned int i = 0; i < count; i++)
502 info[i].syllable() = 0;
503
504 HB_BUFFER_DEALLOCATE_VAR (buffer, myanmar_category);
505 HB_BUFFER_DEALLOCATE_VAR (buffer, myanmar_position);
506 }
507
508
509 /* Uniscribe seems to have a shaper for 'mymr' that is like the
510 * generic shaper, except that it zeros mark advances GDEF_LATE. */
511 const hb_ot_complex_shaper_t _hb_ot_complex_shaper_myanmar_old =
512 {
513 "default",
514 NULL, /* collect_features */
515 NULL, /* override_features */
516 NULL, /* data_create */
517 NULL, /* data_destroy */
518 NULL, /* preprocess_text */
519 NULL, /* postprocess_glyphs */
520 HB_OT_SHAPE_NORMALIZATION_MODE_DEFAULT,
521 NULL, /* decompose */
522 NULL, /* compose */
523 NULL, /* setup_masks */
524 NULL, /* disable_otl */
525 HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_LATE,
526 true, /* fallback_position */
527 };
528
529 const hb_ot_complex_shaper_t _hb_ot_complex_shaper_myanmar =
530 {
531 "myanmar",
532 collect_features_myanmar,
533 override_features_myanmar,
534 NULL, /* data_create */
535 NULL, /* data_destroy */
536 NULL, /* preprocess_text */
537 NULL, /* postprocess_glyphs */
538 HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS_NO_SHORT_CIRCUIT,
539 NULL, /* decompose */
540 NULL, /* compose */
541 setup_masks_myanmar,
542 NULL, /* disable_otl */
543 HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_EARLY,
544 false, /* fallback_position */
545 };