1 /*
2 * Copyright © 2011,2012 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-fallback-private.hh"
28 #include "hb-ot-layout-gsubgpos-private.hh"
29
30 static unsigned int
31 recategorize_combining_class (hb_codepoint_t u,
32 unsigned int klass)
33 {
34 if (klass >= 200)
35 return klass;
36
37 /* Thai / Lao need some per-character work. */
38 if ((u & ~0xFF) == 0x0E00u)
39 {
40 if (unlikely (klass == 0))
41 {
42 switch (u)
43 {
44 case 0x0E31u:
45 case 0x0E34u:
46 case 0x0E35u:
47 case 0x0E36u:
48 case 0x0E37u:
49 case 0x0E47u:
50 case 0x0E4Cu:
51 case 0x0E4Du:
52 case 0x0E4Eu:
53 klass = HB_UNICODE_COMBINING_CLASS_ABOVE_RIGHT;
54 break;
55
56 case 0x0EB1u:
57 case 0x0EB4u:
58 case 0x0EB5u:
59 case 0x0EB6u:
60 case 0x0EB7u:
61 case 0x0EBBu:
62 case 0x0ECCu:
63 case 0x0ECDu:
64 klass = HB_UNICODE_COMBINING_CLASS_ABOVE;
65 break;
66
67 case 0x0EBCu:
68 klass = HB_UNICODE_COMBINING_CLASS_BELOW;
69 break;
70 }
71 } else {
72 /* Thai virama is below-right */
73 if (u == 0x0E3Au)
74 klass = HB_UNICODE_COMBINING_CLASS_BELOW_RIGHT;
75 }
76 }
77
78 switch (klass)
79 {
80
81 /* Hebrew */
82
83 case HB_MODIFIED_COMBINING_CLASS_CCC10: /* sheva */
84 case HB_MODIFIED_COMBINING_CLASS_CCC11: /* hataf segol */
85 case HB_MODIFIED_COMBINING_CLASS_CCC12: /* hataf patah */
86 case HB_MODIFIED_COMBINING_CLASS_CCC13: /* hataf qamats */
87 case HB_MODIFIED_COMBINING_CLASS_CCC14: /* hiriq */
88 case HB_MODIFIED_COMBINING_CLASS_CCC15: /* tsere */
89 case HB_MODIFIED_COMBINING_CLASS_CCC16: /* segol */
90 case HB_MODIFIED_COMBINING_CLASS_CCC17: /* patah */
91 case HB_MODIFIED_COMBINING_CLASS_CCC18: /* qamats */
92 case HB_MODIFIED_COMBINING_CLASS_CCC20: /* qubuts */
93 case HB_MODIFIED_COMBINING_CLASS_CCC22: /* meteg */
94 return HB_UNICODE_COMBINING_CLASS_BELOW;
95
96 case HB_MODIFIED_COMBINING_CLASS_CCC23: /* rafe */
97 return HB_UNICODE_COMBINING_CLASS_ATTACHED_ABOVE;
98
99 case HB_MODIFIED_COMBINING_CLASS_CCC24: /* shin dot */
100 return HB_UNICODE_COMBINING_CLASS_ABOVE_RIGHT;
101
102 case HB_MODIFIED_COMBINING_CLASS_CCC25: /* sin dot */
103 case HB_MODIFIED_COMBINING_CLASS_CCC19: /* holam */
104 return HB_UNICODE_COMBINING_CLASS_ABOVE_LEFT;
105
106 case HB_MODIFIED_COMBINING_CLASS_CCC26: /* point varika */
107 return HB_UNICODE_COMBINING_CLASS_ABOVE;
108
109 case HB_MODIFIED_COMBINING_CLASS_CCC21: /* dagesh */
110 break;
111
112
113 /* Arabic and Syriac */
114
115 case HB_MODIFIED_COMBINING_CLASS_CCC27: /* fathatan */
116 case HB_MODIFIED_COMBINING_CLASS_CCC28: /* dammatan */
117 case HB_MODIFIED_COMBINING_CLASS_CCC30: /* fatha */
118 case HB_MODIFIED_COMBINING_CLASS_CCC31: /* damma */
119 case HB_MODIFIED_COMBINING_CLASS_CCC33: /* shadda */
120 case HB_MODIFIED_COMBINING_CLASS_CCC34: /* sukun */
121 case HB_MODIFIED_COMBINING_CLASS_CCC35: /* superscript alef */
122 case HB_MODIFIED_COMBINING_CLASS_CCC36: /* superscript alaph */
123 return HB_UNICODE_COMBINING_CLASS_ABOVE;
124
125 case HB_MODIFIED_COMBINING_CLASS_CCC29: /* kasratan */
126 case HB_MODIFIED_COMBINING_CLASS_CCC32: /* kasra */
127 return HB_UNICODE_COMBINING_CLASS_BELOW;
128
129
130 /* Thai */
131
132 case HB_MODIFIED_COMBINING_CLASS_CCC103: /* sara u / sara uu */
133 return HB_UNICODE_COMBINING_CLASS_BELOW_RIGHT;
134
135 case HB_MODIFIED_COMBINING_CLASS_CCC107: /* mai */
136 return HB_UNICODE_COMBINING_CLASS_ABOVE_RIGHT;
137
138
139 /* Lao */
140
141 case HB_MODIFIED_COMBINING_CLASS_CCC118: /* sign u / sign uu */
142 return HB_UNICODE_COMBINING_CLASS_BELOW;
143
144 case HB_MODIFIED_COMBINING_CLASS_CCC122: /* mai */
145 return HB_UNICODE_COMBINING_CLASS_ABOVE;
146
147
148 /* Tibetan */
149
150 case HB_MODIFIED_COMBINING_CLASS_CCC129: /* sign aa */
151 return HB_UNICODE_COMBINING_CLASS_BELOW;
152
153 case HB_MODIFIED_COMBINING_CLASS_CCC130: /* sign i*/
154 return HB_UNICODE_COMBINING_CLASS_ABOVE;
155
156 case HB_MODIFIED_COMBINING_CLASS_CCC132: /* sign u */
157 return HB_UNICODE_COMBINING_CLASS_BELOW;
158
159 }
160
161 return klass;
162 }
163
164 void
165 _hb_ot_shape_fallback_position_recategorize_marks (const hb_ot_shape_plan_t *plan HB_UNUSED,
166 hb_font_t *font HB_UNUSED,
167 hb_buffer_t *buffer)
168 {
169 unsigned int count = buffer->len;
170 hb_glyph_info_t *info = buffer->info;
171 for (unsigned int i = 0; i < count; i++)
172 if (_hb_glyph_info_get_general_category (&info[i]) == HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK) {
173 unsigned int combining_class = _hb_glyph_info_get_modified_combining_class (&info[i]);
174 combining_class = recategorize_combining_class (info[i].codepoint, combining_class);
175 _hb_glyph_info_set_modified_combining_class (&info[i], combining_class);
176 }
177 }
178
179
180 static void
181 zero_mark_advances (hb_buffer_t *buffer,
182 unsigned int start,
183 unsigned int end)
184 {
185 hb_glyph_info_t *info = buffer->info;
186 for (unsigned int i = start; i < end; i++)
187 if (_hb_glyph_info_get_general_category (&info[i]) == HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK)
188 {
189 buffer->pos[i].x_advance = 0;
190 buffer->pos[i].y_advance = 0;
191 }
192 }
193
194 static inline void
195 position_mark (const hb_ot_shape_plan_t *plan,
196 hb_font_t *font,
197 hb_buffer_t *buffer,
198 hb_glyph_extents_t &base_extents,
199 unsigned int i,
200 unsigned int combining_class)
201 {
202 hb_glyph_extents_t mark_extents;
203 if (!font->get_glyph_extents (buffer->info[i].codepoint,
204 &mark_extents))
205 return;
206
207 hb_position_t y_gap = font->y_scale / 16;
208
209 hb_glyph_position_t &pos = buffer->pos[i];
210 pos.x_offset = pos.y_offset = 0;
211
212
213 /* We don't position LEFT and RIGHT marks. */
214
215 /* X positioning */
216 switch (combining_class)
217 {
218 case HB_UNICODE_COMBINING_CLASS_DOUBLE_BELOW:
219 case HB_UNICODE_COMBINING_CLASS_DOUBLE_ABOVE:
220 if (buffer->props.direction == HB_DIRECTION_LTR) {
221 pos.x_offset += base_extents.x_bearing - mark_extents.width / 2 - mark_extents.x_bearing;
222 break;
223 } else if (buffer->props.direction == HB_DIRECTION_RTL) {
224 pos.x_offset += base_extents.x_bearing + base_extents.width - mark_extents.width / 2 - mark_extents.x_bearing;
225 break;
226 }
227 HB_FALLTHROUGH;
228
229 default:
230 case HB_UNICODE_COMBINING_CLASS_ATTACHED_BELOW:
231 case HB_UNICODE_COMBINING_CLASS_ATTACHED_ABOVE:
232 case HB_UNICODE_COMBINING_CLASS_BELOW:
233 case HB_UNICODE_COMBINING_CLASS_ABOVE:
234 /* Center align. */
235 pos.x_offset += base_extents.x_bearing + (base_extents.width - mark_extents.width) / 2 - mark_extents.x_bearing;
236 break;
237
238 case HB_UNICODE_COMBINING_CLASS_ATTACHED_BELOW_LEFT:
239 case HB_UNICODE_COMBINING_CLASS_BELOW_LEFT:
240 case HB_UNICODE_COMBINING_CLASS_ABOVE_LEFT:
241 /* Left align. */
242 pos.x_offset += base_extents.x_bearing - mark_extents.x_bearing;
243 break;
244
245 case HB_UNICODE_COMBINING_CLASS_ATTACHED_ABOVE_RIGHT:
246 case HB_UNICODE_COMBINING_CLASS_BELOW_RIGHT:
247 case HB_UNICODE_COMBINING_CLASS_ABOVE_RIGHT:
248 /* Right align. */
249 pos.x_offset += base_extents.x_bearing + base_extents.width - mark_extents.width - mark_extents.x_bearing;
250 break;
251 }
252
253 /* Y positioning */
254 switch (combining_class)
255 {
256 case HB_UNICODE_COMBINING_CLASS_DOUBLE_BELOW:
257 case HB_UNICODE_COMBINING_CLASS_BELOW_LEFT:
258 case HB_UNICODE_COMBINING_CLASS_BELOW:
259 case HB_UNICODE_COMBINING_CLASS_BELOW_RIGHT:
260 /* Add gap, fall-through. */
261 base_extents.height -= y_gap;
262 HB_FALLTHROUGH;
263
264 case HB_UNICODE_COMBINING_CLASS_ATTACHED_BELOW_LEFT:
265 case HB_UNICODE_COMBINING_CLASS_ATTACHED_BELOW:
266 pos.y_offset = base_extents.y_bearing + base_extents.height - mark_extents.y_bearing;
267 /* Never shift up "below" marks. */
268 if ((y_gap > 0) == (pos.y_offset > 0))
269 {
270 base_extents.height -= pos.y_offset;
271 pos.y_offset = 0;
272 }
273 base_extents.height += mark_extents.height;
274 break;
275
276 case HB_UNICODE_COMBINING_CLASS_DOUBLE_ABOVE:
277 case HB_UNICODE_COMBINING_CLASS_ABOVE_LEFT:
278 case HB_UNICODE_COMBINING_CLASS_ABOVE:
279 case HB_UNICODE_COMBINING_CLASS_ABOVE_RIGHT:
280 /* Add gap, fall-through. */
281 base_extents.y_bearing += y_gap;
282 base_extents.height -= y_gap;
283 HB_FALLTHROUGH;
284
285 case HB_UNICODE_COMBINING_CLASS_ATTACHED_ABOVE:
286 case HB_UNICODE_COMBINING_CLASS_ATTACHED_ABOVE_RIGHT:
287 pos.y_offset = base_extents.y_bearing - (mark_extents.y_bearing + mark_extents.height);
288 /* Don't shift down "above" marks too much. */
289 if ((y_gap > 0) != (pos.y_offset > 0))
290 {
291 unsigned int correction = -pos.y_offset / 2;
292 base_extents.y_bearing += correction;
293 base_extents.height -= correction;
294 pos.y_offset += correction;
295 }
296 base_extents.y_bearing -= mark_extents.height;
297 base_extents.height += mark_extents.height;
298 break;
299 }
300 }
301
302 static inline void
303 position_around_base (const hb_ot_shape_plan_t *plan,
304 hb_font_t *font,
305 hb_buffer_t *buffer,
306 unsigned int base,
307 unsigned int end)
308 {
309 hb_direction_t horiz_dir = HB_DIRECTION_INVALID;
310
311 buffer->unsafe_to_break (base, end);
312
313 hb_glyph_extents_t base_extents;
314 if (!font->get_glyph_extents (buffer->info[base].codepoint,
315 &base_extents))
316 {
317 /* If extents don't work, zero marks and go home. */
318 zero_mark_advances (buffer, base + 1, end);
319 return;
320 }
321 base_extents.x_bearing += buffer->pos[base].x_offset;
322 base_extents.y_bearing += buffer->pos[base].y_offset;
323
324 unsigned int lig_id = _hb_glyph_info_get_lig_id (&buffer->info[base]);
325 unsigned int num_lig_components = _hb_glyph_info_get_lig_num_comps (&buffer->info[base]);
326
327 hb_position_t x_offset = 0, y_offset = 0;
328 if (HB_DIRECTION_IS_FORWARD (buffer->props.direction)) {
329 x_offset -= buffer->pos[base].x_advance;
330 y_offset -= buffer->pos[base].y_advance;
331 }
332
333 hb_glyph_extents_t component_extents = base_extents;
334 unsigned int last_lig_component = (unsigned int) -1;
335 unsigned int last_combining_class = 255;
336 hb_glyph_extents_t cluster_extents = base_extents; /* Initialization is just to shut gcc up. */
337 hb_glyph_info_t *info = buffer->info;
338 for (unsigned int i = base + 1; i < end; i++)
339 if (_hb_glyph_info_get_modified_combining_class (&info[i]))
340 {
341 if (num_lig_components > 1) {
342 unsigned int this_lig_id = _hb_glyph_info_get_lig_id (&info[i]);
343 unsigned int this_lig_component = _hb_glyph_info_get_lig_comp (&info[i]) - 1;
344 /* Conditions for attaching to the last component. */
345 if (!lig_id || lig_id != this_lig_id || this_lig_component >= num_lig_components)
346 this_lig_component = num_lig_components - 1;
347 if (last_lig_component != this_lig_component)
348 {
349 last_lig_component = this_lig_component;
350 last_combining_class = 255;
351 component_extents = base_extents;
352 if (unlikely (horiz_dir == HB_DIRECTION_INVALID)) {
353 if (HB_DIRECTION_IS_HORIZONTAL (plan->props.direction))
354 horiz_dir = plan->props.direction;
355 else
356 horiz_dir = hb_script_get_horizontal_direction (plan->props.script);
357 }
358 if (horiz_dir == HB_DIRECTION_LTR)
359 component_extents.x_bearing += (this_lig_component * component_extents.width) / num_lig_components;
360 else
361 component_extents.x_bearing += ((num_lig_components - 1 - this_lig_component) * component_extents.width) / num_lig_components;
362 component_extents.width /= num_lig_components;
363 }
364 }
365
366 unsigned int this_combining_class = _hb_glyph_info_get_modified_combining_class (&info[i]);
367 if (last_combining_class != this_combining_class)
368 {
369 last_combining_class = this_combining_class;
370 cluster_extents = component_extents;
371 }
372
373 position_mark (plan, font, buffer, cluster_extents, i, this_combining_class);
374
375 buffer->pos[i].x_advance = 0;
376 buffer->pos[i].y_advance = 0;
377 buffer->pos[i].x_offset += x_offset;
378 buffer->pos[i].y_offset += y_offset;
379
380 } else {
381 if (HB_DIRECTION_IS_FORWARD (buffer->props.direction)) {
382 x_offset -= buffer->pos[i].x_advance;
383 y_offset -= buffer->pos[i].y_advance;
384 } else {
385 x_offset += buffer->pos[i].x_advance;
386 y_offset += buffer->pos[i].y_advance;
387 }
388 }
389 }
390
391 static inline void
392 position_cluster (const hb_ot_shape_plan_t *plan,
393 hb_font_t *font,
394 hb_buffer_t *buffer,
395 unsigned int start,
396 unsigned int end)
397 {
398 if (end - start < 2)
399 return;
400
401 /* Find the base glyph */
402 hb_glyph_info_t *info = buffer->info;
403 for (unsigned int i = start; i < end; i++)
404 if (!HB_UNICODE_GENERAL_CATEGORY_IS_MARK (_hb_glyph_info_get_general_category (&info[i])))
405 {
406 /* Find mark glyphs */
407 unsigned int j;
408 for (j = i + 1; j < end; j++)
409 if (!HB_UNICODE_GENERAL_CATEGORY_IS_MARK (_hb_glyph_info_get_general_category (&info[j])))
410 break;
411
412 position_around_base (plan, font, buffer, i, j);
413
414 i = j - 1;
415 }
416 }
417
418 void
419 _hb_ot_shape_fallback_position (const hb_ot_shape_plan_t *plan,
420 hb_font_t *font,
421 hb_buffer_t *buffer)
422 {
423 _hb_buffer_assert_gsubgpos_vars (buffer);
424
425 unsigned int start = 0;
426 unsigned int count = buffer->len;
427 hb_glyph_info_t *info = buffer->info;
428 for (unsigned int i = 1; i < count; i++)
429 if (likely (!HB_UNICODE_GENERAL_CATEGORY_IS_MARK (_hb_glyph_info_get_general_category (&info[i])))) {
430 position_cluster (plan, font, buffer, start, i);
431 start = i;
432 }
433 position_cluster (plan, font, buffer, start, count);
434 }
435
436
437 /* Performs old-style TrueType kerning. */
438 void
439 _hb_ot_shape_fallback_kern (const hb_ot_shape_plan_t *plan,
440 hb_font_t *font,
441 hb_buffer_t *buffer)
442 {
443 if (!plan->has_kern) return;
444
445 OT::hb_apply_context_t c (1, font, buffer);
446 c.set_lookup_mask (plan->kern_mask);
447 c.set_lookup_props (OT::LookupFlag::IgnoreMarks);
448 OT::hb_apply_context_t::skipping_iterator_t &skippy_iter = c.iter_input;
449 skippy_iter.init (&c);
450
451 unsigned int count = buffer->len;
452 hb_glyph_info_t *info = buffer->info;
453 hb_glyph_position_t *pos = buffer->pos;
454 for (unsigned int idx = 0; idx < count;)
455 {
456 skippy_iter.reset (idx, 1);
457 if (!skippy_iter.next ())
458 {
459 idx++;
460 continue;
461 }
462
463 hb_position_t x_kern, y_kern;
464 font->get_glyph_kerning_for_direction (info[idx].codepoint,
465 info[skippy_iter.idx].codepoint,
466 buffer->props.direction,
467 &x_kern, &y_kern);
468
469 if (x_kern)
470 {
471 hb_position_t kern1 = x_kern >> 1;
472 hb_position_t kern2 = x_kern - kern1;
473 pos[idx].x_advance += kern1;
474 pos[skippy_iter.idx].x_advance += kern2;
475 pos[skippy_iter.idx].x_offset += kern2;
476 }
477
478 if (y_kern)
479 {
480 hb_position_t kern1 = y_kern >> 1;
481 hb_position_t kern2 = y_kern - kern1;
482 pos[idx].y_advance += kern1;
483 pos[skippy_iter.idx].y_advance += kern2;
484 pos[skippy_iter.idx].y_offset += kern2;
485 }
486
487 idx = skippy_iter.idx;
488 }
489 }
490
491
492 /* Adjusts width of various spaces. */
493 void
494 _hb_ot_shape_fallback_spaces (const hb_ot_shape_plan_t *plan,
495 hb_font_t *font,
496 hb_buffer_t *buffer)
497 {
498 if (!HB_DIRECTION_IS_HORIZONTAL (buffer->props.direction))
499 return;
500
501 hb_glyph_info_t *info = buffer->info;
502 hb_glyph_position_t *pos = buffer->pos;
503 unsigned int count = buffer->len;
504 for (unsigned int i = 0; i < count; i++)
505 if (_hb_glyph_info_is_unicode_space (&info[i]) && !_hb_glyph_info_ligated (&info[i]))
506 {
507 hb_unicode_funcs_t::space_t space_type = _hb_glyph_info_get_unicode_space_fallback_type (&info[i]);
508 hb_codepoint_t glyph;
509 typedef hb_unicode_funcs_t t;
510 switch (space_type)
511 {
512 case t::NOT_SPACE: /* Shouldn't happen. */
513 case t::SPACE:
514 break;
515
516 case t::SPACE_EM:
517 case t::SPACE_EM_2:
518 case t::SPACE_EM_3:
519 case t::SPACE_EM_4:
520 case t::SPACE_EM_5:
521 case t::SPACE_EM_6:
522 case t::SPACE_EM_16:
523 pos[i].x_advance = (font->x_scale + ((int) space_type)/2) / (int) space_type;
524 break;
525
526 case t::SPACE_4_EM_18:
527 pos[i].x_advance = font->x_scale * 4 / 18;
528 break;
529
530 case t::SPACE_FIGURE:
531 for (char u = '0'; u <= '9'; u++)
532 if (font->get_nominal_glyph (u, &glyph))
533 {
534 pos[i].x_advance = font->get_glyph_h_advance (glyph);
535 break;
536 }
537 break;
538
539 case t::SPACE_PUNCTUATION:
540 if (font->get_nominal_glyph ('.', &glyph))
541 pos[i].x_advance = font->get_glyph_h_advance (glyph);
542 else if (font->get_nominal_glyph (',', &glyph))
543 pos[i].x_advance = font->get_glyph_h_advance (glyph);
544 break;
545
546 case t::SPACE_NARROW:
547 /* Half-space?
548 * Unicode doc http://www.unicode.org/charts/PDF/U2000.pdf says ~1/4 or 1/5 of EM.
549 * However, in my testing, many fonts have their regular space being about that
550 * size. To me, a percentage of the space width makes more sense. Half is as
551 * good as any. */
552 pos[i].x_advance /= 2;
553 break;
554 }
555 }
556 }