1 /*
2 * Copyright © 2010,2012 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-private.hh"
28 #include "hb-debug.hh"
29 #include "hb-ot-shape-complex-arabic-private.hh"
30 #include "hb-ot-shape-private.hh"
31
32
33 /* buffer var allocations */
34 #define arabic_shaping_action() complex_var_u8_0() /* arabic shaping action */
35
36 #define HB_BUFFER_SCRATCH_FLAG_ARABIC_HAS_STCH HB_BUFFER_SCRATCH_FLAG_COMPLEX0
37
38 /* See:
39 * https://github.com/harfbuzz/harfbuzz/commit/6e6f82b6f3dde0fc6c3c7d991d9ec6cfff57823d#commitcomment-14248516 */
40 #define HB_ARABIC_GENERAL_CATEGORY_IS_WORD(gen_cat) \
41 (FLAG_UNSAFE (gen_cat) & \
42 (FLAG (HB_UNICODE_GENERAL_CATEGORY_UNASSIGNED) | \
43 FLAG (HB_UNICODE_GENERAL_CATEGORY_PRIVATE_USE) | \
44 /*FLAG (HB_UNICODE_GENERAL_CATEGORY_LOWERCASE_LETTER) |*/ \
45 FLAG (HB_UNICODE_GENERAL_CATEGORY_MODIFIER_LETTER) | \
46 FLAG (HB_UNICODE_GENERAL_CATEGORY_OTHER_LETTER) | \
47 /*FLAG (HB_UNICODE_GENERAL_CATEGORY_TITLECASE_LETTER) |*/ \
48 /*FLAG (HB_UNICODE_GENERAL_CATEGORY_UPPERCASE_LETTER) |*/ \
49 FLAG (HB_UNICODE_GENERAL_CATEGORY_SPACING_MARK) | \
50 FLAG (HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK) | \
51 FLAG (HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK) | \
52 FLAG (HB_UNICODE_GENERAL_CATEGORY_DECIMAL_NUMBER) | \
53 FLAG (HB_UNICODE_GENERAL_CATEGORY_LETTER_NUMBER) | \
54 FLAG (HB_UNICODE_GENERAL_CATEGORY_OTHER_NUMBER) | \
55 FLAG (HB_UNICODE_GENERAL_CATEGORY_CURRENCY_SYMBOL) | \
56 FLAG (HB_UNICODE_GENERAL_CATEGORY_MODIFIER_SYMBOL) | \
57 FLAG (HB_UNICODE_GENERAL_CATEGORY_MATH_SYMBOL) | \
58 FLAG (HB_UNICODE_GENERAL_CATEGORY_OTHER_SYMBOL)))
59
60
61 /*
62 * Joining types:
63 */
64
65 /*
66 * Bits used in the joining tables
67 */
68 enum hb_arabic_joining_type_t {
69 JOINING_TYPE_U = 0,
70 JOINING_TYPE_L = 1,
71 JOINING_TYPE_R = 2,
72 JOINING_TYPE_D = 3,
73 JOINING_TYPE_C = JOINING_TYPE_D,
74 JOINING_GROUP_ALAPH = 4,
75 JOINING_GROUP_DALATH_RISH = 5,
76 NUM_STATE_MACHINE_COLS = 6,
77
78 JOINING_TYPE_T = 7,
79 JOINING_TYPE_X = 8 /* means: use general-category to choose between U or T. */
80 };
81
82 #include "hb-ot-shape-complex-arabic-table.hh"
83
84 static unsigned int get_joining_type (hb_codepoint_t u, hb_unicode_general_category_t gen_cat)
85 {
86 unsigned int j_type = joining_type(u);
87 if (likely (j_type != JOINING_TYPE_X))
88 return j_type;
89
90 return (FLAG_UNSAFE(gen_cat) &
91 (FLAG(HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK) |
92 FLAG(HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK) |
93 FLAG(HB_UNICODE_GENERAL_CATEGORY_FORMAT))
94 ) ? JOINING_TYPE_T : JOINING_TYPE_U;
95 }
96
97 #define FEATURE_IS_SYRIAC(tag) hb_in_range<unsigned char> ((unsigned char) (tag), '2', '3')
98
99 static const hb_tag_t arabic_features[] =
100 {
101 HB_TAG('i','s','o','l'),
102 HB_TAG('f','i','n','a'),
103 HB_TAG('f','i','n','2'),
104 HB_TAG('f','i','n','3'),
105 HB_TAG('m','e','d','i'),
106 HB_TAG('m','e','d','2'),
107 HB_TAG('i','n','i','t'),
108 HB_TAG_NONE
109 };
110
111
112 /* Same order as the feature array */
113 enum arabic_action_t {
114 ISOL,
115 FINA,
116 FIN2,
117 FIN3,
118 MEDI,
119 MED2,
120 INIT,
121
122 NONE,
123
124 ARABIC_NUM_FEATURES = NONE,
125
126 /* We abuse the same byte for other things... */
127 STCH_FIXED,
128 STCH_REPEATING,
129 };
130
131 static const struct arabic_state_table_entry {
132 uint8_t prev_action;
133 uint8_t curr_action;
134 uint16_t next_state;
135 } arabic_state_table[][NUM_STATE_MACHINE_COLS] =
136 {
137 /* jt_U, jt_L, jt_R, jt_D, jg_ALAPH, jg_DALATH_RISH */
138
139 /* State 0: prev was U, not willing to join. */
140 { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,6}, },
141
142 /* State 1: prev was R or ISOL/ALAPH, not willing to join. */
143 { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN2,5}, {NONE,ISOL,6}, },
144
145 /* State 2: prev was D/L in ISOL form, willing to join. */
146 { {NONE,NONE,0}, {NONE,ISOL,2}, {INIT,FINA,1}, {INIT,FINA,3}, {INIT,FINA,4}, {INIT,FINA,6}, },
147
148 /* State 3: prev was D in FINA form, willing to join. */
149 { {NONE,NONE,0}, {NONE,ISOL,2}, {MEDI,FINA,1}, {MEDI,FINA,3}, {MEDI,FINA,4}, {MEDI,FINA,6}, },
150
151 /* State 4: prev was FINA ALAPH, not willing to join. */
152 { {NONE,NONE,0}, {NONE,ISOL,2}, {MED2,ISOL,1}, {MED2,ISOL,2}, {MED2,FIN2,5}, {MED2,ISOL,6}, },
153
154 /* State 5: prev was FIN2/FIN3 ALAPH, not willing to join. */
155 { {NONE,NONE,0}, {NONE,ISOL,2}, {ISOL,ISOL,1}, {ISOL,ISOL,2}, {ISOL,FIN2,5}, {ISOL,ISOL,6}, },
156
157 /* State 6: prev was DALATH/RISH, not willing to join. */
158 { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN3,5}, {NONE,ISOL,6}, }
159 };
160
161
162 static void
163 nuke_joiners (const hb_ot_shape_plan_t *plan,
164 hb_font_t *font,
165 hb_buffer_t *buffer);
166
167 static void
168 arabic_fallback_shape (const hb_ot_shape_plan_t *plan,
169 hb_font_t *font,
170 hb_buffer_t *buffer);
171
172 static void
173 record_stch (const hb_ot_shape_plan_t *plan,
174 hb_font_t *font,
175 hb_buffer_t *buffer);
176
177 static void
178 collect_features_arabic (hb_ot_shape_planner_t *plan)
179 {
180 hb_ot_map_builder_t *map = &plan->map;
181
182 /* We apply features according to the Arabic spec, with pauses
183 * in between most.
184 *
185 * The pause between init/medi/... and rlig is required. See eg:
186 * https://bugzilla.mozilla.org/show_bug.cgi?id=644184
187 *
188 * The pauses between init/medi/... themselves are not necessarily
189 * needed as only one of those features is applied to any character.
190 * The only difference it makes is when fonts have contextual
191 * substitutions. We now follow the order of the spec, which makes
192 * for better experience if that's what Uniscribe is doing.
193 *
194 * At least for Arabic, looks like Uniscribe has a pause between
195 * rlig and calt. Otherwise the IranNastaliq's ALLAH ligature won't
196 * work. However, testing shows that rlig and calt are applied
197 * together for Mongolian in Uniscribe. As such, we only add a
198 * pause for Arabic, not other scripts.
199 *
200 * A pause after calt is required to make KFGQPC Uthmanic Script HAFS
201 * work correctly. See https://github.com/harfbuzz/harfbuzz/issues/505
202 */
203
204 map->add_gsub_pause (nuke_joiners);
205
206 map->add_global_bool_feature (HB_TAG('s','t','c','h'));
207 map->add_gsub_pause (record_stch);
208
209 map->add_global_bool_feature (HB_TAG('c','c','m','p'));
210 map->add_global_bool_feature (HB_TAG('l','o','c','l'));
211
212 map->add_gsub_pause (nullptr);
213
214 for (unsigned int i = 0; i < ARABIC_NUM_FEATURES; i++)
215 {
216 bool has_fallback = plan->props.script == HB_SCRIPT_ARABIC && !FEATURE_IS_SYRIAC (arabic_features[i]);
217 map->add_feature (arabic_features[i], 1, has_fallback ? F_HAS_FALLBACK : F_NONE);
218 map->add_gsub_pause (nullptr);
219 }
220
221 map->add_feature (HB_TAG('r','l','i','g'), 1, F_GLOBAL|F_HAS_FALLBACK);
222 if (plan->props.script == HB_SCRIPT_ARABIC)
223 map->add_gsub_pause (arabic_fallback_shape);
224
225 /* No pause after rclt. See 98460779bae19e4d64d29461ff154b3527bf8420. */
226 map->add_global_bool_feature (HB_TAG('r','c','l','t'));
227 map->add_global_bool_feature (HB_TAG('c','a','l','t'));
228 map->add_gsub_pause (nullptr);
229
230 /* The spec includes 'cswh'. Earlier versions of Windows
231 * used to enable this by default, but testing suggests
232 * that Windows 8 and later do not enable it by default,
233 * and spec now says 'Off by default'.
234 * We disabled this in ae23c24c32.
235 * Note that IranNastaliq uses this feature extensively
236 * to fixup broken glyph sequences. Oh well...
237 * Test case: U+0643,U+0640,U+0631. */
238 //map->add_global_bool_feature (HB_TAG('c','s','w','h'));
239 map->add_global_bool_feature (HB_TAG('m','s','e','t'));
240 }
241
242 #include "hb-ot-shape-complex-arabic-fallback.hh"
243
244 struct arabic_shape_plan_t
245 {
246 ASSERT_POD ();
247
248 /* The "+ 1" in the next array is to accommodate for the "NONE" command,
249 * which is not an OpenType feature, but this simplifies the code by not
250 * having to do a "if (... < NONE) ..." and just rely on the fact that
251 * mask_array[NONE] == 0. */
252 hb_mask_t mask_array[ARABIC_NUM_FEATURES + 1];
253
254 arabic_fallback_plan_t *fallback_plan;
255
256 unsigned int do_fallback : 1;
257 unsigned int has_stch : 1;
258 };
259
260 void *
261 data_create_arabic (const hb_ot_shape_plan_t *plan)
262 {
263 arabic_shape_plan_t *arabic_plan = (arabic_shape_plan_t *) calloc (1, sizeof (arabic_shape_plan_t));
264 if (unlikely (!arabic_plan))
265 return nullptr;
266
267 arabic_plan->do_fallback = plan->props.script == HB_SCRIPT_ARABIC;
268 arabic_plan->has_stch = !!plan->map.get_1_mask (HB_TAG ('s','t','c','h'));
269 for (unsigned int i = 0; i < ARABIC_NUM_FEATURES; i++) {
270 arabic_plan->mask_array[i] = plan->map.get_1_mask (arabic_features[i]);
271 arabic_plan->do_fallback = arabic_plan->do_fallback &&
272 (FEATURE_IS_SYRIAC (arabic_features[i]) ||
273 plan->map.needs_fallback (arabic_features[i]));
274 }
275
276 return arabic_plan;
277 }
278
279 void
280 data_destroy_arabic (void *data)
281 {
282 arabic_shape_plan_t *arabic_plan = (arabic_shape_plan_t *) data;
283
284 arabic_fallback_plan_destroy (arabic_plan->fallback_plan);
285
286 free (data);
287 }
288
289 static void
290 arabic_joining (hb_buffer_t *buffer)
291 {
292 unsigned int count = buffer->len;
293 hb_glyph_info_t *info = buffer->info;
294 unsigned int prev = (unsigned int) -1, state = 0;
295
296 /* Check pre-context */
297 for (unsigned int i = 0; i < buffer->context_len[0]; i++)
298 {
299 unsigned int this_type = get_joining_type (buffer->context[0][i], buffer->unicode->general_category (buffer->context[0][i]));
300
301 if (unlikely (this_type == JOINING_TYPE_T))
302 continue;
303
304 const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
305 state = entry->next_state;
306 break;
307 }
308
309 for (unsigned int i = 0; i < count; i++)
310 {
311 unsigned int this_type = get_joining_type (info[i].codepoint, _hb_glyph_info_get_general_category (&info[i]));
312
313 if (unlikely (this_type == JOINING_TYPE_T)) {
314 info[i].arabic_shaping_action() = NONE;
315 continue;
316 }
317
318 const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
319
320 if (entry->prev_action != NONE && prev != (unsigned int) -1)
321 {
322 info[prev].arabic_shaping_action() = entry->prev_action;
323 buffer->unsafe_to_break (prev, i + 1);
324 }
325
326 info[i].arabic_shaping_action() = entry->curr_action;
327
328 prev = i;
329 state = entry->next_state;
330 }
331
332 for (unsigned int i = 0; i < buffer->context_len[1]; i++)
333 {
334 unsigned int this_type = get_joining_type (buffer->context[1][i], buffer->unicode->general_category (buffer->context[1][i]));
335
336 if (unlikely (this_type == JOINING_TYPE_T))
337 continue;
338
339 const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
340 if (entry->prev_action != NONE && prev != (unsigned int) -1)
341 info[prev].arabic_shaping_action() = entry->prev_action;
342 break;
343 }
344 }
345
346 static void
347 mongolian_variation_selectors (hb_buffer_t *buffer)
348 {
349 /* Copy arabic_shaping_action() from base to Mongolian variation selectors. */
350 unsigned int count = buffer->len;
351 hb_glyph_info_t *info = buffer->info;
352 for (unsigned int i = 1; i < count; i++)
353 if (unlikely (hb_in_range<hb_codepoint_t> (info[i].codepoint, 0x180Bu, 0x180Du)))
354 info[i].arabic_shaping_action() = info[i - 1].arabic_shaping_action();
355 }
356
357 void
358 setup_masks_arabic_plan (const arabic_shape_plan_t *arabic_plan,
359 hb_buffer_t *buffer,
360 hb_script_t script)
361 {
362 HB_BUFFER_ALLOCATE_VAR (buffer, arabic_shaping_action);
363
364 arabic_joining (buffer);
365 if (script == HB_SCRIPT_MONGOLIAN)
366 mongolian_variation_selectors (buffer);
367
368 unsigned int count = buffer->len;
369 hb_glyph_info_t *info = buffer->info;
370 for (unsigned int i = 0; i < count; i++)
371 info[i].mask |= arabic_plan->mask_array[info[i].arabic_shaping_action()];
372 }
373
374 static void
375 setup_masks_arabic (const hb_ot_shape_plan_t *plan,
376 hb_buffer_t *buffer,
377 hb_font_t *font HB_UNUSED)
378 {
379 const arabic_shape_plan_t *arabic_plan = (const arabic_shape_plan_t *) plan->data;
380 setup_masks_arabic_plan (arabic_plan, buffer, plan->props.script);
381 }
382
383
384 static void
385 nuke_joiners (const hb_ot_shape_plan_t *plan HB_UNUSED,
386 hb_font_t *font HB_UNUSED,
387 hb_buffer_t *buffer)
388 {
389 unsigned int count = buffer->len;
390 hb_glyph_info_t *info = buffer->info;
391 for (unsigned int i = 0; i < count; i++)
392 if (_hb_glyph_info_is_zwj (&info[i]))
393 _hb_glyph_info_flip_joiners (&info[i]);
394 }
395
396 static void
397 arabic_fallback_shape (const hb_ot_shape_plan_t *plan,
398 hb_font_t *font,
399 hb_buffer_t *buffer)
400 {
401 const arabic_shape_plan_t *arabic_plan = (const arabic_shape_plan_t *) plan->data;
402
403 if (!arabic_plan->do_fallback)
404 return;
405
406 retry:
407 arabic_fallback_plan_t *fallback_plan = (arabic_fallback_plan_t *) hb_atomic_ptr_get (&arabic_plan->fallback_plan);
408 if (unlikely (!fallback_plan))
409 {
410 /* This sucks. We need a font to build the fallback plan... */
411 fallback_plan = arabic_fallback_plan_create (plan, font);
412 if (unlikely (!hb_atomic_ptr_cmpexch (&(const_cast<arabic_shape_plan_t *> (arabic_plan))->fallback_plan, nullptr, fallback_plan))) {
413 arabic_fallback_plan_destroy (fallback_plan);
414 goto retry;
415 }
416 }
417
418 arabic_fallback_plan_shape (fallback_plan, font, buffer);
419 }
420
421 /*
422 * Stretch feature: "stch".
423 * See example here:
424 * https://docs.microsoft.com/en-us/typography/script-development/syriac
425 * We implement this in a generic way, such that the Arabic subtending
426 * marks can use it as well.
427 */
428
429 static void
430 record_stch (const hb_ot_shape_plan_t *plan,
431 hb_font_t *font,
432 hb_buffer_t *buffer)
433 {
434 const arabic_shape_plan_t *arabic_plan = (const arabic_shape_plan_t *) plan->data;
435 if (!arabic_plan->has_stch)
436 return;
437
438 /* 'stch' feature was just applied. Look for anything that multiplied,
439 * and record it for stch treatment later. Note that rtlm, frac, etc
440 * are applied before stch, but we assume that they didn't result in
441 * anything multiplying into 5 pieces, so it's safe-ish... */
442
443 unsigned int count = buffer->len;
444 hb_glyph_info_t *info = buffer->info;
445 for (unsigned int i = 0; i < count; i++)
446 if (unlikely (_hb_glyph_info_multiplied (&info[i])))
447 {
448 unsigned int comp = _hb_glyph_info_get_lig_comp (&info[i]);
449 info[i].arabic_shaping_action() = comp % 2 ? STCH_REPEATING : STCH_FIXED;
450 buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_ARABIC_HAS_STCH;
451 }
452 }
453
454 static void
455 apply_stch (const hb_ot_shape_plan_t *plan,
456 hb_buffer_t *buffer,
457 hb_font_t *font)
458 {
459 if (likely (!(buffer->scratch_flags & HB_BUFFER_SCRATCH_FLAG_ARABIC_HAS_STCH)))
460 return;
461
462 /* The Arabic shaper currently always processes in RTL mode, so we should
463 * stretch / position the stretched pieces to the left / preceding glyphs. */
464
465 /* We do a two pass implementation:
466 * First pass calculates the exact number of extra glyphs we need,
467 * We then enlarge buffer to have that much room,
468 * Second pass applies the stretch, copying things to the end of buffer.
469 */
470
471 int sign = font->x_scale < 0 ? -1 : +1;
472 unsigned int extra_glyphs_needed = 0; // Set during MEASURE, used during CUT
473 typedef enum { MEASURE, CUT } step_t;
474
475 for (step_t step = MEASURE; step <= CUT; step = (step_t) (step + 1))
476 {
477 unsigned int count = buffer->len;
478 hb_glyph_info_t *info = buffer->info;
479 hb_glyph_position_t *pos = buffer->pos;
480 unsigned int new_len = count + extra_glyphs_needed; // write head during CUT
481 unsigned int j = new_len;
482 for (unsigned int i = count; i; i--)
483 {
484 if (!hb_in_range<unsigned> (info[i - 1].arabic_shaping_action(), STCH_FIXED, STCH_REPEATING))
485 {
486 if (step == CUT)
487 {
488 --j;
489 info[j] = info[i - 1];
490 pos[j] = pos[i - 1];
491 }
492 continue;
493 }
494
495 /* Yay, justification! */
496
497 hb_position_t w_total = 0; // Total to be filled
498 hb_position_t w_fixed = 0; // Sum of fixed tiles
499 hb_position_t w_repeating = 0; // Sum of repeating tiles
500 int n_fixed = 0;
501 int n_repeating = 0;
502
503 unsigned int end = i;
504 while (i &&
505 hb_in_range<unsigned> (info[i - 1].arabic_shaping_action(), STCH_FIXED, STCH_REPEATING))
506 {
507 i--;
508 hb_position_t width = font->get_glyph_h_advance (info[i].codepoint);
509 if (info[i].arabic_shaping_action() == STCH_FIXED)
510 {
511 w_fixed += width;
512 n_fixed++;
513 }
514 else
515 {
516 w_repeating += width;
517 n_repeating++;
518 }
519 }
520 unsigned int start = i;
521 unsigned int context = i;
522 while (context &&
523 !hb_in_range<unsigned> (info[context - 1].arabic_shaping_action(), STCH_FIXED, STCH_REPEATING) &&
524 (_hb_glyph_info_is_default_ignorable (&info[context - 1]) ||
525 HB_ARABIC_GENERAL_CATEGORY_IS_WORD (_hb_glyph_info_get_general_category (&info[context - 1]))))
526 {
527 context--;
528 w_total += pos[context].x_advance;
529 }
530 i++; // Don't touch i again.
531
532 DEBUG_MSG (ARABIC, nullptr, "%s stretch at (%d,%d,%d)",
533 step == MEASURE ? "measuring" : "cutting", context, start, end);
534 DEBUG_MSG (ARABIC, nullptr, "rest of word: count=%d width %d", start - context, w_total);
535 DEBUG_MSG (ARABIC, nullptr, "fixed tiles: count=%d width=%d", n_fixed, w_fixed);
536 DEBUG_MSG (ARABIC, nullptr, "repeating tiles: count=%d width=%d", n_repeating, w_repeating);
537
538 /* Number of additional times to repeat each repeating tile. */
539 int n_copies = 0;
540
541 hb_position_t w_remaining = w_total - w_fixed;
542 if (sign * w_remaining > sign * w_repeating && sign * w_repeating > 0)
543 n_copies = (sign * w_remaining) / (sign * w_repeating) - 1;
544
545 /* See if we can improve the fit by adding an extra repeat and squeezing them together a bit. */
546 hb_position_t extra_repeat_overlap = 0;
547 hb_position_t shortfall = sign * w_remaining - sign * w_repeating * (n_copies + 1);
548 if (shortfall > 0 && n_repeating > 0)
549 {
550 ++n_copies;
551 hb_position_t excess = (n_copies + 1) * sign * w_repeating - sign * w_remaining;
552 if (excess > 0)
553 extra_repeat_overlap = excess / (n_copies * n_repeating);
554 }
555
556 if (step == MEASURE)
557 {
558 extra_glyphs_needed += n_copies * n_repeating;
559 DEBUG_MSG (ARABIC, nullptr, "will add extra %d copies of repeating tiles", n_copies);
560 }
561 else
562 {
563 buffer->unsafe_to_break (context, end);
564 hb_position_t x_offset = 0;
565 for (unsigned int k = end; k > start; k--)
566 {
567 hb_position_t width = font->get_glyph_h_advance (info[k - 1].codepoint);
568
569 unsigned int repeat = 1;
570 if (info[k - 1].arabic_shaping_action() == STCH_REPEATING)
571 repeat += n_copies;
572
573 DEBUG_MSG (ARABIC, nullptr, "appending %d copies of glyph %d; j=%d",
574 repeat, info[k - 1].codepoint, j);
575 for (unsigned int n = 0; n < repeat; n++)
576 {
577 x_offset -= width;
578 if (n > 0)
579 x_offset += extra_repeat_overlap;
580 pos[k - 1].x_offset = x_offset;
581 /* Append copy. */
582 --j;
583 info[j] = info[k - 1];
584 pos[j] = pos[k - 1];
585 }
586 }
587 }
588 }
589
590 if (step == MEASURE)
591 {
592 if (unlikely (!buffer->ensure (count + extra_glyphs_needed)))
593 break;
594 }
595 else
596 {
597 assert (j == 0);
598 buffer->len = new_len;
599 }
600 }
601 }
602
603
604 static void
605 postprocess_glyphs_arabic (const hb_ot_shape_plan_t *plan,
606 hb_buffer_t *buffer,
607 hb_font_t *font)
608 {
609 apply_stch (plan, buffer, font);
610
611 HB_BUFFER_DEALLOCATE_VAR (buffer, arabic_shaping_action);
612 }
613
614 /* https://unicode.org/reports/tr53/tr53-1.pdf */
615
616 static hb_codepoint_t
617 modifier_combining_marks[] =
618 {
619 0x0654u, /* ARABIC HAMZA ABOVE */
620 0x0655u, /* ARABIC HAMZA BELOW */
621 0x0658u, /* ARABIC MARK NOON GHUNNA */
622 0x06DCu, /* ARABIC SMALL HIGH SEEN */
623 0x06E3u, /* ARABIC SMALL LOW SEEN */
624 0x06E7u, /* ARABIC SMALL HIGH YEH */
625 0x06E8u, /* ARABIC SMALL HIGH NOON */
626 0x08F3u, /* ARABIC SMALL HIGH WAW */
627 };
628
629 static inline bool
630 info_is_mcm (const hb_glyph_info_t &info)
631 {
632 hb_codepoint_t u = info.codepoint;
633 for (unsigned int i = 0; i < ARRAY_LENGTH (modifier_combining_marks); i++)
634 if (u == modifier_combining_marks[i])
635 return true;
636 return false;
637 }
638
639 static void
640 reorder_marks_arabic (const hb_ot_shape_plan_t *plan,
641 hb_buffer_t *buffer,
642 unsigned int start,
643 unsigned int end)
644 {
645 hb_glyph_info_t *info = buffer->info;
646
647 DEBUG_MSG (ARABIC, buffer, "Reordering marks from %d to %d", start, end);
648
649 unsigned int i = start;
650 for (unsigned int cc = 220; cc <= 230; cc += 10)
651 {
652 DEBUG_MSG (ARABIC, buffer, "Looking for %d's starting at %d", cc, i);
653 while (i < end && info_cc(info[i]) < cc)
654 i++;
655 DEBUG_MSG (ARABIC, buffer, "Looking for %d's stopped at %d", cc, i);
656
657 if (i == end)
658 break;
659
660 if (info_cc(info[i]) > cc)
661 continue;
662
663 unsigned int j = i;
664 while (j < end && info_cc(info[j]) == cc && info_is_mcm (info[j]))
665 j++;
666
667 if (i == j)
668 continue;
669
670 DEBUG_MSG (ARABIC, buffer, "Found %d's from %d to %d", cc, i, j);
671
672 /* Shift it! */
673 DEBUG_MSG (ARABIC, buffer, "Shifting %d's: %d %d", cc, i, j);
674 hb_glyph_info_t temp[HB_OT_SHAPE_COMPLEX_MAX_COMBINING_MARKS];
675 assert (j - i <= ARRAY_LENGTH (temp));
676 buffer->merge_clusters (start, j);
677 memmove (temp, &info[i], (j - i) * sizeof (hb_glyph_info_t));
678 memmove (&info[start + j - i], &info[start], (i - start) * sizeof (hb_glyph_info_t));
679 memmove (&info[start], temp, (j - i) * sizeof (hb_glyph_info_t));
680
681 /* Renumber CC such that the reordered sequence is still sorted.
682 * 22 and 26 are chosen because they are smaller than all Arabic categories,
683 * and are folded back to 220/230 respectively during fallback mark positioning.
684 *
685 * We do this because the CGJ-handling logic in the normalizer relies on
686 * mark sequences having an increasing order even after this reordering.
687 * https://github.com/harfbuzz/harfbuzz/issues/554
688 * This, however, does break some obscure sequences, where the normalizer
689 * might compose a sequence that it should not. For example, in the seequence
690 * ALEF, HAMZAH, MADDAH, we should NOT try to compose ALEF+MADDAH, but with this
691 * renumbering, we will.
692 */
693 unsigned int new_start = start + j - i;
694 unsigned int new_cc = cc == 220 ? HB_MODIFIED_COMBINING_CLASS_CCC22 : HB_MODIFIED_COMBINING_CLASS_CCC26;
695 while (start < new_start)
696 {
697 _hb_glyph_info_set_modified_combining_class (&info[start], new_cc);
698 start++;
699 }
700
701 i = j;
702 }
703 }
704
705 const hb_ot_complex_shaper_t _hb_ot_complex_shaper_arabic =
706 {
707 collect_features_arabic,
708 nullptr, /* override_features */
709 data_create_arabic,
710 data_destroy_arabic,
711 nullptr, /* preprocess_text */
712 postprocess_glyphs_arabic,
713 HB_OT_SHAPE_NORMALIZATION_MODE_DEFAULT,
714 nullptr, /* decompose */
715 nullptr, /* compose */
716 setup_masks_arabic,
717 nullptr, /* disable_otl */
718 reorder_marks_arabic,
719 HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_LATE,
720 true, /* fallback_position */
721 };