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.hh"
28
29 #ifndef HB_NO_OT_SHAPE
30
31 #include "hb-ot-shape-complex-arabic.hh"
32 #include "hb-ot-shape.hh"
33
34
35 /* buffer var allocations */
36 #define arabic_shaping_action() complex_var_u8_auxiliary() /* arabic shaping action */
37
38 #define HB_BUFFER_SCRATCH_FLAG_ARABIC_HAS_STCH HB_BUFFER_SCRATCH_FLAG_COMPLEX0
39
40 /* See:
41 * https://github.com/harfbuzz/harfbuzz/commit/6e6f82b6f3dde0fc6c3c7d991d9ec6cfff57823d#commitcomment-14248516 */
42 #define HB_ARABIC_GENERAL_CATEGORY_IS_WORD(gen_cat) \
43 (FLAG_UNSAFE (gen_cat) & \
44 (FLAG (HB_UNICODE_GENERAL_CATEGORY_UNASSIGNED) | \
45 FLAG (HB_UNICODE_GENERAL_CATEGORY_PRIVATE_USE) | \
46 /*FLAG (HB_UNICODE_GENERAL_CATEGORY_LOWERCASE_LETTER) |*/ \
47 FLAG (HB_UNICODE_GENERAL_CATEGORY_MODIFIER_LETTER) | \
48 FLAG (HB_UNICODE_GENERAL_CATEGORY_OTHER_LETTER) | \
49 /*FLAG (HB_UNICODE_GENERAL_CATEGORY_TITLECASE_LETTER) |*/ \
50 /*FLAG (HB_UNICODE_GENERAL_CATEGORY_UPPERCASE_LETTER) |*/ \
51 FLAG (HB_UNICODE_GENERAL_CATEGORY_SPACING_MARK) | \
52 FLAG (HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK) | \
53 FLAG (HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK) | \
54 FLAG (HB_UNICODE_GENERAL_CATEGORY_DECIMAL_NUMBER) | \
55 FLAG (HB_UNICODE_GENERAL_CATEGORY_LETTER_NUMBER) | \
56 FLAG (HB_UNICODE_GENERAL_CATEGORY_OTHER_NUMBER) | \
57 FLAG (HB_UNICODE_GENERAL_CATEGORY_CURRENCY_SYMBOL) | \
58 FLAG (HB_UNICODE_GENERAL_CATEGORY_MODIFIER_SYMBOL) | \
59 FLAG (HB_UNICODE_GENERAL_CATEGORY_MATH_SYMBOL) | \
60 FLAG (HB_UNICODE_GENERAL_CATEGORY_OTHER_SYMBOL)))
61
62
63 /*
64 * Joining types:
65 */
66
67 /*
68 * Bits used in the joining tables
69 */
70 enum hb_arabic_joining_type_t {
71 JOINING_TYPE_U = 0,
72 JOINING_TYPE_L = 1,
73 JOINING_TYPE_R = 2,
74 JOINING_TYPE_D = 3,
75 JOINING_TYPE_C = JOINING_TYPE_D,
76 JOINING_GROUP_ALAPH = 4,
77 JOINING_GROUP_DALATH_RISH = 5,
78 NUM_STATE_MACHINE_COLS = 6,
79
80 JOINING_TYPE_T = 7,
81 JOINING_TYPE_X = 8 /* means: use general-category to choose between U or T. */
82 };
83
84 #include "hb-ot-shape-complex-arabic-table.hh"
85
get_joining_type(hb_codepoint_t u,hb_unicode_general_category_t gen_cat)86 static unsigned int get_joining_type (hb_codepoint_t u, hb_unicode_general_category_t gen_cat)
87 {
88 unsigned int j_type = joining_type(u);
89 if (likely (j_type != JOINING_TYPE_X))
90 return j_type;
91
92 return (FLAG_UNSAFE(gen_cat) &
93 (FLAG(HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK) |
94 FLAG(HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK) |
95 FLAG(HB_UNICODE_GENERAL_CATEGORY_FORMAT))
96 ) ? JOINING_TYPE_T : JOINING_TYPE_U;
97 }
98
99 #define FEATURE_IS_SYRIAC(tag) hb_in_range<unsigned char> ((unsigned char) (tag), '2', '3')
100
101 static const hb_tag_t arabic_features[] =
102 {
103 HB_TAG('i','s','o','l'),
104 HB_TAG('f','i','n','a'),
105 HB_TAG('f','i','n','2'),
106 HB_TAG('f','i','n','3'),
107 HB_TAG('m','e','d','i'),
108 HB_TAG('m','e','d','2'),
109 HB_TAG('i','n','i','t'),
110 HB_TAG_NONE
111 };
112
113
114 /* Same order as the feature array */
115 enum arabic_action_t {
116 ISOL,
117 FINA,
118 FIN2,
119 FIN3,
120 MEDI,
121 MED2,
122 INIT,
123
124 NONE,
125
126 ARABIC_NUM_FEATURES = NONE,
127
128 /* We abuse the same byte for other things... */
129 STCH_FIXED,
130 STCH_REPEATING,
131 };
132
133 static const struct arabic_state_table_entry {
134 uint8_t prev_action;
135 uint8_t curr_action;
136 uint16_t next_state;
137 } arabic_state_table[][NUM_STATE_MACHINE_COLS] =
138 {
139 /* jt_U, jt_L, jt_R, jt_D, jg_ALAPH, jg_DALATH_RISH */
140
141 /* State 0: prev was U, not willing to join. */
142 { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,6}, },
143
144 /* State 1: prev was R or ISOL/ALAPH, not willing to join. */
145 { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN2,5}, {NONE,ISOL,6}, },
146
147 /* State 2: prev was D/L in ISOL form, willing to join. */
148 { {NONE,NONE,0}, {NONE,ISOL,2}, {INIT,FINA,1}, {INIT,FINA,3}, {INIT,FINA,4}, {INIT,FINA,6}, },
149
150 /* State 3: prev was D in FINA form, willing to join. */
151 { {NONE,NONE,0}, {NONE,ISOL,2}, {MEDI,FINA,1}, {MEDI,FINA,3}, {MEDI,FINA,4}, {MEDI,FINA,6}, },
152
153 /* State 4: prev was FINA ALAPH, not willing to join. */
154 { {NONE,NONE,0}, {NONE,ISOL,2}, {MED2,ISOL,1}, {MED2,ISOL,2}, {MED2,FIN2,5}, {MED2,ISOL,6}, },
155
156 /* State 5: prev was FIN2/FIN3 ALAPH, not willing to join. */
157 { {NONE,NONE,0}, {NONE,ISOL,2}, {ISOL,ISOL,1}, {ISOL,ISOL,2}, {ISOL,FIN2,5}, {ISOL,ISOL,6}, },
158
159 /* State 6: prev was DALATH/RISH, not willing to join. */
160 { {NONE,NONE,0}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN3,5}, {NONE,ISOL,6}, }
161 };
162
163
164 static void
165 arabic_fallback_shape (const hb_ot_shape_plan_t *plan,
166 hb_font_t *font,
167 hb_buffer_t *buffer);
168
169 static void
170 record_stch (const hb_ot_shape_plan_t *plan,
171 hb_font_t *font,
172 hb_buffer_t *buffer);
173
174 static void
collect_features_arabic(hb_ot_shape_planner_t * plan)175 collect_features_arabic (hb_ot_shape_planner_t *plan)
176 {
177 hb_ot_map_builder_t *map = &plan->map;
178
179 /* We apply features according to the Arabic spec, with pauses
180 * in between most.
181 *
182 * The pause between init/medi/... and rlig is required. See eg:
183 * https://bugzilla.mozilla.org/show_bug.cgi?id=644184
184 *
185 * The pauses between init/medi/... themselves are not necessarily
186 * needed as only one of those features is applied to any character.
187 * The only difference it makes is when fonts have contextual
188 * substitutions. We now follow the order of the spec, which makes
189 * for better experience if that's what Uniscribe is doing.
190 *
191 * At least for Arabic, looks like Uniscribe has a pause between
192 * rlig and calt. Otherwise the IranNastaliq's ALLAH ligature won't
193 * work. However, testing shows that rlig and calt are applied
194 * together for Mongolian in Uniscribe. As such, we only add a
195 * pause for Arabic, not other scripts.
196 *
197 * A pause after calt is required to make KFGQPC Uthmanic Script HAFS
198 * work correctly. See https://github.com/harfbuzz/harfbuzz/issues/505
199 */
200
201
202 map->enable_feature (HB_TAG('s','t','c','h'));
203 map->add_gsub_pause (record_stch);
204
205 map->enable_feature (HB_TAG('c','c','m','p'));
206 map->enable_feature (HB_TAG('l','o','c','l'));
207
208 map->add_gsub_pause (nullptr);
209
210 for (unsigned int i = 0; i < ARABIC_NUM_FEATURES; i++)
211 {
212 bool has_fallback = plan->props.script == HB_SCRIPT_ARABIC && !FEATURE_IS_SYRIAC (arabic_features[i]);
213 map->add_feature (arabic_features[i], has_fallback ? F_HAS_FALLBACK : F_NONE);
214 map->add_gsub_pause (nullptr);
215 }
216
217 /* Normally, Unicode says a ZWNJ means "don't ligate". In Arabic script
218 * however, it says a ZWJ should also mean "don't ligate". So we run
219 * the main ligating features as MANUAL_ZWJ. */
220
221 map->enable_feature (HB_TAG('r','l','i','g'), F_MANUAL_ZWJ | F_HAS_FALLBACK);
222
223 if (plan->props.script == HB_SCRIPT_ARABIC)
224 map->add_gsub_pause (arabic_fallback_shape);
225
226 /* No pause after rclt. See 98460779bae19e4d64d29461ff154b3527bf8420. */
227 map->enable_feature (HB_TAG('r','c','l','t'), F_MANUAL_ZWJ);
228 map->enable_feature (HB_TAG('c','a','l','t'), F_MANUAL_ZWJ);
229 map->add_gsub_pause (nullptr);
230
231 /* The spec includes 'cswh'. Earlier versions of Windows
232 * used to enable this by default, but testing suggests
233 * that Windows 8 and later do not enable it by default,
234 * and spec now says 'Off by default'.
235 * We disabled this in ae23c24c32.
236 * Note that IranNastaliq uses this feature extensively
237 * to fixup broken glyph sequences. Oh well...
238 * Test case: U+0643,U+0640,U+0631. */
239 //map->enable_feature (HB_TAG('c','s','w','h'));
240 map->enable_feature (HB_TAG('m','s','e','t'));
241 }
242
243 #include "hb-ot-shape-complex-arabic-fallback.hh"
244
245 struct arabic_shape_plan_t
246 {
247 /* The "+ 1" in the next array is to accommodate for the "NONE" command,
248 * which is not an OpenType feature, but this simplifies the code by not
249 * having to do a "if (... < NONE) ..." and just rely on the fact that
250 * mask_array[NONE] == 0. */
251 hb_mask_t mask_array[ARABIC_NUM_FEATURES + 1];
252
253 hb_atomic_ptr_t<arabic_fallback_plan_t> fallback_plan;
254
255 unsigned int do_fallback : 1;
256 unsigned int has_stch : 1;
257 };
258
259 void *
data_create_arabic(const hb_ot_shape_plan_t * plan)260 data_create_arabic (const hb_ot_shape_plan_t *plan)
261 {
262 arabic_shape_plan_t *arabic_plan = (arabic_shape_plan_t *) hb_calloc (1, sizeof (arabic_shape_plan_t));
263 if (unlikely (!arabic_plan))
264 return nullptr;
265
266 arabic_plan->do_fallback = plan->props.script == HB_SCRIPT_ARABIC;
267 arabic_plan->has_stch = !!plan->map.get_1_mask (HB_TAG ('s','t','c','h'));
268 for (unsigned int i = 0; i < ARABIC_NUM_FEATURES; i++) {
269 arabic_plan->mask_array[i] = plan->map.get_1_mask (arabic_features[i]);
270 arabic_plan->do_fallback = arabic_plan->do_fallback &&
271 (FEATURE_IS_SYRIAC (arabic_features[i]) ||
272 plan->map.needs_fallback (arabic_features[i]));
273 }
274
275 return arabic_plan;
276 }
277
278 void
data_destroy_arabic(void * data)279 data_destroy_arabic (void *data)
280 {
281 arabic_shape_plan_t *arabic_plan = (arabic_shape_plan_t *) data;
282
283 arabic_fallback_plan_destroy (arabic_plan->fallback_plan);
284
285 hb_free (data);
286 }
287
288 static void
arabic_joining(hb_buffer_t * buffer)289 arabic_joining (hb_buffer_t *buffer)
290 {
291 unsigned int count = buffer->len;
292 hb_glyph_info_t *info = buffer->info;
293 unsigned int prev = UINT_MAX, state = 0;
294
295 /* Check pre-context */
296 for (unsigned int i = 0; i < buffer->context_len[0]; i++)
297 {
298 unsigned int this_type = get_joining_type (buffer->context[0][i], buffer->unicode->general_category (buffer->context[0][i]));
299
300 if (unlikely (this_type == JOINING_TYPE_T))
301 continue;
302
303 const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
304 state = entry->next_state;
305 break;
306 }
307
308 for (unsigned int i = 0; i < count; i++)
309 {
310 unsigned int this_type = get_joining_type (info[i].codepoint, _hb_glyph_info_get_general_category (&info[i]));
311
312 if (unlikely (this_type == JOINING_TYPE_T)) {
313 info[i].arabic_shaping_action() = NONE;
314 continue;
315 }
316
317 const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
318
319 if (entry->prev_action != NONE && prev != UINT_MAX)
320 {
321 info[prev].arabic_shaping_action() = entry->prev_action;
322 buffer->unsafe_to_break (prev, i + 1);
323 }
324 else
325 {
326 if (prev == UINT_MAX)
327 {
328 if (this_type >= JOINING_TYPE_R)
329 buffer->unsafe_to_concat_from_outbuffer (0, i + 1);
330 }
331 else
332 {
333 if (this_type >= JOINING_TYPE_R ||
334 (2 <= state && state <= 5) /* States that have a possible prev_action. */)
335 buffer->unsafe_to_concat (prev, i + 1);
336 }
337 }
338
339 info[i].arabic_shaping_action() = entry->curr_action;
340
341 prev = i;
342 state = entry->next_state;
343 }
344
345 for (unsigned int i = 0; i < buffer->context_len[1]; i++)
346 {
347 unsigned int this_type = get_joining_type (buffer->context[1][i], buffer->unicode->general_category (buffer->context[1][i]));
348
349 if (unlikely (this_type == JOINING_TYPE_T))
350 continue;
351
352 const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
353 if (entry->prev_action != NONE && prev != UINT_MAX)
354 {
355 info[prev].arabic_shaping_action() = entry->prev_action;
356 buffer->unsafe_to_break (prev, buffer->len);
357 }
358 else if (2 <= state && state <= 5) /* States that have a possible prev_action. */
359 {
360 buffer->unsafe_to_concat (prev, buffer->len);
361 }
362 break;
363 }
364 }
365
366 static void
mongolian_variation_selectors(hb_buffer_t * buffer)367 mongolian_variation_selectors (hb_buffer_t *buffer)
368 {
369 /* Copy arabic_shaping_action() from base to Mongolian variation selectors. */
370 unsigned int count = buffer->len;
371 hb_glyph_info_t *info = buffer->info;
372 for (unsigned int i = 1; i < count; i++)
373 if (unlikely (hb_in_ranges<hb_codepoint_t> (info[i].codepoint, 0x180Bu, 0x180Du, 0x180Fu, 0x180Fu)))
374 info[i].arabic_shaping_action() = info[i - 1].arabic_shaping_action();
375 }
376
377 void
setup_masks_arabic_plan(const arabic_shape_plan_t * arabic_plan,hb_buffer_t * buffer,hb_script_t script)378 setup_masks_arabic_plan (const arabic_shape_plan_t *arabic_plan,
379 hb_buffer_t *buffer,
380 hb_script_t script)
381 {
382 HB_BUFFER_ALLOCATE_VAR (buffer, arabic_shaping_action);
383
384 arabic_joining (buffer);
385 if (script == HB_SCRIPT_MONGOLIAN)
386 mongolian_variation_selectors (buffer);
387
388 unsigned int count = buffer->len;
389 hb_glyph_info_t *info = buffer->info;
390 for (unsigned int i = 0; i < count; i++)
391 info[i].mask |= arabic_plan->mask_array[info[i].arabic_shaping_action()];
392 }
393
394 static void
setup_masks_arabic(const hb_ot_shape_plan_t * plan,hb_buffer_t * buffer,hb_font_t * font HB_UNUSED)395 setup_masks_arabic (const hb_ot_shape_plan_t *plan,
396 hb_buffer_t *buffer,
397 hb_font_t *font HB_UNUSED)
398 {
399 const arabic_shape_plan_t *arabic_plan = (const arabic_shape_plan_t *) plan->data;
400 setup_masks_arabic_plan (arabic_plan, buffer, plan->props.script);
401 }
402
403 static void
arabic_fallback_shape(const hb_ot_shape_plan_t * plan,hb_font_t * font,hb_buffer_t * buffer)404 arabic_fallback_shape (const hb_ot_shape_plan_t *plan,
405 hb_font_t *font,
406 hb_buffer_t *buffer)
407 {
408 #ifdef HB_NO_OT_SHAPE_COMPLEX_ARABIC_FALLBACK
409 return;
410 #endif
411
412 const arabic_shape_plan_t *arabic_plan = (const arabic_shape_plan_t *) plan->data;
413
414 if (!arabic_plan->do_fallback)
415 return;
416
417 retry:
418 arabic_fallback_plan_t *fallback_plan = arabic_plan->fallback_plan;
419 if (unlikely (!fallback_plan))
420 {
421 /* This sucks. We need a font to build the fallback plan... */
422 fallback_plan = arabic_fallback_plan_create (plan, font);
423 if (unlikely (!arabic_plan->fallback_plan.cmpexch (nullptr, fallback_plan)))
424 {
425 arabic_fallback_plan_destroy (fallback_plan);
426 goto retry;
427 }
428 }
429
430 arabic_fallback_plan_shape (fallback_plan, font, buffer);
431 }
432
433 /*
434 * Stretch feature: "stch".
435 * See example here:
436 * https://docs.microsoft.com/en-us/typography/script-development/syriac
437 * We implement this in a generic way, such that the Arabic subtending
438 * marks can use it as well.
439 */
440
441 static void
record_stch(const hb_ot_shape_plan_t * plan,hb_font_t * font HB_UNUSED,hb_buffer_t * buffer)442 record_stch (const hb_ot_shape_plan_t *plan,
443 hb_font_t *font HB_UNUSED,
444 hb_buffer_t *buffer)
445 {
446 const arabic_shape_plan_t *arabic_plan = (const arabic_shape_plan_t *) plan->data;
447 if (!arabic_plan->has_stch)
448 return;
449
450 /* 'stch' feature was just applied. Look for anything that multiplied,
451 * and record it for stch treatment later. Note that rtlm, frac, etc
452 * are applied before stch, but we assume that they didn't result in
453 * anything multiplying into 5 pieces, so it's safe-ish... */
454
455 unsigned int count = buffer->len;
456 hb_glyph_info_t *info = buffer->info;
457 for (unsigned int i = 0; i < count; i++)
458 if (unlikely (_hb_glyph_info_multiplied (&info[i])))
459 {
460 unsigned int comp = _hb_glyph_info_get_lig_comp (&info[i]);
461 info[i].arabic_shaping_action() = comp % 2 ? STCH_REPEATING : STCH_FIXED;
462 buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_ARABIC_HAS_STCH;
463 }
464 }
465
466 static void
apply_stch(const hb_ot_shape_plan_t * plan HB_UNUSED,hb_buffer_t * buffer,hb_font_t * font)467 apply_stch (const hb_ot_shape_plan_t *plan HB_UNUSED,
468 hb_buffer_t *buffer,
469 hb_font_t *font)
470 {
471 if (likely (!(buffer->scratch_flags & HB_BUFFER_SCRATCH_FLAG_ARABIC_HAS_STCH)))
472 return;
473
474 /* The Arabic shaper currently always processes in RTL mode, so we should
475 * stretch / position the stretched pieces to the left / preceding glyphs. */
476
477 /* We do a two pass implementation:
478 * First pass calculates the exact number of extra glyphs we need,
479 * We then enlarge buffer to have that much room,
480 * Second pass applies the stretch, copying things to the end of buffer.
481 */
482
483 int sign = font->x_scale < 0 ? -1 : +1;
484 unsigned int extra_glyphs_needed = 0; // Set during MEASURE, used during CUT
485 enum { MEASURE, CUT } /* step_t */;
486
487 for (unsigned int step = MEASURE; step <= CUT; step = step + 1)
488 {
489 unsigned int count = buffer->len;
490 hb_glyph_info_t *info = buffer->info;
491 hb_glyph_position_t *pos = buffer->pos;
492 unsigned int new_len = count + extra_glyphs_needed; // write head during CUT
493 unsigned int j = new_len;
494 for (unsigned int i = count; i; i--)
495 {
496 if (!hb_in_range<uint8_t> (info[i - 1].arabic_shaping_action(), STCH_FIXED, STCH_REPEATING))
497 {
498 if (step == CUT)
499 {
500 --j;
501 info[j] = info[i - 1];
502 pos[j] = pos[i - 1];
503 }
504 continue;
505 }
506
507 /* Yay, justification! */
508
509 hb_position_t w_total = 0; // Total to be filled
510 hb_position_t w_fixed = 0; // Sum of fixed tiles
511 hb_position_t w_repeating = 0; // Sum of repeating tiles
512 int n_fixed = 0;
513 int n_repeating = 0;
514
515 unsigned int end = i;
516 while (i &&
517 hb_in_range<uint8_t> (info[i - 1].arabic_shaping_action(), STCH_FIXED, STCH_REPEATING))
518 {
519 i--;
520 hb_position_t width = font->get_glyph_h_advance (info[i].codepoint);
521 if (info[i].arabic_shaping_action() == STCH_FIXED)
522 {
523 w_fixed += width;
524 n_fixed++;
525 }
526 else
527 {
528 w_repeating += width;
529 n_repeating++;
530 }
531 }
532 unsigned int start = i;
533 unsigned int context = i;
534 while (context &&
535 !hb_in_range<uint8_t> (info[context - 1].arabic_shaping_action(), STCH_FIXED, STCH_REPEATING) &&
536 (_hb_glyph_info_is_default_ignorable (&info[context - 1]) ||
537 HB_ARABIC_GENERAL_CATEGORY_IS_WORD (_hb_glyph_info_get_general_category (&info[context - 1]))))
538 {
539 context--;
540 w_total += pos[context].x_advance;
541 }
542 i++; // Don't touch i again.
543
544 DEBUG_MSG (ARABIC, nullptr, "%s stretch at (%d,%d,%d)",
545 step == MEASURE ? "measuring" : "cutting", context, start, end);
546 DEBUG_MSG (ARABIC, nullptr, "rest of word: count=%d width %d", start - context, w_total);
547 DEBUG_MSG (ARABIC, nullptr, "fixed tiles: count=%d width=%d", n_fixed, w_fixed);
548 DEBUG_MSG (ARABIC, nullptr, "repeating tiles: count=%d width=%d", n_repeating, w_repeating);
549
550 /* Number of additional times to repeat each repeating tile. */
551 int n_copies = 0;
552
553 hb_position_t w_remaining = w_total - w_fixed;
554 if (sign * w_remaining > sign * w_repeating && sign * w_repeating > 0)
555 n_copies = (sign * w_remaining) / (sign * w_repeating) - 1;
556
557 /* See if we can improve the fit by adding an extra repeat and squeezing them together a bit. */
558 hb_position_t extra_repeat_overlap = 0;
559 hb_position_t shortfall = sign * w_remaining - sign * w_repeating * (n_copies + 1);
560 if (shortfall > 0 && n_repeating > 0)
561 {
562 ++n_copies;
563 hb_position_t excess = (n_copies + 1) * sign * w_repeating - sign * w_remaining;
564 if (excess > 0)
565 extra_repeat_overlap = excess / (n_copies * n_repeating);
566 }
567
568 if (step == MEASURE)
569 {
570 extra_glyphs_needed += n_copies * n_repeating;
571 DEBUG_MSG (ARABIC, nullptr, "will add extra %d copies of repeating tiles", n_copies);
572 }
573 else
574 {
575 buffer->unsafe_to_break (context, end);
576 hb_position_t x_offset = 0;
577 for (unsigned int k = end; k > start; k--)
578 {
579 hb_position_t width = font->get_glyph_h_advance (info[k - 1].codepoint);
580
581 unsigned int repeat = 1;
582 if (info[k - 1].arabic_shaping_action() == STCH_REPEATING)
583 repeat += n_copies;
584
585 DEBUG_MSG (ARABIC, nullptr, "appending %d copies of glyph %d; j=%d",
586 repeat, info[k - 1].codepoint, j);
587 for (unsigned int n = 0; n < repeat; n++)
588 {
589 x_offset -= width;
590 if (n > 0)
591 x_offset += extra_repeat_overlap;
592 pos[k - 1].x_offset = x_offset;
593 /* Append copy. */
594 --j;
595 info[j] = info[k - 1];
596 pos[j] = pos[k - 1];
597 }
598 }
599 }
600 }
601
602 if (step == MEASURE)
603 {
604 if (unlikely (!buffer->ensure (count + extra_glyphs_needed)))
605 break;
606 }
607 else
608 {
609 assert (j == 0);
610 buffer->len = new_len;
611 }
612 }
613 }
614
615
616 static void
postprocess_glyphs_arabic(const hb_ot_shape_plan_t * plan,hb_buffer_t * buffer,hb_font_t * font)617 postprocess_glyphs_arabic (const hb_ot_shape_plan_t *plan,
618 hb_buffer_t *buffer,
619 hb_font_t *font)
620 {
621 apply_stch (plan, buffer, font);
622
623 HB_BUFFER_DEALLOCATE_VAR (buffer, arabic_shaping_action);
624 }
625
626 /* https://www.unicode.org/reports/tr53/ */
627
628 static hb_codepoint_t
629 modifier_combining_marks[] =
630 {
631 0x0654u, /* ARABIC HAMZA ABOVE */
632 0x0655u, /* ARABIC HAMZA BELOW */
633 0x0658u, /* ARABIC MARK NOON GHUNNA */
634 0x06DCu, /* ARABIC SMALL HIGH SEEN */
635 0x06E3u, /* ARABIC SMALL LOW SEEN */
636 0x06E7u, /* ARABIC SMALL HIGH YEH */
637 0x06E8u, /* ARABIC SMALL HIGH NOON */
638 0x08CAu, /* ARABIC SMALL HIGH FARSI YEH */
639 0x08CBu, /* ARABIC SMALL HIGH YEH BARREE WITH TWO DOTS BELOW */
640 0x08CDu, /* ARABIC SMALL HIGH ZAH */
641 0x08CEu, /* ARABIC LARGE ROUND DOT ABOVE */
642 0x08CFu, /* ARABIC LARGE ROUND DOT BELOW */
643 0x08D3u, /* ARABIC SMALL LOW WAW */
644 0x08F3u, /* ARABIC SMALL HIGH WAW */
645 };
646
647 static inline bool
info_is_mcm(const hb_glyph_info_t & info)648 info_is_mcm (const hb_glyph_info_t &info)
649 {
650 hb_codepoint_t u = info.codepoint;
651 for (unsigned int i = 0; i < ARRAY_LENGTH (modifier_combining_marks); i++)
652 if (u == modifier_combining_marks[i])
653 return true;
654 return false;
655 }
656
657 static void
reorder_marks_arabic(const hb_ot_shape_plan_t * plan HB_UNUSED,hb_buffer_t * buffer,unsigned int start,unsigned int end)658 reorder_marks_arabic (const hb_ot_shape_plan_t *plan HB_UNUSED,
659 hb_buffer_t *buffer,
660 unsigned int start,
661 unsigned int end)
662 {
663 hb_glyph_info_t *info = buffer->info;
664
665 DEBUG_MSG (ARABIC, buffer, "Reordering marks from %d to %d", start, end);
666
667 unsigned int i = start;
668 for (unsigned int cc = 220; cc <= 230; cc += 10)
669 {
670 DEBUG_MSG (ARABIC, buffer, "Looking for %d's starting at %d", cc, i);
671 while (i < end && info_cc(info[i]) < cc)
672 i++;
673 DEBUG_MSG (ARABIC, buffer, "Looking for %d's stopped at %d", cc, i);
674
675 if (i == end)
676 break;
677
678 if (info_cc(info[i]) > cc)
679 continue;
680
681 unsigned int j = i;
682 while (j < end && info_cc(info[j]) == cc && info_is_mcm (info[j]))
683 j++;
684
685 if (i == j)
686 continue;
687
688 DEBUG_MSG (ARABIC, buffer, "Found %d's from %d to %d", cc, i, j);
689
690 /* Shift it! */
691 DEBUG_MSG (ARABIC, buffer, "Shifting %d's: %d %d", cc, i, j);
692 hb_glyph_info_t temp[HB_OT_SHAPE_COMPLEX_MAX_COMBINING_MARKS];
693 assert (j - i <= ARRAY_LENGTH (temp));
694 buffer->merge_clusters (start, j);
695 memmove (temp, &info[i], (j - i) * sizeof (hb_glyph_info_t));
696 memmove (&info[start + j - i], &info[start], (i - start) * sizeof (hb_glyph_info_t));
697 memmove (&info[start], temp, (j - i) * sizeof (hb_glyph_info_t));
698
699 /* Renumber CC such that the reordered sequence is still sorted.
700 * 22 and 26 are chosen because they are smaller than all Arabic categories,
701 * and are folded back to 220/230 respectively during fallback mark positioning.
702 *
703 * We do this because the CGJ-handling logic in the normalizer relies on
704 * mark sequences having an increasing order even after this reordering.
705 * https://github.com/harfbuzz/harfbuzz/issues/554
706 * This, however, does break some obscure sequences, where the normalizer
707 * might compose a sequence that it should not. For example, in the seequence
708 * ALEF, HAMZAH, MADDAH, we should NOT try to compose ALEF+MADDAH, but with this
709 * renumbering, we will.
710 */
711 unsigned int new_start = start + j - i;
712 unsigned int new_cc = cc == 220 ? HB_MODIFIED_COMBINING_CLASS_CCC22 : HB_MODIFIED_COMBINING_CLASS_CCC26;
713 while (start < new_start)
714 {
715 _hb_glyph_info_set_modified_combining_class (&info[start], new_cc);
716 start++;
717 }
718
719 i = j;
720 }
721 }
722
723 const hb_ot_complex_shaper_t _hb_ot_complex_shaper_arabic =
724 {
725 collect_features_arabic,
726 nullptr, /* override_features */
727 data_create_arabic,
728 data_destroy_arabic,
729 nullptr, /* preprocess_text */
730 postprocess_glyphs_arabic,
731 HB_OT_SHAPE_NORMALIZATION_MODE_DEFAULT,
732 nullptr, /* decompose */
733 nullptr, /* compose */
734 setup_masks_arabic,
735 HB_TAG_NONE, /* gpos_tag */
736 reorder_marks_arabic,
737 HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_LATE,
738 true, /* fallback_position */
739 };
740
741
742 #endif
743