2 * Copyright © 2007,2008,2009 Red Hat, Inc.
3 * Copyright © 2012,2013 Google, Inc.
5 * This is part of HarfBuzz, a text shaping library.
7 * Permission is hereby granted, without written agreement and without
8 * license or royalty fees, to use, copy, modify, and distribute this
9 * software and its documentation for any purpose, provided that the
10 * above copyright notice and the following two paragraphs appear in
11 * all copies of this software.
13 * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
14 * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
15 * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
16 * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
19 * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
20 * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
21 * FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
22 * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
23 * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
25 * Red Hat Author(s): Behdad Esfahbod
26 * Google Author(s): Behdad Esfahbod
29 #ifndef HB_OT_LAYOUT_HH
30 #define HB_OT_LAYOUT_HH
35 #include "hb-buffer.hh"
36 #include "hb-open-type.hh"
37 #include "hb-ot-shape.hh"
38 #include "hb-set-digest.hh"
41 struct hb_ot_shape_plan_t;
49 hb_ot_layout_has_kerning (hb_face_t *face);
52 hb_ot_layout_has_machine_kerning (hb_face_t *face);
55 hb_ot_layout_has_cross_kerning (hb_face_t *face);
58 hb_ot_layout_kern (const hb_ot_shape_plan_t *plan,
63 /* Private API corresponding to hb-ot-layout.h: */
66 hb_ot_layout_table_find_feature (hb_face_t *face,
69 unsigned int *feature_index);
76 enum hb_ot_layout_glyph_props_flags_t
78 /* The following three match LookupFlags::Ignore* numbers. */
79 HB_OT_LAYOUT_GLYPH_PROPS_BASE_GLYPH = 0x02u,
80 HB_OT_LAYOUT_GLYPH_PROPS_LIGATURE = 0x04u,
81 HB_OT_LAYOUT_GLYPH_PROPS_MARK = 0x08u,
83 HB_OT_LAYOUT_GLYPH_PROPS_CLASS_MASK = HB_OT_LAYOUT_GLYPH_PROPS_BASE_GLYPH |
84 HB_OT_LAYOUT_GLYPH_PROPS_LIGATURE |
85 HB_OT_LAYOUT_GLYPH_PROPS_MARK,
87 /* The following are used internally; not derived from GDEF. */
88 HB_OT_LAYOUT_GLYPH_PROPS_SUBSTITUTED = 0x10u,
89 HB_OT_LAYOUT_GLYPH_PROPS_LIGATED = 0x20u,
90 HB_OT_LAYOUT_GLYPH_PROPS_MULTIPLIED = 0x40u,
92 HB_MARK_AS_FLAG_T (hb_ot_layout_glyph_props_flags_t);
100 /* Should be called before all the substitute_lookup's are done. */
102 hb_ot_layout_substitute_start (hb_font_t *font,
103 hb_buffer_t *buffer);
106 hb_ot_layout_delete_glyphs_inplace (hb_buffer_t *buffer,
107 bool (*filter) (const hb_glyph_info_t *info));
110 struct hb_ot_apply_context_t;
112 struct hb_ot_layout_lookup_accelerator_t;
116 hb_ot_layout_substitute_lookup (OT::hb_ot_apply_context_t *c,
117 const OT::SubstLookup &lookup,
118 const OT::hb_ot_layout_lookup_accelerator_t &accel);
121 /* Should be called before all the position_lookup's are done. */
123 hb_ot_layout_position_start (hb_font_t *font,
124 hb_buffer_t *buffer);
126 /* Should be called after all the position_lookup's are done, to fini advances. */
128 hb_ot_layout_position_finish_advances (hb_font_t *font,
129 hb_buffer_t *buffer);
131 /* Should be called after hb_ot_layout_position_finish_advances, to fini offsets. */
133 hb_ot_layout_position_finish_offsets (hb_font_t *font,
134 hb_buffer_t *buffer);
138 * Buffer var routines.
141 /* buffer var allocations, used during the entire shaping process */
142 #define unicode_props() var2.u16[0]
144 /* buffer var allocations, used during the GSUB/GPOS processing */
145 #define glyph_props() var1.u16[0] /* GDEF glyph properties */
146 #define lig_props() var1.u8[2] /* GSUB/GPOS ligature tracking */
147 #define syllable() var1.u8[3] /* GSUB/GPOS shaping boundaries */
150 /* Loop over syllables. Based on foreach_cluster(). */
151 #define foreach_syllable(buffer, start, end) \
153 _count = buffer->len, \
154 start = 0, end = _count ? _hb_next_syllable (buffer, 0) : 0; \
156 start = end, end = _hb_next_syllable (buffer, start))
158 static inline unsigned int
159 _hb_next_syllable (hb_buffer_t *buffer, unsigned int start)
161 hb_glyph_info_t *info = buffer->info;
162 unsigned int count = buffer->len;
164 unsigned int syllable = info[start].syllable();
165 while (++start < count && syllable == info[start].syllable())
172 _hb_clear_syllables (const hb_ot_shape_plan_t *plan HB_UNUSED,
173 hb_font_t *font HB_UNUSED,
176 hb_glyph_info_t *info = buffer->info;
177 unsigned int count = buffer->len;
178 for (unsigned int i = 0; i < count; i++)
179 info[i].syllable() = 0;
186 * unicode_props() is a two-byte number. The low byte includes:
187 * - General_Category: 5 bits.
189 * * Is it Default_Ignorable(); we have a modified Default_Ignorable().
190 * * Whether it's one of the three Mongolian Free Variation Selectors,
191 * CGJ, or other characters that are hidden but should not be ignored
192 * like most other Default_Ignorable()s do during matching.
193 * * Whether it's a grapheme continuation.
195 * The high-byte has different meanings, switched by the Gen-Cat:
196 * - For Mn,Mc,Me: the modified Combining_Class.
197 * - For Cf: whether it's ZWJ, ZWNJ, or something else.
198 * - For Ws: index of which space character this is, if space fallback
199 * is needed, ie. we don't set this by default, only if asked to.
202 enum hb_unicode_props_flags_t {
203 UPROPS_MASK_GEN_CAT = 0x001Fu,
204 UPROPS_MASK_IGNORABLE = 0x0020u,
205 UPROPS_MASK_HIDDEN = 0x0040u, /* MONGOLIAN FREE VARIATION SELECTOR 1..3, or TAG characters */
206 UPROPS_MASK_CONTINUATION=0x0080u,
208 /* If GEN_CAT=FORMAT, top byte masks: */
209 UPROPS_MASK_Cf_ZWJ = 0x0100u,
210 UPROPS_MASK_Cf_ZWNJ = 0x0200u
212 HB_MARK_AS_FLAG_T (hb_unicode_props_flags_t);
215 _hb_glyph_info_set_unicode_props (hb_glyph_info_t *info, hb_buffer_t *buffer)
217 hb_unicode_funcs_t *unicode = buffer->unicode;
218 unsigned int u = info->codepoint;
219 unsigned int gen_cat = (unsigned int) unicode->general_category (u);
220 unsigned int props = gen_cat;
224 buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_NON_ASCII;
226 if (unlikely (unicode->is_default_ignorable (u)))
228 buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_DEFAULT_IGNORABLES;
229 props |= UPROPS_MASK_IGNORABLE;
230 if (u == 0x200Cu) props |= UPROPS_MASK_Cf_ZWNJ;
231 else if (u == 0x200Du) props |= UPROPS_MASK_Cf_ZWJ;
232 /* Mongolian Free Variation Selectors need to be remembered
233 * because although we need to hide them like default-ignorables,
234 * they need to non-ignorable during shaping. This is similar to
235 * what we do for joiners in Indic-like shapers, but since the
236 * FVSes are GC=Mn, we have use a separate bit to remember them.
238 * https://github.com/harfbuzz/harfbuzz/issues/234 */
239 else if (unlikely (hb_in_range<hb_codepoint_t> (u, 0x180Bu, 0x180Du))) props |= UPROPS_MASK_HIDDEN;
240 /* TAG characters need similar treatment. Fixes:
241 * https://github.com/harfbuzz/harfbuzz/issues/463 */
242 else if (unlikely (hb_in_range<hb_codepoint_t> (u, 0xE0020u, 0xE007Fu))) props |= UPROPS_MASK_HIDDEN;
243 /* COMBINING GRAPHEME JOINER should not be skipped; at least some times.
244 * https://github.com/harfbuzz/harfbuzz/issues/554 */
245 else if (unlikely (u == 0x034Fu))
247 buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_CGJ;
248 props |= UPROPS_MASK_HIDDEN;
252 if (unlikely (HB_UNICODE_GENERAL_CATEGORY_IS_MARK (gen_cat)))
254 props |= UPROPS_MASK_CONTINUATION;
255 props |= unicode->modified_combining_class (u)<<8;
259 info->unicode_props() = props;
263 _hb_glyph_info_set_general_category (hb_glyph_info_t *info,
264 hb_unicode_general_category_t gen_cat)
266 /* Clears top-byte. */
267 info->unicode_props() = (unsigned int) gen_cat | (info->unicode_props() & (0xFF & ~UPROPS_MASK_GEN_CAT));
270 static inline hb_unicode_general_category_t
271 _hb_glyph_info_get_general_category (const hb_glyph_info_t *info)
273 return (hb_unicode_general_category_t) (info->unicode_props() & UPROPS_MASK_GEN_CAT);
277 _hb_glyph_info_is_unicode_mark (const hb_glyph_info_t *info)
279 return HB_UNICODE_GENERAL_CATEGORY_IS_MARK (info->unicode_props() & UPROPS_MASK_GEN_CAT);
282 _hb_glyph_info_set_modified_combining_class (hb_glyph_info_t *info,
283 unsigned int modified_class)
285 if (unlikely (!_hb_glyph_info_is_unicode_mark (info)))
287 info->unicode_props() = (modified_class<<8) | (info->unicode_props() & 0xFF);
289 static inline unsigned int
290 _hb_glyph_info_get_modified_combining_class (const hb_glyph_info_t *info)
292 return _hb_glyph_info_is_unicode_mark (info) ? info->unicode_props()>>8 : 0;
294 #define info_cc(info) (_hb_glyph_info_get_modified_combining_class (&(info)))
297 _hb_glyph_info_is_unicode_space (const hb_glyph_info_t *info)
299 return _hb_glyph_info_get_general_category (info) ==
300 HB_UNICODE_GENERAL_CATEGORY_SPACE_SEPARATOR;
303 _hb_glyph_info_set_unicode_space_fallback_type (hb_glyph_info_t *info, hb_unicode_funcs_t::space_t s)
305 if (unlikely (!_hb_glyph_info_is_unicode_space (info)))
307 info->unicode_props() = (((unsigned int) s)<<8) | (info->unicode_props() & 0xFF);
309 static inline hb_unicode_funcs_t::space_t
310 _hb_glyph_info_get_unicode_space_fallback_type (const hb_glyph_info_t *info)
312 return _hb_glyph_info_is_unicode_space (info) ?
313 (hb_unicode_funcs_t::space_t) (info->unicode_props()>>8) :
314 hb_unicode_funcs_t::NOT_SPACE;
317 static inline bool _hb_glyph_info_ligated (const hb_glyph_info_t *info);
320 _hb_glyph_info_is_default_ignorable (const hb_glyph_info_t *info)
322 return (info->unicode_props() & UPROPS_MASK_IGNORABLE) &&
323 !_hb_glyph_info_ligated (info);
326 _hb_glyph_info_is_default_ignorable_and_not_hidden (const hb_glyph_info_t *info)
328 return ((info->unicode_props() & (UPROPS_MASK_IGNORABLE|UPROPS_MASK_HIDDEN))
329 == UPROPS_MASK_IGNORABLE) &&
330 !_hb_glyph_info_ligated (info);
333 _hb_glyph_info_unhide (hb_glyph_info_t *info)
335 info->unicode_props() &= ~ UPROPS_MASK_HIDDEN;
339 _hb_glyph_info_set_continuation (hb_glyph_info_t *info)
341 info->unicode_props() |= UPROPS_MASK_CONTINUATION;
344 _hb_glyph_info_reset_continuation (hb_glyph_info_t *info)
346 info->unicode_props() &= ~ UPROPS_MASK_CONTINUATION;
349 _hb_glyph_info_is_continuation (const hb_glyph_info_t *info)
351 return info->unicode_props() & UPROPS_MASK_CONTINUATION;
353 /* Loop over grapheme. Based on foreach_cluster(). */
354 #define foreach_grapheme(buffer, start, end) \
356 _count = buffer->len, \
357 start = 0, end = _count ? _hb_next_grapheme (buffer, 0) : 0; \
359 start = end, end = _hb_next_grapheme (buffer, start))
361 static inline unsigned int
362 _hb_next_grapheme (hb_buffer_t *buffer, unsigned int start)
364 hb_glyph_info_t *info = buffer->info;
365 unsigned int count = buffer->len;
367 while (++start < count && _hb_glyph_info_is_continuation (&info[start]))
374 _hb_glyph_info_is_unicode_format (const hb_glyph_info_t *info)
376 return _hb_glyph_info_get_general_category (info) ==
377 HB_UNICODE_GENERAL_CATEGORY_FORMAT;
380 _hb_glyph_info_is_zwnj (const hb_glyph_info_t *info)
382 return _hb_glyph_info_is_unicode_format (info) && (info->unicode_props() & UPROPS_MASK_Cf_ZWNJ);
385 _hb_glyph_info_is_zwj (const hb_glyph_info_t *info)
387 return _hb_glyph_info_is_unicode_format (info) && (info->unicode_props() & UPROPS_MASK_Cf_ZWJ);
390 _hb_glyph_info_is_joiner (const hb_glyph_info_t *info)
392 return _hb_glyph_info_is_unicode_format (info) && (info->unicode_props() & (UPROPS_MASK_Cf_ZWNJ|UPROPS_MASK_Cf_ZWJ));
395 _hb_glyph_info_flip_joiners (hb_glyph_info_t *info)
397 if (!_hb_glyph_info_is_unicode_format (info))
399 info->unicode_props() ^= UPROPS_MASK_Cf_ZWNJ | UPROPS_MASK_Cf_ZWJ;
402 /* lig_props: aka lig_id / lig_comp
404 * When a ligature is formed:
406 * - The ligature glyph and any marks in between all the same newly allocated
408 * - The ligature glyph will get lig_num_comps set to the number of components
409 * - The marks get lig_comp > 0, reflecting which component of the ligature
410 * they were applied to.
411 * - This is used in GPOS to attach marks to the right component of a ligature
413 * - Note that when marks are ligated together, much of the above is skipped
414 * and the current lig_id reused.
416 * When a multiple-substitution is done:
418 * - All resulting glyphs will have lig_id = 0,
419 * - The resulting glyphs will have lig_comp = 0, 1, 2, ... respectively.
420 * - This is used in GPOS to attach marks to the first component of a
421 * multiple substitution in MarkBasePos.
423 * The numbers are also used in GPOS to do mark-to-mark positioning only
424 * to marks that belong to the same component of the same ligature.
428 _hb_glyph_info_clear_lig_props (hb_glyph_info_t *info)
430 info->lig_props() = 0;
433 #define IS_LIG_BASE 0x10
436 _hb_glyph_info_set_lig_props_for_ligature (hb_glyph_info_t *info,
438 unsigned int lig_num_comps)
440 info->lig_props() = (lig_id << 5) | IS_LIG_BASE | (lig_num_comps & 0x0F);
444 _hb_glyph_info_set_lig_props_for_mark (hb_glyph_info_t *info,
446 unsigned int lig_comp)
448 info->lig_props() = (lig_id << 5) | (lig_comp & 0x0F);
452 _hb_glyph_info_set_lig_props_for_component (hb_glyph_info_t *info, unsigned int comp)
454 _hb_glyph_info_set_lig_props_for_mark (info, 0, comp);
457 static inline unsigned int
458 _hb_glyph_info_get_lig_id (const hb_glyph_info_t *info)
460 return info->lig_props() >> 5;
464 _hb_glyph_info_ligated_internal (const hb_glyph_info_t *info)
466 return !!(info->lig_props() & IS_LIG_BASE);
469 static inline unsigned int
470 _hb_glyph_info_get_lig_comp (const hb_glyph_info_t *info)
472 if (_hb_glyph_info_ligated_internal (info))
475 return info->lig_props() & 0x0F;
478 static inline unsigned int
479 _hb_glyph_info_get_lig_num_comps (const hb_glyph_info_t *info)
481 if ((info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_LIGATURE) &&
482 _hb_glyph_info_ligated_internal (info))
483 return info->lig_props() & 0x0F;
488 static inline uint8_t
489 _hb_allocate_lig_id (hb_buffer_t *buffer) {
490 uint8_t lig_id = buffer->next_serial () & 0x07;
491 if (unlikely (!lig_id))
492 lig_id = _hb_allocate_lig_id (buffer); /* in case of overflow */
499 _hb_glyph_info_set_glyph_props (hb_glyph_info_t *info, unsigned int props)
501 info->glyph_props() = props;
504 static inline unsigned int
505 _hb_glyph_info_get_glyph_props (const hb_glyph_info_t *info)
507 return info->glyph_props();
511 _hb_glyph_info_is_base_glyph (const hb_glyph_info_t *info)
513 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_BASE_GLYPH);
517 _hb_glyph_info_is_ligature (const hb_glyph_info_t *info)
519 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_LIGATURE);
523 _hb_glyph_info_is_mark (const hb_glyph_info_t *info)
525 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_MARK);
529 _hb_glyph_info_substituted (const hb_glyph_info_t *info)
531 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_SUBSTITUTED);
535 _hb_glyph_info_ligated (const hb_glyph_info_t *info)
537 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_LIGATED);
541 _hb_glyph_info_multiplied (const hb_glyph_info_t *info)
543 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_MULTIPLIED);
547 _hb_glyph_info_ligated_and_didnt_multiply (const hb_glyph_info_t *info)
549 return _hb_glyph_info_ligated (info) && !_hb_glyph_info_multiplied (info);
553 _hb_glyph_info_clear_ligated_and_multiplied (hb_glyph_info_t *info)
555 info->glyph_props() &= ~(HB_OT_LAYOUT_GLYPH_PROPS_LIGATED |
556 HB_OT_LAYOUT_GLYPH_PROPS_MULTIPLIED);
560 _hb_glyph_info_clear_substituted (hb_glyph_info_t *info)
562 info->glyph_props() &= ~(HB_OT_LAYOUT_GLYPH_PROPS_SUBSTITUTED);
566 _hb_clear_substitution_flags (const hb_ot_shape_plan_t *plan HB_UNUSED,
567 hb_font_t *font HB_UNUSED,
570 hb_glyph_info_t *info = buffer->info;
571 unsigned int count = buffer->len;
572 for (unsigned int i = 0; i < count; i++)
573 _hb_glyph_info_clear_substituted (&info[i]);
577 /* Allocation / deallocation. */
580 _hb_buffer_allocate_unicode_vars (hb_buffer_t *buffer)
582 HB_BUFFER_ALLOCATE_VAR (buffer, unicode_props);
586 _hb_buffer_deallocate_unicode_vars (hb_buffer_t *buffer)
588 HB_BUFFER_DEALLOCATE_VAR (buffer, unicode_props);
592 _hb_buffer_assert_unicode_vars (hb_buffer_t *buffer)
594 HB_BUFFER_ASSERT_VAR (buffer, unicode_props);
598 _hb_buffer_allocate_gsubgpos_vars (hb_buffer_t *buffer)
600 HB_BUFFER_ALLOCATE_VAR (buffer, glyph_props);
601 HB_BUFFER_ALLOCATE_VAR (buffer, lig_props);
602 HB_BUFFER_ALLOCATE_VAR (buffer, syllable);
606 _hb_buffer_deallocate_gsubgpos_vars (hb_buffer_t *buffer)
608 HB_BUFFER_DEALLOCATE_VAR (buffer, syllable);
609 HB_BUFFER_DEALLOCATE_VAR (buffer, lig_props);
610 HB_BUFFER_DEALLOCATE_VAR (buffer, glyph_props);
614 _hb_buffer_assert_gsubgpos_vars (hb_buffer_t *buffer)
616 HB_BUFFER_ASSERT_VAR (buffer, glyph_props);
617 HB_BUFFER_ASSERT_VAR (buffer, lig_props);
618 HB_BUFFER_ASSERT_VAR (buffer, syllable);
621 /* Make sure no one directly touches our props... */
622 #undef unicode_props0
623 #undef unicode_props1
627 #endif /* HB_OT_LAYOUT_HH */