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 /* The following are used internally; not derived from GDEF. */
84 HB_OT_LAYOUT_GLYPH_PROPS_SUBSTITUTED = 0x10u,
85 HB_OT_LAYOUT_GLYPH_PROPS_LIGATED = 0x20u,
86 HB_OT_LAYOUT_GLYPH_PROPS_MULTIPLIED = 0x40u,
88 HB_OT_LAYOUT_GLYPH_PROPS_PRESERVE = HB_OT_LAYOUT_GLYPH_PROPS_SUBSTITUTED |
89 HB_OT_LAYOUT_GLYPH_PROPS_LIGATED |
90 HB_OT_LAYOUT_GLYPH_PROPS_MULTIPLIED
92 HB_MARK_AS_FLAG_T (hb_ot_layout_glyph_props_flags_t);
100 hb_ot_layout_lookup_would_substitute_fast (hb_face_t *face,
101 unsigned int lookup_index,
102 const hb_codepoint_t *glyphs,
103 unsigned int glyphs_length,
107 /* Should be called before all the substitute_lookup's are done. */
109 hb_ot_layout_substitute_start (hb_font_t *font,
110 hb_buffer_t *buffer);
113 hb_ot_layout_delete_glyphs_inplace (hb_buffer_t *buffer,
114 bool (*filter) (const hb_glyph_info_t *info));
117 struct hb_ot_apply_context_t;
119 struct hb_ot_layout_lookup_accelerator_t;
123 hb_ot_layout_substitute_lookup (OT::hb_ot_apply_context_t *c,
124 const OT::SubstLookup &lookup,
125 const OT::hb_ot_layout_lookup_accelerator_t &accel);
128 /* Should be called before all the position_lookup's are done. */
130 hb_ot_layout_position_start (hb_font_t *font,
131 hb_buffer_t *buffer);
133 /* Should be called after all the position_lookup's are done, to fini advances. */
135 hb_ot_layout_position_finish_advances (hb_font_t *font,
136 hb_buffer_t *buffer);
138 /* Should be called after hb_ot_layout_position_finish_advances, to fini offsets. */
140 hb_ot_layout_position_finish_offsets (hb_font_t *font,
141 hb_buffer_t *buffer);
145 * Buffer var routines.
148 /* buffer var allocations, used during the entire shaping process */
149 #define unicode_props() var2.u16[0]
151 /* buffer var allocations, used during the GSUB/GPOS processing */
152 #define glyph_props() var1.u16[0] /* GDEF glyph properties */
153 #define lig_props() var1.u8[2] /* GSUB/GPOS ligature tracking */
154 #define syllable() var1.u8[3] /* GSUB/GPOS shaping boundaries */
157 /* Loop over syllables. Based on foreach_cluster(). */
158 #define foreach_syllable(buffer, start, end) \
160 _count = buffer->len, \
161 start = 0, end = _count ? _hb_next_syllable (buffer, 0) : 0; \
163 start = end, end = _hb_next_syllable (buffer, start))
165 static inline unsigned int
166 _hb_next_syllable (hb_buffer_t *buffer, unsigned int start)
168 hb_glyph_info_t *info = buffer->info;
169 unsigned int count = buffer->len;
171 unsigned int syllable = info[start].syllable();
172 while (++start < count && syllable == info[start].syllable())
182 * unicode_props() is a two-byte number. The low byte includes:
183 * - General_Category: 5 bits.
185 * * Is it Default_Ignorable(); we have a modified Default_Ignorable().
186 * * Whether it's one of the three Mongolian Free Variation Selectors,
187 * CGJ, or other characters that are hidden but should not be ignored
188 * like most other Default_Ignorable()s do during matching.
189 * * Whether it's a grapheme continuation.
191 * The high-byte has different meanings, switched by the Gen-Cat:
192 * - For Mn,Mc,Me: the modified Combining_Class.
193 * - For Cf: whether it's ZWJ, ZWNJ, or something else.
194 * - For Ws: index of which space character this is, if space fallback
195 * is needed, ie. we don't set this by default, only if asked to.
198 enum hb_unicode_props_flags_t {
199 UPROPS_MASK_GEN_CAT = 0x001Fu,
200 UPROPS_MASK_IGNORABLE = 0x0020u,
201 UPROPS_MASK_HIDDEN = 0x0040u, /* MONGOLIAN FREE VARIATION SELECTOR 1..3, or TAG characters */
202 UPROPS_MASK_CONTINUATION=0x0080u,
204 /* If GEN_CAT=FORMAT, top byte masks: */
205 UPROPS_MASK_Cf_ZWJ = 0x0100u,
206 UPROPS_MASK_Cf_ZWNJ = 0x0200u
208 HB_MARK_AS_FLAG_T (hb_unicode_props_flags_t);
211 _hb_glyph_info_set_unicode_props (hb_glyph_info_t *info, hb_buffer_t *buffer)
213 hb_unicode_funcs_t *unicode = buffer->unicode;
214 unsigned int u = info->codepoint;
215 unsigned int gen_cat = (unsigned int) unicode->general_category (u);
216 unsigned int props = gen_cat;
220 buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_NON_ASCII;
222 if (unlikely (unicode->is_default_ignorable (u)))
224 buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_DEFAULT_IGNORABLES;
225 props |= UPROPS_MASK_IGNORABLE;
226 if (u == 0x200Cu) props |= UPROPS_MASK_Cf_ZWNJ;
227 else if (u == 0x200Du) props |= UPROPS_MASK_Cf_ZWJ;
228 /* Mongolian Free Variation Selectors need to be remembered
229 * because although we need to hide them like default-ignorables,
230 * they need to non-ignorable during shaping. This is similar to
231 * what we do for joiners in Indic-like shapers, but since the
232 * FVSes are GC=Mn, we have use a separate bit to remember them.
234 * https://github.com/harfbuzz/harfbuzz/issues/234 */
235 else if (unlikely (hb_in_range (u, 0x180Bu, 0x180Du))) props |= UPROPS_MASK_HIDDEN;
236 /* TAG characters need similar treatment. Fixes:
237 * https://github.com/harfbuzz/harfbuzz/issues/463 */
238 else if (unlikely (hb_in_range (u, 0xE0020u, 0xE007Fu))) props |= UPROPS_MASK_HIDDEN;
239 /* COMBINING GRAPHEME JOINER should not be skipped; at least some times.
240 * https://github.com/harfbuzz/harfbuzz/issues/554 */
241 else if (unlikely (u == 0x034Fu))
243 buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_CGJ;
244 props |= UPROPS_MASK_HIDDEN;
248 if (unlikely (HB_UNICODE_GENERAL_CATEGORY_IS_MARK (gen_cat)))
250 props |= UPROPS_MASK_CONTINUATION;
251 props |= unicode->modified_combining_class (u)<<8;
255 info->unicode_props() = props;
259 _hb_glyph_info_set_general_category (hb_glyph_info_t *info,
260 hb_unicode_general_category_t gen_cat)
262 /* Clears top-byte. */
263 info->unicode_props() = (unsigned int) gen_cat | (info->unicode_props() & (0xFF & ~UPROPS_MASK_GEN_CAT));
266 static inline hb_unicode_general_category_t
267 _hb_glyph_info_get_general_category (const hb_glyph_info_t *info)
269 return (hb_unicode_general_category_t) (info->unicode_props() & UPROPS_MASK_GEN_CAT);
273 _hb_glyph_info_is_unicode_mark (const hb_glyph_info_t *info)
275 return HB_UNICODE_GENERAL_CATEGORY_IS_MARK (info->unicode_props() & UPROPS_MASK_GEN_CAT);
278 _hb_glyph_info_set_modified_combining_class (hb_glyph_info_t *info,
279 unsigned int modified_class)
281 if (unlikely (!_hb_glyph_info_is_unicode_mark (info)))
283 info->unicode_props() = (modified_class<<8) | (info->unicode_props() & 0xFF);
285 static inline unsigned int
286 _hb_glyph_info_get_modified_combining_class (const hb_glyph_info_t *info)
288 return _hb_glyph_info_is_unicode_mark (info) ? info->unicode_props()>>8 : 0;
290 #define info_cc(info) (_hb_glyph_info_get_modified_combining_class (&(info)))
293 _hb_glyph_info_is_unicode_space (const hb_glyph_info_t *info)
295 return _hb_glyph_info_get_general_category (info) ==
296 HB_UNICODE_GENERAL_CATEGORY_SPACE_SEPARATOR;
299 _hb_glyph_info_set_unicode_space_fallback_type (hb_glyph_info_t *info, hb_unicode_funcs_t::space_t s)
301 if (unlikely (!_hb_glyph_info_is_unicode_space (info)))
303 info->unicode_props() = (((unsigned int) s)<<8) | (info->unicode_props() & 0xFF);
305 static inline hb_unicode_funcs_t::space_t
306 _hb_glyph_info_get_unicode_space_fallback_type (const hb_glyph_info_t *info)
308 return _hb_glyph_info_is_unicode_space (info) ?
309 (hb_unicode_funcs_t::space_t) (info->unicode_props()>>8) :
310 hb_unicode_funcs_t::NOT_SPACE;
313 static inline bool _hb_glyph_info_ligated (const hb_glyph_info_t *info);
316 _hb_glyph_info_is_default_ignorable (const hb_glyph_info_t *info)
318 return (info->unicode_props() & UPROPS_MASK_IGNORABLE) &&
319 !_hb_glyph_info_ligated (info);
322 _hb_glyph_info_is_default_ignorable_and_not_hidden (const hb_glyph_info_t *info)
324 return ((info->unicode_props() & (UPROPS_MASK_IGNORABLE|UPROPS_MASK_HIDDEN))
325 == UPROPS_MASK_IGNORABLE) &&
326 !_hb_glyph_info_ligated (info);
329 _hb_glyph_info_unhide (hb_glyph_info_t *info)
331 info->unicode_props() &= ~ UPROPS_MASK_HIDDEN;
335 _hb_glyph_info_set_continuation (hb_glyph_info_t *info)
337 info->unicode_props() |= UPROPS_MASK_CONTINUATION;
340 _hb_glyph_info_reset_continuation (hb_glyph_info_t *info)
342 info->unicode_props() &= ~ UPROPS_MASK_CONTINUATION;
345 _hb_glyph_info_is_continuation (const hb_glyph_info_t *info)
347 return info->unicode_props() & UPROPS_MASK_CONTINUATION;
349 /* Loop over grapheme. Based on foreach_cluster(). */
350 #define foreach_grapheme(buffer, start, end) \
352 _count = buffer->len, \
353 start = 0, end = _count ? _hb_next_grapheme (buffer, 0) : 0; \
355 start = end, end = _hb_next_grapheme (buffer, start))
357 static inline unsigned int
358 _hb_next_grapheme (hb_buffer_t *buffer, unsigned int start)
360 hb_glyph_info_t *info = buffer->info;
361 unsigned int count = buffer->len;
363 while (++start < count && _hb_glyph_info_is_continuation (&info[start]))
370 _hb_glyph_info_is_unicode_format (const hb_glyph_info_t *info)
372 return _hb_glyph_info_get_general_category (info) ==
373 HB_UNICODE_GENERAL_CATEGORY_FORMAT;
376 _hb_glyph_info_is_zwnj (const hb_glyph_info_t *info)
378 return _hb_glyph_info_is_unicode_format (info) && (info->unicode_props() & UPROPS_MASK_Cf_ZWNJ);
381 _hb_glyph_info_is_zwj (const hb_glyph_info_t *info)
383 return _hb_glyph_info_is_unicode_format (info) && (info->unicode_props() & UPROPS_MASK_Cf_ZWJ);
386 _hb_glyph_info_is_joiner (const hb_glyph_info_t *info)
388 return _hb_glyph_info_is_unicode_format (info) && (info->unicode_props() & (UPROPS_MASK_Cf_ZWNJ|UPROPS_MASK_Cf_ZWJ));
391 _hb_glyph_info_flip_joiners (hb_glyph_info_t *info)
393 if (!_hb_glyph_info_is_unicode_format (info))
395 info->unicode_props() ^= UPROPS_MASK_Cf_ZWNJ | UPROPS_MASK_Cf_ZWJ;
398 /* lig_props: aka lig_id / lig_comp
400 * When a ligature is formed:
402 * - The ligature glyph and any marks in between all the same newly allocated
404 * - The ligature glyph will get lig_num_comps set to the number of components
405 * - The marks get lig_comp > 0, reflecting which component of the ligature
406 * they were applied to.
407 * - This is used in GPOS to attach marks to the right component of a ligature
409 * - Note that when marks are ligated together, much of the above is skipped
410 * and the current lig_id reused.
412 * When a multiple-substitution is done:
414 * - All resulting glyphs will have lig_id = 0,
415 * - The resulting glyphs will have lig_comp = 0, 1, 2, ... respectively.
416 * - This is used in GPOS to attach marks to the first component of a
417 * multiple substitution in MarkBasePos.
419 * The numbers are also used in GPOS to do mark-to-mark positioning only
420 * to marks that belong to the same component of the same ligature.
424 _hb_glyph_info_clear_lig_props (hb_glyph_info_t *info)
426 info->lig_props() = 0;
429 #define IS_LIG_BASE 0x10
432 _hb_glyph_info_set_lig_props_for_ligature (hb_glyph_info_t *info,
434 unsigned int lig_num_comps)
436 info->lig_props() = (lig_id << 5) | IS_LIG_BASE | (lig_num_comps & 0x0F);
440 _hb_glyph_info_set_lig_props_for_mark (hb_glyph_info_t *info,
442 unsigned int lig_comp)
444 info->lig_props() = (lig_id << 5) | (lig_comp & 0x0F);
448 _hb_glyph_info_set_lig_props_for_component (hb_glyph_info_t *info, unsigned int comp)
450 _hb_glyph_info_set_lig_props_for_mark (info, 0, comp);
453 static inline unsigned int
454 _hb_glyph_info_get_lig_id (const hb_glyph_info_t *info)
456 return info->lig_props() >> 5;
460 _hb_glyph_info_ligated_internal (const hb_glyph_info_t *info)
462 return !!(info->lig_props() & IS_LIG_BASE);
465 static inline unsigned int
466 _hb_glyph_info_get_lig_comp (const hb_glyph_info_t *info)
468 if (_hb_glyph_info_ligated_internal (info))
471 return info->lig_props() & 0x0F;
474 static inline unsigned int
475 _hb_glyph_info_get_lig_num_comps (const hb_glyph_info_t *info)
477 if ((info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_LIGATURE) &&
478 _hb_glyph_info_ligated_internal (info))
479 return info->lig_props() & 0x0F;
484 static inline uint8_t
485 _hb_allocate_lig_id (hb_buffer_t *buffer) {
486 uint8_t lig_id = buffer->next_serial () & 0x07;
487 if (unlikely (!lig_id))
488 lig_id = _hb_allocate_lig_id (buffer); /* in case of overflow */
495 _hb_glyph_info_set_glyph_props (hb_glyph_info_t *info, unsigned int props)
497 info->glyph_props() = props;
500 static inline unsigned int
501 _hb_glyph_info_get_glyph_props (const hb_glyph_info_t *info)
503 return info->glyph_props();
507 _hb_glyph_info_is_base_glyph (const hb_glyph_info_t *info)
509 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_BASE_GLYPH);
513 _hb_glyph_info_is_ligature (const hb_glyph_info_t *info)
515 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_LIGATURE);
519 _hb_glyph_info_is_mark (const hb_glyph_info_t *info)
521 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_MARK);
525 _hb_glyph_info_substituted (const hb_glyph_info_t *info)
527 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_SUBSTITUTED);
531 _hb_glyph_info_ligated (const hb_glyph_info_t *info)
533 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_LIGATED);
537 _hb_glyph_info_multiplied (const hb_glyph_info_t *info)
539 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_MULTIPLIED);
543 _hb_glyph_info_ligated_and_didnt_multiply (const hb_glyph_info_t *info)
545 return _hb_glyph_info_ligated (info) && !_hb_glyph_info_multiplied (info);
549 _hb_glyph_info_clear_ligated_and_multiplied (hb_glyph_info_t *info)
551 info->glyph_props() &= ~(HB_OT_LAYOUT_GLYPH_PROPS_LIGATED |
552 HB_OT_LAYOUT_GLYPH_PROPS_MULTIPLIED);
556 _hb_glyph_info_clear_substituted (hb_glyph_info_t *info)
558 info->glyph_props() &= ~(HB_OT_LAYOUT_GLYPH_PROPS_SUBSTITUTED);
562 /* Allocation / deallocation. */
565 _hb_buffer_allocate_unicode_vars (hb_buffer_t *buffer)
567 HB_BUFFER_ALLOCATE_VAR (buffer, unicode_props);
571 _hb_buffer_deallocate_unicode_vars (hb_buffer_t *buffer)
573 HB_BUFFER_DEALLOCATE_VAR (buffer, unicode_props);
577 _hb_buffer_assert_unicode_vars (hb_buffer_t *buffer)
579 HB_BUFFER_ASSERT_VAR (buffer, unicode_props);
583 _hb_buffer_allocate_gsubgpos_vars (hb_buffer_t *buffer)
585 HB_BUFFER_ALLOCATE_VAR (buffer, glyph_props);
586 HB_BUFFER_ALLOCATE_VAR (buffer, lig_props);
587 HB_BUFFER_ALLOCATE_VAR (buffer, syllable);
591 _hb_buffer_deallocate_gsubgpos_vars (hb_buffer_t *buffer)
593 HB_BUFFER_DEALLOCATE_VAR (buffer, syllable);
594 HB_BUFFER_DEALLOCATE_VAR (buffer, lig_props);
595 HB_BUFFER_DEALLOCATE_VAR (buffer, glyph_props);
599 _hb_buffer_assert_gsubgpos_vars (hb_buffer_t *buffer)
601 HB_BUFFER_ASSERT_VAR (buffer, glyph_props);
602 HB_BUFFER_ASSERT_VAR (buffer, lig_props);
603 HB_BUFFER_ASSERT_VAR (buffer, syllable);
606 /* Make sure no one directly touches our props... */
607 #undef unicode_props0
608 #undef unicode_props1
612 #endif /* HB_OT_LAYOUT_HH */