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_PRIVATE_HH
30 #define HB_OT_LAYOUT_PRIVATE_HH
32 #include "hb-private.hh"
34 #include "hb-font-private.hh"
35 #include "hb-buffer-private.hh"
36 #include "hb-set-private.hh"
39 /* Private API corresponding to hb-ot-layout.h: */
42 hb_ot_layout_table_find_feature (hb_face_t *face,
45 unsigned int *feature_index);
52 enum hb_ot_layout_glyph_props_flags_t
54 /* The following three match LookupFlags::Ignore* numbers. */
55 HB_OT_LAYOUT_GLYPH_PROPS_BASE_GLYPH = 0x02u,
56 HB_OT_LAYOUT_GLYPH_PROPS_LIGATURE = 0x04u,
57 HB_OT_LAYOUT_GLYPH_PROPS_MARK = 0x08u,
59 /* The following are used internally; not derived from GDEF. */
60 HB_OT_LAYOUT_GLYPH_PROPS_SUBSTITUTED = 0x10u,
61 HB_OT_LAYOUT_GLYPH_PROPS_LIGATED = 0x20u,
62 HB_OT_LAYOUT_GLYPH_PROPS_MULTIPLIED = 0x40u,
64 HB_OT_LAYOUT_GLYPH_PROPS_PRESERVE = HB_OT_LAYOUT_GLYPH_PROPS_SUBSTITUTED |
65 HB_OT_LAYOUT_GLYPH_PROPS_LIGATED |
66 HB_OT_LAYOUT_GLYPH_PROPS_MULTIPLIED
68 HB_MARK_AS_FLAG_T (hb_ot_layout_glyph_props_flags_t);
76 hb_ot_layout_lookup_would_substitute_fast (hb_face_t *face,
77 unsigned int lookup_index,
78 const hb_codepoint_t *glyphs,
79 unsigned int glyphs_length,
80 hb_bool_t zero_context);
83 /* Should be called before all the substitute_lookup's are done. */
85 hb_ot_layout_substitute_start (hb_font_t *font,
89 struct hb_ot_layout_lookup_accelerator_t;
92 struct hb_apply_context_t;
97 hb_ot_layout_substitute_lookup (OT::hb_apply_context_t *c,
98 const OT::SubstLookup &lookup,
99 const hb_ot_layout_lookup_accelerator_t &accel);
102 /* Should be called before all the position_lookup's are done. */
104 hb_ot_layout_position_start (hb_font_t *font,
105 hb_buffer_t *buffer);
107 /* Should be called after all the position_lookup's are done, to finish advances. */
109 hb_ot_layout_position_finish_advances (hb_font_t *font,
110 hb_buffer_t *buffer);
112 /* Should be called after hb_ot_layout_position_finish_advances, to finish offsets. */
114 hb_ot_layout_position_finish_offsets (hb_font_t *font,
115 hb_buffer_t *buffer);
129 struct hb_ot_layout_lookup_accelerator_t
131 template <typename TLookup>
132 inline void init (const TLookup &lookup)
135 lookup.add_coverage (&digest);
138 inline void fini (void)
142 inline bool may_have (hb_codepoint_t g) const {
143 return digest.may_have (g);
147 hb_set_digest_t digest;
150 struct hb_ot_layout_t
152 hb_blob_t *gdef_blob;
153 hb_blob_t *gsub_blob;
154 hb_blob_t *gpos_blob;
156 const struct OT::GDEF *gdef;
157 const struct OT::GSUB *gsub;
158 const struct OT::GPOS *gpos;
160 unsigned int gsub_lookup_count;
161 unsigned int gpos_lookup_count;
163 hb_ot_layout_lookup_accelerator_t *gsub_accels;
164 hb_ot_layout_lookup_accelerator_t *gpos_accels;
168 HB_INTERNAL hb_ot_layout_t *
169 _hb_ot_layout_create (hb_face_t *face);
172 _hb_ot_layout_destroy (hb_ot_layout_t *layout);
175 #define hb_ot_layout_from_face(face) ((hb_ot_layout_t *) face->shaper_data.ot)
179 * Buffer var routines.
182 /* buffer var allocations, used during the entire shaping process */
183 #define unicode_props() var2.u16[0]
185 /* buffer var allocations, used during the GSUB/GPOS processing */
186 #define glyph_props() var1.u16[0] /* GDEF glyph properties */
187 #define lig_props() var1.u8[2] /* GSUB/GPOS ligature tracking */
188 #define syllable() var1.u8[3] /* GSUB/GPOS shaping boundaries */
191 /* loop over syllables */
193 #define foreach_syllable(buffer, start, end) \
195 _count = buffer->len, \
196 start = 0, end = _count ? _next_syllable (buffer, 0) : 0; \
198 start = end, end = _next_syllable (buffer, start))
200 static inline unsigned int
201 _next_syllable (hb_buffer_t *buffer, unsigned int start)
203 hb_glyph_info_t *info = buffer->info;
204 unsigned int count = buffer->len;
206 unsigned int syllable = info[start].syllable();
207 while (++start < count && syllable == info[start].syllable())
217 * unicode_props() is a two-byte number. The low byte includes:
218 * - General_Category: 5 bits.
220 * * Is it Default_Ignorable(); we have a modified Default_Ignorable().
221 * * Whether it's one of the three Mongolian Free Variation Selectors.
222 * * One free bit right now.
224 * The high-byte has different meanings, switched by the Gen-Cat:
225 * - For Mn,Mc,Me: the modified Combining_Class.
226 * - For Cf: whether it's ZWJ, ZWNJ, or something else.
227 * - For Ws: index of which space character this is, if space fallback
228 * is needed, ie. we don't set this by default, only if asked to.
231 enum hb_unicode_props_flags_t {
232 UPROPS_MASK_GEN_CAT = 0x001Fu,
233 UPROPS_MASK_IGNORABLE = 0x0020u,
234 UPROPS_MASK_FVS = 0x0040u, /* MONGOLIAN FREE VARIATION SELECTOR 1..3 */
236 /* If GEN_CAT=FORMAT, top byte masks: */
237 UPROPS_MASK_Cf_ZWJ = 0x0100u,
238 UPROPS_MASK_Cf_ZWNJ = 0x0200u
240 HB_MARK_AS_FLAG_T (hb_unicode_props_flags_t);
243 _hb_glyph_info_set_unicode_props (hb_glyph_info_t *info, hb_buffer_t *buffer)
245 hb_unicode_funcs_t *unicode = buffer->unicode;
246 unsigned int u = info->codepoint;
247 unsigned int gen_cat = (unsigned int) unicode->general_category (u);
248 unsigned int props = gen_cat;
252 buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_NON_ASCII;
253 if (unlikely (unicode->is_default_ignorable (u)))
255 buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_DEFAULT_IGNORABLES;
256 props |= UPROPS_MASK_IGNORABLE;
257 if (u == 0x200Cu) props |= UPROPS_MASK_Cf_ZWNJ;
258 if (u == 0x200Du) props |= UPROPS_MASK_Cf_ZWJ;
259 /* Mongolian Free Variation Selectors need to be remembered
260 * because although we need to hide them like default-ignorables,
261 * they need to non-ignorable during shaping. This is similar to
262 * what we do for joiners in Indic-like shapers, but since the
263 * FVSes are GC=Mn, we have use a separate bit to remember them.
265 * https://github.com/behdad/harfbuzz/issues/234
267 if (unlikely (hb_in_range (u, 0x180Bu, 0x180Du))) props |= UPROPS_MASK_FVS;
269 else if (unlikely (HB_UNICODE_GENERAL_CATEGORY_IS_NON_ENCLOSING_MARK_OR_MODIFIER_SYMBOL (gen_cat)))
271 /* The above check is just an optimization to let in only things we need further
274 /* Only Mn and Mc can have non-zero ccc:
275 * http://www.unicode.org/policies/stability_policy.html#Property_Value
277 * Canonical_Combining_Class, General_Category
278 * All characters other than those with General_Category property values
279 * Spacing_Mark (Mc) and Nonspacing_Mark (Mn) have the Canonical_Combining_Class
284 * Also, all Mn's that are Default_Ignorable, have ccc=0, hence
287 props |= unicode->modified_combining_class (info->codepoint)<<8;
289 /* Recategorize emoji skin-tone modifiers as Unicode mark, so they
290 * behave correctly in non-native directionality. They originally
291 * are MODIFIER_SYMBOL. Fixes:
292 * https://github.com/behdad/harfbuzz/issues/169
294 if (unlikely (hb_in_range (u, 0x1F3FBu, 0x1F3FFu)))
296 props = gen_cat = HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK;
301 info->unicode_props() = props;
305 _hb_glyph_info_set_general_category (hb_glyph_info_t *info,
306 hb_unicode_general_category_t gen_cat)
308 /* Clears top-byte. */
309 info->unicode_props() = (unsigned int) gen_cat | (info->unicode_props() & (0xFF & ~UPROPS_MASK_GEN_CAT));
312 static inline hb_unicode_general_category_t
313 _hb_glyph_info_get_general_category (const hb_glyph_info_t *info)
315 return (hb_unicode_general_category_t) (info->unicode_props() & UPROPS_MASK_GEN_CAT);
319 _hb_glyph_info_is_unicode_mark (const hb_glyph_info_t *info)
321 return HB_UNICODE_GENERAL_CATEGORY_IS_MARK (info->unicode_props() & UPROPS_MASK_GEN_CAT);
324 _hb_glyph_info_set_modified_combining_class (hb_glyph_info_t *info,
325 unsigned int modified_class)
327 if (unlikely (!_hb_glyph_info_is_unicode_mark (info)))
329 info->unicode_props() = (modified_class<<8) | (info->unicode_props() & 0xFF);
331 static inline unsigned int
332 _hb_glyph_info_get_modified_combining_class (const hb_glyph_info_t *info)
334 return _hb_glyph_info_is_unicode_mark (info) ? info->unicode_props()>>8 : 0;
338 _hb_glyph_info_is_unicode_space (const hb_glyph_info_t *info)
340 return _hb_glyph_info_get_general_category (info) ==
341 HB_UNICODE_GENERAL_CATEGORY_SPACE_SEPARATOR;
344 _hb_glyph_info_set_unicode_space_fallback_type (hb_glyph_info_t *info, hb_unicode_funcs_t::space_t s)
346 if (unlikely (!_hb_glyph_info_is_unicode_space (info)))
348 info->unicode_props() = (((unsigned int) s)<<8) | (info->unicode_props() & 0xFF);
350 static inline hb_unicode_funcs_t::space_t
351 _hb_glyph_info_get_unicode_space_fallback_type (const hb_glyph_info_t *info)
353 return _hb_glyph_info_is_unicode_space (info) ?
354 (hb_unicode_funcs_t::space_t) (info->unicode_props()>>8) :
355 hb_unicode_funcs_t::NOT_SPACE;
358 static inline bool _hb_glyph_info_ligated (const hb_glyph_info_t *info);
360 static inline hb_bool_t
361 _hb_glyph_info_is_default_ignorable (const hb_glyph_info_t *info)
363 return (info->unicode_props() & UPROPS_MASK_IGNORABLE) &&
364 !_hb_glyph_info_ligated (info);
366 static inline hb_bool_t
367 _hb_glyph_info_is_default_ignorable_and_not_fvs (const hb_glyph_info_t *info)
369 return ((info->unicode_props() & (UPROPS_MASK_IGNORABLE|UPROPS_MASK_FVS))
370 == UPROPS_MASK_IGNORABLE) &&
371 !_hb_glyph_info_ligated (info);
375 _hb_glyph_info_is_unicode_format (const hb_glyph_info_t *info)
377 return _hb_glyph_info_get_general_category (info) ==
378 HB_UNICODE_GENERAL_CATEGORY_FORMAT;
380 static inline hb_bool_t
381 _hb_glyph_info_is_zwnj (const hb_glyph_info_t *info)
383 return _hb_glyph_info_is_unicode_format (info) && (info->unicode_props() & UPROPS_MASK_Cf_ZWNJ);
385 static inline hb_bool_t
386 _hb_glyph_info_is_zwj (const hb_glyph_info_t *info)
388 return _hb_glyph_info_is_unicode_format (info) && (info->unicode_props() & UPROPS_MASK_Cf_ZWJ);
390 static inline hb_bool_t
391 _hb_glyph_info_is_joiner (const hb_glyph_info_t *info)
393 return _hb_glyph_info_is_unicode_format (info) && (info->unicode_props() & (UPROPS_MASK_Cf_ZWNJ|UPROPS_MASK_Cf_ZWJ));
396 _hb_glyph_info_flip_joiners (hb_glyph_info_t *info)
398 if (!_hb_glyph_info_is_unicode_format (info))
400 info->unicode_props() ^= UPROPS_MASK_Cf_ZWNJ | UPROPS_MASK_Cf_ZWJ;
403 /* lig_props: aka lig_id / lig_comp
405 * When a ligature is formed:
407 * - The ligature glyph and any marks in between all the same newly allocated
409 * - The ligature glyph will get lig_num_comps set to the number of components
410 * - The marks get lig_comp > 0, reflecting which component of the ligature
411 * they were applied to.
412 * - This is used in GPOS to attach marks to the right component of a ligature
414 * - Note that when marks are ligated together, much of the above is skipped
415 * and the current lig_id reused.
417 * When a multiple-substitution is done:
419 * - All resulting glyphs will have lig_id = 0,
420 * - The resulting glyphs will have lig_comp = 0, 1, 2, ... respectively.
421 * - This is used in GPOS to attach marks to the first component of a
422 * multiple substitution in MarkBasePos.
424 * The numbers are also used in GPOS to do mark-to-mark positioning only
425 * to marks that belong to the same component of the same ligature.
429 _hb_glyph_info_clear_lig_props (hb_glyph_info_t *info)
431 info->lig_props() = 0;
434 #define IS_LIG_BASE 0x10
437 _hb_glyph_info_set_lig_props_for_ligature (hb_glyph_info_t *info,
439 unsigned int lig_num_comps)
441 info->lig_props() = (lig_id << 5) | IS_LIG_BASE | (lig_num_comps & 0x0F);
445 _hb_glyph_info_set_lig_props_for_mark (hb_glyph_info_t *info,
447 unsigned int lig_comp)
449 info->lig_props() = (lig_id << 5) | (lig_comp & 0x0F);
453 _hb_glyph_info_set_lig_props_for_component (hb_glyph_info_t *info, unsigned int comp)
455 _hb_glyph_info_set_lig_props_for_mark (info, 0, comp);
458 static inline unsigned int
459 _hb_glyph_info_get_lig_id (const hb_glyph_info_t *info)
461 return info->lig_props() >> 5;
465 _hb_glyph_info_ligated_internal (const hb_glyph_info_t *info)
467 return !!(info->lig_props() & IS_LIG_BASE);
470 static inline unsigned int
471 _hb_glyph_info_get_lig_comp (const hb_glyph_info_t *info)
473 if (_hb_glyph_info_ligated_internal (info))
476 return info->lig_props() & 0x0F;
479 static inline unsigned int
480 _hb_glyph_info_get_lig_num_comps (const hb_glyph_info_t *info)
482 if ((info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_LIGATURE) &&
483 _hb_glyph_info_ligated_internal (info))
484 return info->lig_props() & 0x0F;
489 static inline uint8_t
490 _hb_allocate_lig_id (hb_buffer_t *buffer) {
491 uint8_t lig_id = buffer->next_serial () & 0x07;
492 if (unlikely (!lig_id))
493 lig_id = _hb_allocate_lig_id (buffer); /* in case of overflow */
500 _hb_glyph_info_set_glyph_props (hb_glyph_info_t *info, unsigned int props)
502 info->glyph_props() = props;
505 static inline unsigned int
506 _hb_glyph_info_get_glyph_props (const hb_glyph_info_t *info)
508 return info->glyph_props();
512 _hb_glyph_info_is_base_glyph (const hb_glyph_info_t *info)
514 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_BASE_GLYPH);
518 _hb_glyph_info_is_ligature (const hb_glyph_info_t *info)
520 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_LIGATURE);
524 _hb_glyph_info_is_mark (const hb_glyph_info_t *info)
526 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_MARK);
530 _hb_glyph_info_substituted (const hb_glyph_info_t *info)
532 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_SUBSTITUTED);
536 _hb_glyph_info_ligated (const hb_glyph_info_t *info)
538 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_LIGATED);
542 _hb_glyph_info_multiplied (const hb_glyph_info_t *info)
544 return !!(info->glyph_props() & HB_OT_LAYOUT_GLYPH_PROPS_MULTIPLIED);
548 _hb_glyph_info_ligated_and_didnt_multiply (const hb_glyph_info_t *info)
550 return _hb_glyph_info_ligated (info) && !_hb_glyph_info_multiplied (info);
554 _hb_glyph_info_clear_ligated_and_multiplied (hb_glyph_info_t *info)
556 info->glyph_props() &= ~(HB_OT_LAYOUT_GLYPH_PROPS_LIGATED |
557 HB_OT_LAYOUT_GLYPH_PROPS_MULTIPLIED);
561 _hb_glyph_info_clear_substituted (hb_glyph_info_t *info)
563 info->glyph_props() &= ~(HB_OT_LAYOUT_GLYPH_PROPS_SUBSTITUTED);
567 /* Allocation / deallocation. */
570 _hb_buffer_allocate_unicode_vars (hb_buffer_t *buffer)
572 HB_BUFFER_ALLOCATE_VAR (buffer, unicode_props);
576 _hb_buffer_deallocate_unicode_vars (hb_buffer_t *buffer)
578 HB_BUFFER_DEALLOCATE_VAR (buffer, unicode_props);
582 _hb_buffer_assert_unicode_vars (hb_buffer_t *buffer)
584 HB_BUFFER_ASSERT_VAR (buffer, unicode_props);
588 _hb_buffer_allocate_gsubgpos_vars (hb_buffer_t *buffer)
590 HB_BUFFER_ALLOCATE_VAR (buffer, glyph_props);
591 HB_BUFFER_ALLOCATE_VAR (buffer, lig_props);
592 HB_BUFFER_ALLOCATE_VAR (buffer, syllable);
596 _hb_buffer_deallocate_gsubgpos_vars (hb_buffer_t *buffer)
598 HB_BUFFER_DEALLOCATE_VAR (buffer, syllable);
599 HB_BUFFER_DEALLOCATE_VAR (buffer, lig_props);
600 HB_BUFFER_DEALLOCATE_VAR (buffer, glyph_props);
604 _hb_buffer_assert_gsubgpos_vars (hb_buffer_t *buffer)
606 HB_BUFFER_ASSERT_VAR (buffer, glyph_props);
607 HB_BUFFER_ASSERT_VAR (buffer, lig_props);
608 HB_BUFFER_ASSERT_VAR (buffer, syllable);
611 /* Make sure no one directly touches our props... */
612 #undef unicode_props0
613 #undef unicode_props1
618 #endif /* HB_OT_LAYOUT_PRIVATE_HH */