1 /*
2 * Copyright © 2010,2011,2012 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 #ifndef HB_OT_SHAPE_COMPLEX_PRIVATE_HH
28 #define HB_OT_SHAPE_COMPLEX_PRIVATE_HH
29
30 #include "hb-private.hh"
31
32 #include "hb-ot-shape-private.hh"
33 #include "hb-ot-shape-normalize-private.hh"
34
35
36
37 /* buffer var allocations, used by complex shapers */
38 #define complex_var_u8_0() var2.u8[2]
39 #define complex_var_u8_1() var2.u8[3]
40
41
42 enum hb_ot_shape_zero_width_marks_type_t {
43 HB_OT_SHAPE_ZERO_WIDTH_MARKS_NONE,
44 // HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_UNICODE_EARLY,
45 HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_UNICODE_LATE,
46 HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_EARLY,
47 HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_LATE,
48
49 HB_OT_SHAPE_ZERO_WIDTH_MARKS_DEFAULT = HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_UNICODE_LATE
50 };
51
52
53 /* Master OT shaper list */
54 #define HB_COMPLEX_SHAPERS_IMPLEMENT_SHAPERS \
55 HB_COMPLEX_SHAPER_IMPLEMENT (default) /* should be first */ \
56 HB_COMPLEX_SHAPER_IMPLEMENT (arabic) \
57 HB_COMPLEX_SHAPER_IMPLEMENT (hangul) \
58 HB_COMPLEX_SHAPER_IMPLEMENT (hebrew) \
59 HB_COMPLEX_SHAPER_IMPLEMENT (indic) \
60 HB_COMPLEX_SHAPER_IMPLEMENT (myanmar) \
61 HB_COMPLEX_SHAPER_IMPLEMENT (sea) \
62 HB_COMPLEX_SHAPER_IMPLEMENT (thai) \
63 HB_COMPLEX_SHAPER_IMPLEMENT (tibetan) \
64 /* ^--- Add new shapers here */
65
66
67 struct hb_ot_complex_shaper_t
68 {
69 char name[8];
70
71 /* collect_features()
72 * Called during shape_plan().
73 * Shapers should use plan->map to add their features and callbacks.
74 * May be NULL.
75 */
76 void (*collect_features) (hb_ot_shape_planner_t *plan);
77
78 /* override_features()
79 * Called during shape_plan().
80 * Shapers should use plan->map to override features and add callbacks after
81 * common features are added.
82 * May be NULL.
83 */
84 void (*override_features) (hb_ot_shape_planner_t *plan);
85
86
87 /* data_create()
88 * Called at the end of shape_plan().
89 * Whatever shapers return will be accessible through plan->data later.
90 * If NULL is returned, means a plan failure.
91 */
92 void *(*data_create) (const hb_ot_shape_plan_t *plan);
93
94 /* data_destroy()
95 * Called when the shape_plan is being destroyed.
96 * plan->data is passed here for destruction.
97 * If NULL is returned, means a plan failure.
98 * May be NULL.
99 */
100 void (*data_destroy) (void *data);
101
102
103 /* preprocess_text()
104 * Called during shape().
105 * Shapers can use to modify text before shaping starts.
106 * May be NULL.
107 */
108 void (*preprocess_text) (const hb_ot_shape_plan_t *plan,
109 hb_buffer_t *buffer,
110 hb_font_t *font);
111
112
113 hb_ot_shape_normalization_mode_t normalization_preference;
114
115 /* decompose()
116 * Called during shape()'s normalization.
117 * May be NULL.
118 */
119 bool (*decompose) (const hb_ot_shape_normalize_context_t *c,
120 hb_codepoint_t ab,
121 hb_codepoint_t *a,
122 hb_codepoint_t *b);
123
124 /* compose()
125 * Called during shape()'s normalization.
126 * May be NULL.
127 */
128 bool (*compose) (const hb_ot_shape_normalize_context_t *c,
129 hb_codepoint_t a,
130 hb_codepoint_t b,
131 hb_codepoint_t *ab);
132
133 /* setup_masks()
134 * Called during shape().
135 * Shapers should use map to get feature masks and set on buffer.
136 * Shapers may NOT modify characters.
137 * May be NULL.
138 */
139 void (*setup_masks) (const hb_ot_shape_plan_t *plan,
140 hb_buffer_t *buffer,
141 hb_font_t *font);
142
143 hb_ot_shape_zero_width_marks_type_t zero_width_marks;
144
145 bool fallback_position;
146 };
147
148 #define HB_COMPLEX_SHAPER_IMPLEMENT(name) extern HB_INTERNAL const hb_ot_complex_shaper_t _hb_ot_complex_shaper_##name;
149 HB_COMPLEX_SHAPERS_IMPLEMENT_SHAPERS
150 #undef HB_COMPLEX_SHAPER_IMPLEMENT
151
152
153 static inline const hb_ot_complex_shaper_t *
hb_ot_shape_complex_categorize(const hb_ot_shape_planner_t * planner)154 hb_ot_shape_complex_categorize (const hb_ot_shape_planner_t *planner)
155 {
156 switch ((hb_tag_t) planner->props.script)
157 {
158 default:
159 return &_hb_ot_complex_shaper_default;
160
161
162 /* Unicode-1.1 additions */
163 case HB_SCRIPT_ARABIC:
164
165 /* Unicode-3.0 additions */
166 case HB_SCRIPT_MONGOLIAN:
167 case HB_SCRIPT_SYRIAC:
168
169 /* Unicode-5.0 additions */
170 case HB_SCRIPT_NKO:
171 case HB_SCRIPT_PHAGS_PA:
172
173 /* Unicode-6.0 additions */
174 case HB_SCRIPT_MANDAIC:
175
176 /* Unicode-7.0 additions */
177 case HB_SCRIPT_MANICHAEAN:
178 case HB_SCRIPT_PSALTER_PAHLAVI:
179
180 /* For Arabic script, use the Arabic shaper even if no OT script tag was found.
181 * This is because we do fallback shaping for Arabic script (and not others). */
182 if (planner->map.chosen_script[0] != HB_OT_TAG_DEFAULT_SCRIPT ||
183 planner->props.script == HB_SCRIPT_ARABIC)
184 return &_hb_ot_complex_shaper_arabic;
185 else
186 return &_hb_ot_complex_shaper_default;
187
188
189 /* Unicode-1.1 additions */
190 case HB_SCRIPT_THAI:
191 case HB_SCRIPT_LAO:
192
193 return &_hb_ot_complex_shaper_thai;
194
195
196 /* Unicode-1.1 additions */
197 case HB_SCRIPT_HANGUL:
198
199 return &_hb_ot_complex_shaper_hangul;
200
201
202 /* Unicode-2.0 additions */
203 case HB_SCRIPT_TIBETAN:
204
205 return &_hb_ot_complex_shaper_tibetan;
206
207
208 /* Unicode-1.1 additions */
209 case HB_SCRIPT_HEBREW:
210
211 return &_hb_ot_complex_shaper_hebrew;
212
213
214 /* ^--- Add new shapers here */
215
216
217 #if 0
218 /* Note:
219 *
220 * These disabled scripts are listed in ucd/IndicSyllabicCategory.txt, but according
221 * to Martin Hosken and Jonathan Kew do not require complex shaping.
222 *
223 * TODO We should automate figuring out which scripts do not need complex shaping
224 *
225 * TODO We currently keep data for these scripts in our indic table. Need to fix the
226 * generator to not do that.
227 */
228
229
230 /* Simple? */
231
232 /* Unicode-3.2 additions */
233 case HB_SCRIPT_BUHID:
234 case HB_SCRIPT_HANUNOO:
235
236 /* Unicode-5.1 additions */
237 case HB_SCRIPT_SAURASHTRA:
238
239 /* Unicode-6.0 additions */
240 case HB_SCRIPT_BATAK:
241 case HB_SCRIPT_BRAHMI:
242
243
244 /* Simple */
245
246 /* Unicode-1.1 additions */
247 /* These have their own shaper now. */
248 case HB_SCRIPT_LAO:
249 case HB_SCRIPT_THAI:
250
251 /* Unicode-3.2 additions */
252 case HB_SCRIPT_TAGALOG:
253 case HB_SCRIPT_TAGBANWA:
254
255 /* Unicode-4.0 additions */
256 case HB_SCRIPT_LIMBU:
257 case HB_SCRIPT_TAI_LE:
258
259 /* Unicode-4.1 additions */
260 case HB_SCRIPT_KHAROSHTHI:
261 case HB_SCRIPT_SYLOTI_NAGRI:
262
263 /* Unicode-5.1 additions */
264 case HB_SCRIPT_KAYAH_LI:
265
266 /* Unicode-5.2 additions */
267 case HB_SCRIPT_TAI_VIET:
268
269
270 #endif
271
272 /* Unicode-1.1 additions */
273 case HB_SCRIPT_BENGALI:
274 case HB_SCRIPT_DEVANAGARI:
275 case HB_SCRIPT_GUJARATI:
276 case HB_SCRIPT_GURMUKHI:
277 case HB_SCRIPT_KANNADA:
278 case HB_SCRIPT_MALAYALAM:
279 case HB_SCRIPT_ORIYA:
280 case HB_SCRIPT_TAMIL:
281 case HB_SCRIPT_TELUGU:
282
283 /* Unicode-3.0 additions */
284 case HB_SCRIPT_SINHALA:
285
286 /* Unicode-5.0 additions */
287 case HB_SCRIPT_BALINESE:
288
289 /* Unicode-5.1 additions */
290 case HB_SCRIPT_LEPCHA:
291 case HB_SCRIPT_REJANG:
292 case HB_SCRIPT_SUNDANESE:
293
294 /* Unicode-5.2 additions */
295 case HB_SCRIPT_JAVANESE:
296 case HB_SCRIPT_KAITHI:
297 case HB_SCRIPT_MEETEI_MAYEK:
298
299 /* Unicode-6.0 additions */
300
301 /* Unicode-6.1 additions */
302 case HB_SCRIPT_CHAKMA:
303 case HB_SCRIPT_SHARADA:
304 case HB_SCRIPT_TAKRI:
305
306 /* If the designer designed the font for the 'DFLT' script,
307 * use the default shaper. Otherwise, use the Indic shaper.
308 * Note that for some simple scripts, there may not be *any*
309 * GSUB/GPOS needed, so there may be no scripts found! */
310 if (planner->map.chosen_script[0] == HB_TAG ('D','F','L','T'))
311 return &_hb_ot_complex_shaper_default;
312 else
313 return &_hb_ot_complex_shaper_indic;
314
315 case HB_SCRIPT_KHMER:
316 /* A number of Khmer fonts in the wild don't have a 'pref' feature,
317 * and as such won't shape properly via the Indic shaper;
318 * however, they typically have 'liga' / 'clig' features that implement
319 * the necessary "reordering" by means of ligature substitutions.
320 * So we send such pref-less fonts through the generic shaper instead. */
321 if (planner->map.found_script[0] &&
322 hb_ot_layout_language_find_feature (planner->face, HB_OT_TAG_GSUB,
323 planner->map.script_index[0],
324 planner->map.language_index[0],
325 HB_TAG ('p','r','e','f'),
326 NULL))
327 return &_hb_ot_complex_shaper_indic;
328 else
329 return &_hb_ot_complex_shaper_default;
330
331 case HB_SCRIPT_MYANMAR:
332 /* For Myanmar, we only want to use the Myanmar shaper if the "new" script
333 * tag is found. For "old" script tag we want to use the default shaper. */
334 if (planner->map.chosen_script[0] == HB_TAG ('m','y','m','2'))
335 return &_hb_ot_complex_shaper_myanmar;
336 else
337 return &_hb_ot_complex_shaper_default;
338
339 /* Unicode-4.1 additions */
340 case HB_SCRIPT_BUGINESE:
341 case HB_SCRIPT_NEW_TAI_LUE:
342
343 /* Unicode-5.1 additions */
344 case HB_SCRIPT_CHAM:
345
346 /* Unicode-5.2 additions */
347 case HB_SCRIPT_TAI_THAM:
348
349 /* If the designer designed the font for the 'DFLT' script,
350 * use the default shaper. Otherwise, use the Indic shaper.
351 * Note that for some simple scripts, there may not be *any*
352 * GSUB/GPOS needed, so there may be no scripts found! */
353 if (planner->map.chosen_script[0] == HB_TAG ('D','F','L','T'))
354 return &_hb_ot_complex_shaper_default;
355 else
356 return &_hb_ot_complex_shaper_sea;
357 }
358 }
359
360
361 #endif /* HB_OT_SHAPE_COMPLEX_PRIVATE_HH */
362