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 (myanmar_old) \
60 HB_COMPLEX_SHAPER_IMPLEMENT (indic) \
61 HB_COMPLEX_SHAPER_IMPLEMENT (myanmar) \
62 HB_COMPLEX_SHAPER_IMPLEMENT (sea) \
63 HB_COMPLEX_SHAPER_IMPLEMENT (thai) \
64 HB_COMPLEX_SHAPER_IMPLEMENT (tibetan) \
65 /* ^--- Add new shapers here */
66
67
68 struct hb_ot_complex_shaper_t
69 {
70 char name[8];
71
72 /* collect_features()
73 * Called during shape_plan().
74 * Shapers should use plan->map to add their features and callbacks.
75 * May be NULL.
76 */
77 void (*collect_features) (hb_ot_shape_planner_t *plan);
78
79 /* override_features()
80 * Called during shape_plan().
81 * Shapers should use plan->map to override features and add callbacks after
82 * common features are added.
83 * May be NULL.
84 */
85 void (*override_features) (hb_ot_shape_planner_t *plan);
86
87
88 /* data_create()
89 * Called at the end of shape_plan().
90 * Whatever shapers return will be accessible through plan->data later.
91 * If NULL is returned, means a plan failure.
92 */
93 void *(*data_create) (const hb_ot_shape_plan_t *plan);
94
95 /* data_destroy()
96 * Called when the shape_plan is being destroyed.
97 * plan->data is passed here for destruction.
98 * If NULL is returned, means a plan failure.
99 * May be NULL.
100 */
101 void (*data_destroy) (void *data);
102
103
104 /* preprocess_text()
105 * Called during shape().
106 * Shapers can use to modify text before shaping starts.
107 * May be NULL.
108 */
109 void (*preprocess_text) (const hb_ot_shape_plan_t *plan,
110 hb_buffer_t *buffer,
111 hb_font_t *font);
112
113
114 hb_ot_shape_normalization_mode_t normalization_preference;
115
116 /* decompose()
117 * Called during shape()'s normalization.
118 * May be NULL.
119 */
120 bool (*decompose) (const hb_ot_shape_normalize_context_t *c,
121 hb_codepoint_t ab,
122 hb_codepoint_t *a,
123 hb_codepoint_t *b);
124
125 /* compose()
126 * Called during shape()'s normalization.
127 * May be NULL.
128 */
129 bool (*compose) (const hb_ot_shape_normalize_context_t *c,
130 hb_codepoint_t a,
131 hb_codepoint_t b,
132 hb_codepoint_t *ab);
133
134 /* setup_masks()
135 * Called during shape().
136 * Shapers should use map to get feature masks and set on buffer.
137 * Shapers may NOT modify characters.
138 * May be NULL.
139 */
140 void (*setup_masks) (const hb_ot_shape_plan_t *plan,
141 hb_buffer_t *buffer,
142 hb_font_t *font);
143
144 hb_ot_shape_zero_width_marks_type_t zero_width_marks;
145
146 bool fallback_position;
147 };
148
149 #define HB_COMPLEX_SHAPER_IMPLEMENT(name) extern HB_INTERNAL const hb_ot_complex_shaper_t _hb_ot_complex_shaper_##name;
150 HB_COMPLEX_SHAPERS_IMPLEMENT_SHAPERS
151 #undef HB_COMPLEX_SHAPER_IMPLEMENT
152
153
154 static inline const hb_ot_complex_shaper_t *
hb_ot_shape_complex_categorize(const hb_ot_shape_planner_t * planner)155 hb_ot_shape_complex_categorize (const hb_ot_shape_planner_t *planner)
156 {
157 switch ((hb_tag_t) planner->props.script)
158 {
159 default:
160 return &_hb_ot_complex_shaper_default;
161
162
163 /* Unicode-1.1 additions */
164 case HB_SCRIPT_ARABIC:
165
166 /* Unicode-3.0 additions */
167 case HB_SCRIPT_MONGOLIAN:
168 case HB_SCRIPT_SYRIAC:
169
170 /* Unicode-5.0 additions */
171 case HB_SCRIPT_NKO:
172 case HB_SCRIPT_PHAGS_PA:
173
174 /* Unicode-6.0 additions */
175 case HB_SCRIPT_MANDAIC:
176
177 /* Unicode-7.0 additions */
178 case HB_SCRIPT_MANICHAEAN:
179 case HB_SCRIPT_PSALTER_PAHLAVI:
180
181 /* For Arabic script, use the Arabic shaper even if no OT script tag was found.
182 * This is because we do fallback shaping for Arabic script (and not others). */
183 if (planner->map.chosen_script[0] != HB_OT_TAG_DEFAULT_SCRIPT ||
184 planner->props.script == HB_SCRIPT_ARABIC)
185 return &_hb_ot_complex_shaper_arabic;
186 else
187 return &_hb_ot_complex_shaper_default;
188
189
190 /* Unicode-1.1 additions */
191 case HB_SCRIPT_THAI:
192 case HB_SCRIPT_LAO:
193
194 return &_hb_ot_complex_shaper_thai;
195
196
197 /* Unicode-1.1 additions */
198 case HB_SCRIPT_HANGUL:
199
200 return &_hb_ot_complex_shaper_hangul;
201
202
203 /* Unicode-2.0 additions */
204 case HB_SCRIPT_TIBETAN:
205
206 return &_hb_ot_complex_shaper_tibetan;
207
208
209 /* Unicode-1.1 additions */
210 case HB_SCRIPT_HEBREW:
211
212 return &_hb_ot_complex_shaper_hebrew;
213
214
215 /* ^--- Add new shapers here */
216
217
218 #if 0
219 /* Note:
220 *
221 * These disabled scripts are listed in ucd/IndicSyllabicCategory.txt, but according
222 * to Martin Hosken and Jonathan Kew do not require complex shaping.
223 *
224 * TODO We should automate figuring out which scripts do not need complex shaping
225 *
226 * TODO We currently keep data for these scripts in our indic table. Need to fix the
227 * generator to not do that.
228 */
229
230
231 /* Simple? */
232
233 /* Unicode-3.2 additions */
234 case HB_SCRIPT_BUHID:
235 case HB_SCRIPT_HANUNOO:
236
237 /* Unicode-5.1 additions */
238 case HB_SCRIPT_SAURASHTRA:
239
240 /* Unicode-6.0 additions */
241 case HB_SCRIPT_BATAK:
242 case HB_SCRIPT_BRAHMI:
243
244
245 /* Simple */
246
247 /* Unicode-1.1 additions */
248 /* These have their own shaper now. */
249 case HB_SCRIPT_LAO:
250 case HB_SCRIPT_THAI:
251
252 /* Unicode-3.2 additions */
253 case HB_SCRIPT_TAGALOG:
254 case HB_SCRIPT_TAGBANWA:
255
256 /* Unicode-4.0 additions */
257 case HB_SCRIPT_LIMBU:
258 case HB_SCRIPT_TAI_LE:
259
260 /* Unicode-4.1 additions */
261 case HB_SCRIPT_KHAROSHTHI:
262 case HB_SCRIPT_NEW_TAI_LUE:
263 case HB_SCRIPT_SYLOTI_NAGRI:
264
265 /* Unicode-5.1 additions */
266 case HB_SCRIPT_KAYAH_LI:
267
268 /* Unicode-5.2 additions */
269 case HB_SCRIPT_TAI_VIET:
270
271
272 #endif
273
274 /* Unicode-1.1 additions */
275 case HB_SCRIPT_BENGALI:
276 case HB_SCRIPT_DEVANAGARI:
277 case HB_SCRIPT_GUJARATI:
278 case HB_SCRIPT_GURMUKHI:
279 case HB_SCRIPT_KANNADA:
280 case HB_SCRIPT_MALAYALAM:
281 case HB_SCRIPT_ORIYA:
282 case HB_SCRIPT_TAMIL:
283 case HB_SCRIPT_TELUGU:
284
285 /* Unicode-3.0 additions */
286 case HB_SCRIPT_SINHALA:
287
288 /* Unicode-5.0 additions */
289 case HB_SCRIPT_BALINESE:
290
291 /* Unicode-5.1 additions */
292 case HB_SCRIPT_LEPCHA:
293 case HB_SCRIPT_REJANG:
294 case HB_SCRIPT_SUNDANESE:
295
296 /* Unicode-5.2 additions */
297 case HB_SCRIPT_JAVANESE:
298 case HB_SCRIPT_KAITHI:
299 case HB_SCRIPT_MEETEI_MAYEK:
300
301 /* Unicode-6.0 additions */
302
303 /* Unicode-6.1 additions */
304 case HB_SCRIPT_CHAKMA:
305 case HB_SCRIPT_SHARADA:
306 case HB_SCRIPT_TAKRI:
307
308 /* If the designer designed the font for the 'DFLT' script,
309 * use the default shaper. Otherwise, use the Indic shaper.
310 * Note that for some simple scripts, there may not be *any*
311 * GSUB/GPOS needed, so there may be no scripts found! */
312 if (planner->map.chosen_script[0] == HB_TAG ('D','F','L','T'))
313 return &_hb_ot_complex_shaper_default;
314 else
315 return &_hb_ot_complex_shaper_indic;
316
317 case HB_SCRIPT_KHMER:
318 /* A number of Khmer fonts in the wild don't have a 'pref' feature,
319 * and as such won't shape properly via the Indic shaper;
320 * however, they typically have 'liga' / 'clig' features that implement
321 * the necessary "reordering" by means of ligature substitutions.
322 * So we send such pref-less fonts through the generic shaper instead. */
323 if (planner->map.found_script[0] &&
324 hb_ot_layout_language_find_feature (planner->face, HB_OT_TAG_GSUB,
325 planner->map.script_index[0],
326 planner->map.language_index[0],
327 HB_TAG ('p','r','e','f'),
328 NULL))
329 return &_hb_ot_complex_shaper_indic;
330 else
331 return &_hb_ot_complex_shaper_default;
332
333 case HB_SCRIPT_MYANMAR:
334 if (planner->map.chosen_script[0] == HB_TAG ('m','y','m','2'))
335 return &_hb_ot_complex_shaper_myanmar;
336 else if (planner->map.chosen_script[0] == HB_TAG ('m','y','m','r'))
337 return &_hb_ot_complex_shaper_myanmar_old;
338 else
339 return &_hb_ot_complex_shaper_default;
340
341 /* Unicode-4.1 additions */
342 case HB_SCRIPT_BUGINESE:
343
344 /* Unicode-5.1 additions */
345 case HB_SCRIPT_CHAM:
346
347 /* Unicode-5.2 additions */
348 case HB_SCRIPT_TAI_THAM:
349
350 /* If the designer designed the font for the 'DFLT' script,
351 * use the default shaper. Otherwise, use the Indic shaper.
352 * Note that for some simple scripts, there may not be *any*
353 * GSUB/GPOS needed, so there may be no scripts found! */
354 if (planner->map.chosen_script[0] == HB_TAG ('D','F','L','T'))
355 return &_hb_ot_complex_shaper_default;
356 else
357 return &_hb_ot_complex_shaper_sea;
358 }
359 }
360
361
362 #endif /* HB_OT_SHAPE_COMPLEX_PRIVATE_HH */
363