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 #define HB_OT_SHAPE_COMPLEX_MAX_COMBINING_MARKS 32
43 
44 enum hb_ot_shape_zero_width_marks_type_t {
45   HB_OT_SHAPE_ZERO_WIDTH_MARKS_NONE,
46   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_EARLY,
47   HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_LATE
48 };
49 
50 
51 /* Master OT shaper list */
52 #define HB_COMPLEX_SHAPERS_IMPLEMENT_SHAPERS \
53   HB_COMPLEX_SHAPER_IMPLEMENT (default) /* should be first */ \
54   HB_COMPLEX_SHAPER_IMPLEMENT (arabic) \
55   HB_COMPLEX_SHAPER_IMPLEMENT (hangul) \
56   HB_COMPLEX_SHAPER_IMPLEMENT (hebrew) \
57   HB_COMPLEX_SHAPER_IMPLEMENT (myanmar_old) \
58   HB_COMPLEX_SHAPER_IMPLEMENT (indic) \
59   HB_COMPLEX_SHAPER_IMPLEMENT (myanmar) \
60   HB_COMPLEX_SHAPER_IMPLEMENT (thai) \
61   HB_COMPLEX_SHAPER_IMPLEMENT (tibetan) \
62   HB_COMPLEX_SHAPER_IMPLEMENT (use) \
63   /* ^--- Add new shapers here */
64 
65 
66 struct hb_ot_complex_shaper_t
67 {
68   /* collect_features()
69    * Called during shape_plan().
70    * Shapers should use plan->map to add their features and callbacks.
71    * May be nullptr.
72    */
73   void (*collect_features) (hb_ot_shape_planner_t *plan);
74 
75   /* override_features()
76    * Called during shape_plan().
77    * Shapers should use plan->map to override features and add callbacks after
78    * common features are added.
79    * May be nullptr.
80    */
81   void (*override_features) (hb_ot_shape_planner_t *plan);
82 
83 
84   /* data_create()
85    * Called at the end of shape_plan().
86    * Whatever shapers return will be accessible through plan->data later.
87    * If nullptr is returned, means a plan failure.
88    */
89   void *(*data_create) (const hb_ot_shape_plan_t *plan);
90 
91   /* data_destroy()
92    * Called when the shape_plan is being destroyed.
93    * plan->data is passed here for destruction.
94    * If nullptr is returned, means a plan failure.
95    * May be nullptr.
96    */
97   void (*data_destroy) (void *data);
98 
99 
100   /* preprocess_text()
101    * Called during shape().
102    * Shapers can use to modify text before shaping starts.
103    * May be nullptr.
104    */
105   void (*preprocess_text) (const hb_ot_shape_plan_t *plan,
106 			   hb_buffer_t              *buffer,
107 			   hb_font_t                *font);
108 
109   /* postprocess_glyphs()
110    * Called during shape().
111    * Shapers can use to modify glyphs after shaping ends.
112    * May be nullptr.
113    */
114   void (*postprocess_glyphs) (const hb_ot_shape_plan_t *plan,
115 			      hb_buffer_t              *buffer,
116 			      hb_font_t                *font);
117 
118 
119   hb_ot_shape_normalization_mode_t normalization_preference;
120 
121   /* decompose()
122    * Called during shape()'s normalization.
123    * May be nullptr.
124    */
125   bool (*decompose) (const hb_ot_shape_normalize_context_t *c,
126 		     hb_codepoint_t  ab,
127 		     hb_codepoint_t *a,
128 		     hb_codepoint_t *b);
129 
130   /* compose()
131    * Called during shape()'s normalization.
132    * May be nullptr.
133    */
134   bool (*compose) (const hb_ot_shape_normalize_context_t *c,
135 		   hb_codepoint_t  a,
136 		   hb_codepoint_t  b,
137 		   hb_codepoint_t *ab);
138 
139   /* setup_masks()
140    * Called during shape().
141    * Shapers should use map to get feature masks and set on buffer.
142    * Shapers may NOT modify characters.
143    * May be nullptr.
144    */
145   void (*setup_masks) (const hb_ot_shape_plan_t *plan,
146 		       hb_buffer_t              *buffer,
147 		       hb_font_t                *font);
148 
149   /* disable_otl()
150    * Called during shape().
151    * If set and returns true, GDEF/GSUB/GPOS of the font are ignored
152    * and fallback operations used.
153    * May be nullptr.
154    */
155   bool (*disable_otl) (const hb_ot_shape_plan_t *plan);
156 
157   /* reorder_marks()
158    * Called during shape().
159    * Shapers can use to modify ordering of combining marks.
160    * May be nullptr.
161    */
162   void (*reorder_marks) (const hb_ot_shape_plan_t *plan,
163 			 hb_buffer_t              *buffer,
164 			 unsigned int              start,
165 			 unsigned int              end);
166 
167   hb_ot_shape_zero_width_marks_type_t zero_width_marks;
168 
169   bool fallback_position;
170 };
171 
172 #define HB_COMPLEX_SHAPER_IMPLEMENT(name) extern HB_INTERNAL const hb_ot_complex_shaper_t _hb_ot_complex_shaper_##name;
173 HB_COMPLEX_SHAPERS_IMPLEMENT_SHAPERS
174 #undef HB_COMPLEX_SHAPER_IMPLEMENT
175 
176 
177 static inline const hb_ot_complex_shaper_t *
hb_ot_shape_complex_categorize(const hb_ot_shape_planner_t * planner)178 hb_ot_shape_complex_categorize (const hb_ot_shape_planner_t *planner)
179 {
180   switch ((hb_tag_t) planner->props.script)
181   {
182     default:
183       return &_hb_ot_complex_shaper_default;
184 
185 
186     /* Unicode-1.1 additions */
187     case HB_SCRIPT_ARABIC:
188 
189     /* Unicode-3.0 additions */
190     case HB_SCRIPT_MONGOLIAN:
191     case HB_SCRIPT_SYRIAC:
192 
193     /* Unicode-5.0 additions */
194     case HB_SCRIPT_NKO:
195     case HB_SCRIPT_PHAGS_PA:
196 
197     /* Unicode-6.0 additions */
198     case HB_SCRIPT_MANDAIC:
199 
200     /* Unicode-7.0 additions */
201     case HB_SCRIPT_MANICHAEAN:
202     case HB_SCRIPT_PSALTER_PAHLAVI:
203 
204     /* Unicode-9.0 additions */
205     case HB_SCRIPT_ADLAM:
206 
207       /* For Arabic script, use the Arabic shaper even if no OT script tag was found.
208        * This is because we do fallback shaping for Arabic script (and not others).
209        * But note that Arabic shaping is applicable only to horizontal layout; for
210        * vertical text, just use the generic shaper instead. */
211       if ((planner->map.chosen_script[0] != HB_OT_TAG_DEFAULT_SCRIPT ||
212 	   planner->props.script == HB_SCRIPT_ARABIC) &&
213 	  HB_DIRECTION_IS_HORIZONTAL(planner->props.direction))
214 	return &_hb_ot_complex_shaper_arabic;
215       else
216 	return &_hb_ot_complex_shaper_default;
217 
218 
219     /* Unicode-1.1 additions */
220     case HB_SCRIPT_THAI:
221     case HB_SCRIPT_LAO:
222 
223       return &_hb_ot_complex_shaper_thai;
224 
225 
226     /* Unicode-1.1 additions */
227     case HB_SCRIPT_HANGUL:
228 
229       return &_hb_ot_complex_shaper_hangul;
230 
231 
232     /* Unicode-2.0 additions */
233     case HB_SCRIPT_TIBETAN:
234 
235       return &_hb_ot_complex_shaper_tibetan;
236 
237 
238     /* Unicode-1.1 additions */
239     case HB_SCRIPT_HEBREW:
240 
241       return &_hb_ot_complex_shaper_hebrew;
242 
243 
244     /* ^--- Add new shapers here */
245 
246 #if 0
247     /* Unicode-4.1 additions */
248     case HB_SCRIPT_NEW_TAI_LUE:
249 #endif
250 
251     /* Unicode-1.1 additions */
252     case HB_SCRIPT_BENGALI:
253     case HB_SCRIPT_DEVANAGARI:
254     case HB_SCRIPT_GUJARATI:
255     case HB_SCRIPT_GURMUKHI:
256     case HB_SCRIPT_KANNADA:
257     case HB_SCRIPT_MALAYALAM:
258     case HB_SCRIPT_ORIYA:
259     case HB_SCRIPT_TAMIL:
260     case HB_SCRIPT_TELUGU:
261 
262     /* Unicode-3.0 additions */
263     case HB_SCRIPT_SINHALA:
264 
265       /* If the designer designed the font for the 'DFLT' script,
266        * (or we ended up arbitrarily pick 'latn'), use the default shaper.
267        * Otherwise, use the specific shaper.
268        * Note that for some simple scripts, there may not be *any*
269        * GSUB/GPOS needed, so there may be no scripts found! */
270       if (planner->map.chosen_script[0] == HB_TAG ('D','F','L','T') ||
271 	  planner->map.chosen_script[0] == HB_TAG ('l','a','t','n'))
272 	return &_hb_ot_complex_shaper_default;
273       else
274 	return &_hb_ot_complex_shaper_indic;
275 
276     case HB_SCRIPT_KHMER:
277       /* A number of Khmer fonts in the wild don't have a 'pref' feature,
278        * and as such won't shape properly via the Indic shaper;
279        * however, they typically have 'liga' / 'clig' features that implement
280        * the necessary "reordering" by means of ligature substitutions.
281        * So we send such pref-less fonts through the generic shaper instead. */
282       if (planner->map.found_script[0] &&
283 	  hb_ot_layout_language_find_feature (planner->face, HB_OT_TAG_GSUB,
284 					      planner->map.script_index[0],
285 					      planner->map.language_index[0],
286 					      HB_TAG ('p','r','e','f'),
287 					      nullptr))
288 	return &_hb_ot_complex_shaper_indic;
289       else
290 	return &_hb_ot_complex_shaper_default;
291 
292     case HB_SCRIPT_MYANMAR:
293       if (planner->map.chosen_script[0] == HB_TAG ('m','y','m','2'))
294 	return &_hb_ot_complex_shaper_myanmar;
295       else if (planner->map.chosen_script[0] == HB_TAG ('m','y','m','r'))
296 	return &_hb_ot_complex_shaper_myanmar_old;
297       else
298 	return &_hb_ot_complex_shaper_default;
299 
300 
301     /* Unicode-2.0 additions */
302     //case HB_SCRIPT_TIBETAN:
303 
304     /* Unicode-3.0 additions */
305     //case HB_SCRIPT_MONGOLIAN:
306     //case HB_SCRIPT_SINHALA:
307 
308     /* Unicode-3.2 additions */
309     case HB_SCRIPT_BUHID:
310     case HB_SCRIPT_HANUNOO:
311     case HB_SCRIPT_TAGALOG:
312     case HB_SCRIPT_TAGBANWA:
313 
314     /* Unicode-4.0 additions */
315     case HB_SCRIPT_LIMBU:
316     case HB_SCRIPT_TAI_LE:
317 
318     /* Unicode-4.1 additions */
319     case HB_SCRIPT_BUGINESE:
320     case HB_SCRIPT_KHAROSHTHI:
321     case HB_SCRIPT_SYLOTI_NAGRI:
322     case HB_SCRIPT_TIFINAGH:
323 
324     /* Unicode-5.0 additions */
325     case HB_SCRIPT_BALINESE:
326     //case HB_SCRIPT_NKO:
327     //case HB_SCRIPT_PHAGS_PA:
328 
329     /* Unicode-5.1 additions */
330     case HB_SCRIPT_CHAM:
331     case HB_SCRIPT_KAYAH_LI:
332     case HB_SCRIPT_LEPCHA:
333     case HB_SCRIPT_REJANG:
334     case HB_SCRIPT_SAURASHTRA:
335     case HB_SCRIPT_SUNDANESE:
336 
337     /* Unicode-5.2 additions */
338     case HB_SCRIPT_EGYPTIAN_HIEROGLYPHS:
339     case HB_SCRIPT_JAVANESE:
340     case HB_SCRIPT_KAITHI:
341     case HB_SCRIPT_MEETEI_MAYEK:
342     case HB_SCRIPT_TAI_THAM:
343     case HB_SCRIPT_TAI_VIET:
344 
345     /* Unicode-6.0 additions */
346     case HB_SCRIPT_BATAK:
347     case HB_SCRIPT_BRAHMI:
348     //case HB_SCRIPT_MANDAIC:
349 
350     /* Unicode-6.1 additions */
351     case HB_SCRIPT_CHAKMA:
352     case HB_SCRIPT_SHARADA:
353     case HB_SCRIPT_TAKRI:
354 
355     /* Unicode-7.0 additions */
356     case HB_SCRIPT_DUPLOYAN:
357     case HB_SCRIPT_GRANTHA:
358     case HB_SCRIPT_KHOJKI:
359     case HB_SCRIPT_KHUDAWADI:
360     case HB_SCRIPT_MAHAJANI:
361     //case HB_SCRIPT_MANICHAEAN:
362     case HB_SCRIPT_MODI:
363     case HB_SCRIPT_PAHAWH_HMONG:
364     //case HB_SCRIPT_PSALTER_PAHLAVI:
365     case HB_SCRIPT_SIDDHAM:
366     case HB_SCRIPT_TIRHUTA:
367 
368     /* Unicode-8.0 additions */
369     case HB_SCRIPT_AHOM:
370     //case HB_SCRIPT_MULTANI:
371 
372     /* Unicode-9.0 additions */
373     case HB_SCRIPT_BHAIKSUKI:
374     case HB_SCRIPT_MARCHEN:
375     case HB_SCRIPT_NEWA:
376 
377     /* Unicode-10.0 additions */
378     case HB_SCRIPT_MASARAM_GONDI:
379     case HB_SCRIPT_SOYOMBO:
380     case HB_SCRIPT_ZANABAZAR_SQUARE:
381 
382       /* If the designer designed the font for the 'DFLT' script,
383        * (or we ended up arbitrarily pick 'latn'), use the default shaper.
384        * Otherwise, use the specific shaper.
385        * Note that for some simple scripts, there may not be *any*
386        * GSUB/GPOS needed, so there may be no scripts found! */
387       if (planner->map.chosen_script[0] == HB_TAG ('D','F','L','T') ||
388 	  planner->map.chosen_script[0] == HB_TAG ('l','a','t','n'))
389 	return &_hb_ot_complex_shaper_default;
390       else
391 	return &_hb_ot_complex_shaper_use;
392   }
393 }
394 
395 
396 #endif /* HB_OT_SHAPE_COMPLEX_PRIVATE_HH */
397