hb-ot-shape-complex-sea.cc revision ec5448667b30ad662401c2b4f5fc0da524c013fd 
1/*
2 * Copyright © 2011,2012,2013  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#include "hb-ot-shape-complex-indic-private.hh"
28
29/* buffer var allocations */
30#define sea_category() complex_var_u8_0() /* indic_category_t */
31#define sea_position() complex_var_u8_1() /* indic_position_t */
32
33
34/*
35 * South-East Asian shaper.
36 * Loosely based on the Myanmar spec / shaper.
37 * There is no OpenType spec for this.
38 */
39
40static const hb_tag_t
41sea_features[] =
42{
43  /*
44   * Basic features.
45   * These features are applied in order, one at a time, after initial_reordering.
46   */
47  HB_TAG('p','r','e','f'),
48  HB_TAG('a','b','v','f'),
49  HB_TAG('b','l','w','f'),
50  HB_TAG('p','s','t','f'),
51  /*
52   * Other features.
53   * These features are applied all at once, after final_reordering.
54   */
55  HB_TAG('p','r','e','s'),
56  HB_TAG('a','b','v','s'),
57  HB_TAG('b','l','w','s'),
58  HB_TAG('p','s','t','s'),
59  /* Positioning features, though we don't care about the types. */
60  HB_TAG('d','i','s','t'),
61};
62
63/*
64 * Must be in the same order as the sea_features array.
65 */
66enum {
67  _PREF,
68  _ABVF,
69  _BLWF,
70  _PSTF,
71
72  _PRES,
73  _ABVS,
74  _BLWS,
75  _PSTS,
76  _DIST,
77
78  SEA_NUM_FEATURES,
79  SEA_BASIC_FEATURES = _PRES /* Don't forget to update this! */
80};
81
82static void
83setup_syllables (const hb_ot_shape_plan_t *plan,
84		 hb_font_t *font,
85		 hb_buffer_t *buffer);
86static void
87initial_reordering (const hb_ot_shape_plan_t *plan,
88		    hb_font_t *font,
89		    hb_buffer_t *buffer);
90static void
91final_reordering (const hb_ot_shape_plan_t *plan,
92		  hb_font_t *font,
93		  hb_buffer_t *buffer);
94
95static void
96collect_features_sea (hb_ot_shape_planner_t *plan)
97{
98  hb_ot_map_builder_t *map = &plan->map;
99
100  /* Do this before any lookups have been applied. */
101  map->add_gsub_pause (setup_syllables);
102
103  map->add_global_bool_feature (HB_TAG('l','o','c','l'));
104  /* The Indic specs do not require ccmp, but we apply it here since if
105   * there is a use of it, it's typically at the beginning. */
106  map->add_global_bool_feature (HB_TAG('c','c','m','p'));
107
108
109  unsigned int i = 0;
110  map->add_gsub_pause (initial_reordering);
111  for (; i < SEA_BASIC_FEATURES; i++) {
112    map->add_global_bool_feature (sea_features[i]);
113    map->add_gsub_pause (NULL);
114  }
115  map->add_gsub_pause (final_reordering);
116  for (; i < SEA_NUM_FEATURES; i++) {
117    map->add_global_bool_feature (sea_features[i]);
118  }
119}
120
121static void
122override_features_sea (hb_ot_shape_planner_t *plan)
123{
124  plan->map.add_feature (HB_TAG('l','i','g','a'), 0, F_GLOBAL);
125}
126
127
128enum syllable_type_t {
129  consonant_syllable,
130  broken_cluster,
131  non_sea_cluster,
132};
133
134#include "hb-ot-shape-complex-sea-machine.hh"
135
136
137/* Note: This enum is duplicated in the -machine.rl source file.
138 * Not sure how to avoid duplication. */
139enum sea_category_t {
140//  OT_C    = 1,
141  OT_GB   = 12, /* Generic Base XXX DOTTED CIRCLE only for now */
142//  OT_H    = 4,  /* Halant */
143  OT_IV   = 2,  /* Independent Vowel */
144  OT_MR   = 22, /* Medial Ra */
145//  OT_CM   = 17, /* Consonant Medial */
146  OT_VAbv = 26,
147  OT_VBlw = 27,
148  OT_VPre = 28,
149  OT_VPst = 29,
150  OT_T    = 3,  /* Tone Marks */
151//  OT_A    = 10, /* Anusvara */
152};
153
154static inline void
155set_sea_properties (hb_glyph_info_t &info)
156{
157  hb_codepoint_t u = info.codepoint;
158  unsigned int type = hb_indic_get_categories (u);
159  indic_category_t cat = (indic_category_t) (type & 0x7F);
160  indic_position_t pos = (indic_position_t) (type >> 8);
161
162  /* Medial Ra */
163  if (u == 0x1A55 || u == 0xAA34)
164    cat = (indic_category_t) OT_MR;
165
166  if (cat == OT_M)
167  {
168    switch ((int) pos)
169    {
170      case POS_PRE_C:	cat = (indic_category_t) OT_VPre; break;
171      case POS_ABOVE_C:	cat = (indic_category_t) OT_VAbv; break;
172      case POS_BELOW_C:	cat = (indic_category_t) OT_VBlw; break;
173      case POS_POST_C:	cat = (indic_category_t) OT_VPst; break;
174    }
175  }
176
177  info.sea_category() = (sea_category_t) cat;
178  info.sea_position() = pos;
179}
180
181
182static void
183setup_masks_sea (const hb_ot_shape_plan_t *plan HB_UNUSED,
184		   hb_buffer_t              *buffer,
185		   hb_font_t                *font HB_UNUSED)
186{
187  HB_BUFFER_ALLOCATE_VAR (buffer, sea_category);
188  HB_BUFFER_ALLOCATE_VAR (buffer, sea_position);
189
190  /* We cannot setup masks here.  We save information about characters
191   * and setup masks later on in a pause-callback. */
192
193  unsigned int count = buffer->len;
194  for (unsigned int i = 0; i < count; i++)
195    set_sea_properties (buffer->info[i]);
196}
197
198static void
199setup_syllables (const hb_ot_shape_plan_t *plan HB_UNUSED,
200		 hb_font_t *font HB_UNUSED,
201		 hb_buffer_t *buffer)
202{
203  find_syllables (buffer);
204}
205
206static int
207compare_sea_order (const hb_glyph_info_t *pa, const hb_glyph_info_t *pb)
208{
209  int a = pa->sea_position();
210  int b = pb->sea_position();
211
212  return a < b ? -1 : a == b ? 0 : +1;
213}
214
215
216static void
217initial_reordering_consonant_syllable (const hb_ot_shape_plan_t *plan,
218				       hb_face_t *face,
219				       hb_buffer_t *buffer,
220				       unsigned int start, unsigned int end)
221{
222  hb_glyph_info_t *info = buffer->info;
223  unsigned int base = start;
224
225  /* Reorder! */
226  unsigned int i = start;
227  for (; i < base; i++)
228    info[i].sea_position() = POS_PRE_C;
229  if (i < end)
230  {
231    info[i].sea_position() = POS_BASE_C;
232    i++;
233  }
234  for (; i < end; i++)
235  {
236    if (info[i].sea_category() == OT_MR) /* Pre-base reordering */
237    {
238      info[i].sea_position() = POS_PRE_C;
239      continue;
240    }
241    if (info[i].sea_position() < POS_BASE_C) /* Left matra */
242    {
243      continue;
244    }
245
246    info[i].sea_position() = POS_AFTER_MAIN;
247  }
248
249  buffer->merge_clusters (start, end);
250  /* Sit tight, rock 'n roll! */
251  hb_bubble_sort (info + start, end - start, compare_sea_order);
252}
253
254static void
255initial_reordering_broken_cluster (const hb_ot_shape_plan_t *plan,
256				   hb_face_t *face,
257				   hb_buffer_t *buffer,
258				   unsigned int start, unsigned int end)
259{
260  /* We already inserted dotted-circles, so just call the consonant_syllable. */
261  initial_reordering_consonant_syllable (plan, face, buffer, start, end);
262}
263
264static void
265initial_reordering_non_sea_cluster (const hb_ot_shape_plan_t *plan HB_UNUSED,
266					hb_face_t *face HB_UNUSED,
267					hb_buffer_t *buffer HB_UNUSED,
268					unsigned int start HB_UNUSED, unsigned int end HB_UNUSED)
269{
270  /* Nothing to do right now.  If we ever switch to using the output
271   * buffer in the reordering process, we'd need to next_glyph() here. */
272}
273
274
275static void
276initial_reordering_syllable (const hb_ot_shape_plan_t *plan,
277			     hb_face_t *face,
278			     hb_buffer_t *buffer,
279			     unsigned int start, unsigned int end)
280{
281  syllable_type_t syllable_type = (syllable_type_t) (buffer->info[start].syllable() & 0x0F);
282  switch (syllable_type) {
283  case consonant_syllable:	initial_reordering_consonant_syllable  (plan, face, buffer, start, end); return;
284  case broken_cluster:		initial_reordering_broken_cluster      (plan, face, buffer, start, end); return;
285  case non_sea_cluster:	initial_reordering_non_sea_cluster (plan, face, buffer, start, end); return;
286  }
287}
288
289static inline void
290insert_dotted_circles (const hb_ot_shape_plan_t *plan HB_UNUSED,
291		       hb_font_t *font,
292		       hb_buffer_t *buffer)
293{
294  /* Note: This loop is extra overhead, but should not be measurable. */
295  bool has_broken_syllables = false;
296  unsigned int count = buffer->len;
297  for (unsigned int i = 0; i < count; i++)
298    if ((buffer->info[i].syllable() & 0x0F) == broken_cluster) {
299      has_broken_syllables = true;
300      break;
301    }
302  if (likely (!has_broken_syllables))
303    return;
304
305
306  hb_codepoint_t dottedcircle_glyph;
307  if (!font->get_glyph (0x25CC, 0, &dottedcircle_glyph))
308    return;
309
310  hb_glyph_info_t dottedcircle = {0};
311  dottedcircle.codepoint = 0x25CC;
312  set_sea_properties (dottedcircle);
313  dottedcircle.codepoint = dottedcircle_glyph;
314
315  buffer->clear_output ();
316
317  buffer->idx = 0;
318  unsigned int last_syllable = 0;
319  while (buffer->idx < buffer->len)
320  {
321    unsigned int syllable = buffer->cur().syllable();
322    syllable_type_t syllable_type = (syllable_type_t) (syllable & 0x0F);
323    if (unlikely (last_syllable != syllable && syllable_type == broken_cluster))
324    {
325      last_syllable = syllable;
326
327      hb_glyph_info_t info = dottedcircle;
328      info.cluster = buffer->cur().cluster;
329      info.mask = buffer->cur().mask;
330      info.syllable() = buffer->cur().syllable();
331
332      buffer->output_info (info);
333    }
334    else
335      buffer->next_glyph ();
336  }
337
338  buffer->swap_buffers ();
339}
340
341static void
342initial_reordering (const hb_ot_shape_plan_t *plan,
343		    hb_font_t *font,
344		    hb_buffer_t *buffer)
345{
346  insert_dotted_circles (plan, font, buffer);
347
348  hb_glyph_info_t *info = buffer->info;
349  unsigned int count = buffer->len;
350  if (unlikely (!count)) return;
351  unsigned int last = 0;
352  unsigned int last_syllable = info[0].syllable();
353  for (unsigned int i = 1; i < count; i++)
354    if (last_syllable != info[i].syllable()) {
355      initial_reordering_syllable (plan, font->face, buffer, last, i);
356      last = i;
357      last_syllable = info[last].syllable();
358    }
359  initial_reordering_syllable (plan, font->face, buffer, last, count);
360}
361
362static void
363final_reordering (const hb_ot_shape_plan_t *plan,
364		  hb_font_t *font HB_UNUSED,
365		  hb_buffer_t *buffer)
366{
367  hb_glyph_info_t *info = buffer->info;
368  unsigned int count = buffer->len;
369
370  /* Zero syllables now... */
371  for (unsigned int i = 0; i < count; i++)
372    info[i].syllable() = 0;
373
374  HB_BUFFER_DEALLOCATE_VAR (buffer, sea_category);
375  HB_BUFFER_DEALLOCATE_VAR (buffer, sea_position);
376}
377
378
379static hb_ot_shape_normalization_mode_t
380normalization_preference_sea (const hb_segment_properties_t *props HB_UNUSED)
381{
382  return HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS_NO_SHORT_CIRCUIT;
383}
384
385
386const hb_ot_complex_shaper_t _hb_ot_complex_shaper_sea =
387{
388  "sea",
389  collect_features_sea,
390  override_features_sea,
391  NULL, /* data_create */
392  NULL, /* data_destroy */
393  NULL, /* preprocess_text */
394  normalization_preference_sea,
395  NULL, /* decompose */
396  NULL, /* compose */
397  setup_masks_sea,
398  HB_OT_SHAPE_ZERO_WIDTH_MARKS_NONE,
399  false, /* fallback_position */
400};
401