Mercurial > hgrepos > Python2 > PyMuPDF

comparison mupdf-source/thirdparty/harfbuzz/src/graph/pairpos-graph.hh @ 2:b50eed0cc0ef upstream

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4. The directory name has changed: no version number in the expanded directory now.
author Franz Glasner <fzglas.hg@dom66.de>
date Mon, 15 Sep 2025 11:43:07 +0200
parents
children
comparison
equal deleted inserted replaced
1:1d09e1dec1d9 2:b50eed0cc0ef
1 /*
2 * Copyright © 2022 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): Garret Rieger
25 */
26
27 #ifndef GRAPH_PAIRPOS_GRAPH_HH
28 #define GRAPH_PAIRPOS_GRAPH_HH
29
30 #include "split-helpers.hh"
31 #include "coverage-graph.hh"
32 #include "classdef-graph.hh"
33 #include "../OT/Layout/GPOS/PairPos.hh"
34 #include "../OT/Layout/GPOS/PosLookupSubTable.hh"
35
36 namespace graph {
37
38 struct PairPosFormat1 : public OT::Layout::GPOS_impl::PairPosFormat1_3<SmallTypes>
39 {
40 bool sanitize (graph_t::vertex_t& vertex) const
41 {
42 int64_t vertex_len = vertex.obj.tail - vertex.obj.head;
43 unsigned min_size = OT::Layout::GPOS_impl::PairPosFormat1_3<SmallTypes>::min_size;
44 if (vertex_len < min_size) return false;
45
46 return vertex_len >=
47 min_size + pairSet.get_size () - pairSet.len.get_size();
48 }
49
50 hb_vector_t<unsigned> split_subtables (gsubgpos_graph_context_t& c,
51 unsigned parent_index,
52 unsigned this_index)
53 {
54 hb_set_t visited;
55
56 const unsigned coverage_id = c.graph.index_for_offset (this_index, &coverage);
57 const unsigned coverage_size = c.graph.vertices_[coverage_id].table_size ();
58 const unsigned base_size = OT::Layout::GPOS_impl::PairPosFormat1_3<SmallTypes>::min_size;
59
60 unsigned partial_coverage_size = 4;
61 unsigned accumulated = base_size;
62 hb_vector_t<unsigned> split_points;
63 for (unsigned i = 0; i < pairSet.len; i++)
64 {
65 unsigned pair_set_index = pair_set_graph_index (c, this_index, i);
66 unsigned accumulated_delta =
67 c.graph.find_subgraph_size (pair_set_index, visited) +
68 SmallTypes::size; // for PairSet offset.
69 partial_coverage_size += OT::HBUINT16::static_size;
70
71 accumulated += accumulated_delta;
72 unsigned total = accumulated + hb_min (partial_coverage_size, coverage_size);
73
74 if (total >= (1 << 16))
75 {
76 split_points.push (i);
77 accumulated = base_size + accumulated_delta;
78 partial_coverage_size = 6;
79 visited.clear (); // node sharing isn't allowed between splits.
80 }
81 }
82
83 split_context_t split_context {
84 c,
85 this,
86 c.graph.duplicate_if_shared (parent_index, this_index),
87 };
88
89 return actuate_subtable_split<split_context_t> (split_context, split_points);
90 }
91
92 private:
93
94 struct split_context_t {
95 gsubgpos_graph_context_t& c;
96 PairPosFormat1* thiz;
97 unsigned this_index;
98
99 unsigned original_count ()
100 {
101 return thiz->pairSet.len;
102 }
103
104 unsigned clone_range (unsigned start, unsigned end)
105 {
106 return thiz->clone_range (this->c, this->this_index, start, end);
107 }
108
109 bool shrink (unsigned count)
110 {
111 return thiz->shrink (this->c, this->this_index, count);
112 }
113 };
114
115 bool shrink (gsubgpos_graph_context_t& c,
116 unsigned this_index,
117 unsigned count)
118 {
119 DEBUG_MSG (SUBSET_REPACK, nullptr,
120 " Shrinking PairPosFormat1 (%u) to [0, %u).",
121 this_index,
122 count);
123 unsigned old_count = pairSet.len;
124 if (count >= old_count)
125 return true;
126
127 pairSet.len = count;
128 c.graph.vertices_[this_index].obj.tail -= (old_count - count) * SmallTypes::size;
129
130 auto coverage = c.graph.as_mutable_table<Coverage> (this_index, &this->coverage);
131 if (!coverage) return false;
132
133 unsigned coverage_size = coverage.vertex->table_size ();
134 auto new_coverage =
135 + hb_zip (coverage.table->iter (), hb_range ())
136 | hb_filter ([&] (hb_pair_t<unsigned, unsigned> p) {
137 return p.second < count;
138 })
139 | hb_map_retains_sorting (hb_first)
140 ;
141
142 return Coverage::make_coverage (c, new_coverage, coverage.index, coverage_size);
143 }
144
145 // Create a new PairPos including PairSet's from start (inclusive) to end (exclusive).
146 // Returns object id of the new object.
147 unsigned clone_range (gsubgpos_graph_context_t& c,
148 unsigned this_index,
149 unsigned start, unsigned end) const
150 {
151 DEBUG_MSG (SUBSET_REPACK, nullptr,
152 " Cloning PairPosFormat1 (%u) range [%u, %u).", this_index, start, end);
153
154 unsigned num_pair_sets = end - start;
155 unsigned prime_size = OT::Layout::GPOS_impl::PairPosFormat1_3<SmallTypes>::min_size
156 + num_pair_sets * SmallTypes::size;
157
158 unsigned pair_pos_prime_id = c.create_node (prime_size);
159 if (pair_pos_prime_id == (unsigned) -1) return -1;
160
161 PairPosFormat1* pair_pos_prime = (PairPosFormat1*) c.graph.object (pair_pos_prime_id).head;
162 pair_pos_prime->format = this->format;
163 pair_pos_prime->valueFormat[0] = this->valueFormat[0];
164 pair_pos_prime->valueFormat[1] = this->valueFormat[1];
165 pair_pos_prime->pairSet.len = num_pair_sets;
166
167 for (unsigned i = start; i < end; i++)
168 {
169 c.graph.move_child<> (this_index,
170 &pairSet[i],
171 pair_pos_prime_id,
172 &pair_pos_prime->pairSet[i - start]);
173 }
174
175 unsigned coverage_id = c.graph.index_for_offset (this_index, &coverage);
176 if (!Coverage::clone_coverage (c,
177 coverage_id,
178 pair_pos_prime_id,
179 2,
180 start, end))
181 return -1;
182
183 return pair_pos_prime_id;
184 }
185
186
187
188 unsigned pair_set_graph_index (gsubgpos_graph_context_t& c, unsigned this_index, unsigned i) const
189 {
190 return c.graph.index_for_offset (this_index, &pairSet[i]);
191 }
192 };
193
194 struct PairPosFormat2 : public OT::Layout::GPOS_impl::PairPosFormat2_4<SmallTypes>
195 {
196 bool sanitize (graph_t::vertex_t& vertex) const
197 {
198 size_t vertex_len = vertex.table_size ();
199 unsigned min_size = OT::Layout::GPOS_impl::PairPosFormat2_4<SmallTypes>::min_size;
200 if (vertex_len < min_size) return false;
201
202 const unsigned class1_count = class1Count;
203 return vertex_len >=
204 min_size + class1_count * get_class1_record_size ();
205 }
206
207 hb_vector_t<unsigned> split_subtables (gsubgpos_graph_context_t& c,
208 unsigned parent_index,
209 unsigned this_index)
210 {
211 const unsigned base_size = OT::Layout::GPOS_impl::PairPosFormat2_4<SmallTypes>::min_size;
212 const unsigned class_def_2_size = size_of (c, this_index, &classDef2);
213 const Coverage* coverage = get_coverage (c, this_index);
214 const ClassDef* class_def_1 = get_class_def_1 (c, this_index);
215 auto gid_and_class =
216 + coverage->iter ()
217 | hb_map_retains_sorting ([&] (hb_codepoint_t gid) {
218 return hb_pair_t<hb_codepoint_t, hb_codepoint_t> (gid, class_def_1->get_class (gid));
219 })
220 ;
221 class_def_size_estimator_t estimator (gid_and_class);
222
223 const unsigned class1_count = class1Count;
224 const unsigned class2_count = class2Count;
225 const unsigned class1_record_size = get_class1_record_size ();
226
227 const unsigned value_1_len = valueFormat1.get_len ();
228 const unsigned value_2_len = valueFormat2.get_len ();
229 const unsigned total_value_len = value_1_len + value_2_len;
230
231 unsigned accumulated = base_size;
232 unsigned coverage_size = 4;
233 unsigned class_def_1_size = 4;
234 unsigned max_coverage_size = coverage_size;
235 unsigned max_class_def_1_size = class_def_1_size;
236
237 hb_vector_t<unsigned> split_points;
238
239 hb_hashmap_t<unsigned, unsigned> device_tables = get_all_device_tables (c, this_index);
240 hb_vector_t<unsigned> format1_device_table_indices = valueFormat1.get_device_table_indices ();
241 hb_vector_t<unsigned> format2_device_table_indices = valueFormat2.get_device_table_indices ();
242 bool has_device_tables = bool(format1_device_table_indices) || bool(format2_device_table_indices);
243
244 hb_set_t visited;
245 for (unsigned i = 0; i < class1_count; i++)
246 {
247 unsigned accumulated_delta = class1_record_size;
248 coverage_size += estimator.incremental_coverage_size (i);
249 class_def_1_size += estimator.incremental_class_def_size (i);
250 max_coverage_size = hb_max (max_coverage_size, coverage_size);
251 max_class_def_1_size = hb_max (max_class_def_1_size, class_def_1_size);
252
253 if (has_device_tables) {
254 for (unsigned j = 0; j < class2_count; j++)
255 {
256 unsigned value1_index = total_value_len * (class2_count * i + j);
257 unsigned value2_index = value1_index + value_1_len;
258 accumulated_delta += size_of_value_record_children (c,
259 device_tables,
260 format1_device_table_indices,
261 value1_index,
262 visited);
263 accumulated_delta += size_of_value_record_children (c,
264 device_tables,
265 format2_device_table_indices,
266 value2_index,
267 visited);
268 }
269 }
270
271 accumulated += accumulated_delta;
272 unsigned total = accumulated
273 + coverage_size + class_def_1_size + class_def_2_size
274 // The largest object will pack last and can exceed the size limit.
275 - hb_max (hb_max (coverage_size, class_def_1_size), class_def_2_size);
276 if (total >= (1 << 16))
277 {
278 split_points.push (i);
279 // split does not include i, so add the size for i when we reset the size counters.
280 accumulated = base_size + accumulated_delta;
281 coverage_size = 4 + estimator.incremental_coverage_size (i);
282 class_def_1_size = 4 + estimator.incremental_class_def_size (i);
283 visited.clear (); // node sharing isn't allowed between splits.
284 }
285 }
286
287 split_context_t split_context {
288 c,
289 this,
290 c.graph.duplicate_if_shared (parent_index, this_index),
291 class1_record_size,
292 total_value_len,
293 value_1_len,
294 value_2_len,
295 max_coverage_size,
296 max_class_def_1_size,
297 device_tables,
298 format1_device_table_indices,
299 format2_device_table_indices
300 };
301
302 return actuate_subtable_split<split_context_t> (split_context, split_points);
303 }
304 private:
305
306 struct split_context_t
307 {
308 gsubgpos_graph_context_t& c;
309 PairPosFormat2* thiz;
310 unsigned this_index;
311 unsigned class1_record_size;
312 unsigned value_record_len;
313 unsigned value1_record_len;
314 unsigned value2_record_len;
315 unsigned max_coverage_size;
316 unsigned max_class_def_size;
317
318 const hb_hashmap_t<unsigned, unsigned>& device_tables;
319 const hb_vector_t<unsigned>& format1_device_table_indices;
320 const hb_vector_t<unsigned>& format2_device_table_indices;
321
322 unsigned original_count ()
323 {
324 return thiz->class1Count;
325 }
326
327 unsigned clone_range (unsigned start, unsigned end)
328 {
329 return thiz->clone_range (*this, start, end);
330 }
331
332 bool shrink (unsigned count)
333 {
334 return thiz->shrink (*this, count);
335 }
336 };
337
338 size_t get_class1_record_size () const
339 {
340 const size_t class2_count = class2Count;
341 return
342 class2_count * (valueFormat1.get_size () + valueFormat2.get_size ());
343 }
344
345 unsigned clone_range (split_context_t& split_context,
346 unsigned start, unsigned end) const
347 {
348 DEBUG_MSG (SUBSET_REPACK, nullptr,
349 " Cloning PairPosFormat2 (%u) range [%u, %u).", split_context.this_index, start, end);
350
351 graph_t& graph = split_context.c.graph;
352
353 unsigned num_records = end - start;
354 unsigned prime_size = OT::Layout::GPOS_impl::PairPosFormat2_4<SmallTypes>::min_size
355 + num_records * split_context.class1_record_size;
356
357 unsigned pair_pos_prime_id = split_context.c.create_node (prime_size);
358 if (pair_pos_prime_id == (unsigned) -1) return -1;
359
360 PairPosFormat2* pair_pos_prime =
361 (PairPosFormat2*) graph.object (pair_pos_prime_id).head;
362 pair_pos_prime->format = this->format;
363 pair_pos_prime->valueFormat1 = this->valueFormat1;
364 pair_pos_prime->valueFormat2 = this->valueFormat2;
365 pair_pos_prime->class1Count = num_records;
366 pair_pos_prime->class2Count = this->class2Count;
367 clone_class1_records (split_context,
368 pair_pos_prime_id,
369 start,
370 end);
371
372 unsigned coverage_id =
373 graph.index_for_offset (split_context.this_index, &coverage);
374 unsigned class_def_1_id =
375 graph.index_for_offset (split_context.this_index, &classDef1);
376 auto& coverage_v = graph.vertices_[coverage_id];
377 auto& class_def_1_v = graph.vertices_[class_def_1_id];
378 Coverage* coverage_table = (Coverage*) coverage_v.obj.head;
379 ClassDef* class_def_1_table = (ClassDef*) class_def_1_v.obj.head;
380 if (!coverage_table
381 || !coverage_table->sanitize (coverage_v)
382 || !class_def_1_table
383 || !class_def_1_table->sanitize (class_def_1_v))
384 return -1;
385
386 auto klass_map =
387 + coverage_table->iter ()
388 | hb_map_retains_sorting ([&] (hb_codepoint_t gid) {
389 return hb_pair_t<hb_codepoint_t, hb_codepoint_t> (gid, class_def_1_table->get_class (gid));
390 })
391 | hb_filter ([&] (hb_codepoint_t klass) {
392 return klass >= start && klass < end;
393 }, hb_second)
394 | hb_map_retains_sorting ([&] (hb_pair_t<hb_codepoint_t, hb_codepoint_t> gid_and_class) {
395 // Classes must be from 0...N so subtract start
396 return hb_pair_t<hb_codepoint_t, hb_codepoint_t> (gid_and_class.first, gid_and_class.second - start);
397 })
398 ;
399
400 if (!Coverage::add_coverage (split_context.c,
401 pair_pos_prime_id,
402 2,
403 + klass_map | hb_map_retains_sorting (hb_first),
404 split_context.max_coverage_size))
405 return -1;
406
407 // classDef1
408 if (!ClassDef::add_class_def (split_context.c,
409 pair_pos_prime_id,
410 8,
411 + klass_map,
412 split_context.max_class_def_size))
413 return -1;
414
415 // classDef2
416 unsigned class_def_2_id =
417 graph.index_for_offset (split_context.this_index, &classDef2);
418 auto* class_def_link = graph.vertices_[pair_pos_prime_id].obj.real_links.push ();
419 class_def_link->width = SmallTypes::size;
420 class_def_link->objidx = class_def_2_id;
421 class_def_link->position = 10;
422 graph.vertices_[class_def_2_id].parents.push (pair_pos_prime_id);
423 graph.duplicate (pair_pos_prime_id, class_def_2_id);
424
425 return pair_pos_prime_id;
426 }
427
428 void clone_class1_records (split_context_t& split_context,
429 unsigned pair_pos_prime_id,
430 unsigned start, unsigned end) const
431 {
432 PairPosFormat2* pair_pos_prime =
433 (PairPosFormat2*) split_context.c.graph.object (pair_pos_prime_id).head;
434
435 char* start_addr = ((char*)&values[0]) + start * split_context.class1_record_size;
436 unsigned num_records = end - start;
437 hb_memcpy (&pair_pos_prime->values[0],
438 start_addr,
439 num_records * split_context.class1_record_size);
440
441 if (!split_context.format1_device_table_indices
442 && !split_context.format2_device_table_indices)
443 // No device tables to move over.
444 return;
445
446 unsigned class2_count = class2Count;
447 for (unsigned i = start; i < end; i++)
448 {
449 for (unsigned j = 0; j < class2_count; j++)
450 {
451 unsigned value1_index = split_context.value_record_len * (class2_count * i + j);
452 unsigned value2_index = value1_index + split_context.value1_record_len;
453
454 unsigned new_value1_index = split_context.value_record_len * (class2_count * (i - start) + j);
455 unsigned new_value2_index = new_value1_index + split_context.value1_record_len;
456
457 transfer_device_tables (split_context,
458 pair_pos_prime_id,
459 split_context.format1_device_table_indices,
460 value1_index,
461 new_value1_index);
462
463 transfer_device_tables (split_context,
464 pair_pos_prime_id,
465 split_context.format2_device_table_indices,
466 value2_index,
467 new_value2_index);
468 }
469 }
470 }
471
472 void transfer_device_tables (split_context_t& split_context,
473 unsigned pair_pos_prime_id,
474 const hb_vector_t<unsigned>& device_table_indices,
475 unsigned old_value_record_index,
476 unsigned new_value_record_index) const
477 {
478 PairPosFormat2* pair_pos_prime =
479 (PairPosFormat2*) split_context.c.graph.object (pair_pos_prime_id).head;
480
481 for (unsigned i : device_table_indices)
482 {
483 OT::Offset16* record = (OT::Offset16*) &values[old_value_record_index + i];
484 unsigned record_position = ((char*) record) - ((char*) this);
485 if (!split_context.device_tables.has (record_position)) continue;
486
487 split_context.c.graph.move_child (
488 split_context.this_index,
489 record,
490 pair_pos_prime_id,
491 (OT::Offset16*) &pair_pos_prime->values[new_value_record_index + i]);
492 }
493 }
494
495 bool shrink (split_context_t& split_context,
496 unsigned count)
497 {
498 DEBUG_MSG (SUBSET_REPACK, nullptr,
499 " Shrinking PairPosFormat2 (%u) to [0, %u).",
500 split_context.this_index,
501 count);
502 unsigned old_count = class1Count;
503 if (count >= old_count)
504 return true;
505
506 graph_t& graph = split_context.c.graph;
507 class1Count = count;
508 graph.vertices_[split_context.this_index].obj.tail -=
509 (old_count - count) * split_context.class1_record_size;
510
511 auto coverage =
512 graph.as_mutable_table<Coverage> (split_context.this_index, &this->coverage);
513 if (!coverage) return false;
514
515 auto class_def_1 =
516 graph.as_mutable_table<ClassDef> (split_context.this_index, &classDef1);
517 if (!class_def_1) return false;
518
519 auto klass_map =
520 + coverage.table->iter ()
521 | hb_map_retains_sorting ([&] (hb_codepoint_t gid) {
522 return hb_pair_t<hb_codepoint_t, hb_codepoint_t> (gid, class_def_1.table->get_class (gid));
523 })
524 | hb_filter ([&] (hb_codepoint_t klass) {
525 return klass < count;
526 }, hb_second)
527 ;
528
529 auto new_coverage = + klass_map | hb_map_retains_sorting (hb_first);
530 if (!Coverage::make_coverage (split_context.c,
531 + new_coverage,
532 coverage.index,
533 // existing ranges my not be kept, worst case size is a format 1
534 // coverage table.
535 4 + new_coverage.len() * 2))
536 return false;
537
538 return ClassDef::make_class_def (split_context.c,
539 + klass_map,
540 class_def_1.index,
541 class_def_1.vertex->table_size ());
542 }
543
544 hb_hashmap_t<unsigned, unsigned>
545 get_all_device_tables (gsubgpos_graph_context_t& c,
546 unsigned this_index) const
547 {
548 const auto& v = c.graph.vertices_[this_index];
549 return v.position_to_index_map ();
550 }
551
552 const Coverage* get_coverage (gsubgpos_graph_context_t& c,
553 unsigned this_index) const
554 {
555 unsigned coverage_id = c.graph.index_for_offset (this_index, &coverage);
556 auto& coverage_v = c.graph.vertices_[coverage_id];
557
558 Coverage* coverage_table = (Coverage*) coverage_v.obj.head;
559 if (!coverage_table || !coverage_table->sanitize (coverage_v))
560 return &Null(Coverage);
561 return coverage_table;
562 }
563
564 const ClassDef* get_class_def_1 (gsubgpos_graph_context_t& c,
565 unsigned this_index) const
566 {
567 unsigned class_def_1_id = c.graph.index_for_offset (this_index, &classDef1);
568 auto& class_def_1_v = c.graph.vertices_[class_def_1_id];
569
570 ClassDef* class_def_1_table = (ClassDef*) class_def_1_v.obj.head;
571 if (!class_def_1_table || !class_def_1_table->sanitize (class_def_1_v))
572 return &Null(ClassDef);
573 return class_def_1_table;
574 }
575
576 unsigned size_of_value_record_children (gsubgpos_graph_context_t& c,
577 const hb_hashmap_t<unsigned, unsigned>& device_tables,
578 const hb_vector_t<unsigned> device_table_indices,
579 unsigned value_record_index,
580 hb_set_t& visited)
581 {
582 unsigned size = 0;
583 for (unsigned i : device_table_indices)
584 {
585 OT::Layout::GPOS_impl::Value* record = &values[value_record_index + i];
586 unsigned record_position = ((char*) record) - ((char*) this);
587 unsigned* obj_idx;
588 if (!device_tables.has (record_position, &obj_idx)) continue;
589 size += c.graph.find_subgraph_size (*obj_idx, visited);
590 }
591 return size;
592 }
593
594 unsigned size_of (gsubgpos_graph_context_t& c,
595 unsigned this_index,
596 const void* offset) const
597 {
598 const unsigned id = c.graph.index_for_offset (this_index, offset);
599 return c.graph.vertices_[id].table_size ();
600 }
601 };
602
603 struct PairPos : public OT::Layout::GPOS_impl::PairPos
604 {
605 hb_vector_t<unsigned> split_subtables (gsubgpos_graph_context_t& c,
606 unsigned parent_index,
607 unsigned this_index)
608 {
609 switch (u.format) {
610 case 1:
611 return ((PairPosFormat1*)(&u.format1))->split_subtables (c, parent_index, this_index);
612 case 2:
613 return ((PairPosFormat2*)(&u.format2))->split_subtables (c, parent_index, this_index);
614 #ifndef HB_NO_BEYOND_64K
615 case 3: HB_FALLTHROUGH;
616 case 4: HB_FALLTHROUGH;
617 // Don't split 24bit PairPos's.
618 #endif
619 default:
620 return hb_vector_t<unsigned> ();
621 }
622 }
623
624 bool sanitize (graph_t::vertex_t& vertex) const
625 {
626 int64_t vertex_len = vertex.obj.tail - vertex.obj.head;
627 if (vertex_len < u.format.get_size ()) return false;
628
629 switch (u.format) {
630 case 1:
631 return ((PairPosFormat1*)(&u.format1))->sanitize (vertex);
632 case 2:
633 return ((PairPosFormat2*)(&u.format2))->sanitize (vertex);
634 #ifndef HB_NO_BEYOND_64K
635 case 3: HB_FALLTHROUGH;
636 case 4: HB_FALLTHROUGH;
637 #endif
638 default:
639 // We don't handle format 3 and 4 here.
640 return false;
641 }
642 }
643 };
644
645 }
646
647 #endif // GRAPH_PAIRPOS_GRAPH_HH