1 | /* |
2 | * Copyright © 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 OT_NAME_NAME_HH |
28 | #define OT_NAME_NAME_HH |
29 | |
30 | #include "../../hb-open-type.hh" |
31 | #include "../../hb-ot-name-language.hh" |
32 | #include "../../hb-aat-layout.hh" |
33 | #include "../../hb-utf.hh" |
34 | |
35 | |
36 | namespace OT { |
37 | |
38 | template <typename in_utf_t, typename out_utf_t> |
39 | inline unsigned int |
40 | hb_ot_name_convert_utf (hb_bytes_t bytes, |
41 | unsigned int *text_size /* IN/OUT */, |
42 | typename out_utf_t::codepoint_t *text /* OUT */) |
43 | { |
44 | unsigned int src_len = bytes.length / sizeof (typename in_utf_t::codepoint_t); |
45 | const typename in_utf_t::codepoint_t *src = (const typename in_utf_t::codepoint_t *) bytes.arrayZ; |
46 | const typename in_utf_t::codepoint_t *src_end = src + src_len; |
47 | |
48 | typename out_utf_t::codepoint_t *dst = text; |
49 | |
50 | hb_codepoint_t unicode; |
51 | const hb_codepoint_t replacement = HB_BUFFER_REPLACEMENT_CODEPOINT_DEFAULT; |
52 | |
53 | if (text_size && *text_size) |
54 | { |
55 | (*text_size)--; /* Save room for NUL-termination. */ |
56 | const typename out_utf_t::codepoint_t *dst_end = text + *text_size; |
57 | |
58 | while (src < src_end && dst < dst_end) |
59 | { |
60 | const typename in_utf_t::codepoint_t *src_next = in_utf_t::next (src, src_end, &unicode, replacement); |
61 | typename out_utf_t::codepoint_t *dst_next = out_utf_t::encode (dst, dst_end, unicode); |
62 | if (dst_next == dst) |
63 | break; /* Out-of-room. */ |
64 | |
65 | dst = dst_next; |
66 | src = src_next; |
67 | } |
68 | |
69 | *text_size = dst - text; |
70 | *dst = 0; /* NUL-terminate. */ |
71 | } |
72 | |
73 | /* Accumulate length of rest. */ |
74 | unsigned int dst_len = dst - text; |
75 | while (src < src_end) |
76 | { |
77 | src = in_utf_t::next (src, src_end, &unicode, replacement); |
78 | dst_len += out_utf_t::encode_len (unicode); |
79 | } |
80 | return dst_len; |
81 | } |
82 | |
83 | #define entry_score var.u16[0] |
84 | #define entry_index var.u16[1] |
85 | |
86 | |
87 | /* |
88 | * name -- Naming |
89 | * https://docs.microsoft.com/en-us/typography/opentype/spec/name |
90 | */ |
91 | #define HB_OT_TAG_name HB_TAG('n','a','m','e') |
92 | |
93 | #define UNSUPPORTED 42 |
94 | |
95 | struct NameRecord |
96 | { |
97 | hb_language_t language (hb_face_t *face) const |
98 | { |
99 | #ifndef HB_NO_OT_NAME_LANGUAGE |
100 | unsigned int p = platformID; |
101 | unsigned int l = languageID; |
102 | |
103 | if (p == 3) |
104 | return _hb_ot_name_language_for_ms_code (l); |
105 | |
106 | if (p == 1) |
107 | return _hb_ot_name_language_for_mac_code (l); |
108 | |
109 | #ifndef HB_NO_OT_NAME_LANGUAGE_AAT |
110 | if (p == 0) |
111 | return face->table.ltag->get_language (l); |
112 | #endif |
113 | |
114 | #endif |
115 | return HB_LANGUAGE_INVALID; |
116 | } |
117 | |
118 | uint16_t score () const |
119 | { |
120 | /* Same order as in cmap::find_best_subtable(). */ |
121 | unsigned int p = platformID; |
122 | unsigned int e = encodingID; |
123 | |
124 | /* 32-bit. */ |
125 | if (p == 3 && e == 10) return 0; |
126 | if (p == 0 && e == 6) return 1; |
127 | if (p == 0 && e == 4) return 2; |
128 | |
129 | /* 16-bit. */ |
130 | if (p == 3 && e == 1) return 3; |
131 | if (p == 0 && e == 3) return 4; |
132 | if (p == 0 && e == 2) return 5; |
133 | if (p == 0 && e == 1) return 6; |
134 | if (p == 0 && e == 0) return 7; |
135 | |
136 | /* Symbol. */ |
137 | if (p == 3 && e == 0) return 8; |
138 | |
139 | /* We treat all Mac Latin names as ASCII only. */ |
140 | if (p == 1 && e == 0) return 10; /* 10 is magic number :| */ |
141 | |
142 | return UNSUPPORTED; |
143 | } |
144 | |
145 | NameRecord* copy (hb_serialize_context_t *c, const void *base |
146 | #ifdef HB_EXPERIMENTAL_API |
147 | , const hb_hashmap_t<hb_ot_name_record_ids_t, hb_bytes_t> *name_table_overrides |
148 | #endif |
149 | ) const |
150 | { |
151 | TRACE_SERIALIZE (this); |
152 | HB_UNUSED auto snap = c->snapshot (); |
153 | auto *out = c->embed (this); |
154 | if (unlikely (!out)) return_trace (nullptr); |
155 | #ifdef HB_EXPERIMENTAL_API |
156 | hb_ot_name_record_ids_t record_ids (platformID, encodingID, languageID, nameID); |
157 | hb_bytes_t* name_bytes; |
158 | |
159 | if (name_table_overrides->has (record_ids, &name_bytes)) { |
160 | hb_bytes_t encoded_bytes = *name_bytes; |
161 | char *name_str_utf16_be = nullptr; |
162 | |
163 | if (platformID != 1) |
164 | { |
165 | unsigned text_size = hb_ot_name_convert_utf<hb_utf8_t, hb_utf16_be_t> (*name_bytes, nullptr, nullptr); |
166 | |
167 | text_size++; // needs to consider NULL terminator for use in hb_ot_name_convert_utf() |
168 | unsigned byte_len = text_size * hb_utf16_be_t::codepoint_t::static_size; |
169 | name_str_utf16_be = (char *) hb_calloc (byte_len, 1); |
170 | if (!name_str_utf16_be) |
171 | { |
172 | c->revert (snap); |
173 | return_trace (nullptr); |
174 | } |
175 | hb_ot_name_convert_utf<hb_utf8_t, hb_utf16_be_t> (*name_bytes, &text_size, |
176 | (hb_utf16_be_t::codepoint_t *) name_str_utf16_be); |
177 | |
178 | unsigned encoded_byte_len = text_size * hb_utf16_be_t::codepoint_t::static_size; |
179 | if (!encoded_byte_len || !c->check_assign (out->length, encoded_byte_len, HB_SERIALIZE_ERROR_INT_OVERFLOW)) { |
180 | c->revert (snap); |
181 | hb_free (name_str_utf16_be); |
182 | return_trace (nullptr); |
183 | } |
184 | |
185 | encoded_bytes = hb_bytes_t (name_str_utf16_be, encoded_byte_len); |
186 | } |
187 | else |
188 | { |
189 | // mac platform, copy the UTF-8 string(all ascii characters) as is |
190 | if (!c->check_assign (out->length, encoded_bytes.length, HB_SERIALIZE_ERROR_INT_OVERFLOW)) { |
191 | c->revert (snap); |
192 | return_trace (nullptr); |
193 | } |
194 | } |
195 | |
196 | out->offset = 0; |
197 | c->push (); |
198 | encoded_bytes.copy (c); |
199 | c->add_link (out->offset, c->pop_pack (), hb_serialize_context_t::Tail, 0); |
200 | hb_free (name_str_utf16_be); |
201 | } |
202 | else |
203 | #endif |
204 | { |
205 | out->offset.serialize_copy (c, offset, base, 0, hb_serialize_context_t::Tail, length); |
206 | } |
207 | return_trace (out); |
208 | } |
209 | |
210 | bool isUnicode () const |
211 | { |
212 | unsigned int p = platformID; |
213 | unsigned int e = encodingID; |
214 | |
215 | return (p == 0 || |
216 | (p == 3 && (e == 0 || e == 1 || e == 10))); |
217 | } |
218 | |
219 | static int cmp (const void *pa, const void *pb) |
220 | { |
221 | const NameRecord *a = (const NameRecord *)pa; |
222 | const NameRecord *b = (const NameRecord *)pb; |
223 | |
224 | if (a->platformID != b->platformID) |
225 | return a->platformID - b->platformID; |
226 | |
227 | if (a->encodingID != b->encodingID) |
228 | return a->encodingID - b->encodingID; |
229 | |
230 | if (a->languageID != b->languageID) |
231 | return a->languageID - b->languageID; |
232 | |
233 | if (a->nameID != b->nameID) |
234 | return a->nameID - b->nameID; |
235 | |
236 | if (a->length != b->length) |
237 | return a->length - b->length; |
238 | |
239 | return 0; |
240 | } |
241 | |
242 | bool sanitize (hb_sanitize_context_t *c, const void *base) const |
243 | { |
244 | TRACE_SANITIZE (this); |
245 | return_trace (c->check_struct (this) && offset.sanitize (c, base, length)); |
246 | } |
247 | |
248 | HBUINT16 platformID; /* Platform ID. */ |
249 | HBUINT16 encodingID; /* Platform-specific encoding ID. */ |
250 | HBUINT16 languageID; /* Language ID. */ |
251 | HBUINT16 nameID; /* Name ID. */ |
252 | HBUINT16 length; /* String length (in bytes). */ |
253 | NNOffset16To<UnsizedArrayOf<HBUINT8>> |
254 | offset; /* String offset from start of storage area (in bytes). */ |
255 | public: |
256 | DEFINE_SIZE_STATIC (12); |
257 | }; |
258 | |
259 | static int |
260 | _hb_ot_name_entry_cmp_key (const void *pa, const void *pb, bool exact) |
261 | { |
262 | const hb_ot_name_entry_t *a = (const hb_ot_name_entry_t *) pa; |
263 | const hb_ot_name_entry_t *b = (const hb_ot_name_entry_t *) pb; |
264 | |
265 | /* Compare by name_id, then language. */ |
266 | |
267 | if (a->name_id != b->name_id) |
268 | return a->name_id - b->name_id; |
269 | |
270 | if (a->language == b->language) return 0; |
271 | if (!a->language) return -1; |
272 | if (!b->language) return +1; |
273 | |
274 | const char *astr = hb_language_to_string (a->language); |
275 | const char *bstr = hb_language_to_string (b->language); |
276 | |
277 | signed c = strcmp (astr, bstr); |
278 | |
279 | // 'a' is the user request, and 'b' is string in the font. |
280 | // If eg. user asks for "en-us" and font has "en", approve. |
281 | if (!exact && c && |
282 | hb_language_matches (b->language, a->language)) |
283 | return 0; |
284 | |
285 | return c; |
286 | } |
287 | |
288 | static int |
289 | _hb_ot_name_entry_cmp (const void *pa, const void *pb) |
290 | { |
291 | /* Compare by name_id, then language, then score, then index. */ |
292 | |
293 | int v = _hb_ot_name_entry_cmp_key (pa, pb, true); |
294 | if (v) |
295 | return v; |
296 | |
297 | const hb_ot_name_entry_t *a = (const hb_ot_name_entry_t *) pa; |
298 | const hb_ot_name_entry_t *b = (const hb_ot_name_entry_t *) pb; |
299 | |
300 | if (a->entry_score != b->entry_score) |
301 | return a->entry_score - b->entry_score; |
302 | |
303 | if (a->entry_index != b->entry_index) |
304 | return a->entry_index - b->entry_index; |
305 | |
306 | return 0; |
307 | } |
308 | |
309 | struct name |
310 | { |
311 | static constexpr hb_tag_t tableTag = HB_OT_TAG_name; |
312 | |
313 | unsigned int get_size () const |
314 | { return min_size + count * nameRecordZ.item_size; } |
315 | |
316 | template <typename Iterator, |
317 | hb_requires (hb_is_source_of (Iterator, const NameRecord &))> |
318 | bool serialize (hb_serialize_context_t *c, |
319 | Iterator it, |
320 | const void *src_string_pool |
321 | #ifdef HB_EXPERIMENTAL_API |
322 | , const hb_vector_t<hb_ot_name_record_ids_t>& insert_name_records |
323 | , const hb_hashmap_t<hb_ot_name_record_ids_t, hb_bytes_t> *name_table_overrides |
324 | #endif |
325 | ) |
326 | { |
327 | TRACE_SERIALIZE (this); |
328 | |
329 | if (unlikely (!c->extend_min ((*this)))) return_trace (false); |
330 | |
331 | unsigned total_count = it.len () |
332 | #ifdef HB_EXPERIMENTAL_API |
333 | + insert_name_records.length |
334 | #endif |
335 | ; |
336 | this->format = 0; |
337 | if (!c->check_assign (this->count, total_count, HB_SERIALIZE_ERROR_INT_OVERFLOW)) |
338 | return false; |
339 | |
340 | NameRecord *name_records = (NameRecord *) hb_calloc (total_count, NameRecord::static_size); |
341 | if (unlikely (!name_records)) return_trace (false); |
342 | |
343 | hb_array_t<NameRecord> records (name_records, total_count); |
344 | |
345 | for (const NameRecord& record : it) |
346 | { |
347 | hb_memcpy (name_records, &record, NameRecord::static_size); |
348 | name_records++; |
349 | } |
350 | |
351 | #ifdef HB_EXPERIMENTAL_API |
352 | for (unsigned i = 0; i < insert_name_records.length; i++) |
353 | { |
354 | const hb_ot_name_record_ids_t& ids = insert_name_records[i]; |
355 | NameRecord record; |
356 | record.platformID = ids.platform_id; |
357 | record.encodingID = ids.encoding_id; |
358 | record.languageID = ids.language_id; |
359 | record.nameID = ids.name_id; |
360 | record.length = 0; // handled in NameRecord copy() |
361 | record.offset = 0; |
362 | hb_memcpy (name_records, &record, NameRecord::static_size); |
363 | name_records++; |
364 | } |
365 | #endif |
366 | |
367 | records.qsort (); |
368 | |
369 | c->copy_all (records, |
370 | src_string_pool |
371 | #ifdef HB_EXPERIMENTAL_API |
372 | , name_table_overrides |
373 | #endif |
374 | ); |
375 | hb_free (records.arrayZ); |
376 | |
377 | |
378 | if (unlikely (c->ran_out_of_room ())) return_trace (false); |
379 | |
380 | this->stringOffset = c->length (); |
381 | |
382 | return_trace (true); |
383 | } |
384 | |
385 | bool subset (hb_subset_context_t *c) const |
386 | { |
387 | auto *name_prime = c->serializer->start_embed<name> (); |
388 | |
389 | #ifdef HB_EXPERIMENTAL_API |
390 | const hb_hashmap_t<hb_ot_name_record_ids_t, hb_bytes_t> *name_table_overrides = |
391 | &c->plan->name_table_overrides; |
392 | #endif |
393 | |
394 | auto it = |
395 | + nameRecordZ.as_array (count) |
396 | | hb_filter (c->plan->name_ids, &NameRecord::nameID) |
397 | | hb_filter (c->plan->name_languages, &NameRecord::languageID) |
398 | | hb_filter ([&] (const NameRecord& namerecord) { |
399 | return |
400 | (c->plan->flags & HB_SUBSET_FLAGS_NAME_LEGACY) |
401 | || namerecord.isUnicode (); |
402 | }) |
403 | #ifdef HB_EXPERIMENTAL_API |
404 | | hb_filter ([&] (const NameRecord& namerecord) { |
405 | if (name_table_overrides->is_empty ()) |
406 | return true; |
407 | hb_ot_name_record_ids_t rec_ids (namerecord.platformID, |
408 | namerecord.encodingID, |
409 | namerecord.languageID, |
410 | namerecord.nameID); |
411 | |
412 | hb_bytes_t *p; |
413 | if (name_table_overrides->has (rec_ids, &p) && |
414 | (*p).length == 0) |
415 | return false; |
416 | return true; |
417 | }) |
418 | #endif |
419 | ; |
420 | |
421 | #ifdef HB_EXPERIMENTAL_API |
422 | hb_hashmap_t<hb_ot_name_record_ids_t, unsigned> retained_name_record_ids; |
423 | for (const NameRecord& rec : it) |
424 | { |
425 | hb_ot_name_record_ids_t rec_ids (rec.platformID, |
426 | rec.encodingID, |
427 | rec.languageID, |
428 | rec.nameID); |
429 | retained_name_record_ids.set (rec_ids, 1); |
430 | } |
431 | |
432 | hb_vector_t<hb_ot_name_record_ids_t> insert_name_records; |
433 | if (!name_table_overrides->is_empty ()) |
434 | { |
435 | if (unlikely (!insert_name_records.alloc (name_table_overrides->get_population (), true))) |
436 | return false; |
437 | for (const auto& record_ids : name_table_overrides->keys ()) |
438 | { |
439 | if (name_table_overrides->get (record_ids).length == 0) |
440 | continue; |
441 | if (retained_name_record_ids.has (record_ids)) |
442 | continue; |
443 | insert_name_records.push (record_ids); |
444 | } |
445 | } |
446 | #endif |
447 | |
448 | return name_prime->serialize (c->serializer, it, |
449 | std::addressof (this + stringOffset) |
450 | #ifdef HB_EXPERIMENTAL_API |
451 | , insert_name_records |
452 | , name_table_overrides |
453 | #endif |
454 | ); |
455 | } |
456 | |
457 | bool sanitize_records (hb_sanitize_context_t *c) const |
458 | { |
459 | TRACE_SANITIZE (this); |
460 | const void *string_pool = (this+stringOffset).arrayZ; |
461 | return_trace (nameRecordZ.sanitize (c, count, string_pool)); |
462 | } |
463 | |
464 | bool sanitize (hb_sanitize_context_t *c) const |
465 | { |
466 | TRACE_SANITIZE (this); |
467 | return_trace (c->check_struct (this) && |
468 | likely (format == 0 || format == 1) && |
469 | c->check_array (nameRecordZ.arrayZ, count) && |
470 | c->check_range (this, stringOffset) && |
471 | sanitize_records (c)); |
472 | } |
473 | |
474 | struct accelerator_t |
475 | { |
476 | accelerator_t (hb_face_t *face) |
477 | { |
478 | this->table = hb_sanitize_context_t ().reference_table<name> (face); |
479 | assert (this->table.get_length () >= this->table->stringOffset); |
480 | this->pool = (const char *) (const void *) (this->table+this->table->stringOffset); |
481 | this->pool_len = this->table.get_length () - this->table->stringOffset; |
482 | const hb_array_t<const NameRecord> all_names (this->table->nameRecordZ.arrayZ, |
483 | this->table->count); |
484 | |
485 | this->names.alloc (all_names.length, true); |
486 | |
487 | for (unsigned int i = 0; i < all_names.length; i++) |
488 | { |
489 | hb_ot_name_entry_t *entry = this->names.push (); |
490 | |
491 | entry->name_id = all_names[i].nameID; |
492 | entry->language = all_names[i].language (face); |
493 | entry->entry_score = all_names[i].score (); |
494 | entry->entry_index = i; |
495 | } |
496 | |
497 | this->names.qsort (_hb_ot_name_entry_cmp); |
498 | /* Walk and pick best only for each name_id,language pair, |
499 | * while dropping unsupported encodings. */ |
500 | unsigned int j = 0; |
501 | for (unsigned int i = 0; i < this->names.length; i++) |
502 | { |
503 | if (this->names[i].entry_score == UNSUPPORTED || |
504 | this->names[i].language == HB_LANGUAGE_INVALID) |
505 | continue; |
506 | if (i && |
507 | this->names[i - 1].name_id == this->names[i].name_id && |
508 | this->names[i - 1].language == this->names[i].language) |
509 | continue; |
510 | this->names[j++] = this->names[i]; |
511 | } |
512 | this->names.resize (j); |
513 | } |
514 | ~accelerator_t () |
515 | { |
516 | this->table.destroy (); |
517 | } |
518 | |
519 | int get_index (hb_ot_name_id_t name_id, |
520 | hb_language_t language, |
521 | unsigned int *width=nullptr) const |
522 | { |
523 | const hb_ot_name_entry_t key = {name_id, {0}, language}; |
524 | const hb_ot_name_entry_t *entry = hb_bsearch (key, (const hb_ot_name_entry_t *) this->names, |
525 | this->names.length, |
526 | sizeof (hb_ot_name_entry_t), |
527 | _hb_ot_name_entry_cmp_key, |
528 | true); |
529 | |
530 | if (!entry) |
531 | { |
532 | entry = hb_bsearch (key, (const hb_ot_name_entry_t *) this->names, |
533 | this->names.length, |
534 | sizeof (hb_ot_name_entry_t), |
535 | _hb_ot_name_entry_cmp_key, |
536 | false); |
537 | } |
538 | |
539 | if (!entry) |
540 | return -1; |
541 | |
542 | if (width) |
543 | *width = entry->entry_score < 10 ? 2 : 1; |
544 | |
545 | return entry->entry_index; |
546 | } |
547 | |
548 | hb_bytes_t get_name (unsigned int idx) const |
549 | { |
550 | const hb_array_t<const NameRecord> all_names (table->nameRecordZ.arrayZ, table->count); |
551 | const NameRecord &record = all_names[idx]; |
552 | const hb_bytes_t string_pool (pool, pool_len); |
553 | return string_pool.sub_array (record.offset, record.length); |
554 | } |
555 | |
556 | private: |
557 | const char *pool; |
558 | unsigned int pool_len; |
559 | public: |
560 | hb_blob_ptr_t<name> table; |
561 | hb_vector_t<hb_ot_name_entry_t> names; |
562 | }; |
563 | |
564 | public: |
565 | /* We only implement format 0 for now. */ |
566 | HBUINT16 format; /* Format selector (=0/1). */ |
567 | HBUINT16 count; /* Number of name records. */ |
568 | NNOffset16To<UnsizedArrayOf<HBUINT8>> |
569 | stringOffset; /* Offset to start of string storage (from start of table). */ |
570 | UnsizedArrayOf<NameRecord> |
571 | nameRecordZ; /* The name records where count is the number of records. */ |
572 | public: |
573 | DEFINE_SIZE_ARRAY (6, nameRecordZ); |
574 | }; |
575 | |
576 | #undef entry_index |
577 | #undef entry_score |
578 | |
579 | struct name_accelerator_t : name::accelerator_t { |
580 | name_accelerator_t (hb_face_t *face) : name::accelerator_t (face) {} |
581 | }; |
582 | |
583 | } /* namespace OT */ |
584 | |
585 | |
586 | #endif /* OT_NAME_NAME_HH */ |
587 | |