1 /*
2 * Copyright © 1998-2004 David Turner and Werner Lemberg
3 * Copyright © 2004,2007,2009,2010 Red Hat, Inc.
4 * Copyright © 2011,2012 Google, Inc.
5 *
6 * This is part of HarfBuzz, a text shaping library.
7 *
8 * Permission is hereby granted, without written agreement and without
9 * license or royalty fees, to use, copy, modify, and distribute this
10 * software and its documentation for any purpose, provided that the
11 * above copyright notice and the following two paragraphs appear in
12 * all copies of this software.
13 *
14 * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
15 * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
16 * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
17 * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
18 * DAMAGE.
19 *
20 * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
21 * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
22 * FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
23 * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
24 * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
25 *
26 * Red Hat Author(s): Owen Taylor, Behdad Esfahbod
27 * Google Author(s): Behdad Esfahbod
28 */
29
30 #ifndef HB_BUFFER_HH
31 #define HB_BUFFER_HH
32
33 #include "hb.hh"
34 #include "hb-unicode.hh"
35
36
37 #ifndef HB_BUFFER_MAX_LEN_FACTOR
38 #define HB_BUFFER_MAX_LEN_FACTOR 32
39 #endif
40 #ifndef HB_BUFFER_MAX_LEN_MIN
41 #define HB_BUFFER_MAX_LEN_MIN 8192
42 #endif
43 #ifndef HB_BUFFER_MAX_LEN_DEFAULT
44 #define HB_BUFFER_MAX_LEN_DEFAULT 0x3FFFFFFF /* Shaping more than a billion chars? Let us know! */
45 #endif
46
47 #ifndef HB_BUFFER_MAX_OPS_FACTOR
48 #define HB_BUFFER_MAX_OPS_FACTOR 64
49 #endif
50 #ifndef HB_BUFFER_MAX_OPS_MIN
51 #define HB_BUFFER_MAX_OPS_MIN 1024
52 #endif
53 #ifndef HB_BUFFER_MAX_OPS_DEFAULT
54 #define HB_BUFFER_MAX_OPS_DEFAULT 0x1FFFFFFF /* Shaping more than a billion operations? Let us know! */
55 #endif
56
57 static_assert ((sizeof (hb_glyph_info_t) == 20), "");
58 static_assert ((sizeof (hb_glyph_info_t) == sizeof (hb_glyph_position_t)), "");
59
60 HB_MARK_AS_FLAG_T (hb_buffer_flags_t);
61 HB_MARK_AS_FLAG_T (hb_buffer_serialize_flags_t);
62 HB_MARK_AS_FLAG_T (hb_buffer_diff_flags_t);
63
64 enum hb_buffer_scratch_flags_t {
65 HB_BUFFER_SCRATCH_FLAG_DEFAULT = 0x00000000u,
66 HB_BUFFER_SCRATCH_FLAG_HAS_NON_ASCII = 0x00000001u,
67 HB_BUFFER_SCRATCH_FLAG_HAS_DEFAULT_IGNORABLES = 0x00000002u,
68 HB_BUFFER_SCRATCH_FLAG_HAS_SPACE_FALLBACK = 0x00000004u,
69 HB_BUFFER_SCRATCH_FLAG_HAS_GPOS_ATTACHMENT = 0x00000008u,
70 HB_BUFFER_SCRATCH_FLAG_HAS_UNSAFE_TO_BREAK = 0x00000010u,
71 HB_BUFFER_SCRATCH_FLAG_HAS_CGJ = 0x00000020u,
72
73 /* Reserved for complex shapers' internal use. */
74 HB_BUFFER_SCRATCH_FLAG_COMPLEX0 = 0x01000000u,
75 HB_BUFFER_SCRATCH_FLAG_COMPLEX1 = 0x02000000u,
76 HB_BUFFER_SCRATCH_FLAG_COMPLEX2 = 0x04000000u,
77 HB_BUFFER_SCRATCH_FLAG_COMPLEX3 = 0x08000000u,
78 };
79 HB_MARK_AS_FLAG_T (hb_buffer_scratch_flags_t);
80
81
82 /*
83 * hb_buffer_t
84 */
85
86 struct hb_buffer_t
87 {
88 hb_object_header_t header;
89
90 /* Information about how the text in the buffer should be treated */
91 hb_unicode_funcs_t *unicode; /* Unicode functions */
92 hb_buffer_flags_t flags; /* BOT / EOT / etc. */
93 hb_buffer_cluster_level_t cluster_level;
94 hb_codepoint_t replacement; /* U+FFFD or something else. */
95 hb_codepoint_t invisible; /* 0 or something else. */
96 hb_buffer_scratch_flags_t scratch_flags; /* Have space-fallback, etc. */
97 unsigned int max_len; /* Maximum allowed len. */
98 int max_ops; /* Maximum allowed operations. */
99
100 /* Buffer contents */
101 hb_buffer_content_type_t content_type;
102 hb_segment_properties_t props; /* Script, language, direction */
103
104 bool successful; /* Allocations successful */
105 bool have_output; /* Whether we have an output buffer going on */
106 bool have_positions; /* Whether we have positions */
107
108 unsigned int idx; /* Cursor into ->info and ->pos arrays */
109 unsigned int len; /* Length of ->info and ->pos arrays */
110 unsigned int out_len; /* Length of ->out array if have_output */
111
112 unsigned int allocated; /* Length of allocated arrays */
113 hb_glyph_info_t *info;
114 hb_glyph_info_t *out_info;
115 hb_glyph_position_t *pos;
116
117 unsigned int serial;
118
119 /* Text before / after the main buffer contents.
120 * Always in Unicode, and ordered outward.
121 * Index 0 is for "pre-context", 1 for "post-context". */
122 static constexpr unsigned CONTEXT_LENGTH = 5u;
123 hb_codepoint_t context[2][CONTEXT_LENGTH];
124 unsigned int context_len[2];
125
126 /* Debugging API */
127 #ifndef HB_NO_BUFFER_MESSAGE
128 hb_buffer_message_func_t message_func;
129 void *message_data;
130 hb_destroy_func_t message_destroy;
131 #endif
132
133 /* Internal debugging. */
134 /* The bits here reflect current allocations of the bytes in glyph_info_t's var1 and var2. */
135 #ifndef HB_NDEBUG
136 uint8_t allocated_var_bits;
137 #endif
138
139
140 /* Methods */
141
in_errorhb_buffer_t142 bool in_error () const { return !successful; }
143
allocate_varhb_buffer_t144 void allocate_var (unsigned int start, unsigned int count)
145 {
146 #ifndef HB_NDEBUG
147 unsigned int end = start + count;
148 assert (end <= 8);
149 unsigned int bits = (1u<<end) - (1u<<start);
150 assert (0 == (allocated_var_bits & bits));
151 allocated_var_bits |= bits;
152 #endif
153 }
deallocate_varhb_buffer_t154 void deallocate_var (unsigned int start, unsigned int count)
155 {
156 #ifndef HB_NDEBUG
157 unsigned int end = start + count;
158 assert (end <= 8);
159 unsigned int bits = (1u<<end) - (1u<<start);
160 assert (bits == (allocated_var_bits & bits));
161 allocated_var_bits &= ~bits;
162 #endif
163 }
assert_varhb_buffer_t164 void assert_var (unsigned int start, unsigned int count)
165 {
166 #ifndef HB_NDEBUG
167 unsigned int end = start + count;
168 assert (end <= 8);
169 unsigned int bits = (1u<<end) - (1u<<start);
170 assert (bits == (allocated_var_bits & bits));
171 #endif
172 }
deallocate_var_allhb_buffer_t173 void deallocate_var_all ()
174 {
175 #ifndef HB_NDEBUG
176 allocated_var_bits = 0;
177 #endif
178 }
179
curhb_buffer_t180 hb_glyph_info_t &cur (unsigned int i = 0) { return info[idx + i]; }
curhb_buffer_t181 hb_glyph_info_t cur (unsigned int i = 0) const { return info[idx + i]; }
182
cur_poshb_buffer_t183 hb_glyph_position_t &cur_pos (unsigned int i = 0) { return pos[idx + i]; }
cur_poshb_buffer_t184 hb_glyph_position_t cur_pos (unsigned int i = 0) const { return pos[idx + i]; }
185
prevhb_buffer_t186 hb_glyph_info_t &prev () { return out_info[out_len ? out_len - 1 : 0]; }
prevhb_buffer_t187 hb_glyph_info_t prev () const { return out_info[out_len ? out_len - 1 : 0]; }
188
has_separate_outputhb_buffer_t189 bool has_separate_output () const { return info != out_info; }
190
191
192 HB_INTERNAL void reset ();
193 HB_INTERNAL void clear ();
194
backtrack_lenhb_buffer_t195 unsigned int backtrack_len () const { return have_output? out_len : idx; }
lookahead_lenhb_buffer_t196 unsigned int lookahead_len () const { return len - idx; }
next_serialhb_buffer_t197 unsigned int next_serial () { return serial++; }
198
199 HB_INTERNAL void add (hb_codepoint_t codepoint,
200 unsigned int cluster);
201 HB_INTERNAL void add_info (const hb_glyph_info_t &glyph_info);
202
203 HB_INTERNAL void reverse_range (unsigned int start, unsigned int end);
204 HB_INTERNAL void reverse ();
205 HB_INTERNAL void reverse_clusters ();
206 HB_INTERNAL void guess_segment_properties ();
207
208 HB_INTERNAL void swap_buffers ();
209 HB_INTERNAL void remove_output ();
210 HB_INTERNAL void clear_output ();
211 HB_INTERNAL void clear_positions ();
212
213 HB_INTERNAL void replace_glyphs (unsigned int num_in,
214 unsigned int num_out,
215 const hb_codepoint_t *glyph_data);
216
replace_glyphhb_buffer_t217 void replace_glyph (hb_codepoint_t glyph_index)
218 {
219 if (unlikely (out_info != info || out_len != idx)) {
220 if (unlikely (!make_room_for (1, 1))) return;
221 out_info[out_len] = info[idx];
222 }
223 out_info[out_len].codepoint = glyph_index;
224
225 idx++;
226 out_len++;
227 }
228 /* Makes a copy of the glyph at idx to output and replace glyph_index */
output_glyphhb_buffer_t229 hb_glyph_info_t & output_glyph (hb_codepoint_t glyph_index)
230 {
231 if (unlikely (!make_room_for (0, 1))) return Crap(hb_glyph_info_t);
232
233 if (unlikely (idx == len && !out_len))
234 return Crap(hb_glyph_info_t);
235
236 out_info[out_len] = idx < len ? info[idx] : out_info[out_len - 1];
237 out_info[out_len].codepoint = glyph_index;
238
239 out_len++;
240
241 return out_info[out_len - 1];
242 }
output_infohb_buffer_t243 void output_info (const hb_glyph_info_t &glyph_info)
244 {
245 if (unlikely (!make_room_for (0, 1))) return;
246
247 out_info[out_len] = glyph_info;
248
249 out_len++;
250 }
251 /* Copies glyph at idx to output but doesn't advance idx */
copy_glyphhb_buffer_t252 void copy_glyph ()
253 {
254 if (unlikely (!make_room_for (0, 1))) return;
255
256 out_info[out_len] = info[idx];
257
258 out_len++;
259 }
260 /* Copies glyph at idx to output and advance idx.
261 * If there's no output, just advance idx. */
262 void
next_glyphhb_buffer_t263 next_glyph ()
264 {
265 if (have_output)
266 {
267 if (out_info != info || out_len != idx)
268 {
269 if (unlikely (!make_room_for (1, 1))) return;
270 out_info[out_len] = info[idx];
271 }
272 out_len++;
273 }
274
275 idx++;
276 }
277 /* Copies n glyphs at idx to output and advance idx.
278 * If there's no output, just advance idx. */
279 void
next_glyphshb_buffer_t280 next_glyphs (unsigned int n)
281 {
282 if (have_output)
283 {
284 if (out_info != info || out_len != idx)
285 {
286 if (unlikely (!make_room_for (n, n))) return;
287 memmove (out_info + out_len, info + idx, n * sizeof (out_info[0]));
288 }
289 out_len += n;
290 }
291
292 idx += n;
293 }
294 /* Advance idx without copying to output. */
skip_glyphhb_buffer_t295 void skip_glyph () { idx++; }
reset_maskshb_buffer_t296 void reset_masks (hb_mask_t mask)
297 {
298 for (unsigned int j = 0; j < len; j++)
299 info[j].mask = mask;
300 }
add_maskshb_buffer_t301 void add_masks (hb_mask_t mask)
302 {
303 for (unsigned int j = 0; j < len; j++)
304 info[j].mask |= mask;
305 }
306 HB_INTERNAL void set_masks (hb_mask_t value, hb_mask_t mask,
307 unsigned int cluster_start, unsigned int cluster_end);
308
merge_clustershb_buffer_t309 void merge_clusters (unsigned int start, unsigned int end)
310 {
311 if (end - start < 2)
312 return;
313 merge_clusters_impl (start, end);
314 }
315 HB_INTERNAL void merge_clusters_impl (unsigned int start, unsigned int end);
316 HB_INTERNAL void merge_out_clusters (unsigned int start, unsigned int end);
317 /* Merge clusters for deleting current glyph, and skip it. */
318 HB_INTERNAL void delete_glyph ();
319
unsafe_to_breakhb_buffer_t320 void unsafe_to_break (unsigned int start,
321 unsigned int end)
322 {
323 if (end - start < 2)
324 return;
325 unsafe_to_break_impl (start, end);
326 }
327 HB_INTERNAL void unsafe_to_break_impl (unsigned int start, unsigned int end);
328 HB_INTERNAL void unsafe_to_break_from_outbuffer (unsigned int start, unsigned int end);
329
330
331 /* Internal methods */
332 HB_INTERNAL bool move_to (unsigned int i); /* i is output-buffer index. */
333
334 HB_INTERNAL bool enlarge (unsigned int size);
335
ensurehb_buffer_t336 bool ensure (unsigned int size)
337 { return likely (!size || size < allocated) ? true : enlarge (size); }
338
ensure_inplacehb_buffer_t339 bool ensure_inplace (unsigned int size)
340 { return likely (!size || size < allocated); }
341
342 HB_INTERNAL bool make_room_for (unsigned int num_in, unsigned int num_out);
343 HB_INTERNAL bool shift_forward (unsigned int count);
344
345 typedef long scratch_buffer_t;
346 HB_INTERNAL scratch_buffer_t *get_scratch_buffer (unsigned int *size);
347
clear_contexthb_buffer_t348 void clear_context (unsigned int side) { context_len[side] = 0; }
349
350 HB_INTERNAL void sort (unsigned int start, unsigned int end, int(*compar)(const hb_glyph_info_t *, const hb_glyph_info_t *));
351
messaginghb_buffer_t352 bool messaging ()
353 {
354 #ifdef HB_NO_BUFFER_MESSAGE
355 return false;
356 #else
357 return unlikely (message_func);
358 #endif
359 }
messagehb_buffer_t360 bool message (hb_font_t *font, const char *fmt, ...) HB_PRINTF_FUNC(3, 4)
361 {
362 #ifdef HB_NO_BUFFER_MESSAGE
363 return true;
364 #else
365 if (!messaging ())
366 return true;
367 va_list ap;
368 va_start (ap, fmt);
369 bool ret = message_impl (font, fmt, ap);
370 va_end (ap);
371 return ret;
372 #endif
373 }
374 HB_INTERNAL bool message_impl (hb_font_t *font, const char *fmt, va_list ap) HB_PRINTF_FUNC(3, 0);
375
376 static void
set_clusterhb_buffer_t377 set_cluster (hb_glyph_info_t &inf, unsigned int cluster, unsigned int mask = 0)
378 {
379 if (inf.cluster != cluster)
380 {
381 if (mask & HB_GLYPH_FLAG_UNSAFE_TO_BREAK)
382 inf.mask |= HB_GLYPH_FLAG_UNSAFE_TO_BREAK;
383 else
384 inf.mask &= ~HB_GLYPH_FLAG_UNSAFE_TO_BREAK;
385 }
386 inf.cluster = cluster;
387 }
388
389 int
_unsafe_to_break_find_min_clusterhb_buffer_t390 _unsafe_to_break_find_min_cluster (const hb_glyph_info_t *infos,
391 unsigned int start, unsigned int end,
392 unsigned int cluster) const
393 {
394 for (unsigned int i = start; i < end; i++)
395 cluster = hb_min (cluster, infos[i].cluster);
396 return cluster;
397 }
398 void
_unsafe_to_break_set_maskhb_buffer_t399 _unsafe_to_break_set_mask (hb_glyph_info_t *infos,
400 unsigned int start, unsigned int end,
401 unsigned int cluster)
402 {
403 for (unsigned int i = start; i < end; i++)
404 if (cluster != infos[i].cluster)
405 {
406 scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_UNSAFE_TO_BREAK;
407 infos[i].mask |= HB_GLYPH_FLAG_UNSAFE_TO_BREAK;
408 }
409 }
410
unsafe_to_break_allhb_buffer_t411 void unsafe_to_break_all () { unsafe_to_break_impl (0, len); }
safe_to_break_allhb_buffer_t412 void safe_to_break_all ()
413 {
414 for (unsigned int i = 0; i < len; i++)
415 info[i].mask &= ~HB_GLYPH_FLAG_UNSAFE_TO_BREAK;
416 }
417 };
418 DECLARE_NULL_INSTANCE (hb_buffer_t);
419
420
421 /* Loop over clusters. Duplicated in foreach_syllable(). */
422 #define foreach_cluster(buffer, start, end) \
423 for (unsigned int \
424 _count = buffer->len, \
425 start = 0, end = _count ? _next_cluster (buffer, 0) : 0; \
426 start < _count; \
427 start = end, end = _next_cluster (buffer, start))
428
429 static inline unsigned int
_next_cluster(hb_buffer_t * buffer,unsigned int start)430 _next_cluster (hb_buffer_t *buffer, unsigned int start)
431 {
432 hb_glyph_info_t *info = buffer->info;
433 unsigned int count = buffer->len;
434
435 unsigned int cluster = info[start].cluster;
436 while (++start < count && cluster == info[start].cluster)
437 ;
438
439 return start;
440 }
441
442
443 #define HB_BUFFER_XALLOCATE_VAR(b, func, var) \
444 b->func (offsetof (hb_glyph_info_t, var) - offsetof(hb_glyph_info_t, var1), \
445 sizeof (b->info[0].var))
446 #define HB_BUFFER_ALLOCATE_VAR(b, var) HB_BUFFER_XALLOCATE_VAR (b, allocate_var, var ())
447 #define HB_BUFFER_DEALLOCATE_VAR(b, var) HB_BUFFER_XALLOCATE_VAR (b, deallocate_var, var ())
448 #define HB_BUFFER_ASSERT_VAR(b, var) HB_BUFFER_XALLOCATE_VAR (b, assert_var, var ())
449
450
451 #endif /* HB_BUFFER_HH */
452