forked from HandBrake/HandBrake
-
Notifications
You must be signed in to change notification settings - Fork 0
/
decpgssub.c
515 lines (465 loc) · 18.2 KB
/
decpgssub.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
/* decpgssub.c
Copyright (c) 2003-2017 HandBrake Team
This file is part of the HandBrake source code
Homepage: <http://handbrake.fr/>.
It may be used under the terms of the GNU General Public License v2.
For full terms see the file COPYING file or visit http://www.gnu.org/licenses/gpl-2.0.html
*/
#include "hb.h"
#include "hbffmpeg.h"
struct hb_work_private_s
{
AVCodecContext * context;
hb_job_t * job;
// For PGS subs, when doing passthru, we don't know if we need a
// packet until we have processed several packets. So we cache
// all the packets we see until libav returns a subtitle with
// the information we need.
hb_buffer_list_t list_pass;
// It is possible for multiple subtitles to be enncapsulated in
// one packet. This won't happen for PGS subs, but may for other
// types of subtitles. Since I plan to generalize this code to handle
// other than PGS, we will need to keep a list of all subtitles seen
// while parsing an input packet.
hb_buffer_list_t list;
// XXX: we may occasionally see subtitles with broken timestamps
// while this should really get fixed elsewhere,
// dropping subtitles should be avoided as much as possible
int64_t last_pts;
// for PGS subs, we need to pass 'empty' subtitles through (they clear the
// display) - when doing forced-only extraction, only pass empty subtitles
// through if we've seen a forced sub and haven't seen any empty sub since
uint8_t seen_forced_sub;
// if we start encoding partway through the source, we may encounter empty
// subtitles before we see any actual subtitle content - discard them
uint8_t discard_subtitle;
};
static int decsubInit( hb_work_object_t * w, hb_job_t * job )
{
AVCodec *codec = avcodec_find_decoder( AV_CODEC_ID_HDMV_PGS_SUBTITLE );
AVCodecContext *context = avcodec_alloc_context3( codec );
context->codec = codec;
hb_work_private_t * pv;
pv = calloc( 1, sizeof( hb_work_private_t ) );
w->private_data = pv;
hb_buffer_list_clear(&pv->list);
hb_buffer_list_clear(&pv->list_pass);
pv->discard_subtitle = 1;
pv->seen_forced_sub = 0;
pv->last_pts = AV_NOPTS_VALUE;
pv->context = context;
pv->job = job;
// Set decoder opts...
AVDictionary * av_opts = NULL;
// e.g. av_dict_set( &av_opts, "refcounted_frames", "1", 0 );
if (hb_avcodec_open(pv->context, codec, &av_opts, 0))
{
av_dict_free( &av_opts );
hb_log("decsubInit: avcodec_open failed");
return 1;
}
av_dict_free( &av_opts );
return 0;
}
static void make_empty_pgs( hb_buffer_t * buf )
{
hb_buffer_t * b = buf;
uint8_t done = 0;
// Each buffer is composed of 1 or more segments.
// Segment header is:
// type - 1 byte
// length - 2 bytes
// We want to modify the presentation segment which is type 0x16
//
// Note that every pgs display set is required to have a presentation
// segment, so we will only have to look at one display set.
while ( b && !done )
{
int ii = 0;
while (ii + 3 <= b->size)
{
uint8_t type;
int len;
int segment_len_pos;
type = b->data[ii++];
segment_len_pos = ii;
len = ((int)b->data[ii] << 8) + b->data[ii+1];
ii += 2;
if (type == 0x16 && ii + len <= b->size)
{
int obj_count;
int kk, jj = ii;
int obj_start;
// Skip
// video descriptor 5 bytes
// composition descriptor 3 bytes
// palette update flg 1 byte
// palette id ref 1 byte
jj += 10;
// Set number of composition objects to 0
obj_count = b->data[jj];
b->data[jj] = 0;
jj++;
obj_start = jj;
// And remove all the composition objects
for (kk = 0; kk < obj_count; kk++)
{
uint8_t crop;
crop = b->data[jj + 3];
// skip
// object id - 2 bytes
// window id - 1 byte
// object/forced flag - 1 byte
// x pos - 2 bytes
// y pos - 2 bytes
jj += 8;
if (crop & 0x80)
{
// skip
// crop x - 2 bytes
// crop y - 2 bytes
// crop w - 2 bytes
// crop h - 2 bytes
jj += 8;
}
}
if (jj < b->size)
{
memmove(b->data + obj_start, b->data + jj, b->size - jj);
}
b->size = obj_start + ( b->size - jj );
done = 1;
len = obj_start - (segment_len_pos + 2);
b->data[segment_len_pos] = len >> 8;
b->data[segment_len_pos+1] = len & 0xff;
break;
}
ii += len;
}
b = b->next;
}
}
static int decsubWork( hb_work_object_t * w, hb_buffer_t ** buf_in,
hb_buffer_t ** buf_out )
{
hb_work_private_t * pv = w->private_data;
hb_buffer_t * in = *buf_in;
if (in->s.flags & HB_BUF_FLAG_EOF)
{
/* EOF on input stream - send it downstream & say that we're done */
*buf_in = NULL;
hb_buffer_list_append(&pv->list, in);
*buf_out = hb_buffer_list_clear(&pv->list);
return HB_WORK_DONE;
}
if ( !pv->job->indepth_scan &&
w->subtitle->config.dest == PASSTHRUSUB &&
hb_subtitle_can_pass( PGSSUB, pv->job->mux ) )
{
// Append to buffer list. It will be sent to fifo after we determine
// if this is a packet we need.
hb_buffer_list_append(&pv->list_pass, in);
// We are keeping the buffer, so prevent the filter loop from
// deleting it.
*buf_in = NULL;
}
AVSubtitle subtitle;
memset( &subtitle, 0, sizeof(subtitle) );
AVPacket avp;
av_init_packet( &avp );
avp.data = in->data;
avp.size = in->size;
// libav wants pkt pts in AV_TIME_BASE units
if (in->s.start != AV_NOPTS_VALUE)
{
avp.pts = av_rescale(in->s.start, AV_TIME_BASE, 90000);
}
else
{
avp.pts = AV_NOPTS_VALUE;
}
int has_subtitle = 0;
do
{
int usedBytes = avcodec_decode_subtitle2( pv->context, &subtitle, &has_subtitle, &avp );
if (usedBytes < 0)
{
hb_log("unable to decode subtitle with %d bytes.", avp.size);
return HB_WORK_OK;
}
if (usedBytes <= avp.size)
{
avp.data += usedBytes;
avp.size -= usedBytes;
}
else
{
avp.size = 0;
}
/* Subtitles are "usable" if:
* 1. Libav returned a subtitle (has_subtitle) AND
* 2. we're not doing Foreign Audio Search (!pv->job->indepth_scan) AND
* 3. the sub is non-empty or we've seen one such sub before (!pv->discard_subtitle)
* For forced-only extraction, usable subtitles also need to:
* a. be forced (subtitle.rects[0]->flags & AV_SUBTITLE_FLAG_FORCED) OR
* b. follow a forced sub (pv->seen_forced_sub) */
uint8_t forced_sub = 0;
uint8_t useable_sub = 0;
uint8_t clear_subtitle = 0;
if (has_subtitle)
{
// subtitle statistics
if (subtitle.num_rects)
{
w->subtitle->hits++;
if (subtitle.rects[0]->flags & AV_SUBTITLE_FLAG_FORCED)
{
forced_sub = 1;
w->subtitle->forced_hits++;
}
}
else
{
clear_subtitle = 1;
}
// are we doing Foreign Audio Search?
if (!pv->job->indepth_scan)
{
// do we want to discard this subtitle?
pv->discard_subtitle = pv->discard_subtitle && clear_subtitle;
// do we need this subtitle?
useable_sub = (!pv->discard_subtitle &&
(!w->subtitle->config.force ||
forced_sub || pv->seen_forced_sub));
// do we need to create an empty subtitle?
if (w->subtitle->config.force &&
useable_sub && !forced_sub && !clear_subtitle)
{
// We are forced-only and need to output this subtitle, but
// it's neither forced nor empty.
//
// If passthru, create an empty subtitle.
// Also, flag an empty subtitle for subtitle RENDER.
make_empty_pgs(hb_buffer_list_head(&pv->list_pass));
clear_subtitle = 1;
}
// is the subtitle forced?
pv->seen_forced_sub = forced_sub;
}
}
if (useable_sub)
{
int64_t pts = AV_NOPTS_VALUE;
hb_buffer_t * out = NULL;
if (subtitle.pts != AV_NOPTS_VALUE)
{
pts = av_rescale(subtitle.pts, 90000, AV_TIME_BASE);
}
else
{
if (in->s.start >= 0)
{
pts = in->s.start;
}
else
{
// XXX: a broken pts will cause us to drop this subtitle,
// which is bad; use a default duration of 3 seconds
//
// A broken pts is only generated when a pgs packet
// occurs after a discontinuity and before the
// next audio or video packet which re-establishes
// timing (afaik).
if (pv->last_pts == AV_NOPTS_VALUE)
{
pts = 0LL;
}
else
{
pts = pv->last_pts + 3 * 90000LL;
}
hb_log("[warning] decpgssub: track %d, invalid PTS",
w->subtitle->out_track);
}
}
// work around broken timestamps
if (pts < pv->last_pts)
{
// XXX: this should only happen if the prevous pts
// was unknown and our 3 second default duration
// overshot the next pgs pts.
//
// assign a 1 second duration
hb_log("decpgssub: track %d, non-monotically increasing PTS, last %"PRId64" current %"PRId64"",
w->subtitle->out_track,
pv->last_pts, pts);
pts = pv->last_pts + 1 * 90000LL;
}
pv->last_pts = pts;
if ( w->subtitle->config.dest == PASSTHRUSUB &&
hb_subtitle_can_pass( PGSSUB, pv->job->mux ) )
{
/* PGS subtitles are spread across multiple packets,
* 1 per segment.
*
* In the MKV container, all segments are found in the same
* packet (this is expected by some devices, such as the
* WD TV Live). So if there are multiple packets,
* merge them. */
if (hb_buffer_list_count(&pv->list_pass) == 1)
{
// packets already merged (e.g. MKV sources)
out = hb_buffer_list_clear(&pv->list_pass);
}
else
{
int size = 0;
uint8_t * data;
hb_buffer_t * b;
b = hb_buffer_list_head(&pv->list_pass);
while (b != NULL)
{
size += b->size;
b = b->next;
}
out = hb_buffer_init( size );
data = out->data;
b = hb_buffer_list_head(&pv->list_pass);
while (b != NULL)
{
memcpy(data, b->data, b->size);
data += b->size;
b = b->next;
}
hb_buffer_list_close(&pv->list_pass);
out->s = in->s;
}
out->s.frametype = HB_FRAME_SUBTITLE;
out->s.renderOffset = AV_NOPTS_VALUE;
out->s.stop = AV_NOPTS_VALUE;
out->s.start = pts;
}
else
{
if (!clear_subtitle)
{
unsigned ii, x0, y0, x1, y1, w, h;
x0 = subtitle.rects[0]->x;
y0 = subtitle.rects[0]->y;
x1 = subtitle.rects[0]->x + subtitle.rects[0]->w;
y1 = subtitle.rects[0]->y + subtitle.rects[0]->h;
// First, find total bounding rectangle
for (ii = 1; ii < subtitle.num_rects; ii++)
{
if (subtitle.rects[ii]->x < x0)
x0 = subtitle.rects[ii]->x;
if (subtitle.rects[ii]->y < y0)
y0 = subtitle.rects[ii]->y;
if (subtitle.rects[ii]->x + subtitle.rects[ii]->w > x1)
x1 = subtitle.rects[ii]->x + subtitle.rects[ii]->w;
if (subtitle.rects[ii]->y + subtitle.rects[ii]->h > y1)
y1 = subtitle.rects[ii]->y + subtitle.rects[ii]->h;
}
w = x1 - x0;
h = y1 - y0;
out = hb_frame_buffer_init(AV_PIX_FMT_YUVA420P, w, h);
memset(out->data, 0, out->size);
out->s.frametype = HB_FRAME_SUBTITLE;
out->s.id = in->s.id;
out->s.start = pts;
out->s.stop = AV_NOPTS_VALUE;
out->s.renderOffset = AV_NOPTS_VALUE;
out->s.scr_sequence = in->s.scr_sequence;
out->f.x = x0;
out->f.y = y0;
out->f.window_width = pv->context->width;
out->f.window_height = pv->context->height;
for (ii = 0; ii < subtitle.num_rects; ii++)
{
AVSubtitleRect *rect = subtitle.rects[ii];
int off_x = rect->x - x0;
int off_y = rect->y - y0;
uint8_t *lum = out->plane[0].data;
uint8_t *chromaU = out->plane[1].data;
uint8_t *chromaV = out->plane[2].data;
uint8_t *alpha = out->plane[3].data;
lum += off_y * out->plane[0].stride + off_x;
alpha += off_y * out->plane[3].stride + off_x;
chromaU += (off_y >> 1) * out->plane[1].stride + (off_x >> 1);
chromaV += (off_y >> 1) * out->plane[2].stride + (off_x >> 1);
int xx, yy;
for (yy = 0; yy < rect->h; yy++)
{
for (xx = 0; xx < rect->w; xx++)
{
uint32_t argb, yuv;
int pixel;
uint8_t color;
pixel = yy * rect->w + xx;
color = rect->data[0][pixel];
argb = ((uint32_t*)rect->data[1])[color];
yuv = hb_rgb2yuv(argb);
lum[xx] = (yuv >> 16) & 0xff;
alpha[xx] = (argb >> 24) & 0xff;
if ((xx & 1) == 0 && (yy & 1) == 0)
{
chromaV[xx>>1] = (yuv >> 8) & 0xff;
chromaU[xx>>1] = yuv & 0xff;
}
}
lum += out->plane[0].stride;
if ((yy & 1) == 0)
{
chromaU += out->plane[1].stride;
chromaV += out->plane[2].stride;
}
alpha += out->plane[3].stride;
}
}
hb_buffer_list_append(&pv->list, out);
out = NULL;
}
else
{
out = hb_buffer_init( 0 );
out->s.frametype = HB_FRAME_SUBTITLE;
out->s.flags = HB_BUF_FLAG_EOS;
out->s.id = in->s.id;
out->s.start = pts;
out->s.stop = pts;
out->s.renderOffset = AV_NOPTS_VALUE;
out->s.scr_sequence = in->s.scr_sequence;
out->f.x = 0;
out->f.y = 0;
out->f.width = 0;
out->f.height = 0;
}
}
hb_buffer_list_append(&pv->list, out);
}
else if (has_subtitle)
{
hb_buffer_list_close(&pv->list_pass);
}
if (has_subtitle)
{
avsubtitle_free(&subtitle);
}
} while (avp.size > 0);
*buf_out = hb_buffer_list_clear(&pv->list);
return HB_WORK_OK;
}
static void decsubClose( hb_work_object_t * w )
{
hb_work_private_t * pv = w->private_data;
avcodec_flush_buffers( pv->context );
avcodec_close( pv->context );
}
hb_work_object_t hb_decpgssub =
{
WORK_DECPGSSUB,
"PGS decoder",
decsubInit,
decsubWork,
decsubClose
};