Details | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
6148 | serge | 1 | /* |
2 | * General DV muxer/demuxer |
||
3 | * Copyright (c) 2003 Roman Shaposhnik |
||
4 | * |
||
5 | * Many thanks to Dan Dennedy |
||
6 | * of DV technical info. |
||
7 | * |
||
8 | * Raw DV format |
||
9 | * Copyright (c) 2002 Fabrice Bellard |
||
10 | * |
||
11 | * 50 Mbps (DVCPRO50) support |
||
12 | * Copyright (c) 2006 Daniel Maas |
||
13 | * |
||
14 | * This file is part of FFmpeg. |
||
15 | * |
||
16 | * FFmpeg is free software; you can redistribute it and/or |
||
17 | * modify it under the terms of the GNU Lesser General Public |
||
18 | * License as published by the Free Software Foundation; either |
||
19 | * version 2.1 of the License, or (at your option) any later version. |
||
20 | * |
||
21 | * FFmpeg is distributed in the hope that it will be useful, |
||
22 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
||
23 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
||
24 | * Lesser General Public License for more details. |
||
25 | * |
||
26 | * You should have received a copy of the GNU Lesser General Public |
||
27 | * License along with FFmpeg; if not, write to the Free Software |
||
28 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
||
29 | */ |
||
30 | #include |
||
31 | #include |
||
32 | |||
33 | #include "avformat.h" |
||
34 | #include "internal.h" |
||
35 | #include "libavcodec/dv_profile.h" |
||
36 | #include "libavcodec/dvdata.h" |
||
37 | #include "dv.h" |
||
38 | #include "libavutil/fifo.h" |
||
39 | #include "libavutil/mathematics.h" |
||
40 | #include "libavutil/intreadwrite.h" |
||
41 | #include "libavutil/opt.h" |
||
42 | #include "libavutil/timecode.h" |
||
43 | |||
44 | #define MAX_AUDIO_FRAME_SIZE 192000 // 1 second of 48khz 32bit audio |
||
45 | |||
46 | struct DVMuxContext { |
||
47 | AVClass *av_class; |
||
48 | const DVprofile* sys; /* current DV profile, e.g.: 525/60, 625/50 */ |
||
49 | int n_ast; /* number of stereo audio streams (up to 2) */ |
||
50 | AVStream *ast[2]; /* stereo audio streams */ |
||
51 | AVFifoBuffer *audio_data[2]; /* FIFO for storing excessive amounts of PCM */ |
||
52 | int frames; /* current frame number */ |
||
53 | int64_t start_time; /* recording start time */ |
||
54 | int has_audio; /* frame under construction has audio */ |
||
55 | int has_video; /* frame under construction has video */ |
||
56 | uint8_t frame_buf[DV_MAX_FRAME_SIZE]; /* frame under construction */ |
||
57 | AVTimecode tc; /* timecode context */ |
||
58 | }; |
||
59 | |||
60 | static const int dv_aaux_packs_dist[12][9] = { |
||
61 | { 0xff, 0xff, 0xff, 0x50, 0x51, 0x52, 0x53, 0xff, 0xff }, |
||
62 | { 0x50, 0x51, 0x52, 0x53, 0xff, 0xff, 0xff, 0xff, 0xff }, |
||
63 | { 0xff, 0xff, 0xff, 0x50, 0x51, 0x52, 0x53, 0xff, 0xff }, |
||
64 | { 0x50, 0x51, 0x52, 0x53, 0xff, 0xff, 0xff, 0xff, 0xff }, |
||
65 | { 0xff, 0xff, 0xff, 0x50, 0x51, 0x52, 0x53, 0xff, 0xff }, |
||
66 | { 0x50, 0x51, 0x52, 0x53, 0xff, 0xff, 0xff, 0xff, 0xff }, |
||
67 | { 0xff, 0xff, 0xff, 0x50, 0x51, 0x52, 0x53, 0xff, 0xff }, |
||
68 | { 0x50, 0x51, 0x52, 0x53, 0xff, 0xff, 0xff, 0xff, 0xff }, |
||
69 | { 0xff, 0xff, 0xff, 0x50, 0x51, 0x52, 0x53, 0xff, 0xff }, |
||
70 | { 0x50, 0x51, 0x52, 0x53, 0xff, 0xff, 0xff, 0xff, 0xff }, |
||
71 | { 0xff, 0xff, 0xff, 0x50, 0x51, 0x52, 0x53, 0xff, 0xff }, |
||
72 | { 0x50, 0x51, 0x52, 0x53, 0xff, 0xff, 0xff, 0xff, 0xff }, |
||
73 | }; |
||
74 | |||
75 | static int dv_audio_frame_size(const DVprofile* sys, int frame) |
||
76 | { |
||
77 | return sys->audio_samples_dist[frame % (sizeof(sys->audio_samples_dist) / |
||
78 | sizeof(sys->audio_samples_dist[0]))]; |
||
79 | } |
||
80 | |||
81 | static int dv_write_pack(enum dv_pack_type pack_id, DVMuxContext *c, uint8_t* buf, ...) |
||
82 | { |
||
83 | struct tm tc; |
||
84 | time_t ct; |
||
85 | uint32_t timecode; |
||
86 | va_list ap; |
||
87 | |||
88 | buf[0] = (uint8_t)pack_id; |
||
89 | switch (pack_id) { |
||
90 | case dv_timecode: |
||
91 | timecode = av_timecode_get_smpte_from_framenum(&c->tc, c->frames); |
||
92 | timecode |= 1<<23 | 1<<15 | 1<<7 | 1<<6; // biphase and binary group flags |
||
93 | AV_WB32(buf + 1, timecode); |
||
94 | break; |
||
95 | case dv_audio_source: /* AAUX source pack */ |
||
96 | va_start(ap, buf); |
||
97 | buf[1] = (1 << 7) | /* locked mode -- SMPTE only supports locked mode */ |
||
98 | (1 << 6) | /* reserved -- always 1 */ |
||
99 | (dv_audio_frame_size(c->sys, c->frames) - |
||
100 | c->sys->audio_min_samples[0]); |
||
101 | /* # of samples */ |
||
102 | buf[2] = (0 << 7) | /* multi-stereo */ |
||
103 | (0 << 5) | /* #of audio channels per block: 0 -- 1 channel */ |
||
104 | (0 << 4) | /* pair bit: 0 -- one pair of channels */ |
||
105 | !!va_arg(ap, int); /* audio mode */ |
||
106 | buf[3] = (1 << 7) | /* res */ |
||
107 | (1 << 6) | /* multi-language flag */ |
||
108 | (c->sys->dsf << 5) | /* system: 60fields/50fields */ |
||
109 | (c->sys->n_difchan & 2); /* definition: 0 -- 25Mbps, 2 -- 50Mbps */ |
||
110 | buf[4] = (1 << 7) | /* emphasis: 1 -- off */ |
||
111 | (0 << 6) | /* emphasis time constant: 0 -- reserved */ |
||
112 | (0 << 3) | /* frequency: 0 -- 48kHz, 1 -- 44,1kHz, 2 -- 32kHz */ |
||
113 | 0; /* quantization: 0 -- 16bit linear, 1 -- 12bit nonlinear */ |
||
114 | va_end(ap); |
||
115 | break; |
||
116 | case dv_audio_control: |
||
117 | buf[1] = (0 << 6) | /* copy protection: 0 -- unrestricted */ |
||
118 | (1 << 4) | /* input source: 1 -- digital input */ |
||
119 | (3 << 2) | /* compression: 3 -- no information */ |
||
120 | 0; /* misc. info/SMPTE emphasis off */ |
||
121 | buf[2] = (1 << 7) | /* recording start point: 1 -- no */ |
||
122 | (1 << 6) | /* recording end point: 1 -- no */ |
||
123 | (1 << 3) | /* recording mode: 1 -- original */ |
||
124 | 7; |
||
125 | buf[3] = (1 << 7) | /* direction: 1 -- forward */ |
||
126 | (c->sys->pix_fmt == AV_PIX_FMT_YUV420P ? 0x20 : /* speed */ |
||
127 | c->sys->ltc_divisor * 4); |
||
128 | buf[4] = (1 << 7) | /* reserved -- always 1 */ |
||
129 | 0x7f; /* genre category */ |
||
130 | break; |
||
131 | case dv_audio_recdate: |
||
132 | case dv_video_recdate: /* VAUX recording date */ |
||
133 | ct = c->start_time + av_rescale_rnd(c->frames, c->sys->time_base.num, |
||
134 | c->sys->time_base.den, AV_ROUND_DOWN); |
||
135 | ff_brktimegm(ct, &tc); |
||
136 | buf[1] = 0xff; /* ds, tm, tens of time zone, units of time zone */ |
||
137 | /* 0xff is very likely to be "unknown" */ |
||
138 | buf[2] = (3 << 6) | /* reserved -- always 1 */ |
||
139 | ((tc.tm_mday / 10) << 4) | /* Tens of day */ |
||
140 | (tc.tm_mday % 10); /* Units of day */ |
||
141 | buf[3] = /* we set high 4 bits to 0, shouldn't we set them to week? */ |
||
142 | ((tc.tm_mon / 10) << 4) | /* Tens of month */ |
||
143 | (tc.tm_mon % 10); /* Units of month */ |
||
144 | buf[4] = (((tc.tm_year % 100) / 10) << 4) | /* Tens of year */ |
||
145 | (tc.tm_year % 10); /* Units of year */ |
||
146 | break; |
||
147 | case dv_audio_rectime: /* AAUX recording time */ |
||
148 | case dv_video_rectime: /* VAUX recording time */ |
||
149 | ct = c->start_time + av_rescale_rnd(c->frames, c->sys->time_base.num, |
||
150 | c->sys->time_base.den, AV_ROUND_DOWN); |
||
151 | ff_brktimegm(ct, &tc); |
||
152 | buf[1] = (3 << 6) | /* reserved -- always 1 */ |
||
153 | 0x3f; /* tens of frame, units of frame: 0x3f - "unknown" ? */ |
||
154 | buf[2] = (1 << 7) | /* reserved -- always 1 */ |
||
155 | ((tc.tm_sec / 10) << 4) | /* Tens of seconds */ |
||
156 | (tc.tm_sec % 10); /* Units of seconds */ |
||
157 | buf[3] = (1 << 7) | /* reserved -- always 1 */ |
||
158 | ((tc.tm_min / 10) << 4) | /* Tens of minutes */ |
||
159 | (tc.tm_min % 10); /* Units of minutes */ |
||
160 | buf[4] = (3 << 6) | /* reserved -- always 1 */ |
||
161 | ((tc.tm_hour / 10) << 4) | /* Tens of hours */ |
||
162 | (tc.tm_hour % 10); /* Units of hours */ |
||
163 | break; |
||
164 | default: |
||
165 | buf[1] = buf[2] = buf[3] = buf[4] = 0xff; |
||
166 | } |
||
167 | return 5; |
||
168 | } |
||
169 | |||
170 | static void dv_inject_audio(DVMuxContext *c, int channel, uint8_t* frame_ptr) |
||
171 | { |
||
172 | int i, j, d, of, size; |
||
173 | size = 4 * dv_audio_frame_size(c->sys, c->frames); |
||
174 | frame_ptr += channel * c->sys->difseg_size * 150 * 80; |
||
175 | for (i = 0; i < c->sys->difseg_size; i++) { |
||
176 | frame_ptr += 6 * 80; /* skip DIF segment header */ |
||
177 | for (j = 0; j < 9; j++) { |
||
178 | dv_write_pack(dv_aaux_packs_dist[i][j], c, &frame_ptr[3], i >= c->sys->difseg_size/2); |
||
179 | for (d = 8; d < 80; d+=2) { |
||
180 | of = c->sys->audio_shuffle[i][j] + (d - 8)/2 * c->sys->audio_stride; |
||
181 | if (of*2 >= size) |
||
182 | continue; |
||
183 | |||
184 | frame_ptr[d] = *av_fifo_peek2(c->audio_data[channel], of*2+1); // FIXME: maybe we have to admit |
||
185 | frame_ptr[d+1] = *av_fifo_peek2(c->audio_data[channel], of*2); // that DV is a big-endian PCM |
||
186 | } |
||
187 | frame_ptr += 16 * 80; /* 15 Video DIFs + 1 Audio DIF */ |
||
188 | } |
||
189 | } |
||
190 | } |
||
191 | |||
192 | static void dv_inject_metadata(DVMuxContext *c, uint8_t* frame) |
||
193 | { |
||
194 | int j, k; |
||
195 | uint8_t* buf; |
||
196 | |||
197 | for (buf = frame; buf < frame + c->sys->frame_size; buf += 150 * 80) { |
||
198 | /* DV subcode: 2nd and 3d DIFs */ |
||
199 | for (j = 80; j < 80 * 3; j += 80) { |
||
200 | for (k = 6; k < 6 * 8; k += 8) |
||
201 | dv_write_pack(dv_timecode, c, &buf[j+k]); |
||
202 | |||
203 | if (((long)(buf-frame)/(c->sys->frame_size/(c->sys->difseg_size*c->sys->n_difchan))%c->sys->difseg_size) > 5) { /* FIXME: is this really needed ? */ |
||
204 | dv_write_pack(dv_video_recdate, c, &buf[j+14]); |
||
205 | dv_write_pack(dv_video_rectime, c, &buf[j+22]); |
||
206 | dv_write_pack(dv_video_recdate, c, &buf[j+38]); |
||
207 | dv_write_pack(dv_video_rectime, c, &buf[j+46]); |
||
208 | } |
||
209 | } |
||
210 | |||
211 | /* DV VAUX: 4th, 5th and 6th 3DIFs */ |
||
212 | for (j = 80*3 + 3; j < 80*6; j += 80) { |
||
213 | dv_write_pack(dv_video_recdate, c, &buf[j+5*2]); |
||
214 | dv_write_pack(dv_video_rectime, c, &buf[j+5*3]); |
||
215 | dv_write_pack(dv_video_recdate, c, &buf[j+5*11]); |
||
216 | dv_write_pack(dv_video_rectime, c, &buf[j+5*12]); |
||
217 | } |
||
218 | } |
||
219 | } |
||
220 | |||
221 | /* |
||
222 | * The following 3 functions constitute our interface to the world |
||
223 | */ |
||
224 | |||
225 | static int dv_assemble_frame(DVMuxContext *c, AVStream* st, |
||
226 | uint8_t* data, int data_size, uint8_t** frame) |
||
227 | { |
||
228 | int i, reqasize; |
||
229 | |||
230 | *frame = &c->frame_buf[0]; |
||
231 | reqasize = 4 * dv_audio_frame_size(c->sys, c->frames); |
||
232 | |||
233 | switch (st->codec->codec_type) { |
||
234 | case AVMEDIA_TYPE_VIDEO: |
||
235 | /* FIXME: we have to have more sensible approach than this one */ |
||
236 | if (c->has_video) |
||
237 | av_log(st->codec, AV_LOG_ERROR, "Can't process DV frame #%d. Insufficient audio data or severe sync problem.\n", c->frames); |
||
238 | |||
239 | memcpy(*frame, data, c->sys->frame_size); |
||
240 | c->has_video = 1; |
||
241 | break; |
||
242 | case AVMEDIA_TYPE_AUDIO: |
||
243 | for (i = 0; i < c->n_ast && st != c->ast[i]; i++); |
||
244 | |||
245 | /* FIXME: we have to have more sensible approach than this one */ |
||
246 | if (av_fifo_size(c->audio_data[i]) + data_size >= 100*MAX_AUDIO_FRAME_SIZE) |
||
247 | av_log(st->codec, AV_LOG_ERROR, "Can't process DV frame #%d. Insufficient video data or severe sync problem.\n", c->frames); |
||
248 | av_fifo_generic_write(c->audio_data[i], data, data_size, NULL); |
||
249 | |||
250 | /* Let us see if we've got enough audio for one DV frame. */ |
||
251 | c->has_audio |= ((reqasize <= av_fifo_size(c->audio_data[i])) << i); |
||
252 | |||
253 | break; |
||
254 | default: |
||
255 | break; |
||
256 | } |
||
257 | |||
258 | /* Let us see if we have enough data to construct one DV frame. */ |
||
259 | if (c->has_video == 1 && c->has_audio + 1 == 1 << c->n_ast) { |
||
260 | dv_inject_metadata(c, *frame); |
||
261 | c->has_audio = 0; |
||
262 | for (i=0; i < c->n_ast; i++) { |
||
263 | dv_inject_audio(c, i, *frame); |
||
264 | av_fifo_drain(c->audio_data[i], reqasize); |
||
265 | c->has_audio |= ((reqasize <= av_fifo_size(c->audio_data[i])) << i); |
||
266 | } |
||
267 | |||
268 | c->has_video = 0; |
||
269 | |||
270 | c->frames++; |
||
271 | |||
272 | return c->sys->frame_size; |
||
273 | } |
||
274 | |||
275 | return 0; |
||
276 | } |
||
277 | |||
278 | static DVMuxContext* dv_init_mux(AVFormatContext* s) |
||
279 | { |
||
280 | DVMuxContext *c = s->priv_data; |
||
281 | AVStream *vst = NULL; |
||
282 | AVDictionaryEntry *t; |
||
283 | int i; |
||
284 | |||
285 | /* we support at most 1 video and 2 audio streams */ |
||
286 | if (s->nb_streams > 3) |
||
287 | return NULL; |
||
288 | |||
289 | c->n_ast = 0; |
||
290 | c->ast[0] = c->ast[1] = NULL; |
||
291 | |||
292 | /* We have to sort out where audio and where video stream is */ |
||
293 | for (i=0; i |
||
294 | switch (s->streams[i]->codec->codec_type) { |
||
295 | case AVMEDIA_TYPE_VIDEO: |
||
296 | if (vst) return NULL; |
||
297 | vst = s->streams[i]; |
||
298 | break; |
||
299 | case AVMEDIA_TYPE_AUDIO: |
||
300 | if (c->n_ast > 1) return NULL; |
||
301 | c->ast[c->n_ast++] = s->streams[i]; |
||
302 | break; |
||
303 | default: |
||
304 | goto bail_out; |
||
305 | } |
||
306 | } |
||
307 | |||
308 | /* Some checks -- DV format is very picky about its incoming streams */ |
||
309 | if (!vst || vst->codec->codec_id != AV_CODEC_ID_DVVIDEO) |
||
310 | goto bail_out; |
||
311 | for (i=0; i |
||
312 | if (c->ast[i] && (c->ast[i]->codec->codec_id != AV_CODEC_ID_PCM_S16LE || |
||
313 | c->ast[i]->codec->sample_rate != 48000 || |
||
314 | c->ast[i]->codec->channels != 2)) |
||
315 | goto bail_out; |
||
316 | } |
||
317 | c->sys = avpriv_dv_codec_profile(vst->codec); |
||
318 | if (!c->sys) |
||
319 | goto bail_out; |
||
320 | |||
321 | if ((c->n_ast > 1) && (c->sys->n_difchan < 2)) { |
||
322 | /* only 1 stereo pair is allowed in 25Mbps mode */ |
||
323 | goto bail_out; |
||
324 | } |
||
325 | |||
326 | /* Ok, everything seems to be in working order */ |
||
327 | c->frames = 0; |
||
328 | c->has_audio = 0; |
||
329 | c->has_video = 0; |
||
330 | if (t = av_dict_get(s->metadata, "creation_time", NULL, 0)) |
||
331 | c->start_time = ff_iso8601_to_unix_time(t->value); |
||
332 | |||
333 | for (i=0; i < c->n_ast; i++) { |
||
334 | if (c->ast[i] && !(c->audio_data[i]=av_fifo_alloc(100*MAX_AUDIO_FRAME_SIZE))) { |
||
335 | while (i > 0) { |
||
336 | i--; |
||
337 | av_fifo_free(c->audio_data[i]); |
||
338 | } |
||
339 | goto bail_out; |
||
340 | } |
||
341 | } |
||
342 | |||
343 | return c; |
||
344 | |||
345 | bail_out: |
||
346 | return NULL; |
||
347 | } |
||
348 | |||
349 | static void dv_delete_mux(DVMuxContext *c) |
||
350 | { |
||
351 | int i; |
||
352 | for (i=0; i < c->n_ast; i++) |
||
353 | av_fifo_free(c->audio_data[i]); |
||
354 | } |
||
355 | |||
356 | static int dv_write_header(AVFormatContext *s) |
||
357 | { |
||
358 | AVRational rate; |
||
359 | DVMuxContext *dvc = s->priv_data; |
||
360 | AVDictionaryEntry *tcr = av_dict_get(s->metadata, "timecode", NULL, 0); |
||
361 | |||
362 | if (!dv_init_mux(s)) { |
||
363 | av_log(s, AV_LOG_ERROR, "Can't initialize DV format!\n" |
||
364 | "Make sure that you supply exactly two streams:\n" |
||
365 | " video: 25fps or 29.97fps, audio: 2ch/48kHz/PCM\n" |
||
366 | " (50Mbps allows an optional second audio stream)\n"); |
||
367 | return -1; |
||
368 | } |
||
369 | rate.num = dvc->sys->ltc_divisor; |
||
370 | rate.den = 1; |
||
371 | if (!tcr) { // no global timecode, look into the streams |
||
372 | int i; |
||
373 | for (i = 0; i < s->nb_streams; i++) { |
||
374 | tcr = av_dict_get(s->streams[i]->metadata, "timecode", NULL, 0); |
||
375 | if (tcr) |
||
376 | break; |
||
377 | } |
||
378 | } |
||
379 | if (tcr && av_timecode_init_from_string(&dvc->tc, rate, tcr->value, s) >= 0) |
||
380 | return 0; |
||
381 | return av_timecode_init(&dvc->tc, rate, 0, 0, s); |
||
382 | } |
||
383 | |||
384 | static int dv_write_packet(struct AVFormatContext *s, AVPacket *pkt) |
||
385 | { |
||
386 | uint8_t* frame; |
||
387 | int fsize; |
||
388 | |||
389 | fsize = dv_assemble_frame(s->priv_data, s->streams[pkt->stream_index], |
||
390 | pkt->data, pkt->size, &frame); |
||
391 | if (fsize > 0) { |
||
392 | avio_write(s->pb, frame, fsize); |
||
393 | } |
||
394 | return 0; |
||
395 | } |
||
396 | |||
397 | /* |
||
398 | * We might end up with some extra A/V data without matching counterpart. |
||
399 | * E.g. video data without enough audio to write the complete frame. |
||
400 | * Currently we simply drop the last frame. I don't know whether this |
||
401 | * is the best strategy of all |
||
402 | */ |
||
403 | static int dv_write_trailer(struct AVFormatContext *s) |
||
404 | { |
||
405 | dv_delete_mux(s->priv_data); |
||
406 | return 0; |
||
407 | } |
||
408 | |||
409 | AVOutputFormat ff_dv_muxer = { |
||
410 | .name = "dv", |
||
411 | .long_name = NULL_IF_CONFIG_SMALL("DV (Digital Video)"), |
||
412 | .extensions = "dv", |
||
413 | .priv_data_size = sizeof(DVMuxContext), |
||
414 | .audio_codec = AV_CODEC_ID_PCM_S16LE, |
||
415 | .video_codec = AV_CODEC_ID_DVVIDEO, |
||
416 | .write_header = dv_write_header, |
||
417 | .write_packet = dv_write_packet, |
||
418 | .write_trailer = dv_write_trailer, |
||
419 | };>>>>><>=>>><>><>=>>>>>>>>>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>><>6;><6;>7><7>15><15>23><23> |