diff options
Diffstat (limited to 'toxav/media.c')
-rw-r--r-- | toxav/media.c | 397 |
1 files changed, 397 insertions, 0 deletions
diff --git a/toxav/media.c b/toxav/media.c new file mode 100644 index 00000000..e18c1803 --- /dev/null +++ b/toxav/media.c | |||
@@ -0,0 +1,397 @@ | |||
1 | /* AV_codec.c | ||
2 | // * | ||
3 | * Audio and video codec intitialisation, encoding/decoding and playback | ||
4 | * | ||
5 | * Copyright (C) 2013 Tox project All Rights Reserved. | ||
6 | * | ||
7 | * This file is part of Tox. | ||
8 | * | ||
9 | * Tox is free software: you can redistribute it and/or modify | ||
10 | * it under the terms of the GNU General Public License as published by | ||
11 | * the Free Software Foundation, either version 3 of the License, or | ||
12 | * (at your option) any later version. | ||
13 | * | ||
14 | * Tox is distributed in the hope that it will be useful, | ||
15 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | ||
16 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | ||
17 | * GNU General Public License for more details. | ||
18 | * | ||
19 | * You should have received a copy of the GNU General Public License | ||
20 | * along with Tox. If not, see <http://www.gnu.org/licenses/>. | ||
21 | * | ||
22 | */ | ||
23 | |||
24 | /*----------------------------------------------------------------------------------*/ | ||
25 | |||
26 | #ifdef HAVE_CONFIG_H | ||
27 | #include "config.h" | ||
28 | #endif /* HAVE_CONFIG_H */ | ||
29 | |||
30 | #include <stdio.h> | ||
31 | #include <math.h> | ||
32 | #include <libavcodec/avcodec.h> | ||
33 | #include <libavformat/avformat.h> | ||
34 | #include <libswscale/swscale.h> | ||
35 | #include <libavdevice/avdevice.h> | ||
36 | #include <libavutil/opt.h> | ||
37 | #include <opus/opus.h> | ||
38 | #include <assert.h> | ||
39 | |||
40 | #include "rtp.h" | ||
41 | #include "media.h" | ||
42 | |||
43 | struct jitter_buffer { | ||
44 | RTPMessage **queue; | ||
45 | uint16_t capacity; | ||
46 | uint16_t size; | ||
47 | uint16_t front; | ||
48 | uint16_t rear; | ||
49 | uint8_t queue_ready; | ||
50 | uint16_t current_id; | ||
51 | uint32_t current_ts; | ||
52 | uint8_t id_set; | ||
53 | }; | ||
54 | |||
55 | |||
56 | struct jitter_buffer *create_queue(int capacity) | ||
57 | { | ||
58 | struct jitter_buffer *q; | ||
59 | q = (struct jitter_buffer *)calloc(sizeof(struct jitter_buffer),1); | ||
60 | q->queue = (RTPMessage **)calloc(sizeof(RTPMessage*), capacity); | ||
61 | int i = 0; | ||
62 | |||
63 | for (i = 0; i < capacity; ++i) { | ||
64 | q->queue[i] = NULL; | ||
65 | } | ||
66 | |||
67 | q->size = 0; | ||
68 | q->capacity = capacity; | ||
69 | q->front = 0; | ||
70 | q->rear = -1; | ||
71 | q->queue_ready = 0; | ||
72 | q->current_id = 0; | ||
73 | q->current_ts = 0; | ||
74 | q->id_set = 0; | ||
75 | return q; | ||
76 | } | ||
77 | |||
78 | /* returns 1 if 'a' has a higher sequence number than 'b' */ | ||
79 | uint8_t sequence_number_older(uint16_t sn_a, uint16_t sn_b, uint32_t ts_a, uint32_t ts_b) | ||
80 | { | ||
81 | /* TODO: There is already this kind of function in toxrtp.c. | ||
82 | * Maybe merge? | ||
83 | */ | ||
84 | return (sn_a > sn_b || ts_a > ts_b); | ||
85 | } | ||
86 | |||
87 | /* success is 0 when there is nothing to dequeue, 1 when there's a good packet, 2 when there's a lost packet */ | ||
88 | RTPMessage *dequeue(struct jitter_buffer *q, int *success) | ||
89 | { | ||
90 | if (q->size == 0 || q->queue_ready == 0) { | ||
91 | q->queue_ready = 0; | ||
92 | *success = 0; | ||
93 | return NULL; | ||
94 | } | ||
95 | |||
96 | int front = q->front; | ||
97 | |||
98 | if (q->id_set == 0) { | ||
99 | q->current_id = q->queue[front]->header->sequnum; | ||
100 | q->current_ts = q->queue[front]->header->timestamp; | ||
101 | q->id_set = 1; | ||
102 | } else { | ||
103 | int next_id = q->queue[front]->header->sequnum; | ||
104 | int next_ts = q->queue[front]->header->timestamp; | ||
105 | |||
106 | /* if this packet is indeed the expected packet */ | ||
107 | if (next_id == (q->current_id + 1) % MAX_SEQU_NUM) { | ||
108 | q->current_id = next_id; | ||
109 | q->current_ts = next_ts; | ||
110 | } else { | ||
111 | if (sequence_number_older(next_id, q->current_id, next_ts, q->current_ts)) { | ||
112 | printf("nextid: %d current: %d\n", next_id, q->current_id); | ||
113 | q->current_id = (q->current_id + 1) % MAX_SEQU_NUM; | ||
114 | *success = 2; /* tell the decoder the packet is lost */ | ||
115 | return NULL; | ||
116 | } else { | ||
117 | /* packet too old */ | ||
118 | printf("packet too old\n"); | ||
119 | *success = 0; | ||
120 | return NULL; | ||
121 | } | ||
122 | } | ||
123 | } | ||
124 | |||
125 | q->size--; | ||
126 | q->front++; | ||
127 | |||
128 | if (q->front == q->capacity) | ||
129 | q->front = 0; | ||
130 | |||
131 | *success = 1; | ||
132 | q->current_id = q->queue[front]->header->sequnum; | ||
133 | q->current_ts = q->queue[front]->header->timestamp; | ||
134 | return q->queue[front]; | ||
135 | } | ||
136 | |||
137 | int empty_queue(struct jitter_buffer *q) | ||
138 | { | ||
139 | while (q->size > 0) { | ||
140 | q->size--; | ||
141 | rtp_free_msg(NULL, q->queue[q->front]); | ||
142 | q->front++; | ||
143 | |||
144 | if (q->front == q->capacity) | ||
145 | q->front = 0; | ||
146 | } | ||
147 | |||
148 | q->id_set = 0; | ||
149 | q->queue_ready = 0; | ||
150 | return 0; | ||
151 | } | ||
152 | |||
153 | int queue(struct jitter_buffer *q, RTPMessage *pk) | ||
154 | { | ||
155 | if (q->size == q->capacity) { | ||
156 | printf("buffer full, emptying buffer...\n"); | ||
157 | empty_queue(q); | ||
158 | return 0; | ||
159 | } | ||
160 | |||
161 | if (q->size > 8) | ||
162 | q->queue_ready = 1; | ||
163 | |||
164 | ++q->size; | ||
165 | ++q->rear; | ||
166 | |||
167 | if (q->rear == q->capacity) | ||
168 | q->rear = 0; | ||
169 | |||
170 | q->queue[q->rear] = pk; | ||
171 | |||
172 | int a; | ||
173 | int b; | ||
174 | int j; | ||
175 | a = q->rear; | ||
176 | |||
177 | for (j = 0; j < q->size - 1; ++j) { | ||
178 | b = a - 1; | ||
179 | |||
180 | if (b < 0) | ||
181 | b += q->capacity; | ||
182 | |||
183 | if (sequence_number_older(q->queue[b]->header->sequnum, q->queue[a]->header->sequnum, | ||
184 | q->queue[b]->header->timestamp, q->queue[a]->header->timestamp)) { | ||
185 | RTPMessage *temp; | ||
186 | temp = q->queue[a]; | ||
187 | q->queue[a] = q->queue[b]; | ||
188 | q->queue[b] = temp; | ||
189 | printf("had to swap\n"); | ||
190 | } else { | ||
191 | break; | ||
192 | } | ||
193 | |||
194 | a -= 1; | ||
195 | |||
196 | if (a < 0) | ||
197 | a += q->capacity; | ||
198 | } | ||
199 | |||
200 | if (pk) | ||
201 | return 1; | ||
202 | |||
203 | return 0; | ||
204 | } | ||
205 | |||
206 | |||
207 | int init_video_decoder(CodecState *cs) | ||
208 | { | ||
209 | cs->video_decoder = avcodec_find_decoder(VIDEO_CODEC); | ||
210 | |||
211 | if (!cs->video_decoder) { | ||
212 | fprintf(stderr, "Init video_decoder failed!\n"); | ||
213 | return -1; | ||
214 | } | ||
215 | |||
216 | cs->video_decoder_ctx = avcodec_alloc_context3(cs->video_decoder); | ||
217 | |||
218 | if (!cs->video_decoder_ctx) { | ||
219 | fprintf(stderr, "Init video_decoder_ctx failed!\n"); | ||
220 | return -1; | ||
221 | } | ||
222 | |||
223 | if (avcodec_open2(cs->video_decoder_ctx, cs->video_decoder, NULL) < 0) { | ||
224 | fprintf(stderr, "Opening video decoder failed!\n"); | ||
225 | return -1; | ||
226 | } | ||
227 | |||
228 | return 0; | ||
229 | } | ||
230 | |||
231 | int init_audio_decoder(CodecState *cs, uint32_t audio_channels) | ||
232 | { | ||
233 | int rc; | ||
234 | cs->audio_decoder = opus_decoder_create(cs->audio_sample_rate, audio_channels, &rc ); | ||
235 | |||
236 | if ( rc != OPUS_OK ){ | ||
237 | fprintf(stderr, "Error while starting audio decoder!\n"); | ||
238 | return -1; | ||
239 | } | ||
240 | |||
241 | return 0; | ||
242 | } | ||
243 | |||
244 | |||
245 | int init_video_encoder(CodecState *cs, const char* webcam, const char* video_driver, uint32_t video_bitrate) | ||
246 | { | ||
247 | cs->video_input_format = av_find_input_format(video_driver); | ||
248 | |||
249 | if (avformat_open_input(&cs->video_format_ctx, webcam, cs->video_input_format, NULL) != 0) { | ||
250 | fprintf(stderr, "Opening video_input_format failed!\n"); | ||
251 | return -1; | ||
252 | } | ||
253 | |||
254 | avformat_find_stream_info(cs->video_format_ctx, NULL); | ||
255 | av_dump_format(cs->video_format_ctx, 0, webcam, 0); | ||
256 | |||
257 | int i; | ||
258 | |||
259 | for (i = 0; i < cs->video_format_ctx->nb_streams; ++i) { | ||
260 | if (cs->video_format_ctx->streams[i]->codec->codec_type == AVMEDIA_TYPE_VIDEO) { | ||
261 | cs->video_stream = i; | ||
262 | break; | ||
263 | } | ||
264 | } | ||
265 | |||
266 | cs->webcam_decoder_ctx = cs->video_format_ctx->streams[cs->video_stream]->codec; | ||
267 | cs->webcam_decoder = avcodec_find_decoder(cs->webcam_decoder_ctx->codec_id); | ||
268 | |||
269 | if (cs->webcam_decoder == NULL) { | ||
270 | fprintf(stderr, "Unsupported codec!\n"); | ||
271 | return -1; | ||
272 | } | ||
273 | |||
274 | if (cs->webcam_decoder_ctx == NULL) { | ||
275 | fprintf(stderr, "Init webcam_decoder_ctx failed!\n"); | ||
276 | return -1; | ||
277 | } | ||
278 | |||
279 | if (avcodec_open2(cs->webcam_decoder_ctx, cs->webcam_decoder, NULL) < 0) { | ||
280 | fprintf(stderr, "Opening webcam decoder failed!\n"); | ||
281 | return -1; | ||
282 | } | ||
283 | |||
284 | cs->video_encoder = avcodec_find_encoder(VIDEO_CODEC); | ||
285 | |||
286 | if (!cs->video_encoder) { | ||
287 | fprintf(stderr, "Init video_encoder failed!\n"); | ||
288 | return -1; | ||
289 | } | ||
290 | |||
291 | cs->video_encoder_ctx = avcodec_alloc_context3(cs->video_encoder); | ||
292 | |||
293 | if (!cs->video_encoder_ctx) { | ||
294 | fprintf(stderr, "Init video_encoder_ctx failed!\n"); | ||
295 | return -1; | ||
296 | } | ||
297 | |||
298 | cs->video_encoder_ctx->bit_rate = video_bitrate; | ||
299 | cs->video_encoder_ctx->rc_min_rate = cs->video_encoder_ctx->rc_max_rate = cs->video_encoder_ctx->bit_rate; | ||
300 | av_opt_set_double(cs->video_encoder_ctx->priv_data, "max-intra-rate", 90, 0); | ||
301 | av_opt_set(cs->video_encoder_ctx->priv_data, "quality", "realtime", 0); | ||
302 | |||
303 | cs->video_encoder_ctx->thread_count = 4; | ||
304 | cs->video_encoder_ctx->rc_buffer_aggressivity = 0.95; | ||
305 | cs->video_encoder_ctx->rc_buffer_size = video_bitrate * 6; | ||
306 | cs->video_encoder_ctx->profile = 3; | ||
307 | cs->video_encoder_ctx->qmax = 54; | ||
308 | cs->video_encoder_ctx->qmin = 4; | ||
309 | AVRational myrational = {1, 25}; | ||
310 | cs->video_encoder_ctx->time_base = myrational; | ||
311 | cs->video_encoder_ctx->gop_size = 99999; | ||
312 | cs->video_encoder_ctx->pix_fmt = PIX_FMT_YUV420P; | ||
313 | cs->video_encoder_ctx->width = cs->webcam_decoder_ctx->width; | ||
314 | cs->video_encoder_ctx->height = cs->webcam_decoder_ctx->height; | ||
315 | |||
316 | if (avcodec_open2(cs->video_encoder_ctx, cs->video_encoder, NULL) < 0) { | ||
317 | fprintf(stderr, "Opening video encoder failed!\n"); | ||
318 | return -1; | ||
319 | } | ||
320 | |||
321 | return 0; | ||
322 | } | ||
323 | |||
324 | int init_audio_encoder(CodecState *cs) | ||
325 | { | ||
326 | int err = OPUS_OK; | ||
327 | cs->audio_encoder = opus_encoder_create(cs->audio_sample_rate, 1, OPUS_APPLICATION_VOIP, &err); | ||
328 | |||
329 | err = opus_encoder_ctl(cs->audio_encoder, OPUS_SET_BITRATE(cs->audio_bitrate)); | ||
330 | err = opus_encoder_ctl(cs->audio_encoder, OPUS_SET_COMPLEXITY(10)); | ||
331 | err = opus_encoder_ctl(cs->audio_encoder, OPUS_SET_SIGNAL(OPUS_SIGNAL_VOICE)); | ||
332 | |||
333 | /* NOTE: What do we do with this? */ | ||
334 | int nfo; | ||
335 | err = opus_encoder_ctl(cs->audio_encoder, OPUS_GET_LOOKAHEAD(&nfo)); | ||
336 | |||
337 | return err == OPUS_OK ? 0 : -1; | ||
338 | } | ||
339 | |||
340 | |||
341 | CodecState* codec_init_session ( uint32_t audio_bitrate, | ||
342 | uint16_t audio_frame_duration, | ||
343 | uint32_t audio_sample_rate, | ||
344 | uint32_t audio_channels, | ||
345 | uint32_t video_bitrate, | ||
346 | const char* webcam, | ||
347 | const char* webcam_driver ) | ||
348 | { | ||
349 | CodecState* _retu = av_calloc(sizeof(CodecState), 1); | ||
350 | assert(_retu); | ||
351 | |||
352 | |||
353 | avdevice_register_all(); | ||
354 | avcodec_register_all(); | ||
355 | av_register_all(); | ||
356 | |||
357 | |||
358 | _retu->audio_bitrate = audio_bitrate; | ||
359 | _retu->audio_sample_rate = audio_sample_rate; | ||
360 | |||
361 | pthread_mutex_init(&_retu->ctrl_mutex, NULL); | ||
362 | |||
363 | |||
364 | /* Encoders */ | ||
365 | if ( 0 == init_video_encoder(_retu, webcam, webcam_driver, video_bitrate) ) | ||
366 | printf("Video encoder initialized!\n"); | ||
367 | |||
368 | if ( 0 == init_audio_encoder(_retu) ) | ||
369 | printf("Audio encoder initialized!\n"); | ||
370 | |||
371 | |||
372 | /* Decoders */ | ||
373 | if ( 0 == init_video_decoder(_retu) ) | ||
374 | printf("Video decoder initialized!\n"); | ||
375 | |||
376 | if ( 0 == init_audio_decoder(_retu, audio_channels) ) | ||
377 | printf("Audio decoder initialized!\n"); | ||
378 | |||
379 | |||
380 | return _retu; | ||
381 | } | ||
382 | |||
383 | void codec_terminate_session ( CodecState* cs ) | ||
384 | { | ||
385 | if ( cs->audio_encoder ) { | ||
386 | opus_encoder_destroy(cs->audio_encoder); | ||
387 | printf("Terminated encoder!\n"); | ||
388 | } | ||
389 | |||
390 | if ( cs->audio_decoder ) { | ||
391 | opus_decoder_destroy(cs->audio_decoder); | ||
392 | printf("Terminated decoder!\n"); | ||
393 | } | ||
394 | |||
395 | /* TODO: Terminate video */ | ||
396 | |||
397 | } | ||