You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

480 lines
16KB

  1. /*
  2. * FLV muxer
  3. * Copyright (c) 2003 The Libav Project
  4. *
  5. * This file is part of Libav.
  6. *
  7. * Libav is free software; you can redistribute it and/or
  8. * modify it under the terms of the GNU Lesser General Public
  9. * License as published by the Free Software Foundation; either
  10. * version 2.1 of the License, or (at your option) any later version.
  11. *
  12. * Libav is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  15. * Lesser General Public License for more details.
  16. *
  17. * You should have received a copy of the GNU Lesser General Public
  18. * License along with Libav; if not, write to the Free Software
  19. * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20. */
  21. #include "libavutil/intfloat_readwrite.h"
  22. #include "avformat.h"
  23. #include "flv.h"
  24. #include "internal.h"
  25. #include "avc.h"
  26. #include "metadata.h"
  27. #include "libavutil/dict.h"
  28. #undef NDEBUG
  29. #include <assert.h>
  30. static const AVCodecTag flv_video_codec_ids[] = {
  31. {CODEC_ID_FLV1, FLV_CODECID_H263 },
  32. {CODEC_ID_FLASHSV, FLV_CODECID_SCREEN},
  33. {CODEC_ID_FLASHSV2, FLV_CODECID_SCREEN2},
  34. {CODEC_ID_VP6F, FLV_CODECID_VP6 },
  35. {CODEC_ID_VP6, FLV_CODECID_VP6 },
  36. {CODEC_ID_H264, FLV_CODECID_H264 },
  37. {CODEC_ID_NONE, 0}
  38. };
  39. static const AVCodecTag flv_audio_codec_ids[] = {
  40. {CODEC_ID_MP3, FLV_CODECID_MP3 >> FLV_AUDIO_CODECID_OFFSET},
  41. {CODEC_ID_PCM_U8, FLV_CODECID_PCM >> FLV_AUDIO_CODECID_OFFSET},
  42. {CODEC_ID_PCM_S16BE, FLV_CODECID_PCM >> FLV_AUDIO_CODECID_OFFSET},
  43. {CODEC_ID_PCM_S16LE, FLV_CODECID_PCM_LE >> FLV_AUDIO_CODECID_OFFSET},
  44. {CODEC_ID_ADPCM_SWF, FLV_CODECID_ADPCM >> FLV_AUDIO_CODECID_OFFSET},
  45. {CODEC_ID_AAC, FLV_CODECID_AAC >> FLV_AUDIO_CODECID_OFFSET},
  46. {CODEC_ID_NELLYMOSER, FLV_CODECID_NELLYMOSER >> FLV_AUDIO_CODECID_OFFSET},
  47. {CODEC_ID_SPEEX, FLV_CODECID_SPEEX >> FLV_AUDIO_CODECID_OFFSET},
  48. {CODEC_ID_NONE, 0}
  49. };
  50. typedef struct FLVContext {
  51. int reserved;
  52. int64_t duration_offset;
  53. int64_t filesize_offset;
  54. int64_t duration;
  55. } FLVContext;
  56. typedef struct FLVStreamContext {
  57. int delay; ///< first dts delay for each stream (needed for AVC & Speex)
  58. int64_t last_ts; ///< last timestamp for each stream
  59. } FLVStreamContext;
  60. static int get_audio_flags(AVCodecContext *enc){
  61. int flags = (enc->bits_per_coded_sample == 16) ? FLV_SAMPLESSIZE_16BIT : FLV_SAMPLESSIZE_8BIT;
  62. if (enc->codec_id == CODEC_ID_AAC) // specs force these parameters
  63. return FLV_CODECID_AAC | FLV_SAMPLERATE_44100HZ | FLV_SAMPLESSIZE_16BIT | FLV_STEREO;
  64. else if (enc->codec_id == CODEC_ID_SPEEX) {
  65. if (enc->sample_rate != 16000) {
  66. av_log(enc, AV_LOG_ERROR, "flv only supports wideband (16kHz) Speex audio\n");
  67. return -1;
  68. }
  69. if (enc->channels != 1) {
  70. av_log(enc, AV_LOG_ERROR, "flv only supports mono Speex audio\n");
  71. return -1;
  72. }
  73. return FLV_CODECID_SPEEX | FLV_SAMPLERATE_11025HZ | FLV_SAMPLESSIZE_16BIT;
  74. } else {
  75. switch (enc->sample_rate) {
  76. case 44100:
  77. flags |= FLV_SAMPLERATE_44100HZ;
  78. break;
  79. case 22050:
  80. flags |= FLV_SAMPLERATE_22050HZ;
  81. break;
  82. case 11025:
  83. flags |= FLV_SAMPLERATE_11025HZ;
  84. break;
  85. case 8000: //nellymoser only
  86. case 5512: //not mp3
  87. if(enc->codec_id != CODEC_ID_MP3){
  88. flags |= FLV_SAMPLERATE_SPECIAL;
  89. break;
  90. }
  91. default:
  92. av_log(enc, AV_LOG_ERROR, "flv does not support that sample rate, choose from (44100, 22050, 11025).\n");
  93. return -1;
  94. }
  95. }
  96. if (enc->channels > 1) {
  97. flags |= FLV_STEREO;
  98. }
  99. switch(enc->codec_id){
  100. case CODEC_ID_MP3:
  101. flags |= FLV_CODECID_MP3 | FLV_SAMPLESSIZE_16BIT;
  102. break;
  103. case CODEC_ID_PCM_U8:
  104. flags |= FLV_CODECID_PCM | FLV_SAMPLESSIZE_8BIT;
  105. break;
  106. case CODEC_ID_PCM_S16BE:
  107. flags |= FLV_CODECID_PCM | FLV_SAMPLESSIZE_16BIT;
  108. break;
  109. case CODEC_ID_PCM_S16LE:
  110. flags |= FLV_CODECID_PCM_LE | FLV_SAMPLESSIZE_16BIT;
  111. break;
  112. case CODEC_ID_ADPCM_SWF:
  113. flags |= FLV_CODECID_ADPCM | FLV_SAMPLESSIZE_16BIT;
  114. break;
  115. case CODEC_ID_NELLYMOSER:
  116. if (enc->sample_rate == 8000) {
  117. flags |= FLV_CODECID_NELLYMOSER_8KHZ_MONO | FLV_SAMPLESSIZE_16BIT;
  118. } else {
  119. flags |= FLV_CODECID_NELLYMOSER | FLV_SAMPLESSIZE_16BIT;
  120. }
  121. break;
  122. case 0:
  123. flags |= enc->codec_tag<<4;
  124. break;
  125. default:
  126. av_log(enc, AV_LOG_ERROR, "codec not compatible with flv\n");
  127. return -1;
  128. }
  129. return flags;
  130. }
  131. static void put_amf_string(AVIOContext *pb, const char *str)
  132. {
  133. size_t len = strlen(str);
  134. avio_wb16(pb, len);
  135. avio_write(pb, str, len);
  136. }
  137. static void put_avc_eos_tag(AVIOContext *pb, unsigned ts) {
  138. avio_w8(pb, FLV_TAG_TYPE_VIDEO);
  139. avio_wb24(pb, 5); /* Tag Data Size */
  140. avio_wb24(pb, ts); /* lower 24 bits of timestamp in ms*/
  141. avio_w8(pb, (ts >> 24) & 0x7F); /* MSB of ts in ms*/
  142. avio_wb24(pb, 0); /* StreamId = 0 */
  143. avio_w8(pb, 23); /* ub[4] FrameType = 1, ub[4] CodecId = 7 */
  144. avio_w8(pb, 2); /* AVC end of sequence */
  145. avio_wb24(pb, 0); /* Always 0 for AVC EOS. */
  146. avio_wb32(pb, 16); /* Size of FLV tag */
  147. }
  148. static void put_amf_double(AVIOContext *pb, double d)
  149. {
  150. avio_w8(pb, AMF_DATA_TYPE_NUMBER);
  151. avio_wb64(pb, av_dbl2int(d));
  152. }
  153. static void put_amf_bool(AVIOContext *pb, int b) {
  154. avio_w8(pb, AMF_DATA_TYPE_BOOL);
  155. avio_w8(pb, !!b);
  156. }
  157. static int flv_write_header(AVFormatContext *s)
  158. {
  159. AVIOContext *pb = s->pb;
  160. FLVContext *flv = s->priv_data;
  161. AVCodecContext *audio_enc = NULL, *video_enc = NULL;
  162. int i, metadata_count = 0;
  163. double framerate = 0.0;
  164. int64_t metadata_size_pos, data_size, metadata_count_pos;
  165. AVDictionaryEntry *tag = NULL;
  166. for(i=0; i<s->nb_streams; i++){
  167. AVCodecContext *enc = s->streams[i]->codec;
  168. FLVStreamContext *sc;
  169. if (enc->codec_type == AVMEDIA_TYPE_VIDEO) {
  170. if (s->streams[i]->r_frame_rate.den && s->streams[i]->r_frame_rate.num) {
  171. framerate = av_q2d(s->streams[i]->r_frame_rate);
  172. } else {
  173. framerate = 1/av_q2d(s->streams[i]->codec->time_base);
  174. }
  175. video_enc = enc;
  176. if(enc->codec_tag == 0) {
  177. av_log(enc, AV_LOG_ERROR, "video codec not compatible with flv\n");
  178. return -1;
  179. }
  180. } else {
  181. audio_enc = enc;
  182. if(get_audio_flags(enc)<0)
  183. return -1;
  184. }
  185. av_set_pts_info(s->streams[i], 32, 1, 1000); /* 32 bit pts in ms */
  186. sc = av_mallocz(sizeof(FLVStreamContext));
  187. if (!sc)
  188. return AVERROR(ENOMEM);
  189. s->streams[i]->priv_data = sc;
  190. sc->last_ts = -1;
  191. }
  192. avio_write(pb, "FLV", 3);
  193. avio_w8(pb,1);
  194. avio_w8(pb, FLV_HEADER_FLAG_HASAUDIO * !!audio_enc
  195. + FLV_HEADER_FLAG_HASVIDEO * !!video_enc);
  196. avio_wb32(pb,9);
  197. avio_wb32(pb,0);
  198. for(i=0; i<s->nb_streams; i++){
  199. if(s->streams[i]->codec->codec_tag == 5){
  200. avio_w8(pb,8); // message type
  201. avio_wb24(pb,0); // include flags
  202. avio_wb24(pb,0); // time stamp
  203. avio_wb32(pb,0); // reserved
  204. avio_wb32(pb,11); // size
  205. flv->reserved=5;
  206. }
  207. }
  208. /* write meta_tag */
  209. avio_w8(pb, 18); // tag type META
  210. metadata_size_pos= avio_tell(pb);
  211. avio_wb24(pb, 0); // size of data part (sum of all parts below)
  212. avio_wb24(pb, 0); // time stamp
  213. avio_wb32(pb, 0); // reserved
  214. /* now data of data_size size */
  215. /* first event name as a string */
  216. avio_w8(pb, AMF_DATA_TYPE_STRING);
  217. put_amf_string(pb, "onMetaData"); // 12 bytes
  218. /* mixed array (hash) with size and string/type/data tuples */
  219. avio_w8(pb, AMF_DATA_TYPE_MIXEDARRAY);
  220. metadata_count_pos = avio_tell(pb);
  221. metadata_count = 5*!!video_enc + 5*!!audio_enc + 2; // +2 for duration and file size
  222. avio_wb32(pb, metadata_count);
  223. put_amf_string(pb, "duration");
  224. flv->duration_offset= avio_tell(pb);
  225. put_amf_double(pb, s->duration / AV_TIME_BASE); // fill in the guessed duration, it'll be corrected later if incorrect
  226. if(video_enc){
  227. put_amf_string(pb, "width");
  228. put_amf_double(pb, video_enc->width);
  229. put_amf_string(pb, "height");
  230. put_amf_double(pb, video_enc->height);
  231. put_amf_string(pb, "videodatarate");
  232. put_amf_double(pb, video_enc->bit_rate / 1024.0);
  233. put_amf_string(pb, "framerate");
  234. put_amf_double(pb, framerate);
  235. put_amf_string(pb, "videocodecid");
  236. put_amf_double(pb, video_enc->codec_tag);
  237. }
  238. if(audio_enc){
  239. put_amf_string(pb, "audiodatarate");
  240. put_amf_double(pb, audio_enc->bit_rate / 1024.0);
  241. put_amf_string(pb, "audiosamplerate");
  242. put_amf_double(pb, audio_enc->sample_rate);
  243. put_amf_string(pb, "audiosamplesize");
  244. put_amf_double(pb, audio_enc->codec_id == CODEC_ID_PCM_U8 ? 8 : 16);
  245. put_amf_string(pb, "stereo");
  246. put_amf_bool(pb, audio_enc->channels == 2);
  247. put_amf_string(pb, "audiocodecid");
  248. put_amf_double(pb, audio_enc->codec_tag);
  249. }
  250. while ((tag = av_dict_get(s->metadata, "", tag, AV_DICT_IGNORE_SUFFIX))) {
  251. put_amf_string(pb, tag->key);
  252. avio_w8(pb, AMF_DATA_TYPE_STRING);
  253. put_amf_string(pb, tag->value);
  254. metadata_count++;
  255. }
  256. put_amf_string(pb, "filesize");
  257. flv->filesize_offset= avio_tell(pb);
  258. put_amf_double(pb, 0); // delayed write
  259. put_amf_string(pb, "");
  260. avio_w8(pb, AMF_END_OF_OBJECT);
  261. /* write total size of tag */
  262. data_size= avio_tell(pb) - metadata_size_pos - 10;
  263. avio_seek(pb, metadata_count_pos, SEEK_SET);
  264. avio_wb32(pb, metadata_count);
  265. avio_seek(pb, metadata_size_pos, SEEK_SET);
  266. avio_wb24(pb, data_size);
  267. avio_skip(pb, data_size + 10 - 3);
  268. avio_wb32(pb, data_size + 11);
  269. for (i = 0; i < s->nb_streams; i++) {
  270. AVCodecContext *enc = s->streams[i]->codec;
  271. if (enc->codec_id == CODEC_ID_AAC || enc->codec_id == CODEC_ID_H264) {
  272. int64_t pos;
  273. avio_w8(pb, enc->codec_type == AVMEDIA_TYPE_VIDEO ?
  274. FLV_TAG_TYPE_VIDEO : FLV_TAG_TYPE_AUDIO);
  275. avio_wb24(pb, 0); // size patched later
  276. avio_wb24(pb, 0); // ts
  277. avio_w8(pb, 0); // ts ext
  278. avio_wb24(pb, 0); // streamid
  279. pos = avio_tell(pb);
  280. if (enc->codec_id == CODEC_ID_AAC) {
  281. avio_w8(pb, get_audio_flags(enc));
  282. avio_w8(pb, 0); // AAC sequence header
  283. avio_write(pb, enc->extradata, enc->extradata_size);
  284. } else {
  285. avio_w8(pb, enc->codec_tag | FLV_FRAME_KEY); // flags
  286. avio_w8(pb, 0); // AVC sequence header
  287. avio_wb24(pb, 0); // composition time
  288. ff_isom_write_avcc(pb, enc->extradata, enc->extradata_size);
  289. }
  290. data_size = avio_tell(pb) - pos;
  291. avio_seek(pb, -data_size - 10, SEEK_CUR);
  292. avio_wb24(pb, data_size);
  293. avio_skip(pb, data_size + 10 - 3);
  294. avio_wb32(pb, data_size + 11); // previous tag size
  295. }
  296. }
  297. return 0;
  298. }
  299. static int flv_write_trailer(AVFormatContext *s)
  300. {
  301. int64_t file_size;
  302. AVIOContext *pb = s->pb;
  303. FLVContext *flv = s->priv_data;
  304. int i;
  305. /* Add EOS tag */
  306. for (i = 0; i < s->nb_streams; i++) {
  307. AVCodecContext *enc = s->streams[i]->codec;
  308. FLVStreamContext *sc = s->streams[i]->priv_data;
  309. if (enc->codec_type == AVMEDIA_TYPE_VIDEO &&
  310. enc->codec_id == CODEC_ID_H264) {
  311. put_avc_eos_tag(pb, sc->last_ts);
  312. }
  313. }
  314. file_size = avio_tell(pb);
  315. /* update informations */
  316. avio_seek(pb, flv->duration_offset, SEEK_SET);
  317. put_amf_double(pb, flv->duration / (double)1000);
  318. avio_seek(pb, flv->filesize_offset, SEEK_SET);
  319. put_amf_double(pb, file_size);
  320. avio_seek(pb, file_size, SEEK_SET);
  321. return 0;
  322. }
  323. static int flv_write_packet(AVFormatContext *s, AVPacket *pkt)
  324. {
  325. AVIOContext *pb = s->pb;
  326. AVCodecContext *enc = s->streams[pkt->stream_index]->codec;
  327. FLVContext *flv = s->priv_data;
  328. FLVStreamContext *sc = s->streams[pkt->stream_index]->priv_data;
  329. unsigned ts;
  330. int size= pkt->size;
  331. uint8_t *data= NULL;
  332. int flags, flags_size;
  333. // av_log(s, AV_LOG_DEBUG, "type:%d pts: %"PRId64" size:%d\n", enc->codec_type, timestamp, size);
  334. if(enc->codec_id == CODEC_ID_VP6 || enc->codec_id == CODEC_ID_VP6F ||
  335. enc->codec_id == CODEC_ID_AAC)
  336. flags_size= 2;
  337. else if(enc->codec_id == CODEC_ID_H264)
  338. flags_size= 5;
  339. else
  340. flags_size= 1;
  341. if (enc->codec_type == AVMEDIA_TYPE_VIDEO) {
  342. avio_w8(pb, FLV_TAG_TYPE_VIDEO);
  343. flags = enc->codec_tag;
  344. if(flags == 0) {
  345. av_log(enc, AV_LOG_ERROR, "video codec %X not compatible with flv\n",enc->codec_id);
  346. return -1;
  347. }
  348. flags |= pkt->flags & AV_PKT_FLAG_KEY ? FLV_FRAME_KEY : FLV_FRAME_INTER;
  349. } else {
  350. assert(enc->codec_type == AVMEDIA_TYPE_AUDIO);
  351. flags = get_audio_flags(enc);
  352. assert(size);
  353. avio_w8(pb, FLV_TAG_TYPE_AUDIO);
  354. }
  355. if (enc->codec_id == CODEC_ID_H264) {
  356. /* check if extradata looks like mp4 formated */
  357. if (enc->extradata_size > 0 && *(uint8_t*)enc->extradata != 1) {
  358. if (ff_avc_parse_nal_units_buf(pkt->data, &data, &size) < 0)
  359. return -1;
  360. }
  361. }
  362. if (!sc->delay && pkt->dts < 0)
  363. sc->delay = -pkt->dts;
  364. ts = pkt->dts + sc->delay; // add delay to force positive dts
  365. /* check Speex packet duration */
  366. if (enc->codec_id == CODEC_ID_SPEEX && ts - sc->last_ts > 160) {
  367. av_log(s, AV_LOG_WARNING, "Warning: Speex stream has more than "
  368. "8 frames per packet. Adobe Flash "
  369. "Player cannot handle this!\n");
  370. }
  371. if (sc->last_ts < ts)
  372. sc->last_ts = ts;
  373. avio_wb24(pb,size + flags_size);
  374. avio_wb24(pb,ts);
  375. avio_w8(pb,(ts >> 24) & 0x7F); // timestamps are 32bits _signed_
  376. avio_wb24(pb,flv->reserved);
  377. avio_w8(pb,flags);
  378. if (enc->codec_id == CODEC_ID_VP6)
  379. avio_w8(pb,0);
  380. if (enc->codec_id == CODEC_ID_VP6F)
  381. avio_w8(pb, enc->extradata_size ? enc->extradata[0] : 0);
  382. else if (enc->codec_id == CODEC_ID_AAC)
  383. avio_w8(pb,1); // AAC raw
  384. else if (enc->codec_id == CODEC_ID_H264) {
  385. avio_w8(pb,1); // AVC NALU
  386. avio_wb24(pb,pkt->pts - pkt->dts);
  387. }
  388. avio_write(pb, data ? data : pkt->data, size);
  389. avio_wb32(pb,size+flags_size+11); // previous tag size
  390. flv->duration = FFMAX(flv->duration, pkt->pts + sc->delay + pkt->duration);
  391. avio_flush(pb);
  392. av_free(data);
  393. return pb->error;
  394. }
  395. AVOutputFormat ff_flv_muxer = {
  396. .name = "flv",
  397. .long_name = NULL_IF_CONFIG_SMALL("FLV format"),
  398. .mime_type = "video/x-flv",
  399. .extensions = "flv",
  400. .priv_data_size = sizeof(FLVContext),
  401. #if CONFIG_LIBMP3LAME
  402. .audio_codec = CODEC_ID_MP3,
  403. #else // CONFIG_LIBMP3LAME
  404. .audio_codec = CODEC_ID_ADPCM_SWF,
  405. #endif // CONFIG_LIBMP3LAME
  406. .video_codec = CODEC_ID_FLV1,
  407. .write_header = flv_write_header,
  408. .write_packet = flv_write_packet,
  409. .write_trailer = flv_write_trailer,
  410. .codec_tag= (const AVCodecTag* const []){flv_video_codec_ids, flv_audio_codec_ids, 0},
  411. .flags= AVFMT_GLOBALHEADER | AVFMT_VARIABLE_FPS,
  412. };