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.

500 lines
16KB

  1. /*
  2. * Ogg muxer
  3. * Copyright (c) 2007 Baptiste Coudurier <baptiste dot coudurier at free dot fr>
  4. *
  5. * This file is part of FFmpeg.
  6. *
  7. * FFmpeg 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. * FFmpeg 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 FFmpeg; if not, write to the Free Software
  19. * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20. */
  21. #include "libavutil/crc.h"
  22. #include "libavutil/random_seed.h"
  23. #include "libavcodec/xiph.h"
  24. #include "libavcodec/bytestream.h"
  25. #include "libavcodec/flac.h"
  26. #include "avformat.h"
  27. #include "internal.h"
  28. #include "vorbiscomment.h"
  29. #define MAX_PAGE_SIZE 65025
  30. typedef struct {
  31. int64_t granule;
  32. int stream_index;
  33. uint8_t flags;
  34. uint8_t segments_count;
  35. uint8_t segments[255];
  36. uint8_t data[MAX_PAGE_SIZE];
  37. uint16_t size;
  38. } OGGPage;
  39. typedef struct {
  40. unsigned page_counter;
  41. uint8_t *header[3];
  42. int header_len[3];
  43. /** for theora granule */
  44. int kfgshift;
  45. int64_t last_kf_pts;
  46. int vrev;
  47. int eos;
  48. unsigned page_count; ///< number of page buffered
  49. OGGPage page; ///< current page
  50. unsigned serial_num; ///< serial number
  51. } OGGStreamContext;
  52. typedef struct OGGPageList {
  53. OGGPage page;
  54. struct OGGPageList *next;
  55. } OGGPageList;
  56. typedef struct {
  57. OGGPageList *page_list;
  58. } OGGContext;
  59. static void ogg_update_checksum(AVFormatContext *s, ByteIOContext *pb, int64_t crc_offset)
  60. {
  61. int64_t pos = url_ftell(pb);
  62. uint32_t checksum = get_checksum(pb);
  63. url_fseek(pb, crc_offset, SEEK_SET);
  64. put_be32(pb, checksum);
  65. url_fseek(pb, pos, SEEK_SET);
  66. }
  67. static int ogg_write_page(AVFormatContext *s, OGGPage *page, int extra_flags)
  68. {
  69. OGGStreamContext *oggstream = s->streams[page->stream_index]->priv_data;
  70. ByteIOContext *pb;
  71. int64_t crc_offset;
  72. int ret, size;
  73. uint8_t *buf;
  74. ret = url_open_dyn_buf(&pb);
  75. if (ret < 0)
  76. return ret;
  77. init_checksum(pb, ff_crc04C11DB7_update, 0);
  78. put_tag(pb, "OggS");
  79. put_byte(pb, 0);
  80. put_byte(pb, page->flags | extra_flags);
  81. put_le64(pb, page->granule);
  82. put_le32(pb, oggstream->serial_num);
  83. put_le32(pb, oggstream->page_counter++);
  84. crc_offset = url_ftell(pb);
  85. put_le32(pb, 0); // crc
  86. put_byte(pb, page->segments_count);
  87. put_buffer(pb, page->segments, page->segments_count);
  88. put_buffer(pb, page->data, page->size);
  89. ogg_update_checksum(s, pb, crc_offset);
  90. put_flush_packet(pb);
  91. size = url_close_dyn_buf(pb, &buf);
  92. if (size < 0)
  93. return size;
  94. put_buffer(s->pb, buf, size);
  95. put_flush_packet(s->pb);
  96. av_free(buf);
  97. oggstream->page_count--;
  98. return 0;
  99. }
  100. static int64_t ogg_granule_to_timestamp(OGGStreamContext *oggstream, OGGPage *page)
  101. {
  102. if (oggstream->kfgshift)
  103. return (page->granule>>oggstream->kfgshift) +
  104. (page->granule & ((1<<oggstream->kfgshift)-1));
  105. else
  106. return page->granule;
  107. }
  108. static int ogg_compare_granule(AVFormatContext *s, OGGPage *next, OGGPage *page)
  109. {
  110. AVStream *st2 = s->streams[next->stream_index];
  111. AVStream *st = s->streams[page->stream_index];
  112. int64_t next_granule, cur_granule;
  113. if (next->granule == -1 || page->granule == -1)
  114. return 0;
  115. next_granule = av_rescale_q(ogg_granule_to_timestamp(st2->priv_data, next),
  116. st2->time_base, AV_TIME_BASE_Q);
  117. cur_granule = av_rescale_q(ogg_granule_to_timestamp(st->priv_data, page),
  118. st ->time_base, AV_TIME_BASE_Q);
  119. return next_granule > cur_granule;
  120. }
  121. static int ogg_reset_cur_page(OGGStreamContext *oggstream)
  122. {
  123. oggstream->page.granule = -1;
  124. oggstream->page.flags = 0;
  125. oggstream->page.segments_count = 0;
  126. oggstream->page.size = 0;
  127. return 0;
  128. }
  129. static int ogg_buffer_page(AVFormatContext *s, OGGStreamContext *oggstream)
  130. {
  131. OGGContext *ogg = s->priv_data;
  132. OGGPageList **p = &ogg->page_list;
  133. OGGPageList *l = av_mallocz(sizeof(*l));
  134. if (!l)
  135. return AVERROR(ENOMEM);
  136. l->page = oggstream->page;
  137. oggstream->page_count++;
  138. ogg_reset_cur_page(oggstream);
  139. while (*p) {
  140. if (ogg_compare_granule(s, &(*p)->page, &l->page))
  141. break;
  142. p = &(*p)->next;
  143. }
  144. l->next = *p;
  145. *p = l;
  146. return 0;
  147. }
  148. static int ogg_buffer_data(AVFormatContext *s, AVStream *st,
  149. uint8_t *data, unsigned size, int64_t granule)
  150. {
  151. OGGStreamContext *oggstream = st->priv_data;
  152. int total_segments = size / 255 + 1;
  153. uint8_t *p = data;
  154. int i, segments, len;
  155. for (i = 0; i < total_segments; ) {
  156. OGGPage *page = &oggstream->page;
  157. segments = FFMIN(total_segments - i, 255 - page->segments_count);
  158. if (i && !page->segments_count)
  159. page->flags |= 1; // continued packet
  160. memset(page->segments+page->segments_count, 255, segments - 1);
  161. page->segments_count += segments - 1;
  162. len = FFMIN(size, segments*255);
  163. page->segments[page->segments_count++] = len - (segments-1)*255;
  164. memcpy(page->data+page->size, p, len);
  165. p += len;
  166. size -= len;
  167. i += segments;
  168. page->size += len;
  169. if (i == total_segments)
  170. page->granule = granule;
  171. if (page->segments_count == 255) {
  172. ogg_buffer_page(s, oggstream);
  173. }
  174. }
  175. return 0;
  176. }
  177. static uint8_t *ogg_write_vorbiscomment(int offset, int bitexact,
  178. int *header_len, AVMetadata *m, int framing_bit)
  179. {
  180. const char *vendor = bitexact ? "ffmpeg" : LIBAVFORMAT_IDENT;
  181. int size;
  182. uint8_t *p, *p0;
  183. unsigned int count;
  184. size = offset + ff_vorbiscomment_length(m, vendor, &count) + framing_bit;
  185. p = av_mallocz(size);
  186. if (!p)
  187. return NULL;
  188. p0 = p;
  189. p += offset;
  190. ff_vorbiscomment_write(&p, m, vendor, count);
  191. if (framing_bit)
  192. bytestream_put_byte(&p, 1);
  193. *header_len = size;
  194. return p0;
  195. }
  196. static int ogg_build_flac_headers(AVCodecContext *avctx,
  197. OGGStreamContext *oggstream, int bitexact,
  198. AVMetadata *m)
  199. {
  200. enum FLACExtradataFormat format;
  201. uint8_t *streaminfo;
  202. uint8_t *p;
  203. if (!ff_flac_is_extradata_valid(avctx, &format, &streaminfo))
  204. return -1;
  205. // first packet: STREAMINFO
  206. oggstream->header_len[0] = 51;
  207. oggstream->header[0] = av_mallocz(51); // per ogg flac specs
  208. p = oggstream->header[0];
  209. if (!p)
  210. return AVERROR(ENOMEM);
  211. bytestream_put_byte(&p, 0x7F);
  212. bytestream_put_buffer(&p, "FLAC", 4);
  213. bytestream_put_byte(&p, 1); // major version
  214. bytestream_put_byte(&p, 0); // minor version
  215. bytestream_put_be16(&p, 1); // headers packets without this one
  216. bytestream_put_buffer(&p, "fLaC", 4);
  217. bytestream_put_byte(&p, 0x00); // streaminfo
  218. bytestream_put_be24(&p, 34);
  219. bytestream_put_buffer(&p, streaminfo, FLAC_STREAMINFO_SIZE);
  220. // second packet: VorbisComment
  221. p = ogg_write_vorbiscomment(4, bitexact, &oggstream->header_len[1], m, 0);
  222. if (!p)
  223. return AVERROR(ENOMEM);
  224. oggstream->header[1] = p;
  225. bytestream_put_byte(&p, 0x84); // last metadata block and vorbis comment
  226. bytestream_put_be24(&p, oggstream->header_len[1] - 4);
  227. return 0;
  228. }
  229. #define SPEEX_HEADER_SIZE 80
  230. static int ogg_build_speex_headers(AVCodecContext *avctx,
  231. OGGStreamContext *oggstream, int bitexact,
  232. AVMetadata *m)
  233. {
  234. uint8_t *p;
  235. if (avctx->extradata_size < SPEEX_HEADER_SIZE)
  236. return -1;
  237. // first packet: Speex header
  238. p = av_mallocz(SPEEX_HEADER_SIZE);
  239. if (!p)
  240. return AVERROR(ENOMEM);
  241. oggstream->header[0] = p;
  242. oggstream->header_len[0] = SPEEX_HEADER_SIZE;
  243. bytestream_put_buffer(&p, avctx->extradata, SPEEX_HEADER_SIZE);
  244. AV_WL32(&oggstream->header[0][68], 0); // set extra_headers to 0
  245. // second packet: VorbisComment
  246. p = ogg_write_vorbiscomment(0, bitexact, &oggstream->header_len[1], m, 0);
  247. if (!p)
  248. return AVERROR(ENOMEM);
  249. oggstream->header[1] = p;
  250. return 0;
  251. }
  252. static int ogg_write_header(AVFormatContext *s)
  253. {
  254. OGGStreamContext *oggstream;
  255. int i, j;
  256. for (i = 0; i < s->nb_streams; i++) {
  257. AVStream *st = s->streams[i];
  258. unsigned serial_num = i;
  259. if (st->codec->codec_type == AVMEDIA_TYPE_AUDIO)
  260. av_set_pts_info(st, 64, 1, st->codec->sample_rate);
  261. else if (st->codec->codec_type == AVMEDIA_TYPE_VIDEO)
  262. av_set_pts_info(st, 64, st->codec->time_base.num, st->codec->time_base.den);
  263. if (st->codec->codec_id != CODEC_ID_VORBIS &&
  264. st->codec->codec_id != CODEC_ID_THEORA &&
  265. st->codec->codec_id != CODEC_ID_SPEEX &&
  266. st->codec->codec_id != CODEC_ID_FLAC) {
  267. av_log(s, AV_LOG_ERROR, "Unsupported codec id in stream %d\n", i);
  268. return -1;
  269. }
  270. if (!st->codec->extradata || !st->codec->extradata_size) {
  271. av_log(s, AV_LOG_ERROR, "No extradata present\n");
  272. return -1;
  273. }
  274. oggstream = av_mallocz(sizeof(*oggstream));
  275. oggstream->page.stream_index = i;
  276. if (!(st->codec->flags & CODEC_FLAG_BITEXACT))
  277. do {
  278. serial_num = av_get_random_seed();
  279. for (j = 0; j < i; j++) {
  280. OGGStreamContext *sc = s->streams[j]->priv_data;
  281. if (serial_num == sc->serial_num)
  282. break;
  283. }
  284. } while (j < i);
  285. oggstream->serial_num = serial_num;
  286. st->priv_data = oggstream;
  287. if (st->codec->codec_id == CODEC_ID_FLAC) {
  288. int err = ogg_build_flac_headers(st->codec, oggstream,
  289. st->codec->flags & CODEC_FLAG_BITEXACT,
  290. s->metadata);
  291. if (err) {
  292. av_log(s, AV_LOG_ERROR, "Error writing FLAC headers\n");
  293. av_freep(&st->priv_data);
  294. return err;
  295. }
  296. } else if (st->codec->codec_id == CODEC_ID_SPEEX) {
  297. int err = ogg_build_speex_headers(st->codec, oggstream,
  298. st->codec->flags & CODEC_FLAG_BITEXACT,
  299. s->metadata);
  300. if (err) {
  301. av_log(s, AV_LOG_ERROR, "Error writing Speex headers\n");
  302. av_freep(&st->priv_data);
  303. return err;
  304. }
  305. } else {
  306. uint8_t *p;
  307. char *cstr = st->codec->codec_id == CODEC_ID_VORBIS ? "vorbis" : "theora";
  308. int header_type = st->codec->codec_id == CODEC_ID_VORBIS ? 3 : 0x81;
  309. int framing_bit = st->codec->codec_id == CODEC_ID_VORBIS ? 1 : 0;
  310. if (ff_split_xiph_headers(st->codec->extradata, st->codec->extradata_size,
  311. st->codec->codec_id == CODEC_ID_VORBIS ? 30 : 42,
  312. oggstream->header, oggstream->header_len) < 0) {
  313. av_log(s, AV_LOG_ERROR, "Extradata corrupted\n");
  314. av_freep(&st->priv_data);
  315. return -1;
  316. }
  317. p = ogg_write_vorbiscomment(7, st->codec->flags & CODEC_FLAG_BITEXACT,
  318. &oggstream->header_len[1], s->metadata,
  319. framing_bit);
  320. if (!p)
  321. return AVERROR(ENOMEM);
  322. oggstream->header[1] = p;
  323. bytestream_put_byte(&p, header_type);
  324. bytestream_put_buffer(&p, cstr, 6);
  325. if (st->codec->codec_id == CODEC_ID_THEORA) {
  326. /** KFGSHIFT is the width of the less significant section of the granule position
  327. The less significant section is the frame count since the last keyframe */
  328. oggstream->kfgshift = ((oggstream->header[0][40]&3)<<3)|(oggstream->header[0][41]>>5);
  329. oggstream->vrev = oggstream->header[0][9];
  330. av_log(s, AV_LOG_DEBUG, "theora kfgshift %d, vrev %d\n",
  331. oggstream->kfgshift, oggstream->vrev);
  332. }
  333. }
  334. }
  335. for (j = 0; j < s->nb_streams; j++) {
  336. OGGStreamContext *oggstream = s->streams[j]->priv_data;
  337. ogg_buffer_data(s, s->streams[j], oggstream->header[0],
  338. oggstream->header_len[0], 0);
  339. oggstream->page.flags |= 2; // bos
  340. ogg_buffer_page(s, oggstream);
  341. }
  342. for (j = 0; j < s->nb_streams; j++) {
  343. AVStream *st = s->streams[j];
  344. OGGStreamContext *oggstream = st->priv_data;
  345. for (i = 1; i < 3; i++) {
  346. if (oggstream && oggstream->header_len[i])
  347. ogg_buffer_data(s, st, oggstream->header[i],
  348. oggstream->header_len[i], 0);
  349. }
  350. ogg_buffer_page(s, oggstream);
  351. }
  352. return 0;
  353. }
  354. static void ogg_write_pages(AVFormatContext *s, int flush)
  355. {
  356. OGGContext *ogg = s->priv_data;
  357. OGGPageList *next, *p;
  358. if (!ogg->page_list)
  359. return;
  360. for (p = ogg->page_list; p; ) {
  361. OGGStreamContext *oggstream =
  362. s->streams[p->page.stream_index]->priv_data;
  363. if (oggstream->page_count < 2 && !flush)
  364. break;
  365. ogg_write_page(s, &p->page,
  366. flush && oggstream->page_count == 1 ? 4 : 0); // eos
  367. next = p->next;
  368. av_freep(&p);
  369. p = next;
  370. }
  371. ogg->page_list = p;
  372. }
  373. static int ogg_write_packet(AVFormatContext *s, AVPacket *pkt)
  374. {
  375. AVStream *st = s->streams[pkt->stream_index];
  376. OGGStreamContext *oggstream = st->priv_data;
  377. int ret;
  378. int64_t granule;
  379. if (st->codec->codec_id == CODEC_ID_THEORA) {
  380. int64_t pts = oggstream->vrev < 1 ? pkt->pts : pkt->pts + pkt->duration;
  381. int pframe_count;
  382. if (pkt->flags & AV_PKT_FLAG_KEY)
  383. oggstream->last_kf_pts = pts;
  384. pframe_count = pts - oggstream->last_kf_pts;
  385. // prevent frame count from overflow if key frame flag is not set
  386. if (pframe_count >= (1<<oggstream->kfgshift)) {
  387. oggstream->last_kf_pts += pframe_count;
  388. pframe_count = 0;
  389. }
  390. granule = (oggstream->last_kf_pts<<oggstream->kfgshift) | pframe_count;
  391. } else
  392. granule = pkt->pts + pkt->duration;
  393. ret = ogg_buffer_data(s, st, pkt->data, pkt->size, granule);
  394. if (ret < 0)
  395. return ret;
  396. ogg_write_pages(s, 0);
  397. return 0;
  398. }
  399. static int ogg_write_trailer(AVFormatContext *s)
  400. {
  401. int i;
  402. /* flush current page */
  403. for (i = 0; i < s->nb_streams; i++)
  404. ogg_buffer_page(s, s->streams[i]->priv_data);
  405. ogg_write_pages(s, 1);
  406. for (i = 0; i < s->nb_streams; i++) {
  407. AVStream *st = s->streams[i];
  408. OGGStreamContext *oggstream = st->priv_data;
  409. if (st->codec->codec_id == CODEC_ID_FLAC ||
  410. st->codec->codec_id == CODEC_ID_SPEEX) {
  411. av_free(oggstream->header[0]);
  412. av_free(oggstream->header[1]);
  413. }
  414. av_freep(&st->priv_data);
  415. }
  416. return 0;
  417. }
  418. AVOutputFormat ogg_muxer = {
  419. "ogg",
  420. NULL_IF_CONFIG_SMALL("Ogg"),
  421. "application/ogg",
  422. "ogg,ogv,spx",
  423. sizeof(OGGContext),
  424. CODEC_ID_FLAC,
  425. CODEC_ID_THEORA,
  426. ogg_write_header,
  427. ogg_write_packet,
  428. ogg_write_trailer,
  429. .metadata_conv = ff_vorbiscomment_metadata_conv,
  430. };