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.

467 lines
15KB

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