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.

452 lines
14KB

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