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.

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