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.

535 lines
16KB

  1. /*
  2. * MP3 muxer
  3. * Copyright (c) 2003 Fabrice Bellard
  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 "avformat.h"
  22. #include "avio_internal.h"
  23. #include "id3v1.h"
  24. #include "id3v2.h"
  25. #include "rawenc.h"
  26. #include "libavutil/avstring.h"
  27. #include "libavcodec/mpegaudio.h"
  28. #include "libavcodec/mpegaudiodata.h"
  29. #include "libavcodec/mpegaudiodecheader.h"
  30. #include "libavutil/intreadwrite.h"
  31. #include "libavutil/opt.h"
  32. #include "libavcodec/mpegaudio.h"
  33. #include "libavcodec/mpegaudiodata.h"
  34. #include "libavcodec/mpegaudiodecheader.h"
  35. #include "libavformat/avio_internal.h"
  36. #include "libavutil/dict.h"
  37. #include "libavutil/avassert.h"
  38. static int id3v1_set_string(AVFormatContext *s, const char *key,
  39. uint8_t *buf, int buf_size)
  40. {
  41. AVDictionaryEntry *tag;
  42. if ((tag = av_dict_get(s->metadata, key, NULL, 0)))
  43. av_strlcpy(buf, tag->value, buf_size);
  44. return !!tag;
  45. }
  46. static int id3v1_create_tag(AVFormatContext *s, uint8_t *buf)
  47. {
  48. AVDictionaryEntry *tag;
  49. int i, count = 0;
  50. memset(buf, 0, ID3v1_TAG_SIZE); /* fail safe */
  51. buf[0] = 'T';
  52. buf[1] = 'A';
  53. buf[2] = 'G';
  54. /* we knowingly overspecify each tag length by one byte to compensate for the mandatory null byte added by av_strlcpy */
  55. count += id3v1_set_string(s, "TIT2", buf + 3, 30 + 1); //title
  56. count += id3v1_set_string(s, "TPE1", buf + 33, 30 + 1); //author|artist
  57. count += id3v1_set_string(s, "TALB", buf + 63, 30 + 1); //album
  58. count += id3v1_set_string(s, "TDRL", buf + 93, 4 + 1); //date
  59. count += id3v1_set_string(s, "comment", buf + 97, 30 + 1);
  60. if ((tag = av_dict_get(s->metadata, "TRCK", NULL, 0))) { //track
  61. buf[125] = 0;
  62. buf[126] = atoi(tag->value);
  63. count++;
  64. }
  65. buf[127] = 0xFF; /* default to unknown genre */
  66. if ((tag = av_dict_get(s->metadata, "TCON", NULL, 0))) { //genre
  67. for(i = 0; i <= ID3v1_GENRE_MAX; i++) {
  68. if (!av_strcasecmp(tag->value, ff_id3v1_genre_str[i])) {
  69. buf[127] = i;
  70. count++;
  71. break;
  72. }
  73. }
  74. }
  75. return count;
  76. }
  77. #define VBR_NUM_BAGS 400
  78. #define VBR_TOC_SIZE 100
  79. typedef struct MP3Context {
  80. const AVClass *class;
  81. ID3v2EncContext id3;
  82. int id3v2_version;
  83. int write_id3v1;
  84. int64_t frames_offset;
  85. int32_t frames;
  86. int32_t size;
  87. uint32_t want;
  88. uint32_t seen;
  89. uint32_t pos;
  90. uint64_t bag[VBR_NUM_BAGS];
  91. int initial_bitrate;
  92. int has_variable_bitrate;
  93. /* index of the audio stream */
  94. int audio_stream_idx;
  95. /* number of attached pictures we still need to write */
  96. int pics_to_write;
  97. /* audio packets are queued here until we get all the attached pictures */
  98. AVPacketList *queue, *queue_end;
  99. } MP3Context;
  100. static const int64_t xing_offtbl[2][2] = {{32, 17}, {17,9}};
  101. /*
  102. * Write an empty XING header and initialize respective data.
  103. */
  104. static int mp3_write_xing(AVFormatContext *s)
  105. {
  106. MP3Context *mp3 = s->priv_data;
  107. AVCodecContext *codec = s->streams[mp3->audio_stream_idx]->codec;
  108. int bitrate_idx;
  109. int best_bitrate_idx = -1;
  110. int best_bitrate_error= INT_MAX;
  111. int64_t xing_offset;
  112. int32_t header, mask;
  113. MPADecodeHeader c;
  114. int srate_idx, ver = 0, i, channels;
  115. int needed;
  116. const char *vendor = (codec->flags & CODEC_FLAG_BITEXACT) ? "Lavf" : LIBAVFORMAT_IDENT;
  117. if (!s->pb->seekable)
  118. return 0;
  119. for (i = 0; i < FF_ARRAY_ELEMS(avpriv_mpa_freq_tab); i++) {
  120. const uint16_t base_freq = avpriv_mpa_freq_tab[i];
  121. if (codec->sample_rate == base_freq) ver = 0x3; // MPEG 1
  122. else if (codec->sample_rate == base_freq / 2) ver = 0x2; // MPEG 2
  123. else if (codec->sample_rate == base_freq / 4) ver = 0x0; // MPEG 2.5
  124. else continue;
  125. srate_idx = i;
  126. break;
  127. }
  128. if (i == FF_ARRAY_ELEMS(avpriv_mpa_freq_tab)) {
  129. av_log(s, AV_LOG_WARNING, "Unsupported sample rate, not writing Xing header.\n");
  130. return -1;
  131. }
  132. switch (codec->channels) {
  133. case 1: channels = MPA_MONO; break;
  134. case 2: channels = MPA_STEREO; break;
  135. default: av_log(s, AV_LOG_WARNING, "Unsupported number of channels, not writing Xing header.\n"); return -1;
  136. }
  137. /* dummy MPEG audio header */
  138. header = 0xff << 24; // sync
  139. header |= (0x7 << 5 | ver << 3 | 0x1 << 1 | 0x1) << 16; // sync/audio-version/layer 3/no crc*/
  140. header |= (srate_idx << 2) << 8;
  141. header |= channels << 6;
  142. for (bitrate_idx=1; bitrate_idx<15; bitrate_idx++) {
  143. int error;
  144. avpriv_mpegaudio_decode_header(&c, header | (bitrate_idx << (4+8)));
  145. error= FFABS(c.bit_rate - codec->bit_rate);
  146. if(error < best_bitrate_error){
  147. best_bitrate_error= error;
  148. best_bitrate_idx = bitrate_idx;
  149. }
  150. }
  151. av_assert0(best_bitrate_idx >= 0);
  152. for (bitrate_idx= best_bitrate_idx;; bitrate_idx++) {
  153. if (15 == bitrate_idx)
  154. return -1;
  155. mask = bitrate_idx << (4+8);
  156. header |= mask;
  157. avpriv_mpegaudio_decode_header(&c, header);
  158. xing_offset=xing_offtbl[c.lsf == 1][c.nb_channels == 1];
  159. needed = 4 // header
  160. + xing_offset
  161. + 4 // xing tag
  162. + 4 // frames/size/toc flags
  163. + 4 // frames
  164. + 4 // size
  165. + VBR_TOC_SIZE // toc
  166. + 24
  167. ;
  168. if (needed <= c.frame_size)
  169. break;
  170. header &= ~mask;
  171. }
  172. avio_wb32(s->pb, header);
  173. ffio_fill(s->pb, 0, xing_offset);
  174. avio_wb32(s->pb, MKBETAG('X', 'i', 'n', 'g'));
  175. avio_wb32(s->pb, 0x01 | 0x02 | 0x04); // frames/size/toc
  176. mp3->frames_offset = avio_tell(s->pb);
  177. mp3->size = c.frame_size;
  178. mp3->want=1;
  179. mp3->seen=0;
  180. mp3->pos=0;
  181. avio_wb32(s->pb, 0); // frames
  182. avio_wb32(s->pb, 0); // size
  183. // toc
  184. for (i = 0; i < VBR_TOC_SIZE; ++i)
  185. avio_w8(s->pb, (uint8_t)(255 * i / VBR_TOC_SIZE));
  186. for (i = 0; i < strlen(vendor); ++i)
  187. avio_w8(s->pb, vendor[i]);
  188. for (; i < 21; ++i)
  189. avio_w8(s->pb, 0);
  190. avio_wb24(s->pb, FFMAX(codec->delay - 528 - 1, 0)<<12);
  191. ffio_fill(s->pb, 0, c.frame_size - needed);
  192. avio_flush(s->pb);
  193. return 0;
  194. }
  195. /*
  196. * Add a frame to XING data.
  197. * Following lame's "VbrTag.c".
  198. */
  199. static void mp3_xing_add_frame(AVFormatContext *s, AVPacket *pkt)
  200. {
  201. MP3Context *mp3 = s->priv_data;
  202. int i;
  203. ++mp3->frames;
  204. mp3->size += pkt->size;
  205. if (mp3->want == ++mp3->seen) {
  206. mp3->bag[mp3->pos] = mp3->size;
  207. if (VBR_NUM_BAGS == ++mp3->pos) {
  208. /* shrink table to half size by throwing away each second bag. */
  209. for (i = 1; i < VBR_NUM_BAGS; i += 2)
  210. mp3->bag[i >> 1] = mp3->bag[i];
  211. /* double wanted amount per bag. */
  212. mp3->want <<= 1;
  213. /* adjust current position to half of table size. */
  214. mp3->pos >>= 1;
  215. }
  216. mp3->seen = 0;
  217. }
  218. }
  219. static void mp3_fix_xing(AVFormatContext *s)
  220. {
  221. MP3Context *mp3 = s->priv_data;
  222. int i;
  223. avio_flush(s->pb);
  224. /* replace "Xing" identification string with "Info" for CBR files. */
  225. if (!mp3->has_variable_bitrate) {
  226. int64_t tag_offset = mp3->frames_offset
  227. - 4 // frames/size/toc flags
  228. - 4; // xing tag
  229. avio_seek(s->pb, tag_offset, SEEK_SET);
  230. avio_wb32(s->pb, MKBETAG('I', 'n', 'f', 'o'));
  231. }
  232. avio_seek(s->pb, mp3->frames_offset, SEEK_SET);
  233. avio_wb32(s->pb, mp3->frames);
  234. avio_wb32(s->pb, mp3->size);
  235. avio_w8(s->pb, 0); // first toc entry has to be zero.
  236. for (i = 1; i < VBR_TOC_SIZE; ++i) {
  237. int j = i * mp3->pos / VBR_TOC_SIZE;
  238. int seek_point = 256LL * mp3->bag[j] / mp3->size;
  239. avio_w8(s->pb, FFMIN(seek_point, 255));
  240. }
  241. avio_flush(s->pb);
  242. avio_seek(s->pb, 0, SEEK_END);
  243. }
  244. static int mp3_write_packet_internal(AVFormatContext *s, AVPacket *pkt)
  245. {
  246. if (! pkt || ! pkt->data || pkt->size < 4)
  247. return ff_raw_write_packet(s, pkt);
  248. else {
  249. MP3Context *mp3 = s->priv_data;
  250. MPADecodeHeader c;
  251. int av_unused base;
  252. avpriv_mpegaudio_decode_header(&c, AV_RB32(pkt->data));
  253. if (!mp3->initial_bitrate)
  254. mp3->initial_bitrate = c.bit_rate;
  255. if (!mp3->has_variable_bitrate) {
  256. if ((c.bit_rate == 0) || (mp3->initial_bitrate != c.bit_rate))
  257. mp3->has_variable_bitrate = 1;
  258. }
  259. #ifdef FILTER_VBR_HEADERS
  260. /* filter out XING and INFO headers. */
  261. base = 4 + xing_offtbl[c.lsf == 1][c.nb_channels == 1];
  262. if (base + 4 <= pkt->size) {
  263. uint32_t v = AV_RB32(pkt->data + base);
  264. if (MKBETAG('X','i','n','g') == v || MKBETAG('I','n','f','o') == v)
  265. return 0;
  266. }
  267. /* filter out VBRI headers. */
  268. base = 4 + 32;
  269. if (base + 4 <= pkt->size && MKBETAG('V','B','R','I') == AV_RB32(pkt->data + base))
  270. return 0;
  271. #endif
  272. if (mp3->frames_offset)
  273. mp3_xing_add_frame(s, pkt);
  274. return ff_raw_write_packet(s, pkt);
  275. }
  276. }
  277. static int mp3_queue_flush(AVFormatContext *s)
  278. {
  279. MP3Context *mp3 = s->priv_data;
  280. AVPacketList *pktl;
  281. int ret = 0, write = 1;
  282. ff_id3v2_finish(&mp3->id3, s->pb);
  283. mp3_write_xing(s);
  284. while ((pktl = mp3->queue)) {
  285. if (write && (ret = mp3_write_packet_internal(s, &pktl->pkt)) < 0)
  286. write = 0;
  287. av_free_packet(&pktl->pkt);
  288. mp3->queue = pktl->next;
  289. av_freep(&pktl);
  290. }
  291. mp3->queue_end = NULL;
  292. return ret;
  293. }
  294. static int mp2_write_trailer(struct AVFormatContext *s)
  295. {
  296. uint8_t buf[ID3v1_TAG_SIZE];
  297. MP3Context *mp3 = s->priv_data;
  298. if (mp3 && mp3->pics_to_write) {
  299. av_log(s, AV_LOG_WARNING, "No packets were sent for some of the "
  300. "attached pictures.\n");
  301. mp3_queue_flush(s);
  302. }
  303. /* write the id3v1 tag */
  304. if (mp3 && mp3->write_id3v1 && id3v1_create_tag(s, buf) > 0) {
  305. avio_write(s->pb, buf, ID3v1_TAG_SIZE);
  306. }
  307. /* write number of frames */
  308. if (mp3 && mp3->frames_offset) {
  309. avio_seek(s->pb, mp3->frames_offset, SEEK_SET);
  310. avio_wb32(s->pb, s->streams[mp3->audio_stream_idx]->nb_frames);
  311. avio_seek(s->pb, 0, SEEK_END);
  312. }
  313. avio_flush(s->pb);
  314. return 0;
  315. }
  316. static int query_codec(enum AVCodecID id, int std_compliance)
  317. {
  318. CodecMime *cm= ff_id3v2_mime_tags;
  319. while(cm->id != AV_CODEC_ID_NONE) {
  320. if(id == cm->id)
  321. return MKTAG('A', 'P', 'I', 'C');
  322. cm++;
  323. }
  324. return -1;
  325. }
  326. #if CONFIG_MP2_MUXER
  327. AVOutputFormat ff_mp2_muxer = {
  328. .name = "mp2",
  329. .long_name = NULL_IF_CONFIG_SMALL("MP2 (MPEG audio layer 2)"),
  330. .mime_type = "audio/x-mpeg",
  331. .extensions = "mp2,m2a",
  332. .audio_codec = AV_CODEC_ID_MP2,
  333. .video_codec = AV_CODEC_ID_NONE,
  334. .write_packet = ff_raw_write_packet,
  335. .write_trailer = mp2_write_trailer,
  336. .flags = AVFMT_NOTIMESTAMPS,
  337. };
  338. #endif
  339. #if CONFIG_MP3_MUXER
  340. static const AVOption options[] = {
  341. { "id3v2_version", "Select ID3v2 version to write. Currently 3 and 4 are supported.",
  342. offsetof(MP3Context, id3v2_version), AV_OPT_TYPE_INT, {.dbl = 4}, 3, 4, AV_OPT_FLAG_ENCODING_PARAM},
  343. { "write_id3v1", "Enable ID3v1 writing. ID3v1 tags are written in UTF-8 which may not be supported by most software.",
  344. offsetof(MP3Context, write_id3v1), AV_OPT_TYPE_INT, {.dbl = 0}, 0, 1, AV_OPT_FLAG_ENCODING_PARAM},
  345. { NULL },
  346. };
  347. static const AVClass mp3_muxer_class = {
  348. .class_name = "MP3 muxer",
  349. .item_name = av_default_item_name,
  350. .option = options,
  351. .version = LIBAVUTIL_VERSION_INT,
  352. };
  353. static int mp3_write_packet(AVFormatContext *s, AVPacket *pkt)
  354. {
  355. MP3Context *mp3 = s->priv_data;
  356. if (pkt->stream_index == mp3->audio_stream_idx) {
  357. if (mp3->pics_to_write) {
  358. /* buffer audio packets until we get all the pictures */
  359. AVPacketList *pktl = av_mallocz(sizeof(*pktl));
  360. if (!pktl)
  361. return AVERROR(ENOMEM);
  362. pktl->pkt = *pkt;
  363. pkt->destruct = NULL;
  364. if (mp3->queue_end)
  365. mp3->queue_end->next = pktl;
  366. else
  367. mp3->queue = pktl;
  368. mp3->queue_end = pktl;
  369. } else
  370. return mp3_write_packet_internal(s, pkt);
  371. } else {
  372. int ret;
  373. /* warn only once for each stream */
  374. if (s->streams[pkt->stream_index]->nb_frames == 1) {
  375. av_log(s, AV_LOG_WARNING, "Got more than one picture in stream %d,"
  376. " ignoring.\n", pkt->stream_index);
  377. }
  378. if (!mp3->pics_to_write || s->streams[pkt->stream_index]->nb_frames >= 1)
  379. return 0;
  380. if ((ret = ff_id3v2_write_apic(s, &mp3->id3, pkt)) < 0)
  381. return ret;
  382. mp3->pics_to_write--;
  383. /* flush the buffered audio packets */
  384. if (!mp3->pics_to_write &&
  385. (ret = mp3_queue_flush(s)) < 0)
  386. return ret;
  387. }
  388. return 0;
  389. }
  390. /**
  391. * Write an ID3v2 header at beginning of stream
  392. */
  393. static int mp3_write_header(struct AVFormatContext *s)
  394. {
  395. MP3Context *mp3 = s->priv_data;
  396. int ret, i;
  397. /* check the streams -- we want exactly one audio and arbitrary number of
  398. * video (attached pictures) */
  399. mp3->audio_stream_idx = -1;
  400. for (i = 0; i < s->nb_streams; i++) {
  401. AVStream *st = s->streams[i];
  402. if (st->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
  403. if (mp3->audio_stream_idx >= 0 || st->codec->codec_id != AV_CODEC_ID_MP3) {
  404. av_log(s, AV_LOG_ERROR, "Invalid audio stream. Exactly one MP3 "
  405. "audio stream is required.\n");
  406. return AVERROR(EINVAL);
  407. }
  408. mp3->audio_stream_idx = i;
  409. } else if (st->codec->codec_type != AVMEDIA_TYPE_VIDEO) {
  410. av_log(s, AV_LOG_ERROR, "Only audio streams and pictures are allowed in MP3.\n");
  411. return AVERROR(EINVAL);
  412. }
  413. }
  414. if (mp3->audio_stream_idx < 0) {
  415. av_log(s, AV_LOG_ERROR, "No audio stream present.\n");
  416. return AVERROR(EINVAL);
  417. }
  418. mp3->pics_to_write = s->nb_streams - 1;
  419. ff_id3v2_start(&mp3->id3, s->pb, mp3->id3v2_version, ID3v2_DEFAULT_MAGIC);
  420. ret = ff_id3v2_write_metadata(s, &mp3->id3);
  421. if (ret < 0)
  422. return ret;
  423. if (!mp3->pics_to_write) {
  424. ff_id3v2_finish(&mp3->id3, s->pb);
  425. mp3_write_xing(s);
  426. }
  427. return 0;
  428. }
  429. static int mp3_write_trailer(AVFormatContext *s)
  430. {
  431. MP3Context *mp3 = s->priv_data;
  432. int ret=mp2_write_trailer(s);
  433. if (ret < 0)
  434. return ret;
  435. if (mp3->frames_offset)
  436. mp3_fix_xing(s);
  437. return 0;
  438. }
  439. AVOutputFormat ff_mp3_muxer = {
  440. .name = "mp3",
  441. .long_name = NULL_IF_CONFIG_SMALL("MP3 (MPEG audio layer 3)"),
  442. .mime_type = "audio/x-mpeg",
  443. .extensions = "mp3",
  444. .priv_data_size = sizeof(MP3Context),
  445. .audio_codec = AV_CODEC_ID_MP3,
  446. .video_codec = AV_CODEC_ID_PNG,
  447. .write_header = mp3_write_header,
  448. .write_packet = mp3_write_packet,
  449. .write_trailer = mp3_write_trailer,
  450. .query_codec = query_codec,
  451. .flags = AVFMT_NOTIMESTAMPS,
  452. .priv_class = &mp3_muxer_class,
  453. };
  454. #endif