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.

461 lines
13KB

  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 <strings.h>
  22. #include "avformat.h"
  23. #include "id3v1.h"
  24. #include "id3v2.h"
  25. #include "rawenc.h"
  26. #include "libavutil/avstring.h"
  27. #include "libavutil/intreadwrite.h"
  28. #include "libavutil/opt.h"
  29. #include "libavcodec/mpegaudio.h"
  30. #include "libavcodec/mpegaudiodata.h"
  31. #include "libavcodec/mpegaudiodecheader.h"
  32. #include "libavformat/avio_internal.h"
  33. #include "libavutil/dict.h"
  34. static int id3v1_set_string(AVFormatContext *s, const char *key,
  35. uint8_t *buf, int buf_size)
  36. {
  37. AVDictionaryEntry *tag;
  38. if ((tag = av_dict_get(s->metadata, key, NULL, 0)))
  39. av_strlcpy(buf, tag->value, buf_size);
  40. return !!tag;
  41. }
  42. static int id3v1_create_tag(AVFormatContext *s, uint8_t *buf)
  43. {
  44. AVDictionaryEntry *tag;
  45. int i, count = 0;
  46. memset(buf, 0, ID3v1_TAG_SIZE); /* fail safe */
  47. buf[0] = 'T';
  48. buf[1] = 'A';
  49. buf[2] = 'G';
  50. count += id3v1_set_string(s, "TIT2", buf + 3, 30); //title
  51. count += id3v1_set_string(s, "TPE1", buf + 33, 30); //author|artist
  52. count += id3v1_set_string(s, "TALB", buf + 63, 30); //album
  53. count += id3v1_set_string(s, "TDRL", buf + 93, 4); //date
  54. count += id3v1_set_string(s, "comment", buf + 97, 30);
  55. if ((tag = av_dict_get(s->metadata, "TRCK", NULL, 0))) { //track
  56. buf[125] = 0;
  57. buf[126] = atoi(tag->value);
  58. count++;
  59. }
  60. buf[127] = 0xFF; /* default to unknown genre */
  61. if ((tag = av_dict_get(s->metadata, "TCON", NULL, 0))) { //genre
  62. for(i = 0; i <= ID3v1_GENRE_MAX; i++) {
  63. if (!strcasecmp(tag->value, ff_id3v1_genre_str[i])) {
  64. buf[127] = i;
  65. count++;
  66. break;
  67. }
  68. }
  69. }
  70. return count;
  71. }
  72. /* simple formats */
  73. static void id3v2_put_size(AVFormatContext *s, int size)
  74. {
  75. avio_w8(s->pb, size >> 21 & 0x7f);
  76. avio_w8(s->pb, size >> 14 & 0x7f);
  77. avio_w8(s->pb, size >> 7 & 0x7f);
  78. avio_w8(s->pb, size & 0x7f);
  79. }
  80. static int string_is_ascii(const uint8_t *str)
  81. {
  82. while (*str && *str < 128) str++;
  83. return !*str;
  84. }
  85. /**
  86. * Write a text frame with one (normal frames) or two (TXXX frames) strings
  87. * according to encoding (only UTF-8 or UTF-16+BOM supported).
  88. * @return number of bytes written or a negative error code.
  89. */
  90. static int id3v2_put_ttag(AVFormatContext *s, const char *str1, const char *str2,
  91. uint32_t tag, enum ID3v2Encoding enc)
  92. {
  93. int len;
  94. uint8_t *pb;
  95. int (*put)(AVIOContext*, const char*);
  96. AVIOContext *dyn_buf;
  97. if (avio_open_dyn_buf(&dyn_buf) < 0)
  98. return AVERROR(ENOMEM);
  99. /* check if the strings are ASCII-only and use UTF16 only if
  100. * they're not */
  101. if (enc == ID3v2_ENCODING_UTF16BOM && string_is_ascii(str1) &&
  102. (!str2 || string_is_ascii(str2)))
  103. enc = ID3v2_ENCODING_ISO8859;
  104. avio_w8(dyn_buf, enc);
  105. if (enc == ID3v2_ENCODING_UTF16BOM) {
  106. avio_wl16(dyn_buf, 0xFEFF); /* BOM */
  107. put = avio_put_str16le;
  108. } else
  109. put = avio_put_str;
  110. put(dyn_buf, str1);
  111. if (str2)
  112. put(dyn_buf, str2);
  113. len = avio_close_dyn_buf(dyn_buf, &pb);
  114. avio_wb32(s->pb, tag);
  115. id3v2_put_size(s, len);
  116. avio_wb16(s->pb, 0);
  117. avio_write(s->pb, pb, len);
  118. av_freep(&pb);
  119. return len + ID3v2_HEADER_SIZE;
  120. }
  121. static int mp2_write_trailer(struct AVFormatContext *s)
  122. {
  123. uint8_t buf[ID3v1_TAG_SIZE];
  124. /* write the id3v1 tag */
  125. if (id3v1_create_tag(s, buf) > 0) {
  126. avio_write(s->pb, buf, ID3v1_TAG_SIZE);
  127. avio_flush(s->pb);
  128. }
  129. return 0;
  130. }
  131. #if CONFIG_MP2_MUXER
  132. AVOutputFormat ff_mp2_muxer = {
  133. "mp2",
  134. NULL_IF_CONFIG_SMALL("MPEG audio layer 2"),
  135. "audio/x-mpeg",
  136. "mp2,m2a",
  137. 0,
  138. CODEC_ID_MP2,
  139. CODEC_ID_NONE,
  140. NULL,
  141. ff_raw_write_packet,
  142. mp2_write_trailer,
  143. };
  144. #endif
  145. #if CONFIG_MP3_MUXER
  146. #define VBR_NUM_BAGS 400
  147. #define VBR_TOC_SIZE 100
  148. typedef struct MP3Context {
  149. const AVClass *class;
  150. int id3v2_version;
  151. int64_t frames_offset;
  152. int32_t frames;
  153. int32_t size;
  154. uint32_t want;
  155. uint32_t seen;
  156. uint32_t pos;
  157. uint64_t bag[VBR_NUM_BAGS];
  158. } MP3Context;
  159. static const AVOption options[] = {
  160. { "id3v2_version", "Select ID3v2 version to write. Currently 3 and 4 are supported.",
  161. offsetof(MP3Context, id3v2_version), FF_OPT_TYPE_INT, {.dbl = 4}, 3, 4, AV_OPT_FLAG_ENCODING_PARAM},
  162. { NULL },
  163. };
  164. static const AVClass mp3_muxer_class = {
  165. .class_name = "MP3 muxer",
  166. .item_name = av_default_item_name,
  167. .option = options,
  168. .version = LIBAVUTIL_VERSION_INT,
  169. };
  170. static int id3v2_check_write_tag(AVFormatContext *s, AVDictionaryEntry *t, const char table[][4],
  171. enum ID3v2Encoding enc)
  172. {
  173. uint32_t tag;
  174. int i;
  175. if (t->key[0] != 'T' || strlen(t->key) != 4)
  176. return -1;
  177. tag = AV_RB32(t->key);
  178. for (i = 0; *table[i]; i++)
  179. if (tag == AV_RB32(table[i]))
  180. return id3v2_put_ttag(s, t->value, NULL, tag, enc);
  181. return -1;
  182. }
  183. static const int64_t xing_offtbl[2][2] = {{32, 17}, {17,9}};
  184. /*
  185. * Write an empty XING header and initialize respective data.
  186. */
  187. static int mp3_write_xing(AVFormatContext *s)
  188. {
  189. AVCodecContext *codec = s->streams[0]->codec;
  190. MP3Context *mp3 = s->priv_data;
  191. int bitrate_idx = 3;
  192. int64_t xing_offset;
  193. int32_t mask, header;
  194. MPADecodeHeader c;
  195. int srate_idx, i, channels;
  196. int needed;
  197. for (i = 0; i < FF_ARRAY_ELEMS(ff_mpa_freq_tab); i++)
  198. if (ff_mpa_freq_tab[i] == codec->sample_rate) {
  199. srate_idx = i;
  200. break;
  201. }
  202. if (i == FF_ARRAY_ELEMS(ff_mpa_freq_tab)) {
  203. av_log(s, AV_LOG_ERROR, "Unsupported sample rate.\n");
  204. return -1;
  205. }
  206. switch (codec->channels) {
  207. case 1: channels = MPA_MONO; break;
  208. case 2: channels = MPA_STEREO; break;
  209. default: av_log(s, AV_LOG_ERROR, "Unsupported number of channels.\n"); return -1;
  210. }
  211. /* dummy MPEG audio header */
  212. header = 0xff << 24; // sync
  213. header |= (0x7 << 5 | 0x3 << 3 | 0x1 << 1 | 0x1) << 16; // sync/mpeg-1/layer 3/no crc*/
  214. header |= (srate_idx << 2) << 8;
  215. header |= channels << 6;
  216. for (;;) {
  217. if (15 == bitrate_idx)
  218. return -1;
  219. mask = (bitrate_idx << 4) << 8;
  220. header |= mask;
  221. ff_mpegaudio_decode_header(&c, header);
  222. xing_offset=xing_offtbl[c.lsf == 1][c.nb_channels == 1];
  223. needed = 4 // header
  224. + xing_offset
  225. + 4 // xing tag
  226. + 4 // frames/size/toc flags
  227. + 4 // frames
  228. + 4 // size
  229. + VBR_TOC_SIZE; // toc
  230. if (needed <= c.frame_size)
  231. break;
  232. header &= ~mask;
  233. ++bitrate_idx;
  234. }
  235. avio_wb32(s->pb, header);
  236. ffio_fill(s->pb, 0, xing_offset);
  237. avio_wb32(s->pb, MKBETAG('X', 'i', 'n', 'g'));
  238. avio_wb32(s->pb, 0x01 | 0x02 | 0x04); // frames/size/toc
  239. mp3->frames_offset = avio_tell(s->pb);
  240. mp3->size = c.frame_size;
  241. mp3->want=1;
  242. mp3->seen=0;
  243. mp3->pos=0;
  244. avio_wb32(s->pb, 0); // frames
  245. avio_wb32(s->pb, 0); // size
  246. // toc
  247. for (i = 0; i < VBR_TOC_SIZE; ++i)
  248. avio_w8(s->pb, (uint8_t)(255 * i / VBR_TOC_SIZE));
  249. ffio_fill(s->pb, 0, c.frame_size - needed);
  250. avio_flush(s->pb);
  251. return 0;
  252. }
  253. /*
  254. * Add a frame to XING data.
  255. * Following lame's "VbrTag.c".
  256. */
  257. static void mp3_xing_add_frame(AVFormatContext *s, AVPacket *pkt)
  258. {
  259. MP3Context *mp3 = s->priv_data;
  260. int i;
  261. ++mp3->frames;
  262. mp3->size += pkt->size;
  263. if (mp3->want == ++mp3->seen) {
  264. mp3->bag[mp3->pos] = mp3->size;
  265. if (VBR_NUM_BAGS == ++mp3->pos) {
  266. /* shrink table to half size by throwing away each second bag. */
  267. for (i = 1; i < VBR_NUM_BAGS; i += 2)
  268. mp3->bag[i >> 1] = mp3->bag[i];
  269. /* double wanted amount per bag. */
  270. mp3->want <<= 1;
  271. /* adjust current position to half of table size. */
  272. mp3->pos >>= 1;
  273. }
  274. mp3->seen = 0;
  275. }
  276. }
  277. static void mp3_fix_xing(AVFormatContext *s)
  278. {
  279. MP3Context *mp3 = s->priv_data;
  280. int i;
  281. avio_flush(s->pb);
  282. avio_seek(s->pb, mp3->frames_offset, SEEK_SET);
  283. avio_wb32(s->pb, mp3->frames);
  284. avio_wb32(s->pb, mp3->size);
  285. avio_w8(s->pb, 0); // first toc entry has to be zero.
  286. for (i = 1; i < VBR_TOC_SIZE; ++i) {
  287. int j = i * mp3->pos / VBR_TOC_SIZE;
  288. int seek_point = 256LL * mp3->bag[j] / mp3->size;
  289. avio_w8(s->pb, FFMIN(seek_point, 255));
  290. }
  291. avio_flush(s->pb);
  292. avio_seek(s->pb, 0, SEEK_END);
  293. }
  294. /**
  295. * Write an ID3v2 header at beginning of stream
  296. */
  297. static int mp3_write_header(struct AVFormatContext *s)
  298. {
  299. MP3Context *mp3 = s->priv_data;
  300. AVDictionaryEntry *t = NULL;
  301. int totlen = 0, enc = mp3->id3v2_version == 3 ? ID3v2_ENCODING_UTF16BOM :
  302. ID3v2_ENCODING_UTF8;
  303. int64_t size_pos, cur_pos;
  304. avio_wb32(s->pb, MKBETAG('I', 'D', '3', mp3->id3v2_version));
  305. avio_w8(s->pb, 0);
  306. avio_w8(s->pb, 0); /* flags */
  307. /* reserve space for size */
  308. size_pos = avio_tell(s->pb);
  309. avio_wb32(s->pb, 0);
  310. ff_metadata_conv(&s->metadata, ff_id3v2_34_metadata_conv, NULL);
  311. if (mp3->id3v2_version == 4)
  312. ff_metadata_conv(&s->metadata, ff_id3v2_4_metadata_conv, NULL);
  313. while ((t = av_dict_get(s->metadata, "", t, AV_DICT_IGNORE_SUFFIX))) {
  314. int ret;
  315. if ((ret = id3v2_check_write_tag(s, t, ff_id3v2_tags, enc)) > 0) {
  316. totlen += ret;
  317. continue;
  318. }
  319. if ((ret = id3v2_check_write_tag(s, t, mp3->id3v2_version == 3 ?
  320. ff_id3v2_3_tags : ff_id3v2_4_tags, enc)) > 0) {
  321. totlen += ret;
  322. continue;
  323. }
  324. /* unknown tag, write as TXXX frame */
  325. if ((ret = id3v2_put_ttag(s, t->key, t->value, MKBETAG('T', 'X', 'X', 'X'), enc)) < 0)
  326. return ret;
  327. totlen += ret;
  328. }
  329. cur_pos = avio_tell(s->pb);
  330. avio_seek(s->pb, size_pos, SEEK_SET);
  331. id3v2_put_size(s, totlen);
  332. avio_seek(s->pb, cur_pos, SEEK_SET);
  333. if (s->pb->seekable)
  334. mp3_write_xing(s);
  335. return 0;
  336. }
  337. static int mp3_write_packet(AVFormatContext *s, AVPacket *pkt)
  338. {
  339. if (! pkt || ! pkt->data || pkt->size < 4)
  340. return ff_raw_write_packet(s, pkt);
  341. else {
  342. MP3Context *mp3 = s->priv_data;
  343. #ifdef FILTER_VBR_HEADERS
  344. MPADecodeHeader c;
  345. int base;
  346. ff_mpegaudio_decode_header(&c, AV_RB32(pkt->data));
  347. /* filter out XING and INFO headers. */
  348. base = 4 + xing_offtbl[c.lsf == 1][c.nb_channels == 1];
  349. if (base + 4 <= pkt->size) {
  350. uint32_t v = AV_RB32(pkt->data + base);
  351. if (MKBETAG('X','i','n','g') == v || MKBETAG('I','n','f','o') == v)
  352. return 0;
  353. }
  354. /* filter out VBRI headers. */
  355. base = 4 + 32;
  356. if (base + 4 <= pkt->size && MKBETAG('V','B','R','I') == AV_RB32(pkt->data + base))
  357. return 0;
  358. #endif
  359. if (mp3->frames_offset)
  360. mp3_xing_add_frame(s, pkt);
  361. return ff_raw_write_packet(s, pkt);
  362. }
  363. }
  364. static int mp3_write_trailer(AVFormatContext *s)
  365. {
  366. MP3Context *mp3 = s->priv_data;
  367. int ret=mp2_write_trailer(s);
  368. if (ret < 0)
  369. return ret;
  370. if (mp3->frames_offset)
  371. mp3_fix_xing(s);
  372. return 0;
  373. }
  374. AVOutputFormat ff_mp3_muxer = {
  375. "mp3",
  376. NULL_IF_CONFIG_SMALL("MPEG audio layer 3"),
  377. "audio/x-mpeg",
  378. "mp3",
  379. sizeof(MP3Context),
  380. CODEC_ID_MP3,
  381. CODEC_ID_NONE,
  382. mp3_write_header,
  383. mp3_write_packet,
  384. mp3_write_trailer,
  385. AVFMT_NOTIMESTAMPS,
  386. .priv_class = &mp3_muxer_class,
  387. };
  388. #endif