3 * Copyright (c) 2003 Fabrice Bellard
5 * This file is part of FFmpeg.
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.
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.
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
23 #include "avio_internal.h"
27 #include "libavutil/avstring.h"
28 #include "libavcodec/mpegaudio.h"
29 #include "libavcodec/mpegaudiodata.h"
30 #include "libavcodec/mpegaudiodecheader.h"
31 #include "libavutil/intreadwrite.h"
32 #include "libavutil/opt.h"
33 #include "libavcodec/mpegaudio.h"
34 #include "libavcodec/mpegaudiodata.h"
35 #include "libavcodec/mpegaudiodecheader.h"
36 #include "libavformat/avio_internal.h"
37 #include "libavutil/dict.h"
38 #include "libavutil/avassert.h"
40 static int id3v1_set_string(AVFormatContext *s, const char *key,
41 uint8_t *buf, int buf_size)
43 AVDictionaryEntry *tag;
44 if ((tag = av_dict_get(s->metadata, key, NULL, 0)))
45 av_strlcpy(buf, tag->value, buf_size);
49 static int id3v1_create_tag(AVFormatContext *s, uint8_t *buf)
51 AVDictionaryEntry *tag;
54 memset(buf, 0, ID3v1_TAG_SIZE); /* fail safe */
58 /* we knowingly overspecify each tag length by one byte to compensate for the mandatory null byte added by av_strlcpy */
59 count += id3v1_set_string(s, "TIT2", buf + 3, 30 + 1); //title
60 count += id3v1_set_string(s, "TPE1", buf + 33, 30 + 1); //author|artist
61 count += id3v1_set_string(s, "TALB", buf + 63, 30 + 1); //album
62 count += id3v1_set_string(s, "TDRL", buf + 93, 4 + 1); //date
63 count += id3v1_set_string(s, "comment", buf + 97, 30 + 1);
64 if ((tag = av_dict_get(s->metadata, "TRCK", NULL, 0))) { //track
66 buf[126] = atoi(tag->value);
69 buf[127] = 0xFF; /* default to unknown genre */
70 if ((tag = av_dict_get(s->metadata, "TCON", NULL, 0))) { //genre
71 for(i = 0; i <= ID3v1_GENRE_MAX; i++) {
72 if (!av_strcasecmp(tag->value, ff_id3v1_genre_str[i])) {
82 #define XING_NUM_BAGS 400
83 #define XING_TOC_SIZE 100
84 // maximum size of the xing frame: offset/Xing/flags/frames/size/TOC
85 #define XING_MAX_SIZE (32 + 4 + 4 + 4 + 4 + XING_TOC_SIZE)
87 typedef struct MP3Context {
100 uint64_t bag[XING_NUM_BAGS];
102 int has_variable_bitrate;
104 /* index of the audio stream */
105 int audio_stream_idx;
106 /* number of attached pictures we still need to write */
109 /* audio packets are queued here until we get all the attached pictures */
110 AVPacketList *queue, *queue_end;
113 static const uint8_t xing_offtbl[2][2] = {{32, 17}, {17, 9}};
116 * Write an empty XING header and initialize respective data.
118 static int mp3_write_xing(AVFormatContext *s)
120 MP3Context *mp3 = s->priv_data;
121 AVCodecContext *codec = s->streams[mp3->audio_stream_idx]->codec;
123 int best_bitrate_idx = -1;
124 int best_bitrate_error= INT_MAX;
126 int32_t header, mask;
128 int srate_idx, ver = 0, i, channels;
130 const char *vendor = (codec->flags & CODEC_FLAG_BITEXACT) ? "Lavf" : LIBAVFORMAT_IDENT;
132 if (!s->pb->seekable)
135 for (i = 0; i < FF_ARRAY_ELEMS(avpriv_mpa_freq_tab); i++) {
136 const uint16_t base_freq = avpriv_mpa_freq_tab[i];
138 if (codec->sample_rate == base_freq) ver = 0x3; // MPEG 1
139 else if (codec->sample_rate == base_freq / 2) ver = 0x2; // MPEG 2
140 else if (codec->sample_rate == base_freq / 4) ver = 0x0; // MPEG 2.5
146 if (i == FF_ARRAY_ELEMS(avpriv_mpa_freq_tab)) {
147 av_log(s, AV_LOG_WARNING, "Unsupported sample rate, not writing Xing header.\n");
151 switch (codec->channels) {
152 case 1: channels = MPA_MONO; break;
153 case 2: channels = MPA_STEREO; break;
154 default: av_log(s, AV_LOG_WARNING, "Unsupported number of channels, "
155 "not writing Xing header.\n");
159 /* dummy MPEG audio header */
160 header = 0xffU << 24; // sync
161 header |= (0x7 << 5 | ver << 3 | 0x1 << 1 | 0x1) << 16; // sync/audio-version/layer 3/no crc*/
162 header |= (srate_idx << 2) << 8;
163 header |= channels << 6;
165 for (bitrate_idx=1; bitrate_idx<15; bitrate_idx++) {
167 avpriv_mpegaudio_decode_header(&c, header | (bitrate_idx << (4+8)));
168 error= FFABS(c.bit_rate - codec->bit_rate);
169 if(error < best_bitrate_error){
170 best_bitrate_error= error;
171 best_bitrate_idx = bitrate_idx;
174 av_assert0(best_bitrate_idx >= 0);
176 for (bitrate_idx= best_bitrate_idx;; bitrate_idx++) {
177 if (15 == bitrate_idx)
179 mask = bitrate_idx << (4+8);
181 avpriv_mpegaudio_decode_header(&c, header);
182 xing_offset=xing_offtbl[c.lsf == 1][c.nb_channels == 1];
186 + 4 // frames/size/toc flags
189 + XING_TOC_SIZE // toc
193 if (needed <= c.frame_size)
198 avio_wb32(s->pb, header);
200 ffio_fill(s->pb, 0, xing_offset);
201 mp3->xing_offset = avio_tell(s->pb);
202 ffio_wfourcc(s->pb, "Xing");
203 avio_wb32(s->pb, 0x01 | 0x02 | 0x04); // frames / size / TOC
205 mp3->size = c.frame_size;
210 avio_wb32(s->pb, 0); // frames
211 avio_wb32(s->pb, 0); // size
214 for (i = 0; i < XING_TOC_SIZE; ++i)
215 avio_w8(s->pb, (uint8_t)(255 * i / XING_TOC_SIZE));
217 for (i = 0; i < strlen(vendor); ++i)
218 avio_w8(s->pb, vendor[i]);
221 avio_wb24(s->pb, FFMAX(codec->delay - 528 - 1, 0)<<12);
223 ffio_fill(s->pb, 0, c.frame_size - needed);
229 * Add a frame to XING data.
230 * Following lame's "VbrTag.c".
232 static void mp3_xing_add_frame(MP3Context *mp3, AVPacket *pkt)
238 mp3->size += pkt->size;
240 if (mp3->want == mp3->seen) {
241 mp3->bag[mp3->pos] = mp3->size;
243 if (XING_NUM_BAGS == ++mp3->pos) {
244 /* shrink table to half size by throwing away each second bag. */
245 for (i = 1; i < XING_NUM_BAGS; i += 2)
246 mp3->bag[i >> 1] = mp3->bag[i];
248 /* double wanted amount per bag. */
250 /* adjust current position to half of table size. */
251 mp3->pos = XING_NUM_BAGS / 2;
258 static int mp3_write_audio_packet(AVFormatContext *s, AVPacket *pkt)
260 MP3Context *mp3 = s->priv_data;
262 if (pkt->data && pkt->size >= 4) {
266 avpriv_mpegaudio_decode_header(&c, AV_RB32(pkt->data));
268 if (!mp3->initial_bitrate)
269 mp3->initial_bitrate = c.bit_rate;
270 if ((c.bit_rate == 0) || (mp3->initial_bitrate != c.bit_rate))
271 mp3->has_variable_bitrate = 1;
273 #ifdef FILTER_VBR_HEADERS
274 /* filter out XING and INFO headers. */
275 base = 4 + xing_offtbl[c.lsf == 1][c.nb_channels == 1];
277 if (base + 4 <= pkt->size) {
278 uint32_t v = AV_RB32(pkt->data + base);
280 if (MKBETAG('X','i','n','g') == v || MKBETAG('I','n','f','o') == v)
284 /* filter out VBRI headers. */
287 if (base + 4 <= pkt->size && MKBETAG('V','B','R','I') == AV_RB32(pkt->data + base))
291 if (mp3->xing_offset)
292 mp3_xing_add_frame(mp3, pkt);
295 return ff_raw_write_packet(s, pkt);
298 static int mp3_queue_flush(AVFormatContext *s)
300 MP3Context *mp3 = s->priv_data;
302 int ret = 0, write = 1;
304 ff_id3v2_finish(&mp3->id3, s->pb);
307 while ((pktl = mp3->queue)) {
308 if (write && (ret = mp3_write_audio_packet(s, &pktl->pkt)) < 0)
310 av_free_packet(&pktl->pkt);
311 mp3->queue = pktl->next;
314 mp3->queue_end = NULL;
318 static void mp3_update_xing(AVFormatContext *s)
320 MP3Context *mp3 = s->priv_data;
323 /* replace "Xing" identification string with "Info" for CBR files. */
324 if (!mp3->has_variable_bitrate) {
325 avio_seek(s->pb, mp3->xing_offset, SEEK_SET);
326 ffio_wfourcc(s->pb, "Info");
329 avio_seek(s->pb, mp3->xing_offset + 8, SEEK_SET);
330 avio_wb32(s->pb, mp3->frames);
331 avio_wb32(s->pb, mp3->size);
333 avio_w8(s->pb, 0); // first toc entry has to be zero.
335 for (i = 1; i < XING_TOC_SIZE; ++i) {
336 int j = i * mp3->pos / XING_TOC_SIZE;
337 int seek_point = 256LL * mp3->bag[j] / mp3->size;
338 avio_w8(s->pb, FFMIN(seek_point, 255));
341 avio_seek(s->pb, 0, SEEK_END);
344 static int mp3_write_trailer(struct AVFormatContext *s)
346 uint8_t buf[ID3v1_TAG_SIZE];
347 MP3Context *mp3 = s->priv_data;
349 if (mp3->pics_to_write) {
350 av_log(s, AV_LOG_WARNING, "No packets were sent for some of the "
351 "attached pictures.\n");
355 /* write the id3v1 tag */
356 if (mp3->write_id3v1 && id3v1_create_tag(s, buf) > 0) {
357 avio_write(s->pb, buf, ID3v1_TAG_SIZE);
360 if (mp3->xing_offset)
366 static int query_codec(enum AVCodecID id, int std_compliance)
368 const CodecMime *cm= ff_id3v2_mime_tags;
369 while(cm->id != AV_CODEC_ID_NONE) {
371 return MKTAG('A', 'P', 'I', 'C');
378 AVOutputFormat ff_mp2_muxer = {
380 .long_name = NULL_IF_CONFIG_SMALL("MP2 (MPEG audio layer 2)"),
381 .mime_type = "audio/x-mpeg",
382 .extensions = "mp2,m2a",
383 .audio_codec = AV_CODEC_ID_MP2,
384 .video_codec = AV_CODEC_ID_NONE,
385 .write_packet = ff_raw_write_packet,
386 .flags = AVFMT_NOTIMESTAMPS,
392 static const AVOption options[] = {
393 { "id3v2_version", "Select ID3v2 version to write. Currently 3 and 4 are supported.",
394 offsetof(MP3Context, id3v2_version), AV_OPT_TYPE_INT, {.i64 = 4}, 3, 4, AV_OPT_FLAG_ENCODING_PARAM},
395 { "write_id3v1", "Enable ID3v1 writing. ID3v1 tags are written in UTF-8 which may not be supported by most software.",
396 offsetof(MP3Context, write_id3v1), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, AV_OPT_FLAG_ENCODING_PARAM},
400 static const AVClass mp3_muxer_class = {
401 .class_name = "MP3 muxer",
402 .item_name = av_default_item_name,
404 .version = LIBAVUTIL_VERSION_INT,
407 static int mp3_write_packet(AVFormatContext *s, AVPacket *pkt)
409 MP3Context *mp3 = s->priv_data;
411 if (pkt->stream_index == mp3->audio_stream_idx) {
412 if (mp3->pics_to_write) {
413 /* buffer audio packets until we get all the pictures */
414 AVPacketList *pktl = av_mallocz(sizeof(*pktl));
416 return AVERROR(ENOMEM);
419 pkt->destruct = NULL;
422 mp3->queue_end->next = pktl;
425 mp3->queue_end = pktl;
427 return mp3_write_audio_packet(s, pkt);
431 /* warn only once for each stream */
432 if (s->streams[pkt->stream_index]->nb_frames == 1) {
433 av_log(s, AV_LOG_WARNING, "Got more than one picture in stream %d,"
434 " ignoring.\n", pkt->stream_index);
436 if (!mp3->pics_to_write || s->streams[pkt->stream_index]->nb_frames >= 1)
439 if ((ret = ff_id3v2_write_apic(s, &mp3->id3, pkt)) < 0)
441 mp3->pics_to_write--;
443 /* flush the buffered audio packets */
444 if (!mp3->pics_to_write &&
445 (ret = mp3_queue_flush(s)) < 0)
453 * Write an ID3v2 header at beginning of stream
456 static int mp3_write_header(struct AVFormatContext *s)
458 MP3Context *mp3 = s->priv_data;
461 /* check the streams -- we want exactly one audio and arbitrary number of
462 * video (attached pictures) */
463 mp3->audio_stream_idx = -1;
464 for (i = 0; i < s->nb_streams; i++) {
465 AVStream *st = s->streams[i];
466 if (st->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
467 if (mp3->audio_stream_idx >= 0 || st->codec->codec_id != AV_CODEC_ID_MP3) {
468 av_log(s, AV_LOG_ERROR, "Invalid audio stream. Exactly one MP3 "
469 "audio stream is required.\n");
470 return AVERROR(EINVAL);
472 mp3->audio_stream_idx = i;
473 } else if (st->codec->codec_type != AVMEDIA_TYPE_VIDEO) {
474 av_log(s, AV_LOG_ERROR, "Only audio streams and pictures are allowed in MP3.\n");
475 return AVERROR(EINVAL);
478 if (mp3->audio_stream_idx < 0) {
479 av_log(s, AV_LOG_ERROR, "No audio stream present.\n");
480 return AVERROR(EINVAL);
482 mp3->pics_to_write = s->nb_streams - 1;
484 ff_id3v2_start(&mp3->id3, s->pb, mp3->id3v2_version, ID3v2_DEFAULT_MAGIC);
485 ret = ff_id3v2_write_metadata(s, &mp3->id3);
489 if (!mp3->pics_to_write) {
490 ff_id3v2_finish(&mp3->id3, s->pb);
497 AVOutputFormat ff_mp3_muxer = {
499 .long_name = NULL_IF_CONFIG_SMALL("MP3 (MPEG audio layer 3)"),
500 .mime_type = "audio/x-mpeg",
502 .priv_data_size = sizeof(MP3Context),
503 .audio_codec = AV_CODEC_ID_MP3,
504 .video_codec = AV_CODEC_ID_PNG,
505 .write_header = mp3_write_header,
506 .write_packet = mp3_write_packet,
507 .write_trailer = mp3_write_trailer,
508 .query_codec = query_codec,
509 .flags = AVFMT_NOTIMESTAMPS,
510 .priv_class = &mp3_muxer_class,