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 VBR_NUM_BAGS 400
83 #define VBR_TOC_SIZE 100
85 typedef struct MP3Context {
90 int64_t frames_offset;
96 uint64_t bag[VBR_NUM_BAGS];
98 int has_variable_bitrate;
100 /* index of the audio stream */
101 int audio_stream_idx;
102 /* number of attached pictures we still need to write */
105 /* audio packets are queued here until we get all the attached pictures */
106 AVPacketList *queue, *queue_end;
109 static const int64_t xing_offtbl[2][2] = {{32, 17}, {17,9}};
112 * Write an empty XING header and initialize respective data.
114 static int mp3_write_xing(AVFormatContext *s)
116 MP3Context *mp3 = s->priv_data;
117 AVCodecContext *codec = s->streams[mp3->audio_stream_idx]->codec;
119 int best_bitrate_idx = -1;
120 int best_bitrate_error= INT_MAX;
122 int32_t header, mask;
124 int srate_idx, ver = 0, i, channels;
127 if (!s->pb->seekable)
130 for (i = 0; i < FF_ARRAY_ELEMS(avpriv_mpa_freq_tab); i++) {
131 const uint16_t base_freq = avpriv_mpa_freq_tab[i];
132 if (codec->sample_rate == base_freq) ver = 0x3; // MPEG 1
133 else if (codec->sample_rate == base_freq / 2) ver = 0x2; // MPEG 2
134 else if (codec->sample_rate == base_freq / 4) ver = 0x0; // MPEG 2.5
139 if (i == FF_ARRAY_ELEMS(avpriv_mpa_freq_tab)) {
140 av_log(s, AV_LOG_WARNING, "Unsupported sample rate, not writing Xing header.\n");
144 switch (codec->channels) {
145 case 1: channels = MPA_MONO; break;
146 case 2: channels = MPA_STEREO; break;
147 default: av_log(s, AV_LOG_WARNING, "Unsupported number of channels, not writing Xing header.\n"); return -1;
150 /* dummy MPEG audio header */
151 header = 0xff << 24; // sync
152 header |= (0x7 << 5 | ver << 3 | 0x1 << 1 | 0x1) << 16; // sync/audio-version/layer 3/no crc*/
153 header |= (srate_idx << 2) << 8;
154 header |= channels << 6;
156 for (bitrate_idx=1; bitrate_idx<15; bitrate_idx++) {
158 avpriv_mpegaudio_decode_header(&c, header | (bitrate_idx << (4+8)));
159 error= FFABS(c.bit_rate - codec->bit_rate);
160 if(error < best_bitrate_error){
161 best_bitrate_error= error;
162 best_bitrate_idx = bitrate_idx;
165 av_assert0(best_bitrate_idx >= 0);
167 for (bitrate_idx= best_bitrate_idx;; bitrate_idx++) {
168 if (15 == bitrate_idx)
170 mask = bitrate_idx << (4+8);
172 avpriv_mpegaudio_decode_header(&c, header);
173 xing_offset=xing_offtbl[c.lsf == 1][c.nb_channels == 1];
177 + 4 // frames/size/toc flags
180 + VBR_TOC_SIZE; // toc
182 if (needed <= c.frame_size)
187 avio_wb32(s->pb, header);
188 ffio_fill(s->pb, 0, xing_offset);
189 avio_wb32(s->pb, MKBETAG('X', 'i', 'n', 'g'));
190 avio_wb32(s->pb, 0x01 | 0x02 | 0x04); // frames/size/toc
192 mp3->frames_offset = avio_tell(s->pb);
193 mp3->size = c.frame_size;
198 avio_wb32(s->pb, 0); // frames
199 avio_wb32(s->pb, 0); // size
202 for (i = 0; i < VBR_TOC_SIZE; ++i)
203 avio_w8(s->pb, (uint8_t)(255 * i / VBR_TOC_SIZE));
205 ffio_fill(s->pb, 0, c.frame_size - needed);
212 * Add a frame to XING data.
213 * Following lame's "VbrTag.c".
215 static void mp3_xing_add_frame(AVFormatContext *s, AVPacket *pkt)
217 MP3Context *mp3 = s->priv_data;
221 mp3->size += pkt->size;
223 if (mp3->want == ++mp3->seen) {
224 mp3->bag[mp3->pos] = mp3->size;
226 if (VBR_NUM_BAGS == ++mp3->pos) {
227 /* shrink table to half size by throwing away each second bag. */
228 for (i = 1; i < VBR_NUM_BAGS; i += 2)
229 mp3->bag[i >> 1] = mp3->bag[i];
231 /* double wanted amount per bag. */
233 /* adjust current position to half of table size. */
241 static void mp3_fix_xing(AVFormatContext *s)
243 MP3Context *mp3 = s->priv_data;
248 /* replace "Xing" identification string with "Info" for CBR files. */
249 if (!mp3->has_variable_bitrate) {
250 int64_t tag_offset = mp3->frames_offset
251 - 4 // frames/size/toc flags
253 avio_seek(s->pb, tag_offset, SEEK_SET);
254 avio_wb32(s->pb, MKBETAG('I', 'n', 'f', 'o'));
257 avio_seek(s->pb, mp3->frames_offset, SEEK_SET);
258 avio_wb32(s->pb, mp3->frames);
259 avio_wb32(s->pb, mp3->size);
261 avio_w8(s->pb, 0); // first toc entry has to be zero.
263 for (i = 1; i < VBR_TOC_SIZE; ++i) {
264 int j = i * mp3->pos / VBR_TOC_SIZE;
265 int seek_point = 256LL * mp3->bag[j] / mp3->size;
266 avio_w8(s->pb, FFMIN(seek_point, 255));
270 avio_seek(s->pb, 0, SEEK_END);
273 static int mp3_write_packet_internal(AVFormatContext *s, AVPacket *pkt)
275 if (! pkt || ! pkt->data || pkt->size < 4)
276 return ff_raw_write_packet(s, pkt);
278 MP3Context *mp3 = s->priv_data;
282 avpriv_mpegaudio_decode_header(&c, AV_RB32(pkt->data));
284 if (!mp3->initial_bitrate)
285 mp3->initial_bitrate = c.bit_rate;
286 if (!mp3->has_variable_bitrate) {
287 if ((c.bit_rate == 0) || (mp3->initial_bitrate != c.bit_rate))
288 mp3->has_variable_bitrate = 1;
291 #ifdef FILTER_VBR_HEADERS
292 /* filter out XING and INFO headers. */
293 base = 4 + xing_offtbl[c.lsf == 1][c.nb_channels == 1];
295 if (base + 4 <= pkt->size) {
296 uint32_t v = AV_RB32(pkt->data + base);
298 if (MKBETAG('X','i','n','g') == v || MKBETAG('I','n','f','o') == v)
302 /* filter out VBRI headers. */
305 if (base + 4 <= pkt->size && MKBETAG('V','B','R','I') == AV_RB32(pkt->data + base))
309 if (mp3->frames_offset)
310 mp3_xing_add_frame(s, pkt);
312 return ff_raw_write_packet(s, pkt);
316 static int mp3_queue_flush(AVFormatContext *s)
318 MP3Context *mp3 = s->priv_data;
320 int ret = 0, write = 1;
322 ff_id3v2_finish(&mp3->id3, s->pb);
325 while ((pktl = mp3->queue)) {
326 if (write && (ret = mp3_write_packet_internal(s, &pktl->pkt)) < 0)
328 av_free_packet(&pktl->pkt);
329 mp3->queue = pktl->next;
332 mp3->queue_end = NULL;
336 static int mp2_write_trailer(struct AVFormatContext *s)
338 uint8_t buf[ID3v1_TAG_SIZE];
339 MP3Context *mp3 = s->priv_data;
341 if (mp3 && mp3->pics_to_write) {
342 av_log(s, AV_LOG_WARNING, "No packets were sent for some of the "
343 "attached pictures.\n");
347 /* write the id3v1 tag */
348 if (mp3 && mp3->write_id3v1 && id3v1_create_tag(s, buf) > 0) {
349 avio_write(s->pb, buf, ID3v1_TAG_SIZE);
352 /* write number of frames */
353 if (mp3 && mp3->frames_offset) {
354 avio_seek(s->pb, mp3->frames_offset, SEEK_SET);
355 avio_wb32(s->pb, s->streams[mp3->audio_stream_idx]->nb_frames);
356 avio_seek(s->pb, 0, SEEK_END);
365 AVOutputFormat ff_mp2_muxer = {
367 .long_name = NULL_IF_CONFIG_SMALL("MPEG audio layer 2"),
368 .mime_type = "audio/x-mpeg",
369 .extensions = "mp2,m2a",
370 .audio_codec = CODEC_ID_MP2,
371 .video_codec = CODEC_ID_NONE,
372 .write_packet = ff_raw_write_packet,
373 .write_trailer = mp2_write_trailer,
374 .flags = AVFMT_NOTIMESTAMPS,
380 static const AVOption options[] = {
381 { "id3v2_version", "Select ID3v2 version to write. Currently 3 and 4 are supported.",
382 offsetof(MP3Context, id3v2_version), AV_OPT_TYPE_INT, {.dbl = 4}, 3, 4, AV_OPT_FLAG_ENCODING_PARAM},
383 { "write_id3v1", "Enable ID3v1 writing. ID3v1 tags are written in UTF-8 which may not be supported by most software.",
384 offsetof(MP3Context, write_id3v1), AV_OPT_TYPE_INT, {.dbl = 0}, 0, 1, AV_OPT_FLAG_ENCODING_PARAM},
388 static const AVClass mp3_muxer_class = {
389 .class_name = "MP3 muxer",
390 .item_name = av_default_item_name,
392 .version = LIBAVUTIL_VERSION_INT,
395 static int mp3_write_packet(AVFormatContext *s, AVPacket *pkt)
397 MP3Context *mp3 = s->priv_data;
399 if (pkt->stream_index == mp3->audio_stream_idx) {
400 if (mp3->pics_to_write) {
401 /* buffer audio packets until we get all the pictures */
402 AVPacketList *pktl = av_mallocz(sizeof(*pktl));
404 return AVERROR(ENOMEM);
407 pkt->destruct = NULL;
410 mp3->queue_end->next = pktl;
413 mp3->queue_end = pktl;
415 return mp3_write_packet_internal(s, pkt);
419 /* warn only once for each stream */
420 if (s->streams[pkt->stream_index]->nb_frames == 1) {
421 av_log(s, AV_LOG_WARNING, "Got more than one picture in stream %d,"
422 " ignoring.\n", pkt->stream_index);
424 if (!mp3->pics_to_write || s->streams[pkt->stream_index]->nb_frames >= 1)
427 if ((ret = ff_id3v2_write_apic(s, &mp3->id3, pkt)) < 0)
429 mp3->pics_to_write--;
431 /* flush the buffered audio packets */
432 if (!mp3->pics_to_write &&
433 (ret = mp3_queue_flush(s)) < 0)
441 * Write an ID3v2 header at beginning of stream
444 static int mp3_write_header(struct AVFormatContext *s)
446 MP3Context *mp3 = s->priv_data;
449 /* check the streams -- we want exactly one audio and arbitrary number of
450 * video (attached pictures) */
451 mp3->audio_stream_idx = -1;
452 for (i = 0; i < s->nb_streams; i++) {
453 AVStream *st = s->streams[i];
454 if (st->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
455 if (mp3->audio_stream_idx >= 0 || st->codec->codec_id != CODEC_ID_MP3) {
456 av_log(s, AV_LOG_ERROR, "Invalid audio stream. Exactly one MP3 "
457 "audio stream is required.\n");
458 return AVERROR(EINVAL);
460 mp3->audio_stream_idx = i;
461 } else if (st->codec->codec_type != AVMEDIA_TYPE_VIDEO) {
462 av_log(s, AV_LOG_ERROR, "Only audio streams and pictures are allowed in MP3.\n");
463 return AVERROR(EINVAL);
466 if (mp3->audio_stream_idx < 0) {
467 av_log(s, AV_LOG_ERROR, "No audio stream present.\n");
468 return AVERROR(EINVAL);
470 mp3->pics_to_write = s->nb_streams - 1;
472 ff_id3v2_start(&mp3->id3, s->pb, mp3->id3v2_version, ID3v2_DEFAULT_MAGIC);
473 ret = ff_id3v2_write_metadata(s, &mp3->id3);
477 if (!mp3->pics_to_write) {
478 ff_id3v2_finish(&mp3->id3, s->pb);
485 static int mp3_write_trailer(AVFormatContext *s)
487 MP3Context *mp3 = s->priv_data;
488 int ret=mp2_write_trailer(s);
493 if (mp3->frames_offset)
499 AVOutputFormat ff_mp3_muxer = {
501 .long_name = NULL_IF_CONFIG_SMALL("MPEG audio layer 3"),
502 .mime_type = "audio/x-mpeg",
504 .priv_data_size = sizeof(MP3Context),
505 .audio_codec = CODEC_ID_MP3,
506 .video_codec = CODEC_ID_PNG,
507 .write_header = mp3_write_header,
508 .write_packet = mp3_write_packet,
509 .write_trailer = mp3_write_trailer,
510 .flags = AVFMT_NOTIMESTAMPS,
511 .priv_class = &mp3_muxer_class,