3 * Copyright (c) 2003 Fabrice Bellard
5 * This file is part of Libav.
7 * Libav 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 * Libav 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 Libav; 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 "libavutil/dict.h"
34 #include "libavutil/avassert.h"
36 static int id3v1_set_string(AVFormatContext *s, const char *key,
37 uint8_t *buf, int buf_size)
39 AVDictionaryEntry *tag;
40 if ((tag = av_dict_get(s->metadata, key, NULL, 0)))
41 av_strlcpy(buf, tag->value, buf_size);
45 static int id3v1_create_tag(AVFormatContext *s, uint8_t *buf)
47 AVDictionaryEntry *tag;
50 memset(buf, 0, ID3v1_TAG_SIZE); /* fail safe */
54 count += id3v1_set_string(s, "TIT2", buf + 3, 30); //title
55 count += id3v1_set_string(s, "TPE1", buf + 33, 30); //author|artist
56 count += id3v1_set_string(s, "TALB", buf + 63, 30); //album
57 count += id3v1_set_string(s, "TDRL", buf + 93, 4); //date
58 count += id3v1_set_string(s, "comment", buf + 97, 30);
59 if ((tag = av_dict_get(s->metadata, "TRCK", NULL, 0))) { //track
61 buf[126] = atoi(tag->value);
64 buf[127] = 0xFF; /* default to unknown genre */
65 if ((tag = av_dict_get(s->metadata, "TCON", NULL, 0))) { //genre
66 for(i = 0; i <= ID3v1_GENRE_MAX; i++) {
67 if (!av_strcasecmp(tag->value, ff_id3v1_genre_str[i])) {
77 #define XING_NUM_BAGS 400
78 #define XING_TOC_SIZE 100
79 // maximum size of the xing frame: offset/Xing/flags/frames/size/TOC
80 #define XING_MAX_SIZE (32 + 4 + 4 + 4 + 4 + XING_TOC_SIZE)
82 typedef struct MP3Context {
96 uint64_t bag[XING_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 uint8_t xing_offtbl[2][2] = {{32, 17}, {17, 9}};
112 * Write an empty XING header and initialize respective data.
114 static void mp3_write_xing(AVFormatContext *s)
116 MP3Context *mp3 = s->priv_data;
117 AVCodecContext *codec = s->streams[mp3->audio_stream_idx]->codec;
119 MPADecodeHeader mpah;
120 int srate_idx, i, channels;
122 int best_bitrate_idx;
123 int best_bitrate_error = INT_MAX;
126 int lsf, bytes_needed;
128 if (!s->pb->seekable || !mp3->write_xing)
131 for (i = 0; i < FF_ARRAY_ELEMS(avpriv_mpa_freq_tab); i++) {
132 const uint16_t base_freq = avpriv_mpa_freq_tab[i];
134 if (codec->sample_rate == base_freq) ver = 0x3; // MPEG 1
135 else if (codec->sample_rate == base_freq / 2) ver = 0x2; // MPEG 2
136 else if (codec->sample_rate == base_freq / 4) ver = 0x0; // MPEG 2.5
142 if (i == FF_ARRAY_ELEMS(avpriv_mpa_freq_tab)) {
143 av_log(s, AV_LOG_WARNING, "Unsupported sample rate, not writing Xing "
148 switch (codec->channels) {
149 case 1: channels = MPA_MONO; break;
150 case 2: channels = MPA_STEREO; break;
151 default: av_log(s, AV_LOG_WARNING, "Unsupported number of channels, "
152 "not writing Xing header.\n");
156 /* dummy MPEG audio header */
157 header = 0xff << 24; // sync
158 header |= (0x7 << 5 | ver << 3 | 0x1 << 1 | 0x1) << 16; // sync/audio-version/layer 3/no crc*/
159 header |= (srate_idx << 2) << 8;
160 header |= channels << 6;
162 lsf = !((header & (1 << 20) && header & (1 << 19)));
164 xing_offset = xing_offtbl[ver != 3][channels == 1];
165 bytes_needed = 4 // header
168 + 4 // frames/size/toc flags
171 + XING_TOC_SIZE; // toc
173 for (bitrate_idx = 1; bitrate_idx < 15; bitrate_idx++) {
174 int bit_rate = 1000 * avpriv_mpa_bitrate_tab[lsf][3 - 1][bitrate_idx];
175 int error = FFABS(bit_rate - codec->bit_rate);
177 if (error < best_bitrate_error){
178 best_bitrate_error = error;
179 best_bitrate_idx = bitrate_idx;
183 for (bitrate_idx = best_bitrate_idx; bitrate_idx < 15; bitrate_idx++) {
184 int32_t mask = bitrate_idx << (4 + 8);
187 avpriv_mpegaudio_decode_header(&mpah, header);
189 if (bytes_needed <= mpah.frame_size)
195 avio_wb32(s->pb, header);
197 avpriv_mpegaudio_decode_header(&mpah, header);
199 av_assert0(mpah.frame_size >= XING_MAX_SIZE);
201 ffio_fill(s->pb, 0, xing_offset);
202 mp3->xing_offset = avio_tell(s->pb);
203 ffio_wfourcc(s->pb, "Xing");
204 avio_wb32(s->pb, 0x01 | 0x02 | 0x04); // frames / size / TOC
206 mp3->size = mpah.frame_size;
209 avio_wb32(s->pb, 0); // frames
210 avio_wb32(s->pb, 0); // size
213 for (i = 0; i < XING_TOC_SIZE; i++)
214 avio_w8(s->pb, 255 * i / XING_TOC_SIZE);
216 ffio_fill(s->pb, 0, mpah.frame_size - bytes_needed);
220 * Add a frame to XING data.
221 * Following lame's "VbrTag.c".
223 static void mp3_xing_add_frame(MP3Context *mp3, AVPacket *pkt)
229 mp3->size += pkt->size;
231 if (mp3->want == mp3->seen) {
232 mp3->bag[mp3->pos] = mp3->size;
234 if (XING_NUM_BAGS == ++mp3->pos) {
235 /* shrink table to half size by throwing away each second bag. */
236 for (i = 1; i < XING_NUM_BAGS; i += 2)
237 mp3->bag[i / 2] = mp3->bag[i];
239 /* double wanted amount per bag. */
241 /* adjust current position to half of table size. */
242 mp3->pos = XING_NUM_BAGS / 2;
249 static int mp3_write_audio_packet(AVFormatContext *s, AVPacket *pkt)
251 MP3Context *mp3 = s->priv_data;
253 if (mp3->xing_offset && pkt->size >= 4) {
257 h = AV_RB32(pkt->data);
258 if (ff_mpa_check_header(h) == 0) {
259 avpriv_mpegaudio_decode_header(&c, h);
260 if (!mp3->initial_bitrate)
261 mp3->initial_bitrate = c.bit_rate;
262 if ((c.bit_rate == 0) || (mp3->initial_bitrate != c.bit_rate))
263 mp3->has_variable_bitrate = 1;
266 mp3_xing_add_frame(mp3, pkt);
269 return ff_raw_write_packet(s, pkt);
272 static int mp3_queue_flush(AVFormatContext *s)
274 MP3Context *mp3 = s->priv_data;
276 int ret = 0, write = 1;
278 ff_id3v2_finish(&mp3->id3, s->pb);
281 while ((pktl = mp3->queue)) {
282 if (write && (ret = mp3_write_audio_packet(s, &pktl->pkt)) < 0)
284 av_free_packet(&pktl->pkt);
285 mp3->queue = pktl->next;
288 mp3->queue_end = NULL;
292 static void mp3_update_xing(AVFormatContext *s)
294 MP3Context *mp3 = s->priv_data;
297 /* replace "Xing" identification string with "Info" for CBR files. */
298 if (!mp3->has_variable_bitrate) {
299 avio_seek(s->pb, mp3->xing_offset, SEEK_SET);
300 ffio_wfourcc(s->pb, "Info");
303 avio_seek(s->pb, mp3->xing_offset + 8, SEEK_SET);
304 avio_wb32(s->pb, mp3->frames);
305 avio_wb32(s->pb, mp3->size);
307 avio_w8(s->pb, 0); // first toc entry has to be zero.
309 for (i = 1; i < XING_TOC_SIZE; ++i) {
310 int j = i * mp3->pos / XING_TOC_SIZE;
311 int seek_point = 256LL * mp3->bag[j] / mp3->size;
312 avio_w8(s->pb, FFMIN(seek_point, 255));
315 avio_seek(s->pb, 0, SEEK_END);
318 static int mp3_write_trailer(struct AVFormatContext *s)
320 uint8_t buf[ID3v1_TAG_SIZE];
321 MP3Context *mp3 = s->priv_data;
323 if (mp3->pics_to_write) {
324 av_log(s, AV_LOG_WARNING, "No packets were sent for some of the "
325 "attached pictures.\n");
329 /* write the id3v1 tag */
330 if (mp3->write_id3v1 && id3v1_create_tag(s, buf) > 0) {
331 avio_write(s->pb, buf, ID3v1_TAG_SIZE);
334 if (mp3->xing_offset)
341 AVOutputFormat ff_mp2_muxer = {
343 .long_name = NULL_IF_CONFIG_SMALL("MP2 (MPEG audio layer 2)"),
344 .mime_type = "audio/x-mpeg",
345 .extensions = "mp2,m2a,mpa",
346 .audio_codec = AV_CODEC_ID_MP2,
347 .video_codec = AV_CODEC_ID_NONE,
348 .write_packet = ff_raw_write_packet,
349 .flags = AVFMT_NOTIMESTAMPS,
355 static const AVOption options[] = {
356 { "id3v2_version", "Select ID3v2 version to write. Currently 3 and 4 are supported.",
357 offsetof(MP3Context, id3v2_version), AV_OPT_TYPE_INT, {.i64 = 4}, 0, 4, AV_OPT_FLAG_ENCODING_PARAM},
358 { "write_id3v1", "Enable ID3v1 writing. ID3v1 tags are written in UTF-8 which may not be supported by most software.",
359 offsetof(MP3Context, write_id3v1), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, AV_OPT_FLAG_ENCODING_PARAM},
360 { "write_xing", "Write the Xing header containing file duration.",
361 offsetof(MP3Context, write_xing), AV_OPT_TYPE_INT, {.i64 = 1}, 0, 1, AV_OPT_FLAG_ENCODING_PARAM},
365 static const AVClass mp3_muxer_class = {
366 .class_name = "MP3 muxer",
367 .item_name = av_default_item_name,
369 .version = LIBAVUTIL_VERSION_INT,
372 static int mp3_write_packet(AVFormatContext *s, AVPacket *pkt)
374 MP3Context *mp3 = s->priv_data;
376 if (pkt->stream_index == mp3->audio_stream_idx) {
377 if (mp3->pics_to_write) {
378 /* buffer audio packets until we get all the pictures */
379 AVPacketList *pktl = av_mallocz(sizeof(*pktl));
381 return AVERROR(ENOMEM);
384 pktl->pkt.buf = av_buffer_ref(pkt->buf);
385 if (!pktl->pkt.buf) {
387 return AVERROR(ENOMEM);
391 mp3->queue_end->next = pktl;
394 mp3->queue_end = pktl;
396 return mp3_write_audio_packet(s, pkt);
400 /* warn only once for each stream */
401 if (s->streams[pkt->stream_index]->nb_frames == 1) {
402 av_log(s, AV_LOG_WARNING, "Got more than one picture in stream %d,"
403 " ignoring.\n", pkt->stream_index);
405 if (!mp3->pics_to_write || s->streams[pkt->stream_index]->nb_frames >= 1)
408 if ((ret = ff_id3v2_write_apic(s, &mp3->id3, pkt)) < 0)
410 mp3->pics_to_write--;
412 /* flush the buffered audio packets */
413 if (!mp3->pics_to_write &&
414 (ret = mp3_queue_flush(s)) < 0)
422 * Write an ID3v2 header at beginning of stream
425 static int mp3_write_header(struct AVFormatContext *s)
427 MP3Context *mp3 = s->priv_data;
430 if (mp3->id3v2_version &&
431 mp3->id3v2_version != 3 &&
432 mp3->id3v2_version != 4) {
433 av_log(s, AV_LOG_ERROR, "Invalid ID3v2 version requested: %d. Only "
434 "3, 4 or 0 (disabled) are allowed.\n", mp3->id3v2_version);
435 return AVERROR(EINVAL);
438 /* check the streams -- we want exactly one audio and arbitrary number of
439 * video (attached pictures) */
440 mp3->audio_stream_idx = -1;
441 for (i = 0; i < s->nb_streams; i++) {
442 AVStream *st = s->streams[i];
443 if (st->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
444 if (mp3->audio_stream_idx >= 0 || st->codec->codec_id != AV_CODEC_ID_MP3) {
445 av_log(s, AV_LOG_ERROR, "Invalid audio stream. Exactly one MP3 "
446 "audio stream is required.\n");
447 return AVERROR(EINVAL);
449 mp3->audio_stream_idx = i;
450 } else if (st->codec->codec_type != AVMEDIA_TYPE_VIDEO) {
451 av_log(s, AV_LOG_ERROR, "Only audio streams and pictures are allowed in MP3.\n");
452 return AVERROR(EINVAL);
455 if (mp3->audio_stream_idx < 0) {
456 av_log(s, AV_LOG_ERROR, "No audio stream present.\n");
457 return AVERROR(EINVAL);
459 mp3->pics_to_write = s->nb_streams - 1;
461 if (mp3->pics_to_write && !mp3->id3v2_version) {
462 av_log(s, AV_LOG_ERROR, "Attached pictures were requested, but the "
463 "ID3v2 header is disabled.\n");
464 return AVERROR(EINVAL);
467 if (mp3->id3v2_version) {
468 ff_id3v2_start(&mp3->id3, s->pb, mp3->id3v2_version, ID3v2_DEFAULT_MAGIC);
469 ret = ff_id3v2_write_metadata(s, &mp3->id3);
474 if (!mp3->pics_to_write) {
475 if (mp3->id3v2_version)
476 ff_id3v2_finish(&mp3->id3, s->pb);
483 AVOutputFormat ff_mp3_muxer = {
485 .long_name = NULL_IF_CONFIG_SMALL("MP3 (MPEG audio layer 3)"),
486 .mime_type = "audio/x-mpeg",
488 .priv_data_size = sizeof(MP3Context),
489 .audio_codec = AV_CODEC_ID_MP3,
490 .video_codec = AV_CODEC_ID_PNG,
491 .write_header = mp3_write_header,
492 .write_packet = mp3_write_packet,
493 .write_trailer = mp3_write_trailer,
494 .flags = AVFMT_NOTIMESTAMPS,
495 .priv_class = &mp3_muxer_class,