mp3enc: add an option for disabling the Xing frame.
[libav.git] / libavformat / mp3enc.c
1 /*
2 * MP3 muxer
3 * Copyright (c) 2003 Fabrice Bellard
4 *
5 * This file is part of Libav.
6 *
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.
11 *
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.
16 *
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
20 */
21
22 #include "avformat.h"
23 #include "avio_internal.h"
24 #include "id3v1.h"
25 #include "id3v2.h"
26 #include "rawenc.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"
35
36 static int id3v1_set_string(AVFormatContext *s, const char *key,
37 uint8_t *buf, int buf_size)
38 {
39 AVDictionaryEntry *tag;
40 if ((tag = av_dict_get(s->metadata, key, NULL, 0)))
41 av_strlcpy(buf, tag->value, buf_size);
42 return !!tag;
43 }
44
45 static int id3v1_create_tag(AVFormatContext *s, uint8_t *buf)
46 {
47 AVDictionaryEntry *tag;
48 int i, count = 0;
49
50 memset(buf, 0, ID3v1_TAG_SIZE); /* fail safe */
51 buf[0] = 'T';
52 buf[1] = 'A';
53 buf[2] = 'G';
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
60 buf[125] = 0;
61 buf[126] = atoi(tag->value);
62 count++;
63 }
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])) {
68 buf[127] = i;
69 count++;
70 break;
71 }
72 }
73 }
74 return count;
75 }
76
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)
81
82 typedef struct MP3Context {
83 const AVClass *class;
84 ID3v2EncContext id3;
85 int id3v2_version;
86 int write_id3v1;
87 int write_xing;
88
89 /* xing header */
90 int64_t xing_offset;
91 int32_t frames;
92 int32_t size;
93 uint32_t want;
94 uint32_t seen;
95 uint32_t pos;
96 uint64_t bag[XING_NUM_BAGS];
97 int initial_bitrate;
98 int has_variable_bitrate;
99
100 /* index of the audio stream */
101 int audio_stream_idx;
102 /* number of attached pictures we still need to write */
103 int pics_to_write;
104
105 /* audio packets are queued here until we get all the attached pictures */
106 AVPacketList *queue, *queue_end;
107 } MP3Context;
108
109 static const uint8_t xing_offtbl[2][2] = {{32, 17}, {17, 9}};
110
111 /*
112 * Write an empty XING header and initialize respective data.
113 */
114 static void mp3_write_xing(AVFormatContext *s)
115 {
116 MP3Context *mp3 = s->priv_data;
117 AVCodecContext *codec = s->streams[mp3->audio_stream_idx]->codec;
118 int32_t header;
119 MPADecodeHeader mpah;
120 int srate_idx, i, channels;
121 int bitrate_idx;
122 int xing_offset;
123 int ver = 0;
124
125 if (!s->pb->seekable || !mp3->write_xing)
126 return;
127
128 for (i = 0; i < FF_ARRAY_ELEMS(avpriv_mpa_freq_tab); i++) {
129 const uint16_t base_freq = avpriv_mpa_freq_tab[i];
130
131 if (codec->sample_rate == base_freq) ver = 0x3; // MPEG 1
132 else if (codec->sample_rate == base_freq / 2) ver = 0x2; // MPEG 2
133 else if (codec->sample_rate == base_freq / 4) ver = 0x0; // MPEG 2.5
134 else continue;
135
136 srate_idx = i;
137 break;
138 }
139 if (i == FF_ARRAY_ELEMS(avpriv_mpa_freq_tab)) {
140 av_log(s, AV_LOG_WARNING, "Unsupported sample rate, not writing Xing "
141 "header.\n");
142 return;
143 }
144
145 switch (codec->channels) {
146 case 1: channels = MPA_MONO; break;
147 case 2: channels = MPA_STEREO; break;
148 default: av_log(s, AV_LOG_WARNING, "Unsupported number of channels, "
149 "not writing Xing header.\n");
150 return;
151 }
152
153 /* 64 kbps frame, should be large enough */
154 bitrate_idx = (ver == 3) ? 5 : 8;
155
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 |= (bitrate_idx << 4 | srate_idx << 2) << 8;
160 header |= channels << 6;
161 avio_wb32(s->pb, header);
162
163 avpriv_mpegaudio_decode_header(&mpah, header);
164
165 av_assert0(mpah.frame_size >= XING_MAX_SIZE);
166
167 xing_offset = xing_offtbl[ver != 3][codec->channels == 1];
168 ffio_fill(s->pb, 0, xing_offset);
169 mp3->xing_offset = avio_tell(s->pb);
170 ffio_wfourcc(s->pb, "Xing");
171 avio_wb32(s->pb, 0x01 | 0x02 | 0x04); // frames / size / TOC
172
173 mp3->size = mpah.frame_size;
174 mp3->want = 1;
175
176 avio_wb32(s->pb, 0); // frames
177 avio_wb32(s->pb, 0); // size
178
179 // TOC
180 for (i = 0; i < XING_TOC_SIZE; i++)
181 avio_w8(s->pb, 255 * i / XING_TOC_SIZE);
182
183 mpah.frame_size -= 4 + xing_offset + 4 + 4 + 4 + 4 + XING_TOC_SIZE;
184 ffio_fill(s->pb, 0, mpah.frame_size);
185 }
186
187 /*
188 * Add a frame to XING data.
189 * Following lame's "VbrTag.c".
190 */
191 static void mp3_xing_add_frame(MP3Context *mp3, AVPacket *pkt)
192 {
193 int i;
194
195 mp3->frames++;
196 mp3->seen++;
197 mp3->size += pkt->size;
198
199 if (mp3->want == mp3->seen) {
200 mp3->bag[mp3->pos] = mp3->size;
201
202 if (XING_NUM_BAGS == ++mp3->pos) {
203 /* shrink table to half size by throwing away each second bag. */
204 for (i = 1; i < XING_NUM_BAGS; i += 2)
205 mp3->bag[i / 2] = mp3->bag[i];
206
207 /* double wanted amount per bag. */
208 mp3->want *= 2;
209 /* adjust current position to half of table size. */
210 mp3->pos = XING_NUM_BAGS / 2;
211 }
212
213 mp3->seen = 0;
214 }
215 }
216
217 static int mp3_write_audio_packet(AVFormatContext *s, AVPacket *pkt)
218 {
219 MP3Context *mp3 = s->priv_data;
220
221 if (mp3->xing_offset && pkt->size >= 4) {
222 MPADecodeHeader c;
223
224 avpriv_mpegaudio_decode_header(&c, AV_RB32(pkt->data));
225
226 if (!mp3->initial_bitrate)
227 mp3->initial_bitrate = c.bit_rate;
228 if ((c.bit_rate == 0) || (mp3->initial_bitrate != c.bit_rate))
229 mp3->has_variable_bitrate = 1;
230
231 mp3_xing_add_frame(mp3, pkt);
232 }
233
234 return ff_raw_write_packet(s, pkt);
235 }
236
237 static int mp3_queue_flush(AVFormatContext *s)
238 {
239 MP3Context *mp3 = s->priv_data;
240 AVPacketList *pktl;
241 int ret = 0, write = 1;
242
243 ff_id3v2_finish(&mp3->id3, s->pb);
244 mp3_write_xing(s);
245
246 while ((pktl = mp3->queue)) {
247 if (write && (ret = mp3_write_audio_packet(s, &pktl->pkt)) < 0)
248 write = 0;
249 av_free_packet(&pktl->pkt);
250 mp3->queue = pktl->next;
251 av_freep(&pktl);
252 }
253 mp3->queue_end = NULL;
254 return ret;
255 }
256
257 static void mp3_update_xing(AVFormatContext *s)
258 {
259 MP3Context *mp3 = s->priv_data;
260 int i;
261
262 /* replace "Xing" identification string with "Info" for CBR files. */
263 if (!mp3->has_variable_bitrate) {
264 avio_seek(s->pb, mp3->xing_offset, SEEK_SET);
265 ffio_wfourcc(s->pb, "Info");
266 }
267
268 avio_seek(s->pb, mp3->xing_offset + 8, SEEK_SET);
269 avio_wb32(s->pb, mp3->frames);
270 avio_wb32(s->pb, mp3->size);
271
272 avio_w8(s->pb, 0); // first toc entry has to be zero.
273
274 for (i = 1; i < XING_TOC_SIZE; ++i) {
275 int j = i * mp3->pos / XING_TOC_SIZE;
276 int seek_point = 256LL * mp3->bag[j] / mp3->size;
277 avio_w8(s->pb, FFMIN(seek_point, 255));
278 }
279
280 avio_seek(s->pb, 0, SEEK_END);
281 }
282
283 static int mp3_write_trailer(struct AVFormatContext *s)
284 {
285 uint8_t buf[ID3v1_TAG_SIZE];
286 MP3Context *mp3 = s->priv_data;
287
288 if (mp3->pics_to_write) {
289 av_log(s, AV_LOG_WARNING, "No packets were sent for some of the "
290 "attached pictures.\n");
291 mp3_queue_flush(s);
292 }
293
294 /* write the id3v1 tag */
295 if (mp3->write_id3v1 && id3v1_create_tag(s, buf) > 0) {
296 avio_write(s->pb, buf, ID3v1_TAG_SIZE);
297 }
298
299 if (mp3->xing_offset)
300 mp3_update_xing(s);
301
302 return 0;
303 }
304
305 #if CONFIG_MP2_MUXER
306 AVOutputFormat ff_mp2_muxer = {
307 .name = "mp2",
308 .long_name = NULL_IF_CONFIG_SMALL("MP2 (MPEG audio layer 2)"),
309 .mime_type = "audio/x-mpeg",
310 .extensions = "mp2,m2a,mpa",
311 .audio_codec = AV_CODEC_ID_MP2,
312 .video_codec = AV_CODEC_ID_NONE,
313 .write_packet = ff_raw_write_packet,
314 .flags = AVFMT_NOTIMESTAMPS,
315 };
316 #endif
317
318 #if CONFIG_MP3_MUXER
319
320 static const AVOption options[] = {
321 { "id3v2_version", "Select ID3v2 version to write. Currently 3 and 4 are supported.",
322 offsetof(MP3Context, id3v2_version), AV_OPT_TYPE_INT, {.i64 = 4}, 3, 4, AV_OPT_FLAG_ENCODING_PARAM},
323 { "write_id3v1", "Enable ID3v1 writing. ID3v1 tags are written in UTF-8 which may not be supported by most software.",
324 offsetof(MP3Context, write_id3v1), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, AV_OPT_FLAG_ENCODING_PARAM},
325 { "write_xing", "Write the Xing header containing file duration.",
326 offsetof(MP3Context, write_xing), AV_OPT_TYPE_INT, {.i64 = 1}, 0, 1, AV_OPT_FLAG_ENCODING_PARAM},
327 { NULL },
328 };
329
330 static const AVClass mp3_muxer_class = {
331 .class_name = "MP3 muxer",
332 .item_name = av_default_item_name,
333 .option = options,
334 .version = LIBAVUTIL_VERSION_INT,
335 };
336
337 static int mp3_write_packet(AVFormatContext *s, AVPacket *pkt)
338 {
339 MP3Context *mp3 = s->priv_data;
340
341 if (pkt->stream_index == mp3->audio_stream_idx) {
342 if (mp3->pics_to_write) {
343 /* buffer audio packets until we get all the pictures */
344 AVPacketList *pktl = av_mallocz(sizeof(*pktl));
345 if (!pktl)
346 return AVERROR(ENOMEM);
347
348 pktl->pkt = *pkt;
349 pktl->pkt.buf = av_buffer_ref(pkt->buf);
350 if (!pktl->pkt.buf) {
351 av_freep(&pktl);
352 return AVERROR(ENOMEM);
353 }
354
355 if (mp3->queue_end)
356 mp3->queue_end->next = pktl;
357 else
358 mp3->queue = pktl;
359 mp3->queue_end = pktl;
360 } else
361 return mp3_write_audio_packet(s, pkt);
362 } else {
363 int ret;
364
365 /* warn only once for each stream */
366 if (s->streams[pkt->stream_index]->nb_frames == 1) {
367 av_log(s, AV_LOG_WARNING, "Got more than one picture in stream %d,"
368 " ignoring.\n", pkt->stream_index);
369 }
370 if (!mp3->pics_to_write || s->streams[pkt->stream_index]->nb_frames >= 1)
371 return 0;
372
373 if ((ret = ff_id3v2_write_apic(s, &mp3->id3, pkt)) < 0)
374 return ret;
375 mp3->pics_to_write--;
376
377 /* flush the buffered audio packets */
378 if (!mp3->pics_to_write &&
379 (ret = mp3_queue_flush(s)) < 0)
380 return ret;
381 }
382
383 return 0;
384 }
385
386 /**
387 * Write an ID3v2 header at beginning of stream
388 */
389
390 static int mp3_write_header(struct AVFormatContext *s)
391 {
392 MP3Context *mp3 = s->priv_data;
393 int ret, i;
394
395 /* check the streams -- we want exactly one audio and arbitrary number of
396 * video (attached pictures) */
397 mp3->audio_stream_idx = -1;
398 for (i = 0; i < s->nb_streams; i++) {
399 AVStream *st = s->streams[i];
400 if (st->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
401 if (mp3->audio_stream_idx >= 0 || st->codec->codec_id != AV_CODEC_ID_MP3) {
402 av_log(s, AV_LOG_ERROR, "Invalid audio stream. Exactly one MP3 "
403 "audio stream is required.\n");
404 return AVERROR(EINVAL);
405 }
406 mp3->audio_stream_idx = i;
407 } else if (st->codec->codec_type != AVMEDIA_TYPE_VIDEO) {
408 av_log(s, AV_LOG_ERROR, "Only audio streams and pictures are allowed in MP3.\n");
409 return AVERROR(EINVAL);
410 }
411 }
412 if (mp3->audio_stream_idx < 0) {
413 av_log(s, AV_LOG_ERROR, "No audio stream present.\n");
414 return AVERROR(EINVAL);
415 }
416 mp3->pics_to_write = s->nb_streams - 1;
417
418 ff_id3v2_start(&mp3->id3, s->pb, mp3->id3v2_version, ID3v2_DEFAULT_MAGIC);
419 ret = ff_id3v2_write_metadata(s, &mp3->id3);
420 if (ret < 0)
421 return ret;
422
423 if (!mp3->pics_to_write) {
424 ff_id3v2_finish(&mp3->id3, s->pb);
425 mp3_write_xing(s);
426 }
427
428 return 0;
429 }
430
431 AVOutputFormat ff_mp3_muxer = {
432 .name = "mp3",
433 .long_name = NULL_IF_CONFIG_SMALL("MP3 (MPEG audio layer 3)"),
434 .mime_type = "audio/x-mpeg",
435 .extensions = "mp3",
436 .priv_data_size = sizeof(MP3Context),
437 .audio_codec = AV_CODEC_ID_MP3,
438 .video_codec = AV_CODEC_ID_PNG,
439 .write_header = mp3_write_header,
440 .write_packet = mp3_write_packet,
441 .write_trailer = mp3_write_trailer,
442 .flags = AVFMT_NOTIMESTAMPS,
443 .priv_class = &mp3_muxer_class,
444 };
445 #endif