574a3b2e951679262df33fb54da816d69ea6294b
[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
88 /* xing header */
89 int64_t xing_offset;
90 int32_t frames;
91 int32_t size;
92 uint32_t want;
93 uint32_t seen;
94 uint32_t pos;
95 uint64_t bag[XING_NUM_BAGS];
96 int initial_bitrate;
97 int has_variable_bitrate;
98
99 /* index of the audio stream */
100 int audio_stream_idx;
101 /* number of attached pictures we still need to write */
102 int pics_to_write;
103
104 /* audio packets are queued here until we get all the attached pictures */
105 AVPacketList *queue, *queue_end;
106 } MP3Context;
107
108 static const uint8_t xing_offtbl[2][2] = {{32, 17}, {17, 9}};
109
110 /*
111 * Write an empty XING header and initialize respective data.
112 */
113 static void mp3_write_xing(AVFormatContext *s)
114 {
115 MP3Context *mp3 = s->priv_data;
116 AVCodecContext *codec = s->streams[mp3->audio_stream_idx]->codec;
117 int32_t header;
118 MPADecodeHeader mpah;
119 int srate_idx, i, channels;
120 int bitrate_idx;
121 int xing_offset;
122 int ver = 0;
123
124 if (!s->pb->seekable)
125 return;
126
127 for (i = 0; i < FF_ARRAY_ELEMS(avpriv_mpa_freq_tab); i++) {
128 const uint16_t base_freq = avpriv_mpa_freq_tab[i];
129
130 if (codec->sample_rate == base_freq) ver = 0x3; // MPEG 1
131 else if (codec->sample_rate == base_freq / 2) ver = 0x2; // MPEG 2
132 else if (codec->sample_rate == base_freq / 4) ver = 0x0; // MPEG 2.5
133 else continue;
134
135 srate_idx = i;
136 break;
137 }
138 if (i == FF_ARRAY_ELEMS(avpriv_mpa_freq_tab)) {
139 av_log(s, AV_LOG_WARNING, "Unsupported sample rate, not writing Xing "
140 "header.\n");
141 return;
142 }
143
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, "
148 "not writing Xing header.\n");
149 return;
150 }
151
152 /* 64 kbps frame, should be large enough */
153 bitrate_idx = (ver == 3) ? 5 : 8;
154
155 /* dummy MPEG audio header */
156 header = 0xff << 24; // sync
157 header |= (0x7 << 5 | ver << 3 | 0x1 << 1 | 0x1) << 16; // sync/audio-version/layer 3/no crc*/
158 header |= (bitrate_idx << 4 | srate_idx << 2) << 8;
159 header |= channels << 6;
160 avio_wb32(s->pb, header);
161
162 avpriv_mpegaudio_decode_header(&mpah, header);
163
164 av_assert0(mpah.frame_size >= XING_MAX_SIZE);
165
166 xing_offset = xing_offtbl[ver != 3][codec->channels == 1];
167 ffio_fill(s->pb, 0, xing_offset);
168 mp3->xing_offset = avio_tell(s->pb);
169 ffio_wfourcc(s->pb, "Xing");
170 avio_wb32(s->pb, 0x01 | 0x02 | 0x04); // frames / size / TOC
171
172 mp3->size = mpah.frame_size;
173 mp3->want = 1;
174
175 avio_wb32(s->pb, 0); // frames
176 avio_wb32(s->pb, 0); // size
177
178 // TOC
179 for (i = 0; i < XING_TOC_SIZE; i++)
180 avio_w8(s->pb, 255 * i / XING_TOC_SIZE);
181
182 mpah.frame_size -= 4 + xing_offset + 4 + 4 + 4 + 4 + XING_TOC_SIZE;
183 ffio_fill(s->pb, 0, mpah.frame_size);
184 }
185
186 /*
187 * Add a frame to XING data.
188 * Following lame's "VbrTag.c".
189 */
190 static void mp3_xing_add_frame(MP3Context *mp3, AVPacket *pkt)
191 {
192 int i;
193
194 mp3->frames++;
195 mp3->seen++;
196 mp3->size += pkt->size;
197
198 if (mp3->want == mp3->seen) {
199 mp3->bag[mp3->pos] = mp3->size;
200
201 if (XING_NUM_BAGS == ++mp3->pos) {
202 /* shrink table to half size by throwing away each second bag. */
203 for (i = 1; i < XING_NUM_BAGS; i += 2)
204 mp3->bag[i / 2] = mp3->bag[i];
205
206 /* double wanted amount per bag. */
207 mp3->want *= 2;
208 /* adjust current position to half of table size. */
209 mp3->pos = XING_NUM_BAGS / 2;
210 }
211
212 mp3->seen = 0;
213 }
214 }
215
216 static int mp3_write_audio_packet(AVFormatContext *s, AVPacket *pkt)
217 {
218 MP3Context *mp3 = s->priv_data;
219
220 if (mp3->xing_offset && pkt->size >= 4) {
221 MPADecodeHeader c;
222
223 avpriv_mpegaudio_decode_header(&c, AV_RB32(pkt->data));
224
225 if (!mp3->initial_bitrate)
226 mp3->initial_bitrate = c.bit_rate;
227 if ((c.bit_rate == 0) || (mp3->initial_bitrate != c.bit_rate))
228 mp3->has_variable_bitrate = 1;
229
230 mp3_xing_add_frame(mp3, pkt);
231 }
232
233 return ff_raw_write_packet(s, pkt);
234 }
235
236 static int mp3_queue_flush(AVFormatContext *s)
237 {
238 MP3Context *mp3 = s->priv_data;
239 AVPacketList *pktl;
240 int ret = 0, write = 1;
241
242 ff_id3v2_finish(&mp3->id3, s->pb);
243 mp3_write_xing(s);
244
245 while ((pktl = mp3->queue)) {
246 if (write && (ret = mp3_write_audio_packet(s, &pktl->pkt)) < 0)
247 write = 0;
248 av_free_packet(&pktl->pkt);
249 mp3->queue = pktl->next;
250 av_freep(&pktl);
251 }
252 mp3->queue_end = NULL;
253 return ret;
254 }
255
256 static void mp3_update_xing(AVFormatContext *s)
257 {
258 MP3Context *mp3 = s->priv_data;
259 int i;
260
261 /* replace "Xing" identification string with "Info" for CBR files. */
262 if (!mp3->has_variable_bitrate) {
263 avio_seek(s->pb, mp3->xing_offset, SEEK_SET);
264 ffio_wfourcc(s->pb, "Info");
265 }
266
267 avio_seek(s->pb, mp3->xing_offset + 8, SEEK_SET);
268 avio_wb32(s->pb, mp3->frames);
269 avio_wb32(s->pb, mp3->size);
270
271 avio_w8(s->pb, 0); // first toc entry has to be zero.
272
273 for (i = 1; i < XING_TOC_SIZE; ++i) {
274 int j = i * mp3->pos / XING_TOC_SIZE;
275 int seek_point = 256LL * mp3->bag[j] / mp3->size;
276 avio_w8(s->pb, FFMIN(seek_point, 255));
277 }
278
279 avio_seek(s->pb, 0, SEEK_END);
280 }
281
282 static int mp3_write_trailer(struct AVFormatContext *s)
283 {
284 uint8_t buf[ID3v1_TAG_SIZE];
285 MP3Context *mp3 = s->priv_data;
286
287 if (mp3->pics_to_write) {
288 av_log(s, AV_LOG_WARNING, "No packets were sent for some of the "
289 "attached pictures.\n");
290 mp3_queue_flush(s);
291 }
292
293 /* write the id3v1 tag */
294 if (mp3->write_id3v1 && id3v1_create_tag(s, buf) > 0) {
295 avio_write(s->pb, buf, ID3v1_TAG_SIZE);
296 }
297
298 if (mp3->xing_offset)
299 mp3_update_xing(s);
300
301 return 0;
302 }
303
304 #if CONFIG_MP2_MUXER
305 AVOutputFormat ff_mp2_muxer = {
306 .name = "mp2",
307 .long_name = NULL_IF_CONFIG_SMALL("MP2 (MPEG audio layer 2)"),
308 .mime_type = "audio/x-mpeg",
309 .extensions = "mp2,m2a,mpa",
310 .audio_codec = AV_CODEC_ID_MP2,
311 .video_codec = AV_CODEC_ID_NONE,
312 .write_packet = ff_raw_write_packet,
313 .flags = AVFMT_NOTIMESTAMPS,
314 };
315 #endif
316
317 #if CONFIG_MP3_MUXER
318
319 static const AVOption options[] = {
320 { "id3v2_version", "Select ID3v2 version to write. Currently 3 and 4 are supported.",
321 offsetof(MP3Context, id3v2_version), AV_OPT_TYPE_INT, {.i64 = 4}, 3, 4, AV_OPT_FLAG_ENCODING_PARAM},
322 { "write_id3v1", "Enable ID3v1 writing. ID3v1 tags are written in UTF-8 which may not be supported by most software.",
323 offsetof(MP3Context, write_id3v1), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, AV_OPT_FLAG_ENCODING_PARAM},
324 { NULL },
325 };
326
327 static const AVClass mp3_muxer_class = {
328 .class_name = "MP3 muxer",
329 .item_name = av_default_item_name,
330 .option = options,
331 .version = LIBAVUTIL_VERSION_INT,
332 };
333
334 static int mp3_write_packet(AVFormatContext *s, AVPacket *pkt)
335 {
336 MP3Context *mp3 = s->priv_data;
337
338 if (pkt->stream_index == mp3->audio_stream_idx) {
339 if (mp3->pics_to_write) {
340 /* buffer audio packets until we get all the pictures */
341 AVPacketList *pktl = av_mallocz(sizeof(*pktl));
342 if (!pktl)
343 return AVERROR(ENOMEM);
344
345 pktl->pkt = *pkt;
346 pktl->pkt.buf = av_buffer_ref(pkt->buf);
347 if (!pktl->pkt.buf) {
348 av_freep(&pktl);
349 return AVERROR(ENOMEM);
350 }
351
352 if (mp3->queue_end)
353 mp3->queue_end->next = pktl;
354 else
355 mp3->queue = pktl;
356 mp3->queue_end = pktl;
357 } else
358 return mp3_write_audio_packet(s, pkt);
359 } else {
360 int ret;
361
362 /* warn only once for each stream */
363 if (s->streams[pkt->stream_index]->nb_frames == 1) {
364 av_log(s, AV_LOG_WARNING, "Got more than one picture in stream %d,"
365 " ignoring.\n", pkt->stream_index);
366 }
367 if (!mp3->pics_to_write || s->streams[pkt->stream_index]->nb_frames >= 1)
368 return 0;
369
370 if ((ret = ff_id3v2_write_apic(s, &mp3->id3, pkt)) < 0)
371 return ret;
372 mp3->pics_to_write--;
373
374 /* flush the buffered audio packets */
375 if (!mp3->pics_to_write &&
376 (ret = mp3_queue_flush(s)) < 0)
377 return ret;
378 }
379
380 return 0;
381 }
382
383 /**
384 * Write an ID3v2 header at beginning of stream
385 */
386
387 static int mp3_write_header(struct AVFormatContext *s)
388 {
389 MP3Context *mp3 = s->priv_data;
390 int ret, i;
391
392 /* check the streams -- we want exactly one audio and arbitrary number of
393 * video (attached pictures) */
394 mp3->audio_stream_idx = -1;
395 for (i = 0; i < s->nb_streams; i++) {
396 AVStream *st = s->streams[i];
397 if (st->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
398 if (mp3->audio_stream_idx >= 0 || st->codec->codec_id != AV_CODEC_ID_MP3) {
399 av_log(s, AV_LOG_ERROR, "Invalid audio stream. Exactly one MP3 "
400 "audio stream is required.\n");
401 return AVERROR(EINVAL);
402 }
403 mp3->audio_stream_idx = i;
404 } else if (st->codec->codec_type != AVMEDIA_TYPE_VIDEO) {
405 av_log(s, AV_LOG_ERROR, "Only audio streams and pictures are allowed in MP3.\n");
406 return AVERROR(EINVAL);
407 }
408 }
409 if (mp3->audio_stream_idx < 0) {
410 av_log(s, AV_LOG_ERROR, "No audio stream present.\n");
411 return AVERROR(EINVAL);
412 }
413 mp3->pics_to_write = s->nb_streams - 1;
414
415 ff_id3v2_start(&mp3->id3, s->pb, mp3->id3v2_version, ID3v2_DEFAULT_MAGIC);
416 ret = ff_id3v2_write_metadata(s, &mp3->id3);
417 if (ret < 0)
418 return ret;
419
420 if (!mp3->pics_to_write) {
421 ff_id3v2_finish(&mp3->id3, s->pb);
422 mp3_write_xing(s);
423 }
424
425 return 0;
426 }
427
428 AVOutputFormat ff_mp3_muxer = {
429 .name = "mp3",
430 .long_name = NULL_IF_CONFIG_SMALL("MP3 (MPEG audio layer 3)"),
431 .mime_type = "audio/x-mpeg",
432 .extensions = "mp3",
433 .priv_data_size = sizeof(MP3Context),
434 .audio_codec = AV_CODEC_ID_MP3,
435 .video_codec = AV_CODEC_ID_PNG,
436 .write_header = mp3_write_header,
437 .write_packet = mp3_write_packet,
438 .write_trailer = mp3_write_trailer,
439 .flags = AVFMT_NOTIMESTAMPS,
440 .priv_class = &mp3_muxer_class,
441 };
442 #endif