* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
-#include <strings.h>
#include "avformat.h"
+#include "avio_internal.h"
#include "id3v1.h"
#include "id3v2.h"
#include "rawenc.h"
#include "libavutil/avstring.h"
+#include "libavcodec/mpegaudio.h"
+#include "libavcodec/mpegaudiodata.h"
+#include "libavcodec/mpegaudiodecheader.h"
#include "libavutil/intreadwrite.h"
#include "libavutil/opt.h"
+#include "libavutil/dict.h"
+#include "libavutil/avassert.h"
static int id3v1_set_string(AVFormatContext *s, const char *key,
uint8_t *buf, int buf_size)
{
- AVMetadataTag *tag;
- if ((tag = av_metadata_get(s->metadata, key, NULL, 0)))
+ AVDictionaryEntry *tag;
+ if ((tag = av_dict_get(s->metadata, key, NULL, 0)))
av_strlcpy(buf, tag->value, buf_size);
return !!tag;
}
static int id3v1_create_tag(AVFormatContext *s, uint8_t *buf)
{
- AVMetadataTag *tag;
+ AVDictionaryEntry *tag;
int i, count = 0;
memset(buf, 0, ID3v1_TAG_SIZE); /* fail safe */
count += id3v1_set_string(s, "TALB", buf + 63, 30); //album
count += id3v1_set_string(s, "TDRL", buf + 93, 4); //date
count += id3v1_set_string(s, "comment", buf + 97, 30);
- if ((tag = av_metadata_get(s->metadata, "TRCK", NULL, 0))) { //track
+ if ((tag = av_dict_get(s->metadata, "TRCK", NULL, 0))) { //track
buf[125] = 0;
buf[126] = atoi(tag->value);
count++;
}
buf[127] = 0xFF; /* default to unknown genre */
- if ((tag = av_metadata_get(s->metadata, "TCON", NULL, 0))) { //genre
+ if ((tag = av_dict_get(s->metadata, "TCON", NULL, 0))) { //genre
for(i = 0; i <= ID3v1_GENRE_MAX; i++) {
- if (!strcasecmp(tag->value, ff_id3v1_genre_str[i])) {
+ if (!av_strcasecmp(tag->value, ff_id3v1_genre_str[i])) {
buf[127] = i;
count++;
break;
return count;
}
-/* simple formats */
+#define XING_NUM_BAGS 400
+#define XING_TOC_SIZE 100
+// maximum size of the xing frame: offset/Xing/flags/frames/size/TOC
+#define XING_MAX_SIZE (32 + 4 + 4 + 4 + 4 + XING_TOC_SIZE)
-static void id3v2_put_size(AVFormatContext *s, int size)
+typedef struct MP3Context {
+ const AVClass *class;
+ ID3v2EncContext id3;
+ int id3v2_version;
+ int write_id3v1;
+ int write_xing;
+
+ /* xing header */
+ int64_t xing_offset;
+ int32_t frames;
+ int32_t size;
+ uint32_t want;
+ uint32_t seen;
+ uint32_t pos;
+ uint64_t bag[XING_NUM_BAGS];
+ int initial_bitrate;
+ int has_variable_bitrate;
+
+ /* index of the audio stream */
+ int audio_stream_idx;
+ /* number of attached pictures we still need to write */
+ int pics_to_write;
+
+ /* audio packets are queued here until we get all the attached pictures */
+ AVPacketList *queue, *queue_end;
+} MP3Context;
+
+static const uint8_t xing_offtbl[2][2] = {{32, 17}, {17, 9}};
+
+/*
+ * Write an empty XING header and initialize respective data.
+ */
+static void mp3_write_xing(AVFormatContext *s)
{
- avio_w8(s->pb, size >> 21 & 0x7f);
- avio_w8(s->pb, size >> 14 & 0x7f);
- avio_w8(s->pb, size >> 7 & 0x7f);
- avio_w8(s->pb, size & 0x7f);
+ MP3Context *mp3 = s->priv_data;
+ AVCodecContext *codec = s->streams[mp3->audio_stream_idx]->codec;
+ int32_t header;
+ MPADecodeHeader mpah;
+ int srate_idx, i, channels;
+ int bitrate_idx;
+ int best_bitrate_idx;
+ int best_bitrate_error = INT_MAX;
+ int xing_offset;
+ int ver = 0;
+ int lsf, bytes_needed;
+
+ if (!s->pb->seekable || !mp3->write_xing)
+ return;
+
+ for (i = 0; i < FF_ARRAY_ELEMS(avpriv_mpa_freq_tab); i++) {
+ const uint16_t base_freq = avpriv_mpa_freq_tab[i];
+
+ if (codec->sample_rate == base_freq) ver = 0x3; // MPEG 1
+ else if (codec->sample_rate == base_freq / 2) ver = 0x2; // MPEG 2
+ else if (codec->sample_rate == base_freq / 4) ver = 0x0; // MPEG 2.5
+ else continue;
+
+ srate_idx = i;
+ break;
+ }
+ if (i == FF_ARRAY_ELEMS(avpriv_mpa_freq_tab)) {
+ av_log(s, AV_LOG_WARNING, "Unsupported sample rate, not writing Xing "
+ "header.\n");
+ return;
+ }
+
+ switch (codec->channels) {
+ case 1: channels = MPA_MONO; break;
+ case 2: channels = MPA_STEREO; break;
+ default: av_log(s, AV_LOG_WARNING, "Unsupported number of channels, "
+ "not writing Xing header.\n");
+ return;
+ }
+
+ /* dummy MPEG audio header */
+ header = 0xff << 24; // sync
+ header |= (0x7 << 5 | ver << 3 | 0x1 << 1 | 0x1) << 16; // sync/audio-version/layer 3/no crc*/
+ header |= (srate_idx << 2) << 8;
+ header |= channels << 6;
+
+ lsf = !((header & (1 << 20) && header & (1 << 19)));
+
+ xing_offset = xing_offtbl[ver != 3][channels == 1];
+ bytes_needed = 4 // header
+ + xing_offset
+ + 4 // xing tag
+ + 4 // frames/size/toc flags
+ + 4 // frames
+ + 4 // size
+ + XING_TOC_SIZE; // toc
+
+ for (bitrate_idx = 1; bitrate_idx < 15; bitrate_idx++) {
+ int bit_rate = 1000 * avpriv_mpa_bitrate_tab[lsf][3 - 1][bitrate_idx];
+ int error = FFABS(bit_rate - codec->bit_rate);
+
+ if (error < best_bitrate_error){
+ best_bitrate_error = error;
+ best_bitrate_idx = bitrate_idx;
+ }
+ }
+
+ for (bitrate_idx = best_bitrate_idx; bitrate_idx < 15; bitrate_idx++) {
+ int32_t mask = bitrate_idx << (4 + 8);
+ header |= mask;
+
+ avpriv_mpegaudio_decode_header(&mpah, header);
+
+ if (bytes_needed <= mpah.frame_size)
+ break;
+
+ header &= ~mask;
+ }
+
+ avio_wb32(s->pb, header);
+
+ avpriv_mpegaudio_decode_header(&mpah, header);
+
+ av_assert0(mpah.frame_size >= XING_MAX_SIZE);
+
+ ffio_fill(s->pb, 0, xing_offset);
+ mp3->xing_offset = avio_tell(s->pb);
+ ffio_wfourcc(s->pb, "Xing");
+ avio_wb32(s->pb, 0x01 | 0x02 | 0x04); // frames / size / TOC
+
+ mp3->size = mpah.frame_size;
+ mp3->want = 1;
+
+ avio_wb32(s->pb, 0); // frames
+ avio_wb32(s->pb, 0); // size
+
+ // TOC
+ for (i = 0; i < XING_TOC_SIZE; i++)
+ avio_w8(s->pb, 255 * i / XING_TOC_SIZE);
+
+ ffio_fill(s->pb, 0, mpah.frame_size - bytes_needed);
}
-static int string_is_ascii(const uint8_t *str)
+/*
+ * Add a frame to XING data.
+ * Following lame's "VbrTag.c".
+ */
+static void mp3_xing_add_frame(MP3Context *mp3, AVPacket *pkt)
{
- while (*str && *str < 128) str++;
- return !*str;
+ int i;
+
+ mp3->frames++;
+ mp3->seen++;
+ mp3->size += pkt->size;
+
+ if (mp3->want == mp3->seen) {
+ mp3->bag[mp3->pos] = mp3->size;
+
+ if (XING_NUM_BAGS == ++mp3->pos) {
+ /* shrink table to half size by throwing away each second bag. */
+ for (i = 1; i < XING_NUM_BAGS; i += 2)
+ mp3->bag[i / 2] = mp3->bag[i];
+
+ /* double wanted amount per bag. */
+ mp3->want *= 2;
+ /* adjust current position to half of table size. */
+ mp3->pos = XING_NUM_BAGS / 2;
+ }
+
+ mp3->seen = 0;
+ }
}
-/**
- * Write a text frame with one (normal frames) or two (TXXX frames) strings
- * according to encoding (only UTF-8 or UTF-16+BOM supported).
- * @return number of bytes written or a negative error code.
- */
-static int id3v2_put_ttag(AVFormatContext *s, const char *str1, const char *str2,
- uint32_t tag, enum ID3v2Encoding enc)
+static int mp3_write_audio_packet(AVFormatContext *s, AVPacket *pkt)
+{
+ MP3Context *mp3 = s->priv_data;
+
+ if (mp3->xing_offset && pkt->size >= 4) {
+ MPADecodeHeader c;
+
+ avpriv_mpegaudio_decode_header(&c, AV_RB32(pkt->data));
+
+ if (!mp3->initial_bitrate)
+ mp3->initial_bitrate = c.bit_rate;
+ if ((c.bit_rate == 0) || (mp3->initial_bitrate != c.bit_rate))
+ mp3->has_variable_bitrate = 1;
+
+ mp3_xing_add_frame(mp3, pkt);
+ }
+
+ return ff_raw_write_packet(s, pkt);
+}
+
+static int mp3_queue_flush(AVFormatContext *s)
+{
+ MP3Context *mp3 = s->priv_data;
+ AVPacketList *pktl;
+ int ret = 0, write = 1;
+
+ ff_id3v2_finish(&mp3->id3, s->pb);
+ mp3_write_xing(s);
+
+ while ((pktl = mp3->queue)) {
+ if (write && (ret = mp3_write_audio_packet(s, &pktl->pkt)) < 0)
+ write = 0;
+ av_free_packet(&pktl->pkt);
+ mp3->queue = pktl->next;
+ av_freep(&pktl);
+ }
+ mp3->queue_end = NULL;
+ return ret;
+}
+
+static void mp3_update_xing(AVFormatContext *s)
{
- int len;
- uint8_t *pb;
- int (*put)(AVIOContext*, const char*);
- AVIOContext *dyn_buf;
- if (avio_open_dyn_buf(&dyn_buf) < 0)
- return AVERROR(ENOMEM);
-
- /* check if the strings are ASCII-only and use UTF16 only if
- * they're not */
- if (enc == ID3v2_ENCODING_UTF16BOM && string_is_ascii(str1) &&
- (!str2 || string_is_ascii(str2)))
- enc = ID3v2_ENCODING_ISO8859;
-
- avio_w8(dyn_buf, enc);
- if (enc == ID3v2_ENCODING_UTF16BOM) {
- avio_wl16(dyn_buf, 0xFEFF); /* BOM */
- put = avio_put_str16le;
- } else
- put = avio_put_str;
-
- put(dyn_buf, str1);
- if (str2)
- put(dyn_buf, str2);
- len = avio_close_dyn_buf(dyn_buf, &pb);
-
- avio_wb32(s->pb, tag);
- id3v2_put_size(s, len);
- avio_wb16(s->pb, 0);
- avio_write(s->pb, pb, len);
-
- av_freep(&pb);
- return len + ID3v2_HEADER_SIZE;
+ MP3Context *mp3 = s->priv_data;
+ int i;
+
+ /* replace "Xing" identification string with "Info" for CBR files. */
+ if (!mp3->has_variable_bitrate) {
+ avio_seek(s->pb, mp3->xing_offset, SEEK_SET);
+ ffio_wfourcc(s->pb, "Info");
+ }
+
+ avio_seek(s->pb, mp3->xing_offset + 8, SEEK_SET);
+ avio_wb32(s->pb, mp3->frames);
+ avio_wb32(s->pb, mp3->size);
+
+ avio_w8(s->pb, 0); // first toc entry has to be zero.
+
+ for (i = 1; i < XING_TOC_SIZE; ++i) {
+ int j = i * mp3->pos / XING_TOC_SIZE;
+ int seek_point = 256LL * mp3->bag[j] / mp3->size;
+ avio_w8(s->pb, FFMIN(seek_point, 255));
+ }
+
+ avio_seek(s->pb, 0, SEEK_END);
}
static int mp3_write_trailer(struct AVFormatContext *s)
{
uint8_t buf[ID3v1_TAG_SIZE];
+ MP3Context *mp3 = s->priv_data;
+
+ if (mp3->pics_to_write) {
+ av_log(s, AV_LOG_WARNING, "No packets were sent for some of the "
+ "attached pictures.\n");
+ mp3_queue_flush(s);
+ }
/* write the id3v1 tag */
- if (id3v1_create_tag(s, buf) > 0) {
+ if (mp3->write_id3v1 && id3v1_create_tag(s, buf) > 0) {
avio_write(s->pb, buf, ID3v1_TAG_SIZE);
- avio_flush(s->pb);
}
+
+ if (mp3->xing_offset)
+ mp3_update_xing(s);
+
return 0;
}
#if CONFIG_MP2_MUXER
AVOutputFormat ff_mp2_muxer = {
- "mp2",
- NULL_IF_CONFIG_SMALL("MPEG audio layer 2"),
- "audio/x-mpeg",
- "mp2,m2a",
- 0,
- CODEC_ID_MP2,
- CODEC_ID_NONE,
- NULL,
- ff_raw_write_packet,
- mp3_write_trailer,
+ .name = "mp2",
+ .long_name = NULL_IF_CONFIG_SMALL("MP2 (MPEG audio layer 2)"),
+ .mime_type = "audio/x-mpeg",
+ .extensions = "mp2,m2a,mpa",
+ .audio_codec = AV_CODEC_ID_MP2,
+ .video_codec = AV_CODEC_ID_NONE,
+ .write_packet = ff_raw_write_packet,
+ .flags = AVFMT_NOTIMESTAMPS,
};
#endif
#if CONFIG_MP3_MUXER
-typedef struct MP3Context {
- const AVClass *class;
- int id3v2_version;
-} MP3Context;
static const AVOption options[] = {
{ "id3v2_version", "Select ID3v2 version to write. Currently 3 and 4 are supported.",
- offsetof(MP3Context, id3v2_version), FF_OPT_TYPE_INT, {.dbl = 4}, 3, 4, AV_OPT_FLAG_ENCODING_PARAM},
+ offsetof(MP3Context, id3v2_version), AV_OPT_TYPE_INT, {.i64 = 4}, 0, 4, AV_OPT_FLAG_ENCODING_PARAM},
+ { "write_id3v1", "Enable ID3v1 writing. ID3v1 tags are written in UTF-8 which may not be supported by most software.",
+ offsetof(MP3Context, write_id3v1), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, AV_OPT_FLAG_ENCODING_PARAM},
+ { "write_xing", "Write the Xing header containing file duration.",
+ offsetof(MP3Context, write_xing), AV_OPT_TYPE_INT, {.i64 = 1}, 0, 1, AV_OPT_FLAG_ENCODING_PARAM},
{ NULL },
};
.version = LIBAVUTIL_VERSION_INT,
};
-static int id3v2_check_write_tag(AVFormatContext *s, AVMetadataTag *t, const char table[][4],
- enum ID3v2Encoding enc)
+static int mp3_write_packet(AVFormatContext *s, AVPacket *pkt)
{
- uint32_t tag;
- int i;
+ MP3Context *mp3 = s->priv_data;
+
+ if (pkt->stream_index == mp3->audio_stream_idx) {
+ if (mp3->pics_to_write) {
+ /* buffer audio packets until we get all the pictures */
+ AVPacketList *pktl = av_mallocz(sizeof(*pktl));
+ if (!pktl)
+ return AVERROR(ENOMEM);
+
+ pktl->pkt = *pkt;
+ pktl->pkt.buf = av_buffer_ref(pkt->buf);
+ if (!pktl->pkt.buf) {
+ av_freep(&pktl);
+ return AVERROR(ENOMEM);
+ }
+
+ if (mp3->queue_end)
+ mp3->queue_end->next = pktl;
+ else
+ mp3->queue = pktl;
+ mp3->queue_end = pktl;
+ } else
+ return mp3_write_audio_packet(s, pkt);
+ } else {
+ int ret;
+
+ /* warn only once for each stream */
+ if (s->streams[pkt->stream_index]->nb_frames == 1) {
+ av_log(s, AV_LOG_WARNING, "Got more than one picture in stream %d,"
+ " ignoring.\n", pkt->stream_index);
+ }
+ if (!mp3->pics_to_write || s->streams[pkt->stream_index]->nb_frames >= 1)
+ return 0;
+
+ if ((ret = ff_id3v2_write_apic(s, &mp3->id3, pkt)) < 0)
+ return ret;
+ mp3->pics_to_write--;
+
+ /* flush the buffered audio packets */
+ if (!mp3->pics_to_write &&
+ (ret = mp3_queue_flush(s)) < 0)
+ return ret;
+ }
- if (t->key[0] != 'T' || strlen(t->key) != 4)
- return -1;
- tag = AV_RB32(t->key);
- for (i = 0; *table[i]; i++)
- if (tag == AV_RB32(table[i]))
- return id3v2_put_ttag(s, t->value, NULL, tag, enc);
- return -1;
+ return 0;
}
/**
static int mp3_write_header(struct AVFormatContext *s)
{
MP3Context *mp3 = s->priv_data;
- AVMetadataTag *t = NULL;
- int totlen = 0, enc = mp3->id3v2_version == 3 ? ID3v2_ENCODING_UTF16BOM :
- ID3v2_ENCODING_UTF8;
- int64_t size_pos, cur_pos;
-
- avio_wb32(s->pb, MKBETAG('I', 'D', '3', mp3->id3v2_version));
- avio_w8(s->pb, 0);
- avio_w8(s->pb, 0); /* flags */
-
- /* reserve space for size */
- size_pos = avio_tell(s->pb);
- avio_wb32(s->pb, 0);
-
- ff_metadata_conv(&s->metadata, ff_id3v2_34_metadata_conv, NULL);
- if (mp3->id3v2_version == 4)
- ff_metadata_conv(&s->metadata, ff_id3v2_4_metadata_conv, NULL);
-
- while ((t = av_metadata_get(s->metadata, "", t, AV_METADATA_IGNORE_SUFFIX))) {
- int ret;
+ int ret, i;
+
+ if (mp3->id3v2_version &&
+ mp3->id3v2_version != 3 &&
+ mp3->id3v2_version != 4) {
+ av_log(s, AV_LOG_ERROR, "Invalid ID3v2 version requested: %d. Only "
+ "3, 4 or 0 (disabled) are allowed.\n", mp3->id3v2_version);
+ return AVERROR(EINVAL);
+ }
- if ((ret = id3v2_check_write_tag(s, t, ff_id3v2_tags, enc)) > 0) {
- totlen += ret;
- continue;
- }
- if ((ret = id3v2_check_write_tag(s, t, mp3->id3v2_version == 3 ?
- ff_id3v2_3_tags : ff_id3v2_4_tags, enc)) > 0) {
- totlen += ret;
- continue;
+ /* check the streams -- we want exactly one audio and arbitrary number of
+ * video (attached pictures) */
+ mp3->audio_stream_idx = -1;
+ for (i = 0; i < s->nb_streams; i++) {
+ AVStream *st = s->streams[i];
+ if (st->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
+ if (mp3->audio_stream_idx >= 0 || st->codec->codec_id != AV_CODEC_ID_MP3) {
+ av_log(s, AV_LOG_ERROR, "Invalid audio stream. Exactly one MP3 "
+ "audio stream is required.\n");
+ return AVERROR(EINVAL);
+ }
+ mp3->audio_stream_idx = i;
+ } else if (st->codec->codec_type != AVMEDIA_TYPE_VIDEO) {
+ av_log(s, AV_LOG_ERROR, "Only audio streams and pictures are allowed in MP3.\n");
+ return AVERROR(EINVAL);
}
+ }
+ if (mp3->audio_stream_idx < 0) {
+ av_log(s, AV_LOG_ERROR, "No audio stream present.\n");
+ return AVERROR(EINVAL);
+ }
+ mp3->pics_to_write = s->nb_streams - 1;
- /* unknown tag, write as TXXX frame */
- if ((ret = id3v2_put_ttag(s, t->key, t->value, MKBETAG('T', 'X', 'X', 'X'), enc)) < 0)
+ if (mp3->pics_to_write && !mp3->id3v2_version) {
+ av_log(s, AV_LOG_ERROR, "Attached pictures were requested, but the "
+ "ID3v2 header is disabled.\n");
+ return AVERROR(EINVAL);
+ }
+
+ if (mp3->id3v2_version) {
+ ff_id3v2_start(&mp3->id3, s->pb, mp3->id3v2_version, ID3v2_DEFAULT_MAGIC);
+ ret = ff_id3v2_write_metadata(s, &mp3->id3);
+ if (ret < 0)
return ret;
- totlen += ret;
}
- cur_pos = avio_tell(s->pb);
- avio_seek(s->pb, size_pos, SEEK_SET);
- id3v2_put_size(s, totlen);
- avio_seek(s->pb, cur_pos, SEEK_SET);
+ if (!mp3->pics_to_write) {
+ if (mp3->id3v2_version)
+ ff_id3v2_finish(&mp3->id3, s->pb);
+ mp3_write_xing(s);
+ }
return 0;
}
AVOutputFormat ff_mp3_muxer = {
- "mp3",
- NULL_IF_CONFIG_SMALL("MPEG audio layer 3"),
- "audio/x-mpeg",
- "mp3",
- sizeof(MP3Context),
- CODEC_ID_MP3,
- CODEC_ID_NONE,
- mp3_write_header,
- ff_raw_write_packet,
- mp3_write_trailer,
- AVFMT_NOTIMESTAMPS,
- .priv_class = &mp3_muxer_class,
+ .name = "mp3",
+ .long_name = NULL_IF_CONFIG_SMALL("MP3 (MPEG audio layer 3)"),
+ .mime_type = "audio/x-mpeg",
+ .extensions = "mp3",
+ .priv_data_size = sizeof(MP3Context),
+ .audio_codec = AV_CODEC_ID_MP3,
+ .video_codec = AV_CODEC_ID_PNG,
+ .write_header = mp3_write_header,
+ .write_packet = mp3_write_packet,
+ .write_trailer = mp3_write_trailer,
+ .flags = AVFMT_NOTIMESTAMPS,
+ .priv_class = &mp3_muxer_class,
};
#endif