2 * WebM DASH Manifest XML muxer
3 * Copyright (c) 2014 Vignesh Venkatasubramanian
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 * WebM DASH Specification:
24 * https://sites.google.com/a/webmproject.org/wiki/adaptive-streaming/webm-dash-specification
31 #include "avio_internal.h"
34 #include "libavutil/avstring.h"
35 #include "libavutil/dict.h"
36 #include "libavutil/opt.h"
38 typedef struct AdaptationSet {
44 typedef struct WebMDashMuxContext {
46 char *adaptation_sets;
51 static const char *get_codec_name(int codec_id)
58 case AV_CODEC_ID_VORBIS:
60 case AV_CODEC_ID_OPUS:
66 static double get_duration(AVFormatContext *s)
70 for (i = 0; i < s->nb_streams; i++) {
71 AVDictionaryEntry *duration = av_dict_get(s->streams[i]->metadata,
73 if (!duration || atof(duration->value) < 0) continue;
74 if (atof(duration->value) > max) max = atof(duration->value);
79 static void write_header(AVFormatContext *s)
81 double min_buffer_time = 1.0;
82 avio_printf(s->pb, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
83 avio_printf(s->pb, "<MPD\n");
84 avio_printf(s->pb, " xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\"\n");
85 avio_printf(s->pb, " xmlns=\"urn:mpeg:DASH:schema:MPD:2011\"\n");
86 avio_printf(s->pb, " xsi:schemaLocation=\"urn:mpeg:DASH:schema:MPD:2011\"\n");
87 avio_printf(s->pb, " type=\"static\"\n");
88 avio_printf(s->pb, " mediaPresentationDuration=\"PT%gS\"\n",
90 avio_printf(s->pb, " minBufferTime=\"PT%gS\"\n",
92 avio_printf(s->pb, " profiles=\"urn:webm:dash:profile:webm-on-demand:2012\"");
93 avio_printf(s->pb, ">\n");
96 static void write_footer(AVFormatContext *s)
98 avio_printf(s->pb, "</MPD>\n");
101 static int subsegment_alignment(AVFormatContext *s, AdaptationSet *as) {
103 AVDictionaryEntry *gold = av_dict_get(s->streams[as->streams[0]]->metadata,
104 CUE_TIMESTAMPS, NULL, 0);
106 for (i = 1; i < as->nb_streams; i++) {
107 AVDictionaryEntry *ts = av_dict_get(s->streams[as->streams[i]]->metadata,
108 CUE_TIMESTAMPS, NULL, 0);
109 if (!ts || strncmp(gold->value, ts->value, strlen(gold->value))) return 0;
114 static int bitstream_switching(AVFormatContext *s, AdaptationSet *as) {
116 AVDictionaryEntry *gold_track_num = av_dict_get(s->streams[as->streams[0]]->metadata,
117 TRACK_NUMBER, NULL, 0);
118 AVCodecContext *gold_codec = s->streams[as->streams[0]]->codec;
119 if (!gold_track_num) return 0;
120 for (i = 1; i < as->nb_streams; i++) {
121 AVDictionaryEntry *track_num = av_dict_get(s->streams[as->streams[i]]->metadata,
122 TRACK_NUMBER, NULL, 0);
123 AVCodecContext *codec = s->streams[as->streams[i]]->codec;
125 strncmp(gold_track_num->value, track_num->value, strlen(gold_track_num->value)) ||
126 gold_codec->codec_id != codec->codec_id ||
127 gold_codec->extradata_size != codec->extradata_size ||
128 memcmp(gold_codec->extradata, codec->extradata, codec->extradata_size)) {
136 * Writes a Representation within an Adaptation Set. Returns 0 on success and
139 static int write_representation(AVFormatContext *s, AVStream *stream, int id,
140 int output_width, int output_height,
141 int output_sample_rate) {
142 AVDictionaryEntry *irange = av_dict_get(stream->metadata, INITIALIZATION_RANGE, NULL, 0);
143 AVDictionaryEntry *cues_start = av_dict_get(stream->metadata, CUES_START, NULL, 0);
144 AVDictionaryEntry *cues_end = av_dict_get(stream->metadata, CUES_END, NULL, 0);
145 AVDictionaryEntry *filename = av_dict_get(stream->metadata, FILENAME, NULL, 0);
146 AVDictionaryEntry *bandwidth = av_dict_get(stream->metadata, BANDWIDTH, NULL, 0);
147 if (!irange || cues_start == NULL || cues_end == NULL || filename == NULL ||
151 avio_printf(s->pb, "<Representation id=\"%d\"", id);
152 avio_printf(s->pb, " bandwidth=\"%s\"", bandwidth->value);
153 if (stream->codec->codec_type == AVMEDIA_TYPE_VIDEO && output_width)
154 avio_printf(s->pb, " width=\"%d\"", stream->codec->width);
155 if (stream->codec->codec_type == AVMEDIA_TYPE_VIDEO && output_height)
156 avio_printf(s->pb, " height=\"%d\"", stream->codec->height);
157 if (stream->codec->codec_type = AVMEDIA_TYPE_AUDIO && output_sample_rate)
158 avio_printf(s->pb, " audioSamplingRate=\"%d\"", stream->codec->sample_rate);
159 avio_printf(s->pb, ">\n");
160 avio_printf(s->pb, "<BaseURL>%s</BaseURL>\n", filename->value);
161 avio_printf(s->pb, "<SegmentBase\n");
162 avio_printf(s->pb, " indexRange=\"%s-%s\">\n", cues_start->value, cues_end->value);
163 avio_printf(s->pb, "<Initialization\n");
164 avio_printf(s->pb, " range=\"0-%s\" />\n", irange->value);
165 avio_printf(s->pb, "</SegmentBase>\n");
166 avio_printf(s->pb, "</Representation>\n");
171 * Checks if width of all streams are the same. Returns 1 if true, 0 otherwise.
173 static int check_matching_width(AVFormatContext *s, AdaptationSet *as) {
175 if (as->nb_streams < 2) return 1;
176 first_width = s->streams[as->streams[0]]->codec->width;
177 for (i = 1; i < as->nb_streams; i++)
178 if (first_width != s->streams[as->streams[i]]->codec->width)
184 * Checks if height of all streams are the same. Returns 1 if true, 0 otherwise.
186 static int check_matching_height(AVFormatContext *s, AdaptationSet *as) {
188 if (as->nb_streams < 2) return 1;
189 first_height = s->streams[as->streams[0]]->codec->height;
190 for (i = 1; i < as->nb_streams; i++)
191 if (first_height != s->streams[as->streams[i]]->codec->height)
197 * Checks if sample rate of all streams are the same. Returns 1 if true, 0 otherwise.
199 static int check_matching_sample_rate(AVFormatContext *s, AdaptationSet *as) {
200 int first_sample_rate, i;
201 if (as->nb_streams < 2) return 1;
202 first_sample_rate = s->streams[as->streams[0]]->codec->sample_rate;
203 for (i = 1; i < as->nb_streams; i++)
204 if (first_sample_rate != s->streams[as->streams[i]]->codec->sample_rate)
210 * Writes an Adaptation Set. Returns 0 on success and < 0 on failure.
212 static int write_adaptation_set(AVFormatContext *s, int as_index)
214 WebMDashMuxContext *w = s->priv_data;
215 AdaptationSet *as = &w->as[as_index];
216 AVCodecContext *codec = s->streams[as->streams[0]]->codec;
217 AVDictionaryEntry *lang;
219 static const char boolean[2][6] = { "false", "true" };
220 int subsegmentStartsWithSAP = 1;
222 // Width, Height and Sample Rate will go in the AdaptationSet tag if they
223 // are the same for all contained Representations. otherwise, they will go
224 // on their respective Representation tag.
225 int width_in_as = 1, height_in_as = 1, sample_rate_in_as = 1;
226 if (codec->codec_type == AVMEDIA_TYPE_VIDEO) {
227 width_in_as = check_matching_width(s, as);
228 height_in_as = check_matching_height(s, as);
230 sample_rate_in_as = check_matching_sample_rate(s, as);
233 avio_printf(s->pb, "<AdaptationSet id=\"%s\"", as->id);
234 avio_printf(s->pb, " mimeType=\"%s/webm\"",
235 codec->codec_type == AVMEDIA_TYPE_VIDEO ? "video" : "audio");
236 avio_printf(s->pb, " codecs=\"%s\"", get_codec_name(codec->codec_id));
238 lang = av_dict_get(s->streams[as->streams[0]]->metadata, "language", NULL, 0);
239 if (lang) avio_printf(s->pb, " lang=\"%s\"", lang->value);
241 if (codec->codec_type == AVMEDIA_TYPE_VIDEO && width_in_as)
242 avio_printf(s->pb, " width=\"%d\"", codec->width);
243 if (codec->codec_type == AVMEDIA_TYPE_VIDEO && height_in_as)
244 avio_printf(s->pb, " height=\"%d\"", codec->height);
245 if (codec->codec_type == AVMEDIA_TYPE_AUDIO && sample_rate_in_as)
246 avio_printf(s->pb, " audioSamplingRate=\"%d\"", codec->sample_rate);
248 avio_printf(s->pb, " bitstreamSwitching=\"%s\"",
249 boolean[bitstream_switching(s, as)]);
250 avio_printf(s->pb, " subsegmentAlignment=\"%s\"",
251 boolean[subsegment_alignment(s, as)]);
253 for (i = 0; i < as->nb_streams; i++) {
254 AVDictionaryEntry *kf = av_dict_get(s->streams[as->streams[i]]->metadata,
255 CLUSTER_KEYFRAME, NULL, 0);
256 if (!kf || !strncmp(kf->value, "0", 1)) subsegmentStartsWithSAP = 0;
258 avio_printf(s->pb, " subsegmentStartsWithSAP=\"%d\"", subsegmentStartsWithSAP);
259 avio_printf(s->pb, ">\n");
261 for (i = 0; i < as->nb_streams; i++) {
262 write_representation(s, s->streams[as->streams[i]], i,
263 !width_in_as, !height_in_as, !sample_rate_in_as);
265 avio_printf(s->pb, "</AdaptationSet>\n");
269 static int to_integer(char *p, int len)
272 char *q = av_malloc(sizeof(char) * len);
274 av_strlcpy(q, p, len);
280 static int parse_adaptation_sets(AVFormatContext *s)
282 WebMDashMuxContext *w = s->priv_data;
283 char *p = w->adaptation_sets;
285 enum { new_set, parsed_id, parsing_streams } state;
286 // syntax id=0,streams=0,1,2 id=1,streams=3,4 and so on
288 while (p < w->adaptation_sets + strlen(w->adaptation_sets)) {
291 else if (state == new_set && !strncmp(p, "id=", 3)) {
292 w->as = av_realloc(w->as, sizeof(*w->as) * ++w->nb_as);
293 if (w->as == NULL) return -1;
294 w->as[w->nb_as - 1].nb_streams = 0;
295 w->as[w->nb_as - 1].streams = NULL;
296 p += 3; // consume "id="
297 q = w->as[w->nb_as - 1].id;
298 while (*p != ',') *q++ = *p++;
302 } else if (state == parsed_id && !strncmp(p, "streams=", 8)) {
303 p += 8; // consume "streams="
304 state = parsing_streams;
305 } else if (state == parsing_streams) {
306 struct AdaptationSet *as = &w->as[w->nb_as - 1];
308 while (*q != '\0' && *q != ',' && *q != ' ') q++;
309 as->streams = av_realloc(as->streams, sizeof(*as->streams) * ++as->nb_streams);
310 if (as->streams == NULL) return -1;
311 as->streams[as->nb_streams - 1] = to_integer(p, q - p + 1);
312 if (as->streams[as->nb_streams - 1] < 0) return -1;
313 if (*q == '\0') break;
314 if (*q == ' ') state = new_set;
323 static int webm_dash_manifest_write_header(AVFormatContext *s)
327 WebMDashMuxContext *w = s->priv_data;
328 parse_adaptation_sets(s);
330 avio_printf(s->pb, "<Period id=\"0\"");
331 avio_printf(s->pb, " start=\"PT%gS\"", start);
332 avio_printf(s->pb, " duration=\"PT%gS\"", get_duration(s));
333 avio_printf(s->pb, " >\n");
335 for (i = 0; i < w->nb_as; i++) {
336 if (write_adaptation_set(s, i) < 0) return -1;
339 avio_printf(s->pb, "</Period>\n");
344 static int webm_dash_manifest_write_packet(AVFormatContext *s, AVPacket *pkt)
349 static int webm_dash_manifest_write_trailer(AVFormatContext *s)
351 WebMDashMuxContext *w = s->priv_data;
353 for (i = 0; i < w->nb_as; i++) {
354 av_freep(&w->as[i].streams);
360 #define OFFSET(x) offsetof(WebMDashMuxContext, x)
361 static const AVOption options[] = {
362 { "adaptation_sets", "Adaptation sets. Syntax: id=0,streams=0,1,2 id=1,streams=3,4 and so on", OFFSET(adaptation_sets), AV_OPT_TYPE_STRING, { 0 }, 0, 0, AV_OPT_FLAG_ENCODING_PARAM },
366 #if CONFIG_WEBM_DASH_MANIFEST_MUXER
367 static const AVClass webm_dash_class = {
368 .class_name = "WebM DASH Manifest muxer",
369 .item_name = av_default_item_name,
371 .version = LIBAVUTIL_VERSION_INT,
374 AVOutputFormat ff_webm_dash_manifest_muxer = {
375 .name = "webm_dash_manifest",
376 .long_name = NULL_IF_CONFIG_SMALL("WebM DASH Manifest"),
377 .mime_type = "application/xml",
379 .priv_data_size = sizeof(WebMDashMuxContext),
380 .write_header = webm_dash_manifest_write_header,
381 .write_packet = webm_dash_manifest_write_packet,
382 .write_trailer = webm_dash_manifest_write_trailer,
383 .priv_class = &webm_dash_class,