X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavformat%2Fwebmdashenc.c;h=1304c1a8c3c2f597252152f721ec638c8c8cd3d4;hb=1d5d666601b6dc9319f8a539c40d016e3aef932e;hp=fd07b3e34a1568f5662216ce23693bf39cd0d86f;hpb=9b614826275e346ac17b9bc7ef5a58dded5b1855;p=ffmpeg diff --git a/libavformat/webmdashenc.c b/libavformat/webmdashenc.c index fd07b3e34a1..1304c1a8c3c 100644 --- a/libavformat/webmdashenc.c +++ b/libavformat/webmdashenc.c @@ -79,19 +79,20 @@ static double get_duration(AVFormatContext *s) static int write_header(AVFormatContext *s) { WebMDashMuxContext *w = s->priv_data; + AVIOContext *pb = s->pb; double min_buffer_time = 1.0; - avio_printf(s->pb, "\n"); - avio_printf(s->pb, "pb, " xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\"\n"); - avio_printf(s->pb, " xmlns=\"urn:mpeg:DASH:schema:MPD:2011\"\n"); - avio_printf(s->pb, " xsi:schemaLocation=\"urn:mpeg:DASH:schema:MPD:2011\"\n"); - avio_printf(s->pb, " type=\"%s\"\n", w->is_live ? "dynamic" : "static"); + avio_printf(pb, "\n"); + avio_printf(pb, "is_live ? "dynamic" : "static"); if (!w->is_live) { - avio_printf(s->pb, " mediaPresentationDuration=\"PT%gS\"\n", + avio_printf(pb, " mediaPresentationDuration=\"PT%gS\"\n", get_duration(s)); } - avio_printf(s->pb, " minBufferTime=\"PT%gS\"\n", min_buffer_time); - avio_printf(s->pb, " profiles=\"%s\"%s", + avio_printf(pb, " minBufferTime=\"PT%gS\"\n", min_buffer_time); + avio_printf(pb, " profiles=\"%s\"%s", w->is_live ? "urn:mpeg:dash:profile:isoff-live:2011" : "urn:webm:dash:profile:webm-on-demand:2012", w->is_live ? "\n" : ">\n"); if (w->is_live) { @@ -105,14 +106,14 @@ static int write_header(AVFormatContext *s) if (s->flags & AVFMT_FLAG_BITEXACT) { av_strlcpy(gmt_iso, "", 1); } - avio_printf(s->pb, " availabilityStartTime=\"%s\"\n", gmt_iso); - avio_printf(s->pb, " timeShiftBufferDepth=\"PT%gS\"\n", w->time_shift_buffer_depth); - avio_printf(s->pb, " minimumUpdatePeriod=\"PT%dS\"", w->minimum_update_period); - avio_printf(s->pb, ">\n"); + avio_printf(pb, " availabilityStartTime=\"%s\"\n", gmt_iso); + avio_printf(pb, " timeShiftBufferDepth=\"PT%gS\"\n", w->time_shift_buffer_depth); + avio_printf(pb, " minimumUpdatePeriod=\"PT%dS\"", w->minimum_update_period); + avio_printf(pb, ">\n"); if (w->utc_timing_url) { - avio_printf(s->pb, "pb, " schemeIdUri=\"urn:mpeg:dash:utc:http-iso:2014\"\n"); - avio_printf(s->pb, " value=\"%s\"/>\n", w->utc_timing_url); + avio_printf(pb, "\n", w->utc_timing_url); } } return 0; @@ -123,7 +124,8 @@ static void write_footer(AVFormatContext *s) avio_printf(s->pb, "\n"); } -static int subsegment_alignment(AVFormatContext *s, AdaptationSet *as) { +static int subsegment_alignment(AVFormatContext *s, const AdaptationSet *as) +{ int i; AVDictionaryEntry *gold = av_dict_get(s->streams[as->streams[0]]->metadata, CUE_TIMESTAMPS, NULL, 0); @@ -131,26 +133,30 @@ static int subsegment_alignment(AVFormatContext *s, AdaptationSet *as) { for (i = 1; i < as->nb_streams; i++) { AVDictionaryEntry *ts = av_dict_get(s->streams[as->streams[i]]->metadata, CUE_TIMESTAMPS, NULL, 0); - if (!ts || strncmp(gold->value, ts->value, strlen(gold->value))) return 0; + if (!ts || !av_strstart(ts->value, gold->value, NULL)) return 0; } return 1; } -static int bitstream_switching(AVFormatContext *s, AdaptationSet *as) { +static int bitstream_switching(AVFormatContext *s, const AdaptationSet *as) +{ int i; - AVDictionaryEntry *gold_track_num = av_dict_get(s->streams[as->streams[0]]->metadata, + const AVStream *gold_st = s->streams[as->streams[0]]; + AVDictionaryEntry *gold_track_num = av_dict_get(gold_st->metadata, TRACK_NUMBER, NULL, 0); - AVCodecParameters *gold_par = s->streams[as->streams[0]]->codecpar; + AVCodecParameters *gold_par = gold_st->codecpar; if (!gold_track_num) return 0; for (i = 1; i < as->nb_streams; i++) { - AVDictionaryEntry *track_num = av_dict_get(s->streams[as->streams[i]]->metadata, + const AVStream *st = s->streams[as->streams[i]]; + AVDictionaryEntry *track_num = av_dict_get(st->metadata, TRACK_NUMBER, NULL, 0); - AVCodecParameters *par = s->streams[as->streams[i]]->codecpar; + AVCodecParameters *par = st->codecpar; if (!track_num || - strncmp(gold_track_num->value, track_num->value, strlen(gold_track_num->value)) || + !av_strstart(track_num->value, gold_track_num->value, NULL) || gold_par->codec_id != par->codec_id || gold_par->extradata_size != par->extradata_size || - memcmp(gold_par->extradata, par->extradata, par->extradata_size)) { + (par->extradata_size > 0 && + memcmp(gold_par->extradata, par->extradata, par->extradata_size))) { return 0; } } @@ -161,60 +167,65 @@ static int bitstream_switching(AVFormatContext *s, AdaptationSet *as) { * Writes a Representation within an Adaptation Set. Returns 0 on success and * < 0 on failure. */ -static int write_representation(AVFormatContext *s, AVStream *stream, char *id, +static int write_representation(AVFormatContext *s, AVStream *st, char *id, int output_width, int output_height, - int output_sample_rate) { + int output_sample_rate) +{ WebMDashMuxContext *w = s->priv_data; - AVDictionaryEntry *irange = av_dict_get(stream->metadata, INITIALIZATION_RANGE, NULL, 0); - AVDictionaryEntry *cues_start = av_dict_get(stream->metadata, CUES_START, NULL, 0); - AVDictionaryEntry *cues_end = av_dict_get(stream->metadata, CUES_END, NULL, 0); - AVDictionaryEntry *filename = av_dict_get(stream->metadata, FILENAME, NULL, 0); - AVDictionaryEntry *bandwidth = av_dict_get(stream->metadata, BANDWIDTH, NULL, 0); + AVIOContext *pb = s->pb; + const AVCodecParameters *par = st->codecpar; + AVDictionaryEntry *bandwidth = av_dict_get(st->metadata, BANDWIDTH, NULL, 0); const char *bandwidth_str; - if ((w->is_live && (!filename)) || - (!w->is_live && (!irange || !cues_start || !cues_end || !filename || !bandwidth))) { - return AVERROR_INVALIDDATA; - } - avio_printf(s->pb, "is_live && !bandwidth) { - bandwidth_str = (stream->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) ? "128000" : "1000000"; - } else { + avio_printf(pb, "value; + } else if (w->is_live) { + // if bandwidth for live was not provided, use a default + bandwidth_str = (par->codec_type == AVMEDIA_TYPE_AUDIO) ? "128000" : "1000000"; + } else { + return AVERROR(EINVAL); } - avio_printf(s->pb, " bandwidth=\"%s\"", bandwidth_str); - if (stream->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && output_width) - avio_printf(s->pb, " width=\"%d\"", stream->codecpar->width); - if (stream->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && output_height) - avio_printf(s->pb, " height=\"%d\"", stream->codecpar->height); - if (stream->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && output_sample_rate) - avio_printf(s->pb, " audioSamplingRate=\"%d\"", stream->codecpar->sample_rate); + avio_printf(pb, " bandwidth=\"%s\"", bandwidth_str); + if (par->codec_type == AVMEDIA_TYPE_VIDEO && output_width) + avio_printf(pb, " width=\"%d\"", par->width); + if (par->codec_type == AVMEDIA_TYPE_VIDEO && output_height) + avio_printf(pb, " height=\"%d\"", par->height); + if (par->codec_type == AVMEDIA_TYPE_AUDIO && output_sample_rate) + avio_printf(pb, " audioSamplingRate=\"%d\"", par->sample_rate); if (w->is_live) { // For live streams, Codec and Mime Type always go in the Representation tag. - avio_printf(s->pb, " codecs=\"%s\"", get_codec_name(stream->codecpar->codec_id)); - avio_printf(s->pb, " mimeType=\"%s/webm\"", - stream->codecpar->codec_type == AVMEDIA_TYPE_VIDEO ? "video" : "audio"); + avio_printf(pb, " codecs=\"%s\"", get_codec_name(par->codec_id)); + avio_printf(pb, " mimeType=\"%s/webm\"", + par->codec_type == AVMEDIA_TYPE_VIDEO ? "video" : "audio"); // For live streams, subsegments always start with key frames. So this // is always 1. - avio_printf(s->pb, " startsWithSAP=\"1\""); - avio_printf(s->pb, ">"); + avio_printf(pb, " startsWithSAP=\"1\""); + avio_printf(pb, ">"); } else { - avio_printf(s->pb, ">\n"); - avio_printf(s->pb, "%s\n", filename->value); - avio_printf(s->pb, "pb, " indexRange=\"%s-%s\">\n", cues_start->value, cues_end->value); - avio_printf(s->pb, "pb, " range=\"0-%s\" />\n", irange->value); - avio_printf(s->pb, "\n"); + AVDictionaryEntry *irange = av_dict_get(st->metadata, INITIALIZATION_RANGE, NULL, 0); + AVDictionaryEntry *cues_start = av_dict_get(st->metadata, CUES_START, NULL, 0); + AVDictionaryEntry *cues_end = av_dict_get(st->metadata, CUES_END, NULL, 0); + AVDictionaryEntry *filename = av_dict_get(st->metadata, FILENAME, NULL, 0); + if (!irange || !cues_start || !cues_end || !filename) + return AVERROR(EINVAL); + + avio_printf(pb, ">\n"); + avio_printf(pb, "%s\n", filename->value); + avio_printf(pb, "\n", cues_start->value, cues_end->value); + avio_printf(pb, "\n", irange->value); + avio_printf(pb, "\n"); } - avio_printf(s->pb, "\n"); + avio_printf(pb, "\n"); return 0; } /* * Checks if width of all streams are the same. Returns 1 if true, 0 otherwise. */ -static int check_matching_width(AVFormatContext *s, AdaptationSet *as) { +static int check_matching_width(AVFormatContext *s, const AdaptationSet *as) +{ int first_width, i; if (as->nb_streams < 2) return 1; first_width = s->streams[as->streams[0]]->codecpar->width; @@ -227,7 +238,8 @@ static int check_matching_width(AVFormatContext *s, AdaptationSet *as) { /* * Checks if height of all streams are the same. Returns 1 if true, 0 otherwise. */ -static int check_matching_height(AVFormatContext *s, AdaptationSet *as) { +static int check_matching_height(AVFormatContext *s, const AdaptationSet *as) +{ int first_height, i; if (as->nb_streams < 2) return 1; first_height = s->streams[as->streams[0]]->codecpar->height; @@ -240,7 +252,8 @@ static int check_matching_height(AVFormatContext *s, AdaptationSet *as) { /* * Checks if sample rate of all streams are the same. Returns 1 if true, 0 otherwise. */ -static int check_matching_sample_rate(AVFormatContext *s, AdaptationSet *as) { +static int check_matching_sample_rate(AVFormatContext *s, const AdaptationSet *as) +{ int first_sample_rate, i; if (as->nb_streams < 2) return 1; first_sample_rate = s->streams[as->streams[0]]->codecpar->sample_rate; @@ -250,7 +263,8 @@ static int check_matching_sample_rate(AVFormatContext *s, AdaptationSet *as) { return 1; } -static void free_adaptation_sets(AVFormatContext *s) { +static void free_adaptation_sets(AVFormatContext *s) +{ WebMDashMuxContext *w = s->priv_data; int i; for (i = 0; i < w->nb_as; i++) { @@ -261,70 +275,24 @@ static void free_adaptation_sets(AVFormatContext *s) { } /* - * Parses a live header filename and computes the representation id, - * initialization pattern and the media pattern. Pass NULL if you don't want to - * compute any of those 3. Returns 0 on success and non-zero on failure. + * Parses a live header filename and returns the position of the '_' and '.' + * delimiting and . * * Name of the header file should conform to the following pattern: * _.hdr where can be * anything. The chunks should be named according to the following pattern: * __.chk */ -static int parse_filename(char *filename, char **representation_id, - char **initialization_pattern, char **media_pattern) { - char *underscore_pos = NULL; - char *period_pos = NULL; - char *temp_pos = NULL; - char *filename_str = av_strdup(filename); - int ret = 0; - - if (!filename_str) { - ret = AVERROR(ENOMEM); - goto end; - } - temp_pos = av_stristr(filename_str, "_"); - while (temp_pos) { - underscore_pos = temp_pos + 1; - temp_pos = av_stristr(temp_pos + 1, "_"); - } - if (!underscore_pos) { - ret = AVERROR_INVALIDDATA; - goto end; - } - period_pos = av_stristr(underscore_pos, "."); - if (!period_pos) { - ret = AVERROR_INVALIDDATA; - goto end; - } - *(underscore_pos - 1) = 0; - if (representation_id) { - *representation_id = av_malloc(period_pos - underscore_pos + 1); - if (!(*representation_id)) { - ret = AVERROR(ENOMEM); - goto end; - } - av_strlcpy(*representation_id, underscore_pos, period_pos - underscore_pos + 1); - } - if (initialization_pattern) { - *initialization_pattern = av_asprintf("%s_$RepresentationID$.hdr", - filename_str); - if (!(*initialization_pattern)) { - ret = AVERROR(ENOMEM); - goto end; - } - } - if (media_pattern) { - *media_pattern = av_asprintf("%s_$RepresentationID$_$Number$.chk", - filename_str); - if (!(*media_pattern)) { - ret = AVERROR(ENOMEM); - goto end; - } - } - -end: - av_freep(&filename_str); - return ret; +static int split_filename(char *filename, char **underscore_pos, + char **period_pos) +{ + *underscore_pos = strrchr(filename, '_'); + if (!*underscore_pos) + return AVERROR(EINVAL); + *period_pos = strchr(*underscore_pos, '.'); + if (!*period_pos) + return AVERROR(EINVAL); + return 0; } /* @@ -334,8 +302,10 @@ static int write_adaptation_set(AVFormatContext *s, int as_index) { WebMDashMuxContext *w = s->priv_data; AdaptationSet *as = &w->as[as_index]; - AVCodecParameters *par = s->streams[as->streams[0]]->codecpar; + const AVStream *st = s->streams[as->streams[0]]; + AVCodecParameters *par = st->codecpar; AVDictionaryEntry *lang; + AVIOContext *pb = s->pb; int i; static const char boolean[2][6] = { "false", "true" }; int subsegmentStartsWithSAP = 1; @@ -346,30 +316,31 @@ static int write_adaptation_set(AVFormatContext *s, int as_index) // in the Representation tag. int width_in_as = 1, height_in_as = 1, sample_rate_in_as = 1; if (par->codec_type == AVMEDIA_TYPE_VIDEO) { - width_in_as = !w->is_live && check_matching_width(s, as); + width_in_as = !w->is_live && check_matching_width (s, as); height_in_as = !w->is_live && check_matching_height(s, as); } else { sample_rate_in_as = !w->is_live && check_matching_sample_rate(s, as); } - avio_printf(s->pb, "id); - avio_printf(s->pb, " mimeType=\"%s/webm\"", + avio_printf(pb, "id); + avio_printf(pb, " mimeType=\"%s/webm\"", par->codec_type == AVMEDIA_TYPE_VIDEO ? "video" : "audio"); - avio_printf(s->pb, " codecs=\"%s\"", get_codec_name(par->codec_id)); + avio_printf(pb, " codecs=\"%s\"", get_codec_name(par->codec_id)); - lang = av_dict_get(s->streams[as->streams[0]]->metadata, "language", NULL, 0); - if (lang) avio_printf(s->pb, " lang=\"%s\"", lang->value); + lang = av_dict_get(st->metadata, "language", NULL, 0); + if (lang) + avio_printf(pb, " lang=\"%s\"", lang->value); if (par->codec_type == AVMEDIA_TYPE_VIDEO && width_in_as) - avio_printf(s->pb, " width=\"%d\"", par->width); + avio_printf(pb, " width=\"%d\"", par->width); if (par->codec_type == AVMEDIA_TYPE_VIDEO && height_in_as) - avio_printf(s->pb, " height=\"%d\"", par->height); + avio_printf(pb, " height=\"%d\"", par->height); if (par->codec_type == AVMEDIA_TYPE_AUDIO && sample_rate_in_as) - avio_printf(s->pb, " audioSamplingRate=\"%d\"", par->sample_rate); + avio_printf(pb, " audioSamplingRate=\"%d\"", par->sample_rate); - avio_printf(s->pb, " bitstreamSwitching=\"%s\"", + avio_printf(pb, " bitstreamSwitching=\"%s\"", boolean[bitstream_switching(s, as)]); - avio_printf(s->pb, " subsegmentAlignment=\"%s\"", + avio_printf(pb, " subsegmentAlignment=\"%s\"", boolean[w->is_live || subsegment_alignment(s, as)]); for (i = 0; i < as->nb_streams; i++) { @@ -377,49 +348,55 @@ static int write_adaptation_set(AVFormatContext *s, int as_index) CLUSTER_KEYFRAME, NULL, 0); if (!w->is_live && (!kf || !strncmp(kf->value, "0", 1))) subsegmentStartsWithSAP = 0; } - avio_printf(s->pb, " subsegmentStartsWithSAP=\"%d\"", subsegmentStartsWithSAP); - avio_printf(s->pb, ">\n"); + avio_printf(pb, " subsegmentStartsWithSAP=\"%d\"", subsegmentStartsWithSAP); + avio_printf(pb, ">\n"); if (w->is_live) { AVDictionaryEntry *filename = - av_dict_get(s->streams[as->streams[0]]->metadata, FILENAME, NULL, 0); - char *initialization_pattern = NULL; - char *media_pattern = NULL; - int ret = parse_filename(filename->value, NULL, &initialization_pattern, - &media_pattern); + av_dict_get(st->metadata, FILENAME, NULL, 0); + char *underscore_pos, *period_pos; + int ret; + if (!filename) + return AVERROR(EINVAL); + ret = split_filename(filename->value, &underscore_pos, &period_pos); if (ret) return ret; - avio_printf(s->pb, "\n", + *underscore_pos = '\0'; + avio_printf(pb, "\n", par->codec_type == AVMEDIA_TYPE_VIDEO ? "video" : "audio"); - avio_printf(s->pb, "pb, " timescale=\"1000\""); - avio_printf(s->pb, " duration=\"%d\"", w->chunk_duration); - avio_printf(s->pb, " media=\"%s\"", media_pattern); - avio_printf(s->pb, " startNumber=\"%d\"", w->chunk_start_index); - avio_printf(s->pb, " initialization=\"%s\"", initialization_pattern); - avio_printf(s->pb, "/>\n"); - av_free(initialization_pattern); - av_free(media_pattern); + avio_printf(pb, "chunk_duration); + avio_printf(pb, " media=\"%s_$RepresentationID$_$Number$.chk\"", + filename->value); + avio_printf(pb, " startNumber=\"%d\"", w->chunk_start_index); + avio_printf(pb, " initialization=\"%s_$RepresentationID$.hdr\"", + filename->value); + avio_printf(pb, "/>\n"); + *underscore_pos = '_'; } for (i = 0; i < as->nb_streams; i++) { - char *representation_id = NULL; + char buf[25], *representation_id = buf, *underscore_pos, *period_pos; + AVStream *st = s->streams[as->streams[i]]; int ret; if (w->is_live) { AVDictionaryEntry *filename = - av_dict_get(s->streams[as->streams[i]]->metadata, FILENAME, NULL, 0); + av_dict_get(st->metadata, FILENAME, NULL, 0); if (!filename) return AVERROR(EINVAL); - if (ret = parse_filename(filename->value, &representation_id, NULL, NULL)) + ret = split_filename(filename->value, &underscore_pos, &period_pos); + if (ret < 0) return ret; + representation_id = underscore_pos + 1; + *period_pos = '\0'; } else { - representation_id = av_asprintf("%d", w->representation_id++); - if (!representation_id) return AVERROR(ENOMEM); + snprintf(buf, sizeof(buf), "%d", w->representation_id++); } - ret = write_representation(s, s->streams[as->streams[i]], - representation_id, !width_in_as, + ret = write_representation(s, st, representation_id, !width_in_as, !height_in_as, !sample_rate_in_as); - av_free(representation_id); if (ret) return ret; + if (w->is_live) + *period_pos = '.'; } avio_printf(s->pb, "\n"); return 0; @@ -437,8 +414,13 @@ static int parse_adaptation_sets(AVFormatContext *s) } // syntax id=0,streams=0,1,2 id=1,streams=3,4 and so on state = new_set; - while (p < w->adaptation_sets + strlen(w->adaptation_sets)) { - if (state == new_set && *p == ' ') { + while (1) { + if (*p == '\0') { + if (state == new_set) + break; + else + return AVERROR(EINVAL); + } else if (state == new_set && *p == ' ') { p++; continue; } else if (state == new_set && !strncmp(p, "id=", 3)) {