libavformat/webvttdec.c

   1 /*
   2  * Copyright (c) 2012 Clément Bœsch
   3  *
   4  * This file is part of FFmpeg.
   5  *
   6  * FFmpeg is free software; you can redistribute it and/or
   7  * modify it under the terms of the GNU Lesser General Public
   8  * License as published by the Free Software Foundation; either
   9  * version 2.1 of the License, or (at your option) any later version.
  10  *
  11  * FFmpeg is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14  * Lesser General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU Lesser General Public
  17  * License along with FFmpeg; if not, write to the Free Software
  18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  19  */
  20
  21 /**
  22  * @file
  23  * WebVTT subtitle demuxer
  24  * @see https://www.w3.org/TR/webvtt1/
  25  */
  26
  27 #include "avformat.h"
  28 #include "demux.h"
  29 #include "internal.h"
  30 #include "subtitles.h"
  31 #include "libavutil/bprint.h"
  32 #include "libavutil/intreadwrite.h"
  33 #include "libavutil/opt.h"
  34
  35 typedef struct {
  36     const AVClass *class;
  37     FFDemuxSubtitlesQueue q;
  38     int kind;
  39 } WebVTTContext;
  40
  41 static int webvtt_probe(const AVProbeData *p)
  42 {
  43     const uint8_t *ptr = p->buf;
  44
  45     if (AV_RB24(ptr) == 0xEFBBBF)
  46         ptr += 3;  /* skip UTF-8 BOM */
  47     if (!strncmp(ptr, "WEBVTT", 6) &&
  48         (!ptr[6] || strchr("\n\r\t ", ptr[6])))
  49         return AVPROBE_SCORE_MAX;
  50     return 0;
  51 }
  52
  53 static int64_t read_ts(const char *s)
  54 {
  55     int hh, mm, ss, ms;
  56     if (sscanf(s, "%u:%u:%u.%u", &hh, &mm, &ss, &ms) == 4) return (hh*3600LL + mm*60LL + ss) * 1000LL + ms;
  57     if (sscanf(s,    "%u:%u.%u",      &mm, &ss, &ms) == 3) return (            mm*60LL + ss) * 1000LL + ms;
  58     return AV_NOPTS_VALUE;
  59 }
  60
  61 static int webvtt_read_header(AVFormatContext *s)
  62 {
  63     WebVTTContext *webvtt = s->priv_data;
  64     AVBPrint cue;
  65     int res = 0;
  66     AVStream *st = avformat_new_stream(s, NULL);
  67
  68     if (!st)
  69         return AVERROR(ENOMEM);
  70     avpriv_set_pts_info(st, 64, 1, 1000);
  71     st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
  72     st->codecpar->codec_id   = AV_CODEC_ID_WEBVTT;
  73     st->disposition |= webvtt->kind;
  74
  75     av_bprint_init(&cue,    0, AV_BPRINT_SIZE_UNLIMITED);
  76
  77     for (;;) {
  78         int i;
  79         int64_t pos;
  80         AVPacket *sub;
  81         const char *p, *identifier, *settings;
  82         size_t identifier_len, settings_len;
  83         int64_t ts_start, ts_end;
  84
  85         res = ff_subtitles_read_chunk(s->pb, &cue);
  86         if (res < 0)
  87             goto end;
  88
  89         if (!cue.len)
  90             break;
  91
  92         p = identifier = cue.str;
  93         pos = avio_tell(s->pb);
  94
  95         /* ignore header chunk */
  96         if (!strncmp(p, "\xEF\xBB\xBFWEBVTT", 9) ||
  97             !strncmp(p, "WEBVTT", 6) ||
  98             !strncmp(p, "STYLE", 5) ||
  99             !strncmp(p, "REGION", 6) ||
 100             !strncmp(p, "NOTE", 4))
 101             continue;
 102
 103         /* optional cue identifier (can be a number like in SRT or some kind of
 104          * chaptering id) */
 105         for (i = 0; p[i] && p[i] != '\n' && p[i] != '\r'; i++) {
 106             if (!strncmp(p + i, "-->", 3)) {
 107                 identifier = NULL;
 108                 break;
 109             }
 110         }
 111         if (!identifier)
 112             identifier_len = 0;
 113         else {
 114             identifier_len = strcspn(p, "\r\n");
 115             p += identifier_len;
 116             if (*p == '\r')
 117                 p++;
 118             if (*p == '\n')
 119                 p++;
 120         }
 121
 122         /* cue timestamps */
 123         if ((ts_start = read_ts(p)) == AV_NOPTS_VALUE)
 124             break;
 125         if (!(p = strstr(p, "-->")))
 126             break;
 127         p += 2;
 128         do p++; while (*p == ' ' || *p == '\t');
 129         if ((ts_end = read_ts(p)) == AV_NOPTS_VALUE)
 130             break;
 131
 132         /* optional cue settings */
 133         p += strcspn(p, "\n\r\t ");
 134         while (*p == '\t' || *p == ' ')
 135             p++;
 136         settings = p;
 137         settings_len = strcspn(p, "\r\n");
 138         p += settings_len;
 139         if (*p == '\r')
 140             p++;
 141         if (*p == '\n')
 142             p++;
 143
 144         /* create packet */
 145         sub = ff_subtitles_queue_insert(&webvtt->q, p, strlen(p), 0);
 146         if (!sub) {
 147             res = AVERROR(ENOMEM);
 148             goto end;
 149         }
 150         sub->pos = pos;
 151         sub->pts = ts_start;
 152         sub->duration = ts_end - ts_start;
 153
 154 #define SET_SIDE_DATA(name, type) do {                                  \
 155     if (name##_len) {                                                   \
 156         uint8_t *buf = av_packet_new_side_data(sub, type, name##_len);  \
 157         if (!buf) {                                                     \
 158             res = AVERROR(ENOMEM);                                      \
 159             goto end;                                                   \
 160         }                                                               \
 161         memcpy(buf, name, name##_len);                                  \
 162     }                                                                   \
 163 } while (0)
 164
 165         SET_SIDE_DATA(identifier, AV_PKT_DATA_WEBVTT_IDENTIFIER);
 166         SET_SIDE_DATA(settings,   AV_PKT_DATA_WEBVTT_SETTINGS);
 167     }
 168
 169     ff_subtitles_queue_finalize(s, &webvtt->q);
 170
 171 end:
 172     av_bprint_finalize(&cue,    NULL);
 173     return res;
 174 }
 175
 176 static int webvtt_read_packet(AVFormatContext *s, AVPacket *pkt)
 177 {
 178     WebVTTContext *webvtt = s->priv_data;
 179     return ff_subtitles_queue_read_packet(&webvtt->q, pkt);
 180 }
 181
 182 static int webvtt_read_seek(AVFormatContext *s, int stream_index,
 183                             int64_t min_ts, int64_t ts, int64_t max_ts, int flags)
 184 {
 185     WebVTTContext *webvtt = s->priv_data;
 186     return ff_subtitles_queue_seek(&webvtt->q, s, stream_index,
 187                                    min_ts, ts, max_ts, flags);
 188 }
 189
 190 static int webvtt_read_close(AVFormatContext *s)
 191 {
 192     WebVTTContext *webvtt = s->priv_data;
 193     ff_subtitles_queue_clean(&webvtt->q);
 194     return 0;
 195 }
 196
 197 #define OFFSET(x) offsetof(WebVTTContext, x)
 198 #define KIND_FLAGS AV_OPT_FLAG_SUBTITLE_PARAM|AV_OPT_FLAG_DECODING_PARAM
 199
 200 static const AVOption options[] = {
 201     { "kind", "Set kind of WebVTT track", OFFSET(kind), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, KIND_FLAGS, .unit = "webvtt_kind" },
 202         { "subtitles",    "WebVTT subtitles kind",    0, AV_OPT_TYPE_CONST, { .i64 = 0 },                           INT_MIN, INT_MAX, KIND_FLAGS, .unit = "webvtt_kind" },
 203         { "captions",     "WebVTT captions kind",     0, AV_OPT_TYPE_CONST, { .i64 = AV_DISPOSITION_CAPTIONS },     INT_MIN, INT_MAX, KIND_FLAGS, .unit = "webvtt_kind" },
 204         { "descriptions", "WebVTT descriptions kind", 0, AV_OPT_TYPE_CONST, { .i64 = AV_DISPOSITION_DESCRIPTIONS }, INT_MIN, INT_MAX, KIND_FLAGS, .unit = "webvtt_kind" },
 205         { "metadata",     "WebVTT metadata kind",     0, AV_OPT_TYPE_CONST, { .i64 = AV_DISPOSITION_METADATA },     INT_MIN, INT_MAX, KIND_FLAGS, .unit = "webvtt_kind" },
 206     { NULL }
 207 };
 208
 209 static const AVClass webvtt_demuxer_class = {
 210     .class_name  = "WebVTT demuxer",
 211     .item_name   = av_default_item_name,
 212     .option      = options,
 213     .version     = LIBAVUTIL_VERSION_INT,
 214 };
 215
 216 const FFInputFormat ff_webvtt_demuxer = {
 217     .p.name         = "webvtt",
 218     .p.long_name    = NULL_IF_CONFIG_SMALL("WebVTT subtitle"),
 219     .p.mime_type    = "text/vtt",
 220     .p.extensions   = "vtt,webvtt",
 221     .p.priv_class   = &webvtt_demuxer_class,
 222     .priv_data_size = sizeof(WebVTTContext),
 223     .flags_internal = FF_INFMT_FLAG_INIT_CLEANUP,
 224     .read_probe     = webvtt_probe,
 225     .read_header    = webvtt_read_header,
 226     .read_packet    = webvtt_read_packet,
 227     .read_seek2     = webvtt_read_seek,
 228     .read_close     = webvtt_read_close,
 229 };