libavformat/hdsenc.c

   1 /*
   2  * Live HDS fragmenter
   3  * Copyright (c) 2013 Martin Storsjo
   4  *
   5  * This file is part of FFmpeg.
   6  *
   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.
  11  *
  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.
  16  *
  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
  20  */
  21
  22 #include "config.h"
  23 #if HAVE_UNISTD_H
  24 #include <unistd.h>
  25 #endif
  26
  27 #include "avformat.h"
  28 #include "internal.h"
  29 #include "mux.h"
  30 #include "os_support.h"
  31
  32 #include "libavutil/avstring.h"
  33 #include "libavutil/base64.h"
  34 #include "libavutil/intreadwrite.h"
  35 #include "libavutil/mathematics.h"
  36 #include "libavutil/mem.h"
  37 #include "libavutil/opt.h"
  38
  39 typedef struct Fragment {
  40     char file[1024];
  41     int64_t start_time, duration;
  42     int n;
  43 } Fragment;
  44
  45 typedef struct OutputStream {
  46     int bitrate;
  47     int first_stream;
  48     AVFormatContext *ctx;
  49     int ctx_inited;
  50     uint8_t iobuf[32768];
  51     char temp_filename[1024];
  52     int64_t frag_start_ts, last_ts;
  53     AVIOContext *out;
  54     int packets_written;
  55     int nb_fragments, fragments_size, fragment_index;
  56     Fragment **fragments;
  57
  58     int has_audio, has_video;
  59
  60     uint8_t *metadata;
  61     int metadata_size;
  62
  63     uint8_t *extra_packets[2];
  64     int extra_packet_sizes[2];
  65     int nb_extra_packets;
  66 } OutputStream;
  67
  68 typedef struct HDSContext {
  69     const AVClass *class;  /* Class for private options. */
  70     int window_size;
  71     int extra_window_size;
  72     int min_frag_duration;
  73     int remove_at_exit;
  74
  75     OutputStream *streams;
  76     int nb_streams;
  77 } HDSContext;
  78
  79 static int parse_header(OutputStream *os, const uint8_t *buf, int buf_size)
  80 {
  81     if (buf_size < 13)
  82         return AVERROR_INVALIDDATA;
  83     if (memcmp(buf, "FLV", 3))
  84         return AVERROR_INVALIDDATA;
  85     buf      += 13;
  86     buf_size -= 13;
  87     while (buf_size >= 11 + 4) {
  88         int type = buf[0];
  89         int size = AV_RB24(&buf[1]) + 11 + 4;
  90         if (size > buf_size)
  91             return AVERROR_INVALIDDATA;
  92         if (type == 8 || type == 9) {
  93             if (os->nb_extra_packets >= FF_ARRAY_ELEMS(os->extra_packets))
  94                 return AVERROR_INVALIDDATA;
  95             os->extra_packet_sizes[os->nb_extra_packets] = size;
  96             os->extra_packets[os->nb_extra_packets] = av_memdup(buf, size);
  97             if (!os->extra_packets[os->nb_extra_packets])
  98                 return AVERROR(ENOMEM);
  99             os->nb_extra_packets++;
 100         } else if (type == 0x12) {
 101             if (os->metadata)
 102                 return AVERROR_INVALIDDATA;
 103             os->metadata_size = size - 11 - 4;
 104             os->metadata      = av_memdup(buf + 11, os->metadata_size);
 105             if (!os->metadata)
 106                 return AVERROR(ENOMEM);
 107         }
 108         buf      += size;
 109         buf_size -= size;
 110     }
 111     if (!os->metadata)
 112         return AVERROR_INVALIDDATA;
 113     return 0;
 114 }
 115
 116 static int hds_write(void *opaque, const uint8_t *buf, int buf_size)
 117 {
 118     OutputStream *os = opaque;
 119     if (os->out) {
 120         avio_write(os->out, buf, buf_size);
 121     } else {
 122         if (!os->metadata_size) {
 123             int ret;
 124             // Assuming the IO buffer is large enough to fit the
 125             // FLV header and all metadata and extradata packets
 126             if ((ret = parse_header(os, buf, buf_size)) < 0)
 127                 return ret;
 128         }
 129     }
 130     return buf_size;
 131 }
 132
 133 static void hds_free(AVFormatContext *s)
 134 {
 135     HDSContext *c = s->priv_data;
 136     int i, j;
 137     if (!c->streams)
 138         return;
 139     for (i = 0; i < s->nb_streams; i++) {
 140         OutputStream *os = &c->streams[i];
 141         if (os->out)
 142             ff_format_io_close(s, &os->out);
 143         if (os->ctx && os->ctx_inited)
 144             av_write_trailer(os->ctx);
 145         if (os->ctx)
 146             avio_context_free(&os->ctx->pb);
 147         avformat_free_context(os->ctx);
 148         av_freep(&os->metadata);
 149         for (j = 0; j < os->nb_extra_packets; j++)
 150             av_freep(&os->extra_packets[j]);
 151         for (j = 0; j < os->nb_fragments; j++)
 152             av_freep(&os->fragments[j]);
 153         av_freep(&os->fragments);
 154     }
 155     av_freep(&c->streams);
 156 }
 157
 158 static int write_manifest(AVFormatContext *s, int final)
 159 {
 160     HDSContext *c = s->priv_data;
 161     AVIOContext *out;
 162     char filename[1024], temp_filename[1024];
 163     int ret, i;
 164     double duration = 0;
 165
 166     if (c->nb_streams > 0)
 167         duration = c->streams[0].last_ts * av_q2d(s->streams[0]->time_base);
 168
 169     snprintf(filename, sizeof(filename), "%s/index.f4m", s->url);
 170     snprintf(temp_filename, sizeof(temp_filename), "%s/index.f4m.tmp", s->url);
 171     ret = s->io_open(s, &out, temp_filename, AVIO_FLAG_WRITE, NULL);
 172     if (ret < 0) {
 173         av_log(s, AV_LOG_ERROR, "Unable to open %s for writing\n", temp_filename);
 174         return ret;
 175     }
 176     avio_printf(out, "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n");
 177     avio_printf(out, "<manifest xmlns=\"http://ns.adobe.com/f4m/1.0\">\n");
 178     avio_printf(out, "\t<id>%s</id>\n", av_basename(s->url));
 179     avio_printf(out, "\t<streamType>%s</streamType>\n",
 180                      final ? "recorded" : "live");
 181     avio_printf(out, "\t<deliveryType>streaming</deliveryType>\n");
 182     if (final)
 183         avio_printf(out, "\t<duration>%f</duration>\n", duration);
 184     for (i = 0; i < c->nb_streams; i++) {
 185         OutputStream *os = &c->streams[i];
 186         int b64_size = AV_BASE64_SIZE(os->metadata_size);
 187         char *base64 = av_malloc(b64_size);
 188         if (!base64) {
 189             ff_format_io_close(s, &out);
 190             return AVERROR(ENOMEM);
 191         }
 192         av_base64_encode(base64, b64_size, os->metadata, os->metadata_size);
 193
 194         avio_printf(out, "\t<bootstrapInfo profile=\"named\" url=\"stream%d.abst\" id=\"bootstrap%d\" />\n", i, i);
 195         avio_printf(out, "\t<media bitrate=\"%d\" url=\"stream%d\" bootstrapInfoId=\"bootstrap%d\">\n", os->bitrate/1000, i, i);
 196         avio_printf(out, "\t\t<metadata>%s</metadata>\n", base64);
 197         avio_printf(out, "\t</media>\n");
 198         av_free(base64);
 199     }
 200     avio_printf(out, "</manifest>\n");
 201     avio_flush(out);
 202     ff_format_io_close(s, &out);
 203     return ff_rename(temp_filename, filename, s);
 204 }
 205
 206 static void update_size(AVIOContext *out, int64_t pos)
 207 {
 208     int64_t end = avio_tell(out);
 209     avio_seek(out, pos, SEEK_SET);
 210     avio_wb32(out, end - pos);
 211     avio_seek(out, end, SEEK_SET);
 212 }
 213
 214 /* Note, the .abst files need to be served with the "binary/octet"
 215  * mime type, otherwise at least the OSMF player can easily fail
 216  * with "stream not found" when polling for the next fragment. */
 217 static int write_abst(AVFormatContext *s, OutputStream *os, int final)
 218 {
 219     HDSContext *c = s->priv_data;
 220     AVIOContext *out;
 221     char filename[1024], temp_filename[1024];
 222     int i, ret;
 223     int64_t asrt_pos, afrt_pos;
 224     int start = 0, fragments;
 225     int index = s->streams[os->first_stream]->id;
 226     int64_t cur_media_time = 0;
 227     if (c->window_size)
 228         start = FFMAX(os->nb_fragments - c->window_size, 0);
 229     fragments = os->nb_fragments - start;
 230     if (final)
 231         cur_media_time = os->last_ts;
 232     else if (os->nb_fragments)
 233         cur_media_time = os->fragments[os->nb_fragments - 1]->start_time;
 234
 235     snprintf(filename, sizeof(filename),
 236              "%s/stream%d.abst", s->url, index);
 237     snprintf(temp_filename, sizeof(temp_filename),
 238              "%s/stream%d.abst.tmp", s->url, index);
 239     ret = s->io_open(s, &out, temp_filename, AVIO_FLAG_WRITE, NULL);
 240     if (ret < 0) {
 241         av_log(s, AV_LOG_ERROR, "Unable to open %s for writing\n", temp_filename);
 242         return ret;
 243     }
 244     avio_wb32(out, 0); // abst size
 245     avio_wl32(out, MKTAG('a','b','s','t'));
 246     avio_wb32(out, 0); // version + flags
 247     avio_wb32(out, os->fragment_index - 1); // BootstrapinfoVersion
 248     avio_w8(out, final ? 0 : 0x20); // profile, live, update
 249     avio_wb32(out, 1000); // timescale
 250     avio_wb64(out, cur_media_time);
 251     avio_wb64(out, 0); // SmpteTimeCodeOffset
 252     avio_w8(out, 0); // MovieIdentifer (null string)
 253     avio_w8(out, 0); // ServerEntryCount
 254     avio_w8(out, 0); // QualityEntryCount
 255     avio_w8(out, 0); // DrmData (null string)
 256     avio_w8(out, 0); // MetaData (null string)
 257     avio_w8(out, 1); // SegmentRunTableCount
 258     asrt_pos = avio_tell(out);
 259     avio_wb32(out, 0); // asrt size
 260     avio_wl32(out, MKTAG('a','s','r','t'));
 261     avio_wb32(out, 0); // version + flags
 262     avio_w8(out, 0); // QualityEntryCount
 263     avio_wb32(out, 1); // SegmentRunEntryCount
 264     avio_wb32(out, 1); // FirstSegment
 265     avio_wb32(out, final ? (os->fragment_index - 1) : 0xffffffff); // FragmentsPerSegment
 266     update_size(out, asrt_pos);
 267     avio_w8(out, 1); // FragmentRunTableCount
 268     afrt_pos = avio_tell(out);
 269     avio_wb32(out, 0); // afrt size
 270     avio_wl32(out, MKTAG('a','f','r','t'));
 271     avio_wb32(out, 0); // version + flags
 272     avio_wb32(out, 1000); // timescale
 273     avio_w8(out, 0); // QualityEntryCount
 274     avio_wb32(out, fragments); // FragmentRunEntryCount
 275     for (i = start; i < os->nb_fragments; i++) {
 276         avio_wb32(out, os->fragments[i]->n);
 277         avio_wb64(out, os->fragments[i]->start_time);
 278         avio_wb32(out, os->fragments[i]->duration);
 279     }
 280     update_size(out, afrt_pos);
 281     update_size(out, 0);
 282     ff_format_io_close(s, &out);
 283     return ff_rename(temp_filename, filename, s);
 284 }
 285
 286 static int init_file(AVFormatContext *s, OutputStream *os, int64_t start_ts)
 287 {
 288     int ret, i;
 289     ret = s->io_open(s, &os->out, os->temp_filename, AVIO_FLAG_WRITE, NULL);
 290     if (ret < 0)
 291         return ret;
 292     avio_wb32(os->out, 0);
 293     avio_wl32(os->out, MKTAG('m','d','a','t'));
 294     for (i = 0; i < os->nb_extra_packets; i++) {
 295         AV_WB24(os->extra_packets[i] + 4, start_ts);
 296         os->extra_packets[i][7] = (start_ts >> 24) & 0x7f;
 297         avio_write(os->out, os->extra_packets[i], os->extra_packet_sizes[i]);
 298     }
 299     return 0;
 300 }
 301
 302 static void close_file(AVFormatContext *s, OutputStream *os)
 303 {
 304     int64_t pos = avio_tell(os->out);
 305     avio_seek(os->out, 0, SEEK_SET);
 306     avio_wb32(os->out, pos);
 307     avio_flush(os->out);
 308     ff_format_io_close(s, &os->out);
 309 }
 310
 311 static int hds_write_header(AVFormatContext *s)
 312 {
 313     HDSContext *c = s->priv_data;
 314     const AVOutputFormat *oformat;
 315     int ret = 0, i;
 316
 317     if (mkdir(s->url, 0777) == -1 && errno != EEXIST) {
 318         av_log(s, AV_LOG_ERROR , "Failed to create directory %s\n", s->url);
 319         return AVERROR(errno);
 320     }
 321
 322     oformat = av_guess_format("flv", NULL, NULL);
 323     if (!oformat) {
 324         return AVERROR_MUXER_NOT_FOUND;
 325     }
 326
 327     c->streams = av_calloc(s->nb_streams, sizeof(*c->streams));
 328     if (!c->streams) {
 329         return AVERROR(ENOMEM);
 330     }
 331
 332     for (i = 0; i < s->nb_streams; i++) {
 333         OutputStream *os = &c->streams[c->nb_streams];
 334         AVFormatContext *ctx;
 335         AVStream *st = s->streams[i];
 336
 337         if (!st->codecpar->bit_rate) {
 338             av_log(s, AV_LOG_ERROR, "No bit rate set for stream %d\n", i);
 339             return AVERROR(EINVAL);
 340         }
 341         if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
 342             if (os->has_video) {
 343                 c->nb_streams++;
 344                 os++;
 345             }
 346             os->has_video = 1;
 347         } else if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
 348             if (os->has_audio) {
 349                 c->nb_streams++;
 350                 os++;
 351             }
 352             os->has_audio = 1;
 353         } else {
 354             av_log(s, AV_LOG_ERROR, "Unsupported stream type in stream %d\n", i);
 355             return AVERROR(EINVAL);
 356         }
 357         os->bitrate += s->streams[i]->codecpar->bit_rate;
 358
 359         if (!os->ctx) {
 360             os->first_stream = i;
 361             ctx = avformat_alloc_context();
 362             if (!ctx) {
 363                 return AVERROR(ENOMEM);
 364             }
 365             os->ctx = ctx;
 366             ctx->oformat = oformat;
 367             ctx->interrupt_callback = s->interrupt_callback;
 368             ctx->flags = s->flags;
 369
 370             ctx->pb = avio_alloc_context(os->iobuf, sizeof(os->iobuf),
 371                                          1, os,
 372                                          NULL, hds_write, NULL);
 373             if (!ctx->pb) {
 374                 return AVERROR(ENOMEM);
 375             }
 376         } else {
 377             ctx = os->ctx;
 378         }
 379         s->streams[i]->id = c->nb_streams;
 380
 381         if (!(st = avformat_new_stream(ctx, NULL))) {
 382             return AVERROR(ENOMEM);
 383         }
 384         avcodec_parameters_copy(st->codecpar, s->streams[i]->codecpar);
 385         st->codecpar->codec_tag = 0;
 386         st->sample_aspect_ratio = s->streams[i]->sample_aspect_ratio;
 387         st->time_base = s->streams[i]->time_base;
 388     }
 389     if (c->streams[c->nb_streams].ctx)
 390         c->nb_streams++;
 391
 392     for (i = 0; i < c->nb_streams; i++) {
 393         OutputStream *os = &c->streams[i];
 394         int j;
 395         if ((ret = avformat_write_header(os->ctx, NULL)) < 0) {
 396              return ret;
 397         }
 398         os->ctx_inited = 1;
 399         avio_flush(os->ctx->pb);
 400         for (j = 0; j < os->ctx->nb_streams; j++)
 401             s->streams[os->first_stream + j]->time_base = os->ctx->streams[j]->time_base;
 402
 403         snprintf(os->temp_filename, sizeof(os->temp_filename),
 404                  "%s/stream%d_temp", s->url, i);
 405         ret = init_file(s, os, 0);
 406         if (ret < 0)
 407             return ret;
 408
 409         if (!os->has_video && c->min_frag_duration <= 0) {
 410             av_log(s, AV_LOG_WARNING,
 411                    "No video stream in output stream %d and no min frag duration set\n", i);
 412         }
 413         os->fragment_index = 1;
 414         write_abst(s, os, 0);
 415     }
 416     ret = write_manifest(s, 0);
 417
 418     return ret;
 419 }
 420
 421 static int add_fragment(OutputStream *os, const char *file,
 422                         int64_t start_time, int64_t duration)
 423 {
 424     Fragment *frag;
 425     if (duration == 0)
 426         duration = 1;
 427     if (os->nb_fragments >= os->fragments_size) {
 428         int ret;
 429         os->fragments_size = (os->fragments_size + 1) * 2;
 430         if ((ret = av_reallocp_array(&os->fragments, os->fragments_size,
 431                                      sizeof(*os->fragments))) < 0) {
 432             os->fragments_size = 0;
 433             os->nb_fragments   = 0;
 434             return ret;
 435         }
 436     }
 437     frag = av_mallocz(sizeof(*frag));
 438     if (!frag)
 439         return AVERROR(ENOMEM);
 440     av_strlcpy(frag->file, file, sizeof(frag->file));
 441     frag->start_time = start_time;
 442     frag->duration   = duration;
 443     frag->n          = os->fragment_index;
 444     os->fragments[os->nb_fragments++] = frag;
 445     os->fragment_index++;
 446     return 0;
 447 }
 448
 449 static int hds_flush(AVFormatContext *s, OutputStream *os, int final,
 450                      int64_t end_ts)
 451 {
 452     HDSContext *c = s->priv_data;
 453     int i, ret = 0;
 454     char target_filename[1024];
 455     int index = s->streams[os->first_stream]->id;
 456
 457     if (!os->packets_written)
 458         return 0;
 459
 460     avio_flush(os->ctx->pb);
 461     os->packets_written = 0;
 462     close_file(s, os);
 463
 464     snprintf(target_filename, sizeof(target_filename),
 465              "%s/stream%dSeg1-Frag%d", s->url, index, os->fragment_index);
 466     ret = ff_rename(os->temp_filename, target_filename, s);
 467     if (ret < 0)
 468         return ret;
 469     add_fragment(os, target_filename, os->frag_start_ts, end_ts - os->frag_start_ts);
 470
 471     if (!final) {
 472         ret = init_file(s, os, end_ts);
 473         if (ret < 0)
 474             return ret;
 475     }
 476
 477     if (c->window_size || (final && c->remove_at_exit)) {
 478         int remove = os->nb_fragments - c->window_size - c->extra_window_size;
 479         if (final && c->remove_at_exit)
 480             remove = os->nb_fragments;
 481         if (remove > 0) {
 482             for (i = 0; i < remove; i++) {
 483                 unlink(os->fragments[i]->file);
 484                 av_freep(&os->fragments[i]);
 485             }
 486             os->nb_fragments -= remove;
 487             memmove(os->fragments, os->fragments + remove,
 488                     os->nb_fragments * sizeof(*os->fragments));
 489         }
 490     }
 491
 492     if (ret >= 0)
 493         ret = write_abst(s, os, final);
 494     return ret;
 495 }
 496
 497 static int hds_write_packet(AVFormatContext *s, AVPacket *pkt)
 498 {
 499     HDSContext *c = s->priv_data;
 500     AVStream *st = s->streams[pkt->stream_index];
 501     FFStream *const sti = ffstream(st);
 502     OutputStream *os = &c->streams[s->streams[pkt->stream_index]->id];
 503     int64_t end_dts = os->fragment_index * (int64_t)c->min_frag_duration;
 504     int ret;
 505
 506     if (sti->first_dts == AV_NOPTS_VALUE)
 507         sti->first_dts = pkt->dts;
 508
 509     if ((!os->has_video || st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) &&
 510         av_compare_ts(pkt->dts - sti->first_dts, st->time_base,
 511                       end_dts, AV_TIME_BASE_Q) >= 0 &&
 512         pkt->flags & AV_PKT_FLAG_KEY && os->packets_written) {
 513
 514         if ((ret = hds_flush(s, os, 0, pkt->dts)) < 0)
 515             return ret;
 516     }
 517
 518     // Note, these fragment start timestamps, that represent a whole
 519     // OutputStream, assume all streams in it have the same time base.
 520     if (!os->packets_written)
 521         os->frag_start_ts = pkt->dts;
 522     os->last_ts = pkt->dts;
 523
 524     os->packets_written++;
 525     return ff_write_chained(os->ctx, pkt->stream_index - os->first_stream, pkt, s, 0);
 526 }
 527
 528 static int hds_write_trailer(AVFormatContext *s)
 529 {
 530     HDSContext *c = s->priv_data;
 531     int i;
 532
 533     for (i = 0; i < c->nb_streams; i++)
 534         hds_flush(s, &c->streams[i], 1, c->streams[i].last_ts);
 535     write_manifest(s, 1);
 536
 537     if (c->remove_at_exit) {
 538         char filename[1024];
 539         snprintf(filename, sizeof(filename), "%s/index.f4m", s->url);
 540         unlink(filename);
 541         for (i = 0; i < c->nb_streams; i++) {
 542             snprintf(filename, sizeof(filename), "%s/stream%d.abst", s->url, i);
 543             unlink(filename);
 544         }
 545         rmdir(s->url);
 546     }
 547
 548     return 0;
 549 }
 550
 551 #define OFFSET(x) offsetof(HDSContext, x)
 552 #define E AV_OPT_FLAG_ENCODING_PARAM
 553 static const AVOption options[] = {
 554     { "window_size", "number of fragments kept in the manifest", OFFSET(window_size), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, E },
 555     { "extra_window_size", "number of fragments kept outside of the manifest before removing from disk", OFFSET(extra_window_size), AV_OPT_TYPE_INT, { .i64 = 5 }, 0, INT_MAX, E },
 556     { "min_frag_duration", "minimum fragment duration (in microseconds)", OFFSET(min_frag_duration), AV_OPT_TYPE_INT64, { .i64 = 10000000 }, 0, INT_MAX, E },
 557     { "remove_at_exit", "remove all fragments when finished", OFFSET(remove_at_exit), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, E },
 558     { NULL },
 559 };
 560
 561 static const AVClass hds_class = {
 562     .class_name = "HDS muxer",
 563     .item_name  = av_default_item_name,
 564     .option     = options,
 565     .version    = LIBAVUTIL_VERSION_INT,
 566 };
 567
 568 const FFOutputFormat ff_hds_muxer = {
 569     .p.name         = "hds",
 570     .p.long_name    = NULL_IF_CONFIG_SMALL("HDS Muxer"),
 571     .p.audio_codec  = AV_CODEC_ID_AAC,
 572     .p.video_codec  = AV_CODEC_ID_H264,
 573     .p.flags        = AVFMT_GLOBALHEADER | AVFMT_NOFILE,
 574     .p.priv_class   = &hds_class,
 575     .priv_data_size = sizeof(HDSContext),
 576     .write_header   = hds_write_header,
 577     .write_packet   = hds_write_packet,
 578     .write_trailer  = hds_write_trailer,
 579     .deinit         = hds_free,
 580 };