libavformat/amvenc.c

   1 /*
   2  * AMV muxer
   3  *
   4  * Copyright (C) 2020 Zane van Iperen (zane@zanevaniperen.com)
   5  *
   6  * This file is part of FFmpeg.
   7  *
   8  * FFmpeg is free software; you can redistribute it and/or
   9  * modify it under the terms of the GNU Lesser General Public
  10  * License as published by the Free Software Foundation; either
  11  * version 2.1 of the License, or (at your option) any later version.
  12  *
  13  * FFmpeg is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  16  * Lesser General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU Lesser General Public
  19  * License along with FFmpeg; if not, write to the Free Software
  20  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  21  */
  22 #include "avformat.h"
  23 #include "mux.h"
  24 #include "riff.h"
  25 #include "internal.h"
  26 #include "avio_internal.h"
  27 #include "libavutil/intreadwrite.h"
  28 #include "libavutil/avassert.h"
  29
  30 /*
  31  * Things to note:
  32  * - AMV is a hard-coded (and broken) subset of AVI. It's not worth sullying the
  33  *   existing AVI muxer with its filth.
  34  * - No separate demuxer as the existing AVI demuxer can handle these.
  35  * - The sizes of certain tags are deliberately set to 0 as some players break
  36  *   when they're set correctly. Ditto with some header fields.
  37  * - There is no index.
  38  * - Players are **very** sensitive to the frame order and sizes.
  39  *   - Frames must be strictly interleaved as V-A, any V-V or A-A will
  40  *     cause crashes.
  41  *   - Variable video frame sizes seem to be handled fine.
  42  *   - Variable audio frame sizes cause crashes.
  43  *   - If audio is shorter than video, it's padded with silence.
  44  *   - If video is shorter than audio, the most recent frame is repeated.
  45  */
  46
  47 #define AMV_STREAM_COUNT     2
  48 #define AMV_STREAM_VIDEO     0
  49 #define AMV_STREAM_AUDIO     1
  50 #define AMV_VIDEO_STRH_SIZE 56
  51 #define AMV_VIDEO_STRF_SIZE 36
  52 #define AMV_AUDIO_STRH_SIZE 48
  53 #define AMV_AUDIO_STRF_SIZE 20 /* sizeof(WAVEFORMATEX) + 2 */
  54
  55 typedef struct AMVContext
  56 {
  57     int64_t riff_start;
  58     int64_t movi_list;
  59     int64_t offset_duration;
  60     int     last_stream;
  61
  62     int32_t us_per_frame; /* Microseconds per frame.         */
  63
  64     int32_t aframe_size;  /* Expected audio frame size.      */
  65     int32_t ablock_align; /* Expected audio block align.     */
  66     AVPacket *apad;       /* Dummy audio packet for padding; not owned by us. */
  67     AVPacket *vpad;       /* Most recent video frame, for padding. */
  68
  69     /*
  70      * Cumulative PTS values for each stream, used for the final
  71      * duration calculation.
  72      */
  73     int64_t lastpts[AMV_STREAM_COUNT];
  74 } AMVContext;
  75
  76 /* ff_{start,end}_tag(), but sets the size to 0. */
  77 static int64_t amv_start_tag(AVIOContext *pb, const char *tag)
  78 {
  79     ffio_wfourcc(pb, tag);
  80     avio_wl32(pb, 0);
  81     return avio_tell(pb);
  82 }
  83
  84 static void amv_end_tag(AVIOContext *pb, int64_t start)
  85 {
  86     int64_t pos;
  87     av_assert0((start&1) == 0);
  88
  89     pos = avio_tell(pb);
  90     if (pos & 1)
  91         avio_w8(pb, 0);
  92 }
  93
  94 static av_cold int amv_init(AVFormatContext *s)
  95 {
  96     AMVContext *amv = s->priv_data;
  97     AVStream   *vst, *ast;
  98     int ret;
  99
 100     amv->last_stream  = -1;
 101
 102     if (s->nb_streams != AMV_STREAM_COUNT) {
 103         av_log(s, AV_LOG_ERROR, "AMV files only support 2 streams\n");
 104         return AVERROR(EINVAL);
 105     }
 106
 107     vst = s->streams[AMV_STREAM_VIDEO];
 108     ast = s->streams[AMV_STREAM_AUDIO];
 109
 110     if (vst->codecpar->codec_id != AV_CODEC_ID_AMV) {
 111         av_log(s, AV_LOG_ERROR, "First AMV stream must be %s\n",
 112                 avcodec_get_name(AV_CODEC_ID_AMV));
 113         return AVERROR(EINVAL);
 114     }
 115
 116     av_assert1(ast->codecpar->codec_id == AV_CODEC_ID_ADPCM_IMA_AMV);
 117
 118     /* These files are broken-enough as they are. They shouldn't be streamed. */
 119     if (!(s->pb->seekable & AVIO_SEEKABLE_NORMAL)) {
 120         av_log(s, AV_LOG_ERROR, "Stream not seekable, unable to write output file\n");
 121         return AVERROR(EINVAL);
 122     }
 123
 124     amv->us_per_frame = av_rescale(AV_TIME_BASE, vst->time_base.num, vst->time_base.den);
 125     amv->aframe_size  = av_rescale(ast->codecpar->sample_rate, amv->us_per_frame, AV_TIME_BASE);
 126     amv->ablock_align = 8 + (FFALIGN(amv->aframe_size, 2) / 2);
 127
 128     av_log(s, AV_LOG_TRACE, "us_per_frame = %d\n", amv->us_per_frame);
 129     av_log(s, AV_LOG_TRACE, "aframe_size  = %d\n", amv->aframe_size);
 130     av_log(s, AV_LOG_TRACE, "ablock_align = %d\n", amv->ablock_align);
 131
 132     /*
 133      * Bail if the framerate's too high. Prevents the audio frame size from
 134      * getting too small. 63fps is the closest value to 60fps that divides
 135      * cleanly, so cap it there.
 136      */
 137     if (amv->us_per_frame < 15873) {
 138         av_log(s, AV_LOG_ERROR, "Refusing to mux >63fps video\n");
 139         return AVERROR(EINVAL);
 140     }
 141
 142     /*
 143      * frame_size will be set if coming from the encoder.
 144      * Make sure the its been configured correctly. The audio frame duration
 145      * needs to match that of the video.
 146      */
 147     if (ast->codecpar->frame_size) {
 148         AVCodecParameters *par = ast->codecpar;
 149         int bad = 0;
 150
 151         if (par->frame_size != amv->aframe_size) {
 152             av_log(s, AV_LOG_ERROR, "Invalid audio frame size. Got %d, wanted %d\n",
 153                    par->frame_size, amv->aframe_size);
 154             bad = 1;
 155         }
 156
 157         if (par->block_align != amv->ablock_align) {
 158             av_log(s, AV_LOG_ERROR, "Invalid audio block align. Got %d, wanted %d\n",
 159                    par->block_align, amv->ablock_align);
 160             bad = 1;
 161         }
 162
 163         if (bad) {
 164             av_log(s, AV_LOG_ERROR, "Try -block_size %d\n", amv->aframe_size);
 165             return AVERROR(EINVAL);
 166         }
 167
 168         if (ast->codecpar->sample_rate % amv->aframe_size) {
 169             av_log(s, AV_LOG_ERROR, "Audio sample rate not a multiple of the frame size.\n"
 170                 "Please change video frame rate. Suggested rates: 10,14,15,18,21,25,30\n");
 171             return AVERROR(EINVAL);
 172         }
 173     } else {
 174         /* If remuxing from the same source, then this will match the video. */
 175         int32_t aus = av_rescale(AV_TIME_BASE, ast->time_base.num, ast->time_base.den);
 176         if (aus != amv->us_per_frame) {
 177             av_log(s, AV_LOG_ERROR, "Cannot remux streams with a different time base\n");
 178             return AVERROR(EINVAL);
 179         }
 180     }
 181
 182     /* Allocate and fill dummy packet so we can pad the audio. */
 183     amv->apad = ffformatcontext(s)->pkt;
 184     if ((ret = av_new_packet(amv->apad, amv->ablock_align)) < 0) {
 185         return ret;
 186     }
 187
 188     amv->apad->stream_index = AMV_STREAM_AUDIO;
 189     memset(amv->apad->data, 0, amv->ablock_align);
 190     AV_WL32(amv->apad->data + 4, amv->aframe_size);
 191
 192     amv->vpad = av_packet_alloc();
 193     if (!amv->vpad) {
 194         return AVERROR(ENOMEM);
 195     }
 196     amv->vpad->stream_index = AMV_STREAM_VIDEO;
 197     amv->vpad->duration     = 1;
 198     return 0;
 199 }
 200
 201 static void amv_deinit(AVFormatContext *s)
 202 {
 203     AMVContext *amv = s->priv_data;
 204
 205     av_packet_free(&amv->vpad);
 206 }
 207
 208 static void amv_write_vlist(AVFormatContext *s, AVCodecParameters *par)
 209 {
 210     int64_t tag_list, tag_str;
 211
 212     av_assert0(par->codec_id == AV_CODEC_ID_AMV);
 213
 214     tag_list = amv_start_tag(s->pb, "LIST");
 215     ffio_wfourcc(s->pb, "strl");
 216     tag_str = ff_start_tag(s->pb, "strh");
 217     ffio_fill(s->pb, 0, AMV_VIDEO_STRH_SIZE);
 218     ff_end_tag(s->pb, tag_str);
 219
 220     tag_str = ff_start_tag(s->pb, "strf");
 221     ffio_fill(s->pb, 0, AMV_VIDEO_STRF_SIZE);
 222     ff_end_tag(s->pb, tag_str);
 223
 224     amv_end_tag(s->pb, tag_list);
 225 }
 226
 227 static void amv_write_alist(AVFormatContext *s, AVCodecParameters *par)
 228 {
 229     uint8_t buf[AMV_AUDIO_STRF_SIZE];
 230     AVIOContext *pb = s->pb;
 231     int64_t tag_list, tag_str;
 232
 233     av_assert0(par->codec_id == AV_CODEC_ID_ADPCM_IMA_AMV);
 234
 235     tag_list = amv_start_tag(pb, "LIST");
 236     ffio_wfourcc(pb, "strl");
 237     tag_str = ff_start_tag(pb, "strh");
 238     ffio_fill(s->pb, 0, AMV_AUDIO_STRH_SIZE);
 239     ff_end_tag(pb, tag_str);
 240
 241     /* Bodge an (incorrect) WAVEFORMATEX (+2 pad bytes) */
 242     tag_str = ff_start_tag(pb, "strf");
 243     AV_WL16(buf +  0, 1);
 244     AV_WL16(buf +  2, par->ch_layout.nb_channels);
 245     AV_WL32(buf +  4, par->sample_rate);
 246     AV_WL32(buf +  8, par->sample_rate * par->ch_layout.nb_channels * 2);
 247     AV_WL16(buf + 12, 2);
 248     AV_WL16(buf + 14, 16);
 249     AV_WL16(buf + 16, 0);
 250     AV_WL16(buf + 18, 0);
 251     avio_write(pb, buf, AMV_AUDIO_STRF_SIZE);
 252     ff_end_tag(pb, tag_str);
 253
 254     amv_end_tag(pb, tag_list);
 255 }
 256
 257 static int amv_write_header(AVFormatContext *s)
 258 {
 259     AMVContext *amv = s->priv_data;
 260     AVIOContext *pb = s->pb;
 261     AVStream *vst   = s->streams[AMV_STREAM_VIDEO];
 262     AVStream *ast   = s->streams[AMV_STREAM_AUDIO];
 263     uint8_t amvh[56] = {0};
 264     int64_t list1;
 265
 266     amv->riff_start = amv_start_tag(pb, "RIFF");
 267     ffio_wfourcc(pb, "AMV ");
 268     list1 = amv_start_tag(pb, "LIST");
 269     ffio_wfourcc(pb, "hdrl");
 270
 271     ffio_wfourcc(pb, "amvh");
 272     avio_wl32(pb, 56);
 273
 274     AV_WL32(amvh +  0, amv->us_per_frame);
 275     AV_WL32(amvh + 32, vst->codecpar->width);
 276     AV_WL32(amvh + 36, vst->codecpar->height);
 277     AV_WL32(amvh + 40, vst->time_base.den);
 278     AV_WL32(amvh + 44, vst->time_base.num);
 279     AV_WL32(amvh + 48, 0);
 280     AV_WL32(amvh + 52, 0); /* duration, filled in later. */
 281
 282     avio_write(pb, amvh, sizeof(amvh));
 283     amv->offset_duration = avio_tell(pb) - 4;
 284
 285     amv_write_vlist(s, vst->codecpar);
 286     amv_write_alist(s, ast->codecpar);
 287     amv_end_tag(pb, list1);
 288
 289     amv->movi_list = amv_start_tag(pb, "LIST");
 290     ffio_wfourcc(pb, "movi");
 291     return 0;
 292 }
 293
 294 static int amv_write_packet_internal(AVFormatContext *s, AVPacket *pkt)
 295 {
 296     AMVContext *amv = s->priv_data;
 297
 298     if (pkt->stream_index == AMV_STREAM_VIDEO)
 299         ffio_wfourcc(s->pb, "00dc");
 300     else if (pkt->stream_index == AMV_STREAM_AUDIO)
 301         ffio_wfourcc(s->pb, "01wb");
 302     else
 303         av_assert0(0);
 304
 305     if (pkt->stream_index == AMV_STREAM_AUDIO && pkt->size != amv->ablock_align) {
 306         /* Can happen when remuxing files produced by another encoder. */
 307         av_log(s, AV_LOG_WARNING, "Invalid audio packet size (%d != %d)\n",
 308                pkt->size, amv->ablock_align);
 309     }
 310
 311     avio_wl32(s->pb, pkt->size);
 312     avio_write(s->pb, pkt->data, pkt->size);
 313
 314     amv->lastpts[pkt->stream_index] += pkt->duration;
 315     amv->last_stream = pkt->stream_index;
 316     return 0;
 317 }
 318
 319 static int amv_pad(AVFormatContext *s, AVPacket *pkt)
 320 {
 321     AMVContext *amv = s->priv_data;
 322     int stream_index = pkt->stream_index;
 323
 324     if (stream_index != amv->last_stream)
 325         return 0;
 326
 327     stream_index = (stream_index + 1) % s->nb_streams;
 328     if (stream_index == AMV_STREAM_VIDEO)
 329         return amv_write_packet_internal(s, amv->vpad);
 330     else if (stream_index == AMV_STREAM_AUDIO)
 331         return amv_write_packet_internal(s, amv->apad);
 332     else
 333         av_assert0(0);
 334
 335     return AVERROR(EINVAL);
 336 }
 337
 338 static int amv_write_packet(AVFormatContext *s, AVPacket *pkt)
 339 {
 340     AMVContext *amv = s->priv_data;
 341     int ret;
 342
 343     /* Add a dummy frame if we've received two of the same index. */
 344     if ((ret = amv_pad(s, pkt)) < 0)
 345         return ret;
 346
 347     if ((ret = amv_write_packet_internal(s, pkt)) < 0)
 348         return ret;
 349
 350     if (pkt->stream_index == AMV_STREAM_VIDEO) {
 351         /* Save the last packet for padding. */
 352         av_packet_unref(amv->vpad);
 353         if ((ret = av_packet_ref(amv->vpad, pkt)) < 0)
 354             return ret;
 355     }
 356
 357     return 0;
 358 }
 359
 360 static int amv_write_trailer(AVFormatContext *s)
 361 {
 362     AMVContext *amv = s->priv_data;
 363     AVStream   *vst = s->streams[AMV_STREAM_VIDEO];
 364     AVStream   *ast = s->streams[AMV_STREAM_AUDIO];
 365     int64_t maxpts, ret;
 366     int hh, mm, ss;
 367
 368     /* Pad-out one last audio frame if needed. */
 369     if (amv->last_stream == AMV_STREAM_VIDEO) {
 370         if ((ret = amv_write_packet_internal(s, amv->apad)) < 0)
 371             return ret;
 372     }
 373
 374     amv_end_tag(s->pb, amv->movi_list);
 375     amv_end_tag(s->pb, amv->riff_start);
 376
 377     ffio_wfourcc(s->pb, "AMV_");
 378     ffio_wfourcc(s->pb, "END_");
 379
 380     if ((ret = avio_seek(s->pb, amv->offset_duration, SEEK_SET)) < 0)
 381         return ret;
 382
 383     /* Go back and write the duration. */
 384     maxpts = FFMAX(
 385         av_rescale_q(amv->lastpts[AMV_STREAM_VIDEO], vst->time_base, AV_TIME_BASE_Q),
 386         av_rescale_q(amv->lastpts[AMV_STREAM_AUDIO], ast->time_base, AV_TIME_BASE_Q)
 387     );
 388
 389     ss  = maxpts / AV_TIME_BASE;
 390     mm  = ss / 60;
 391     hh  = mm / 60;
 392     ss %= 60;
 393     mm %= 60;
 394
 395     avio_w8(s->pb, ss);
 396     avio_w8(s->pb, mm);
 397     avio_wl16(s->pb, hh);
 398     return 0;
 399 }
 400
 401 const FFOutputFormat ff_amv_muxer = {
 402     .p.name         = "amv",
 403     .p.long_name    = NULL_IF_CONFIG_SMALL("AMV"),
 404     .p.mime_type    = "video/amv",
 405     .p.extensions   = "amv",
 406     .priv_data_size = sizeof(AMVContext),
 407     .p.audio_codec  = AV_CODEC_ID_ADPCM_IMA_AMV,
 408     .p.video_codec  = AV_CODEC_ID_AMV,
 409     .p.subtitle_codec = AV_CODEC_ID_NONE,
 410     .flags_internal   = FF_OFMT_FLAG_MAX_ONE_OF_EACH |
 411                         FF_OFMT_FLAG_ONLY_DEFAULT_CODECS,
 412     .init           = amv_init,
 413     .deinit         = amv_deinit,
 414     .write_header   = amv_write_header,
 415     .write_packet   = amv_write_packet,
 416     .write_trailer  = amv_write_trailer,
 417 };