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 "riff.h"
24 #include "internal.h"
25 #include "avio_internal.h"
26 #include "libavutil/intreadwrite.h"
27 #include "libavutil/avassert.h"
28
29 /*
30 * Things to note:
31 * - AMV is a hard-coded (and broken) subset of AVI. It's not worth sullying the
32 * existing AVI muxer with its filth.
33 * - No separate demuxer as the existing AVI demuxer can handle these.
34 * - The sizes of certain tags are deliberately set to 0 as some players break
35 * when they're set correctly. Ditto with some header fields.
36 * - There is no index.
37 * - Players are **very** sensitive to the frame order and sizes.
38 * - Frames must be strictly interleaved as V-A, any V-V or A-A will
39 * cause crashes.
40 * - Variable video frame sizes seem to be handled fine.
41 * - Variable audio frame sizes cause crashes.
42 * - If audio is shorter than video, it's padded with silence.
43 * - If video is shorter than audio, the most recent frame is repeated.
44 */
45
46 #define AMV_STREAM_COUNT 2
47 #define AMV_STREAM_VIDEO 0
48 #define AMV_STREAM_AUDIO 1
49 #define AMV_VIDEO_STRH_SIZE 56
50 #define AMV_VIDEO_STRF_SIZE 36
51 #define AMV_AUDIO_STRH_SIZE 48
52 #define AMV_AUDIO_STRF_SIZE 20 /* sizeof(WAVEFORMATEX) + 2 */
53
54 typedef struct AMVContext
55 {
56 int64_t riff_start;
57 int64_t movi_list;
58 int64_t offset_duration;
59 int last_stream;
60
61 int32_t us_per_frame; /* Microseconds per frame. */
62
63 int32_t aframe_size; /* Expected audio frame size. */
64 int32_t ablock_align; /* Expected audio block align. */
65 AVPacket *apad; /* Dummy audio packet for padding; not owned by us. */
66 AVPacket *vpad; /* Most recent video frame, for padding. */
67
68 /*
69 * Cumulative PTS values for each stream, used for the final
70 * duration calculcation.
71 */
72 int64_t lastpts[AMV_STREAM_COUNT];
73 } AMVContext;
74
75 /* ff_{start,end}_tag(), but sets the size to 0. */
amv_start_tag(AVIOContext * pb,const char * tag)76 static int64_t amv_start_tag(AVIOContext *pb, const char *tag)
77 {
78 ffio_wfourcc(pb, tag);
79 avio_wl32(pb, 0);
80 return avio_tell(pb);
81 }
82
amv_end_tag(AVIOContext * pb,int64_t start)83 static void amv_end_tag(AVIOContext *pb, int64_t start)
84 {
85 int64_t pos;
86 av_assert0((start&1) == 0);
87
88 pos = avio_tell(pb);
89 if (pos & 1)
90 avio_w8(pb, 0);
91 }
92
amv_init(AVFormatContext * s)93 static av_cold int amv_init(AVFormatContext *s)
94 {
95 AMVContext *amv = s->priv_data;
96 AVStream *vst, *ast;
97 int ret;
98
99 amv->last_stream = -1;
100
101 if (s->nb_streams != AMV_STREAM_COUNT) {
102 av_log(s, AV_LOG_ERROR, "AMV files only support 2 streams\n");
103 return AVERROR(EINVAL);
104 }
105
106 vst = s->streams[AMV_STREAM_VIDEO];
107 ast = s->streams[AMV_STREAM_AUDIO];
108
109 if (vst->codecpar->codec_id != AV_CODEC_ID_AMV) {
110 av_log(s, AV_LOG_ERROR, "First AMV stream must be %s\n",
111 avcodec_get_name(AV_CODEC_ID_AMV));
112 return AVERROR(EINVAL);
113 }
114
115 if (ast->codecpar->codec_id != AV_CODEC_ID_ADPCM_IMA_AMV) {
116 av_log(s, AV_LOG_ERROR, "Second AMV stream must be %s\n",
117 avcodec_get_name(AV_CODEC_ID_ADPCM_IMA_AMV));
118 return AVERROR(EINVAL);
119 }
120
121 /* These files are broken-enough as they are. They shouldn't be streamed. */
122 if (!(s->pb->seekable & AVIO_SEEKABLE_NORMAL)) {
123 av_log(s, AV_LOG_ERROR, "Stream not seekable, unable to write output file\n");
124 return AVERROR(EINVAL);
125 }
126
127 amv->us_per_frame = av_rescale(AV_TIME_BASE, vst->time_base.num, vst->time_base.den);
128 amv->aframe_size = av_rescale(ast->codecpar->sample_rate, amv->us_per_frame, AV_TIME_BASE);
129 amv->ablock_align = 8 + (FFALIGN(amv->aframe_size, 2) / 2);
130
131 av_log(s, AV_LOG_TRACE, "us_per_frame = %d\n", amv->us_per_frame);
132 av_log(s, AV_LOG_TRACE, "aframe_size = %d\n", amv->aframe_size);
133 av_log(s, AV_LOG_TRACE, "ablock_align = %d\n", amv->ablock_align);
134
135 /*
136 * Bail if the framerate's too high. Prevents the audio frame size from
137 * getting too small. 63fps is the closest value to 60fps that divides
138 * cleanly, so cap it there.
139 */
140 if (amv->us_per_frame < 15873) {
141 av_log(s, AV_LOG_ERROR, "Refusing to mux >63fps video\n");
142 return AVERROR(EINVAL);
143 }
144
145 /*
146 * frame_size will be set if coming from the encoder.
147 * Make sure the its been configured correctly. The audio frame duration
148 * needs to match that of the video.
149 */
150 if (ast->codecpar->frame_size) {
151 AVCodecParameters *par = ast->codecpar;
152 int bad = 0;
153
154 if (par->frame_size != amv->aframe_size) {
155 av_log(s, AV_LOG_ERROR, "Invalid audio frame size. Got %d, wanted %d\n",
156 par->frame_size, amv->aframe_size);
157 bad = 1;
158 }
159
160 if (par->block_align != amv->ablock_align) {
161 av_log(s, AV_LOG_ERROR, "Invalid audio block align. Got %d, wanted %d\n",
162 par->block_align, amv->ablock_align);
163 bad = 1;
164 }
165
166 if (bad) {
167 av_log(s, AV_LOG_ERROR, "Try -block_size %d\n", amv->aframe_size);
168 return AVERROR(EINVAL);
169 }
170
171 if (ast->codecpar->sample_rate % amv->aframe_size) {
172 av_log(s, AV_LOG_ERROR, "Audio sample rate not a multiple of the frame size.\n"
173 "Please change video frame rate. Suggested rates: 10,14,15,18,21,25,30\n");
174 return AVERROR(EINVAL);
175 }
176 } else {
177 /* If remuxing from the same source, then this will match the video. */
178 int32_t aus = av_rescale(AV_TIME_BASE, ast->time_base.num, ast->time_base.den);
179 if (aus != amv->us_per_frame) {
180 av_log(s, AV_LOG_ERROR, "Cannot remux streams with a different time base\n");
181 return AVERROR(EINVAL);
182 }
183 }
184
185 /* Allocate and fill dummy packet so we can pad the audio. */
186 amv->apad = ffformatcontext(s)->pkt;
187 if ((ret = av_new_packet(amv->apad, amv->ablock_align)) < 0) {
188 return ret;
189 }
190
191 amv->apad->stream_index = AMV_STREAM_AUDIO;
192 memset(amv->apad->data, 0, amv->ablock_align);
193 AV_WL32(amv->apad->data + 4, amv->aframe_size);
194
195 amv->vpad = av_packet_alloc();
196 if (!amv->vpad) {
197 return AVERROR(ENOMEM);
198 }
199 amv->vpad->stream_index = AMV_STREAM_VIDEO;
200 amv->vpad->duration = 1;
201 return 0;
202 }
203
amv_deinit(AVFormatContext * s)204 static void amv_deinit(AVFormatContext *s)
205 {
206 AMVContext *amv = s->priv_data;
207
208 av_packet_free(&amv->vpad);
209 }
210
amv_write_vlist(AVFormatContext * s,AVCodecParameters * par)211 static void amv_write_vlist(AVFormatContext *s, AVCodecParameters *par)
212 {
213 int64_t tag_list, tag_str;
214
215 av_assert0(par->codec_id == AV_CODEC_ID_AMV);
216
217 tag_list = amv_start_tag(s->pb, "LIST");
218 ffio_wfourcc(s->pb, "strl");
219 tag_str = ff_start_tag(s->pb, "strh");
220 ffio_fill(s->pb, 0, AMV_VIDEO_STRH_SIZE);
221 ff_end_tag(s->pb, tag_str);
222
223 tag_str = ff_start_tag(s->pb, "strf");
224 ffio_fill(s->pb, 0, AMV_VIDEO_STRF_SIZE);
225 ff_end_tag(s->pb, tag_str);
226
227 amv_end_tag(s->pb, tag_list);
228 }
229
amv_write_alist(AVFormatContext * s,AVCodecParameters * par)230 static void amv_write_alist(AVFormatContext *s, AVCodecParameters *par)
231 {
232 uint8_t buf[AMV_AUDIO_STRF_SIZE];
233 AVIOContext *pb = s->pb;
234 int64_t tag_list, tag_str;
235
236 av_assert0(par->codec_id == AV_CODEC_ID_ADPCM_IMA_AMV);
237
238 tag_list = amv_start_tag(pb, "LIST");
239 ffio_wfourcc(pb, "strl");
240 tag_str = ff_start_tag(pb, "strh");
241 ffio_fill(s->pb, 0, AMV_AUDIO_STRH_SIZE);
242 ff_end_tag(pb, tag_str);
243
244 /* Bodge an (incorrect) WAVEFORMATEX (+2 pad bytes) */
245 tag_str = ff_start_tag(pb, "strf");
246 AV_WL16(buf + 0, 1);
247 AV_WL16(buf + 2, par->ch_layout.nb_channels);
248 AV_WL32(buf + 4, par->sample_rate);
249 AV_WL32(buf + 8, par->sample_rate * par->ch_layout.nb_channels * 2);
250 AV_WL16(buf + 12, 2);
251 AV_WL16(buf + 14, 16);
252 AV_WL16(buf + 16, 0);
253 AV_WL16(buf + 18, 0);
254 avio_write(pb, buf, AMV_AUDIO_STRF_SIZE);
255 ff_end_tag(pb, tag_str);
256
257 amv_end_tag(pb, tag_list);
258 }
259
amv_write_header(AVFormatContext * s)260 static int amv_write_header(AVFormatContext *s)
261 {
262 AMVContext *amv = s->priv_data;
263 AVIOContext *pb = s->pb;
264 AVStream *vst = s->streams[AMV_STREAM_VIDEO];
265 AVStream *ast = s->streams[AMV_STREAM_AUDIO];
266 uint8_t amvh[56] = {0};
267 int64_t list1;
268
269 amv->riff_start = amv_start_tag(pb, "RIFF");
270 ffio_wfourcc(pb, "AMV ");
271 list1 = amv_start_tag(pb, "LIST");
272 ffio_wfourcc(pb, "hdrl");
273
274 ffio_wfourcc(pb, "amvh");
275 avio_wl32(pb, 56);
276
277 AV_WL32(amvh + 0, amv->us_per_frame);
278 AV_WL32(amvh + 32, vst->codecpar->width);
279 AV_WL32(amvh + 36, vst->codecpar->height);
280 AV_WL32(amvh + 40, vst->time_base.den);
281 AV_WL32(amvh + 44, vst->time_base.num);
282 AV_WL32(amvh + 48, 0);
283 AV_WL32(amvh + 52, 0); /* duration, filled in later. */
284
285 avio_write(pb, amvh, sizeof(amvh));
286 amv->offset_duration = avio_tell(pb) - 4;
287
288 amv_write_vlist(s, vst->codecpar);
289 amv_write_alist(s, ast->codecpar);
290 amv_end_tag(pb, list1);
291
292 amv->movi_list = amv_start_tag(pb, "LIST");
293 ffio_wfourcc(pb, "movi");
294 return 0;
295 }
296
amv_write_packet_internal(AVFormatContext * s,AVPacket * pkt)297 static int amv_write_packet_internal(AVFormatContext *s, AVPacket *pkt)
298 {
299 AMVContext *amv = s->priv_data;
300
301 if (pkt->stream_index == AMV_STREAM_VIDEO)
302 ffio_wfourcc(s->pb, "00dc");
303 else if (pkt->stream_index == AMV_STREAM_AUDIO)
304 ffio_wfourcc(s->pb, "01wb");
305 else
306 av_assert0(0);
307
308 if (pkt->stream_index == AMV_STREAM_AUDIO && pkt->size != amv->ablock_align) {
309 /* Can happen when remuxing files produced by another encoder. */
310 av_log(s, AV_LOG_WARNING, "Invalid audio packet size (%d != %d)\n",
311 pkt->size, amv->ablock_align);
312 }
313
314 avio_wl32(s->pb, pkt->size);
315 avio_write(s->pb, pkt->data, pkt->size);
316
317 amv->lastpts[pkt->stream_index] += pkt->duration;
318 amv->last_stream = pkt->stream_index;
319 return 0;
320 }
321
amv_pad(AVFormatContext * s,AVPacket * pkt)322 static int amv_pad(AVFormatContext *s, AVPacket *pkt)
323 {
324 AMVContext *amv = s->priv_data;
325 int stream_index = pkt->stream_index;
326
327 if (stream_index != amv->last_stream)
328 return 0;
329
330 stream_index = (stream_index + 1) % s->nb_streams;
331 if (stream_index == AMV_STREAM_VIDEO)
332 return amv_write_packet_internal(s, amv->vpad);
333 else if (stream_index == AMV_STREAM_AUDIO)
334 return amv_write_packet_internal(s, amv->apad);
335 else
336 av_assert0(0);
337
338 return AVERROR(EINVAL);
339 }
340
amv_write_packet(AVFormatContext * s,AVPacket * pkt)341 static int amv_write_packet(AVFormatContext *s, AVPacket *pkt)
342 {
343 AMVContext *amv = s->priv_data;
344 int ret;
345
346 /* Add a dummy frame if we've received two of the same index. */
347 if ((ret = amv_pad(s, pkt)) < 0)
348 return ret;
349
350 if ((ret = amv_write_packet_internal(s, pkt)) < 0)
351 return ret;
352
353 if (pkt->stream_index == AMV_STREAM_VIDEO) {
354 /* Save the last packet for padding. */
355 av_packet_unref(amv->vpad);
356 if ((ret = av_packet_ref(amv->vpad, pkt)) < 0)
357 return ret;
358 }
359
360 return 0;
361 }
362
amv_write_trailer(AVFormatContext * s)363 static int amv_write_trailer(AVFormatContext *s)
364 {
365 AMVContext *amv = s->priv_data;
366 AVStream *vst = s->streams[AMV_STREAM_VIDEO];
367 AVStream *ast = s->streams[AMV_STREAM_AUDIO];
368 int64_t maxpts, ret;
369 int hh, mm, ss;
370
371 /* Pad-out one last audio frame if needed. */
372 if (amv->last_stream == AMV_STREAM_VIDEO) {
373 if ((ret = amv_write_packet_internal(s, amv->apad)) < 0)
374 return ret;
375 }
376
377 amv_end_tag(s->pb, amv->movi_list);
378 amv_end_tag(s->pb, amv->riff_start);
379
380 ffio_wfourcc(s->pb, "AMV_");
381 ffio_wfourcc(s->pb, "END_");
382
383 if ((ret = avio_seek(s->pb, amv->offset_duration, SEEK_SET)) < 0)
384 return ret;
385
386 /* Go back and write the duration. */
387 maxpts = FFMAX(
388 av_rescale_q(amv->lastpts[AMV_STREAM_VIDEO], vst->time_base, AV_TIME_BASE_Q),
389 av_rescale_q(amv->lastpts[AMV_STREAM_AUDIO], ast->time_base, AV_TIME_BASE_Q)
390 );
391
392 ss = maxpts / AV_TIME_BASE;
393 mm = ss / 60;
394 hh = mm / 60;
395 ss %= 60;
396 mm %= 60;
397
398 avio_w8(s->pb, ss);
399 avio_w8(s->pb, mm);
400 avio_wl16(s->pb, hh);
401 return 0;
402 }
403
404 const AVOutputFormat ff_amv_muxer = {
405 .name = "amv",
406 .long_name = NULL_IF_CONFIG_SMALL("AMV"),
407 .mime_type = "video/amv",
408 .extensions = "amv",
409 .priv_data_size = sizeof(AMVContext),
410 .audio_codec = AV_CODEC_ID_ADPCM_IMA_AMV,
411 .video_codec = AV_CODEC_ID_AMV,
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 };
418