FFmpeg coverage

Line	Branch	Exec	Source
1			/*
2			* Microsoft XMV demuxer
3			* Copyright (c) 2011 Sven Hesse <drmccoy@drmccoy.de>
4			* Copyright (c) 2011 Matthew Hoops <clone2727@gmail.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
23			/**
24			* @file
25			* Microsoft XMV demuxer
26			*/
27
28			#include <inttypes.h>
29
30			#include "libavutil/intreadwrite.h"
31			#include "libavutil/mem.h"
32
33			#include "avformat.h"
34			#include "avio_internal.h"
35			#include "demux.h"
36			#include "internal.h"
37			#include "riff.h"
38			#include "libavutil/avassert.h"
39
40			/** The min size of an XMV header. */
41			#define XMV_MIN_HEADER_SIZE 36
42
43			/** Audio flag: ADPCM'd 5.1 stream, front left / right channels */
44			#define XMV_AUDIO_ADPCM51_FRONTLEFTRIGHT 1
45			/** Audio flag: ADPCM'd 5.1 stream, front center / low frequency channels */
46			#define XMV_AUDIO_ADPCM51_FRONTCENTERLOW 2
47			/** Audio flag: ADPCM'd 5.1 stream, rear left / right channels */
48			#define XMV_AUDIO_ADPCM51_REARLEFTRIGHT 4
49
50			/** Audio flag: Any of the ADPCM'd 5.1 stream flags. */
51			#define XMV_AUDIO_ADPCM51 (XMV_AUDIO_ADPCM51_FRONTLEFTRIGHT | \
52			XMV_AUDIO_ADPCM51_FRONTCENTERLOW | \
53			XMV_AUDIO_ADPCM51_REARLEFTRIGHT)
54
55			#define XMV_BLOCK_ALIGN_SIZE 36
56
57			/** A video packet with an XMV file. */
58			typedef struct XMVVideoPacket {
59			int created;
60			int stream_index; ///< The decoder stream index for this video packet.
61
62			uint32_t data_size; ///< The size of the remaining video data.
63			uint64_t data_offset; ///< The offset of the video data within the file.
64
65			uint32_t current_frame; ///< The current frame within this video packet.
66			uint32_t frame_count; ///< The amount of frames within this video packet.
67
68			int has_extradata; ///< Does the video packet contain extra data?
69			uint8_t extradata[4]; ///< The extra data
70
71			int64_t last_pts; ///< PTS of the last video frame.
72			int64_t pts; ///< PTS of the most current video frame.
73			} XMVVideoPacket;
74
75			/** An audio packet with an XMV file. */
76			typedef struct XMVAudioPacket {
77			int created;
78			int stream_index; ///< The decoder stream index for this audio packet.
79
80			/* Stream format properties. */
81			uint16_t compression; ///< The type of compression.
82			uint16_t channels; ///< Number of channels.
83			int32_t sample_rate; ///< Sampling rate.
84			uint16_t bits_per_sample; ///< Bits per compressed sample.
85			uint64_t bit_rate; ///< Bits of compressed data per second.
86			uint16_t flags; ///< Flags
87			unsigned block_align; ///< Bytes per compressed block.
88
89			enum AVCodecID codec_id; ///< The codec ID of the compression scheme.
90
91			uint32_t data_size; ///< The size of the remaining audio data.
92			uint64_t data_offset; ///< The offset of the audio data within the file.
93
94			uint32_t frame_size; ///< Number of bytes to put into an audio frame.
95			} XMVAudioPacket;
96
97			/** Context for demuxing an XMV file. */
98			typedef struct XMVDemuxContext {
99			uint16_t audio_track_count; ///< Number of audio track in this file.
100
101			uint32_t this_packet_size; ///< Size of the current packet.
102			uint32_t next_packet_size; ///< Size of the next packet.
103
104			uint64_t this_packet_offset; ///< Offset of the current packet.
105			uint64_t next_packet_offset; ///< Offset of the next packet.
106
107			uint16_t current_stream; ///< The index of the stream currently handling.
108			uint16_t stream_count; ///< The number of streams in this file.
109
110			uint32_t video_duration;
111			uint32_t video_width;
112			uint32_t video_height;
113
114			XMVVideoPacket video; ///< The video packet contained in each packet.
115			XMVAudioPacket *audio; ///< The audio packets contained in each packet.
116			} XMVDemuxContext;
117
118		✗	static int xmv_probe(const AVProbeData *p)
119			{
120			uint32_t file_version;
121
122		✗	if (p->buf_size < XMV_MIN_HEADER_SIZE)
123		✗	return 0;
124
125		✗	file_version = AV_RL32(p->buf + 16);
126		✗	if ((file_version == 0) || (file_version > 4))
127		✗	return 0;
128
129		✗	if (!memcmp(p->buf + 12, "xobX", 4))
130		✗	return AVPROBE_SCORE_MAX;
131
132		✗	return 0;
133			}
134
135		✗	static int xmv_read_close(AVFormatContext *s)
136			{
137		✗	XMVDemuxContext *xmv = s->priv_data;
138
139		✗	av_freep(&xmv->audio);
140
141		✗	return 0;
142			}
143
144		✗	static int xmv_read_header(AVFormatContext *s)
145			{
146		✗	XMVDemuxContext *xmv = s->priv_data;
147		✗	AVIOContext *pb = s->pb;
148
149			uint32_t file_version;
150			uint32_t this_packet_size;
151			uint16_t audio_track;
152
153		✗	s->ctx_flags |= AVFMTCTX_NOHEADER;
154
155		✗	avio_skip(pb, 4); /* Next packet size */
156
157		✗	this_packet_size = avio_rl32(pb);
158
159		✗	avio_skip(pb, 4); /* Max packet size */
160		✗	avio_skip(pb, 4); /* "xobX" */
161
162		✗	file_version = avio_rl32(pb);
163		✗	if ((file_version != 4) && (file_version != 2))
164		✗	avpriv_request_sample(s, "Uncommon version %"PRIu32"", file_version);
165
166			/* Video tracks */
167
168		✗	xmv->video_width = avio_rl32(pb);
169		✗	xmv->video_height = avio_rl32(pb);
170		✗	xmv->video_duration = avio_rl32(pb);
171
172			/* Audio tracks */
173
174		✗	xmv->audio_track_count = avio_rl16(pb);
175
176		✗	avio_skip(pb, 2); /* Unknown (padding?) */
177
178		✗	xmv->audio = av_calloc(xmv->audio_track_count, sizeof(*xmv->audio));
179		✗	if (!xmv->audio)
180		✗	return AVERROR(ENOMEM);
181
182		✗	for (audio_track = 0; audio_track < xmv->audio_track_count; audio_track++) {
183		✗	XMVAudioPacket *packet = &xmv->audio[audio_track];
184
185		✗	packet->compression = avio_rl16(pb);
186		✗	packet->channels = avio_rl16(pb);
187		✗	packet->sample_rate = avio_rl32(pb);
188		✗	packet->bits_per_sample = avio_rl16(pb);
189		✗	packet->flags = avio_rl16(pb);
190
191		✗	packet->bit_rate = (uint64_t)packet->bits_per_sample *
192		✗	packet->sample_rate *
193		✗	packet->channels;
194		✗	packet->block_align = XMV_BLOCK_ALIGN_SIZE * packet->channels;
195		✗	packet->codec_id = ff_wav_codec_get_id(packet->compression,
196		✗	packet->bits_per_sample);
197
198		✗	packet->stream_index = -1;
199
200		✗	packet->frame_size = 0;
201
202			/* TODO: ADPCM'd 5.1 sound is encoded in three separate streams.
203			* Those need to be interleaved to a proper 5.1 stream. */
204		✗	if (packet->flags & XMV_AUDIO_ADPCM51)
205		✗	av_log(s, AV_LOG_WARNING, "Unsupported 5.1 ADPCM audio stream "
206		✗	"(0x%04X)\n", packet->flags);
207
208		✗	if (!packet->channels || packet->sample_rate <= 0 ||
209		✗	packet->channels >= UINT16_MAX / XMV_BLOCK_ALIGN_SIZE) {
210		✗	av_log(s, AV_LOG_ERROR, "Invalid parameters for audio track %"PRIu16".\n",
211			audio_track);
212		✗	return AVERROR_INVALIDDATA;
213			}
214			}
215
216
217			/* Initialize the packet context */
218
219		✗	xmv->next_packet_offset = avio_tell(pb);
220		✗	if (this_packet_size < xmv->next_packet_offset)
221		✗	return AVERROR_INVALIDDATA;
222		✗	xmv->next_packet_size = this_packet_size - xmv->next_packet_offset;
223		✗	xmv->stream_count = xmv->audio_track_count + 1;
224
225		✗	return 0;
226			}
227
228		✗	static void xmv_read_extradata(uint8_t *extradata, AVIOContext *pb)
229			{
230			/* Read the XMV extradata */
231
232		✗	uint32_t data = avio_rl32(pb);
233
234		✗	int mspel_bit = !!(data & 0x01);
235		✗	int loop_filter = !!(data & 0x02);
236		✗	int abt_flag = !!(data & 0x04);
237		✗	int j_type_bit = !!(data & 0x08);
238		✗	int top_left_mv_flag = !!(data & 0x10);
239		✗	int per_mb_rl_bit = !!(data & 0x20);
240		✗	int slice_count = (data >> 6) & 7;
241
242			/* Write it back as standard WMV2 extradata */
243
244		✗	data = 0;
245
246		✗	data |= mspel_bit << 15;
247		✗	data |= loop_filter << 14;
248		✗	data |= abt_flag << 13;
249		✗	data |= j_type_bit << 12;
250		✗	data |= top_left_mv_flag << 11;
251		✗	data |= per_mb_rl_bit << 10;
252		✗	data |= slice_count << 7;
253
254		✗	AV_WB32(extradata, data);
255		✗	}
256
257		✗	static int xmv_process_packet_header(AVFormatContext *s)
258			{
259		✗	XMVDemuxContext *xmv = s->priv_data;
260		✗	AVIOContext *pb = s->pb;
261			int ret;
262
263			uint8_t data[8];
264			uint16_t audio_track;
265			uint64_t data_offset;
266
267			/* Next packet size */
268		✗	xmv->next_packet_size = avio_rl32(pb);
269
270			/* Packet video header */
271
272		✗	if ((ret = ffio_read_size(pb, data, 8)) < 0)
273		✗	return ret;
274
275		✗	xmv->video.data_size = AV_RL32(data) & 0x007FFFFF;
276
277		✗	xmv->video.current_frame = 0;
278		✗	xmv->video.frame_count = (AV_RL32(data) >> 23) & 0xFF;
279
280		✗	xmv->video.has_extradata = (data[3] & 0x80) != 0;
281
282		✗	if (!xmv->video.created) {
283		✗	AVStream *vst = avformat_new_stream(s, NULL);
284		✗	if (!vst)
285		✗	return AVERROR(ENOMEM);
286
287		✗	avpriv_set_pts_info(vst, 32, 1, 1000);
288
289		✗	vst->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
290		✗	vst->codecpar->codec_id = AV_CODEC_ID_WMV2;
291		✗	vst->codecpar->codec_tag = MKBETAG('W', 'M', 'V', '2');
292		✗	vst->codecpar->width = xmv->video_width;
293		✗	vst->codecpar->height = xmv->video_height;
294
295		✗	vst->duration = xmv->video_duration;
296
297		✗	xmv->video.stream_index = vst->index;
298
299		✗	xmv->video.created = 1;
300			}
301
302			/* Adding the audio data sizes and the video data size keeps you 4 bytes
303			* short for every audio track. But as playing around with XMV files with
304			* ADPCM audio showed, taking the extra 4 bytes from the audio data gives
305			* you either completely distorted audio or click (when skipping the
306			* remaining 68 bytes of the ADPCM block). Subtracting 4 bytes for every
307			* audio track from the video data works at least for the audio. Probably
308			* some alignment thing?
309			* The video data has (always?) lots of padding, so it should work out...
310			*/
311		✗	xmv->video.data_size -= xmv->audio_track_count * 4;
312
313		✗	xmv->current_stream = 0;
314		✗	if (!xmv->video.frame_count) {
315		✗	xmv->video.frame_count = 1;
316		✗	xmv->current_stream = xmv->stream_count > 1;
317			}
318
319			/* Packet audio header */
320
321		✗	for (audio_track = 0; audio_track < xmv->audio_track_count; audio_track++) {
322		✗	XMVAudioPacket *packet = &xmv->audio[audio_track];
323
324		✗	if ((ret = ffio_read_size(pb, data, 4)) < 0)
325		✗	return ret;
326
327		✗	if (!packet->created) {
328		✗	AVStream *ast = avformat_new_stream(s, NULL);
329		✗	if (!ast)
330		✗	return AVERROR(ENOMEM);
331
332		✗	ast->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
333		✗	ast->codecpar->codec_id = packet->codec_id;
334		✗	ast->codecpar->codec_tag = packet->compression;
335		✗	ast->codecpar->ch_layout.nb_channels = packet->channels;
336		✗	ast->codecpar->sample_rate = packet->sample_rate;
337		✗	ast->codecpar->bits_per_coded_sample = packet->bits_per_sample;
338		✗	ast->codecpar->bit_rate = packet->bit_rate;
339		✗	ast->codecpar->block_align = 36 * packet->channels;
340
341		✗	avpriv_set_pts_info(ast, 32, 1, packet->sample_rate);
342
343		✗	packet->stream_index = ast->index;
344
345		✗	ast->duration = xmv->video_duration;
346
347		✗	packet->created = 1;
348			}
349
350		✗	packet->data_size = AV_RL32(data) & 0x007FFFFF;
351		✗	if ((packet->data_size == 0) && (audio_track != 0))
352			/* This happens when I create an XMV with several identical audio
353			* streams. From the size calculations, duplicating the previous
354			* stream's size works out, but the track data itself is silent.
355			* Maybe this should also redirect the offset to the previous track?
356			*/
357		✗	packet->data_size = xmv->audio[audio_track - 1].data_size;
358
359			/* Carve up the audio data in frame_count slices */
360		✗	packet->frame_size = packet->data_size / xmv->video.frame_count;
361		✗	packet->frame_size -= packet->frame_size % packet->block_align;
362			}
363
364			/* Packet data offsets */
365
366		✗	data_offset = avio_tell(pb);
367
368		✗	xmv->video.data_offset = data_offset;
369		✗	data_offset += xmv->video.data_size;
370
371		✗	for (audio_track = 0; audio_track < xmv->audio_track_count; audio_track++) {
372		✗	xmv->audio[audio_track].data_offset = data_offset;
373		✗	data_offset += xmv->audio[audio_track].data_size;
374			}
375
376			/* Video frames header */
377
378			/* Read new video extra data */
379		✗	if (xmv->video.data_size > 0) {
380		✗	if (xmv->video.has_extradata) {
381		✗	xmv_read_extradata(xmv->video.extradata, pb);
382
383		✗	xmv->video.data_size -= 4;
384		✗	xmv->video.data_offset += 4;
385
386		✗	if (xmv->video.stream_index >= 0) {
387		✗	AVStream *vst = s->streams[xmv->video.stream_index];
388
389		✗	av_assert0(xmv->video.stream_index < s->nb_streams);
390
391		✗	if (vst->codecpar->extradata_size < 4) {
392		✗	if ((ret = ff_alloc_extradata(vst->codecpar, 4)) < 0)
393		✗	return ret;
394			}
395
396		✗	memcpy(vst->codecpar->extradata, xmv->video.extradata, 4);
397			}
398			}
399			}
400
401		✗	return 0;
402			}
403
404		✗	static int xmv_fetch_new_packet(AVFormatContext *s)
405			{
406		✗	XMVDemuxContext *xmv = s->priv_data;
407		✗	AVIOContext *pb = s->pb;
408			int result;
409
410		✗	if (xmv->next_packet_size == 0)
411		✗	return AVERROR_EOF;
412
413			/* Seek to it */
414		✗	xmv->this_packet_offset = xmv->next_packet_offset;
415		✗	if (avio_seek(pb, xmv->this_packet_offset, SEEK_SET) != xmv->this_packet_offset)
416		✗	return AVERROR_INVALIDDATA;
417
418			/* Update the size */
419		✗	xmv->this_packet_size = xmv->next_packet_size;
420		✗	if (xmv->this_packet_size < (12 + xmv->audio_track_count * 4))
421		✗	return AVERROR_INVALIDDATA;
422
423			/* Process the header */
424		✗	result = xmv_process_packet_header(s);
425		✗	if (result)
426		✗	return result;
427
428			/* Update the offset */
429		✗	xmv->next_packet_offset = xmv->this_packet_offset + xmv->this_packet_size;
430
431		✗	return 0;
432			}
433
434		✗	static int xmv_fetch_audio_packet(AVFormatContext *s,
435			AVPacket *pkt, uint32_t stream)
436			{
437		✗	XMVDemuxContext *xmv = s->priv_data;
438		✗	AVIOContext *pb = s->pb;
439		✗	XMVAudioPacket *audio = &xmv->audio[stream];
440
441			uint32_t data_size;
442			int result;
443
444			/* Seek to it */
445		✗	if (avio_seek(pb, audio->data_offset, SEEK_SET) != audio->data_offset)
446		✗	return AVERROR_INVALIDDATA;
447
448		✗	if ((xmv->video.current_frame + 1) < xmv->video.frame_count)
449			/* Not the last frame, get at most frame_size bytes. */
450		✗	data_size = FFMIN(audio->frame_size, audio->data_size);
451			else
452			/* Last frame, get the rest. */
453		✗	data_size = audio->data_size;
454
455			/* Read the packet */
456		✗	result = av_get_packet(pb, pkt, data_size);
457		✗	if (result <= 0)
458		✗	return result;
459
460		✗	pkt->stream_index = audio->stream_index;
461
462			/* Advance offset */
463		✗	audio->data_size -= data_size;
464		✗	audio->data_offset += data_size;
465
466		✗	return 0;
467			}
468
469		✗	static int xmv_fetch_video_packet(AVFormatContext *s,
470			AVPacket *pkt)
471			{
472		✗	XMVDemuxContext *xmv = s->priv_data;
473		✗	AVIOContext *pb = s->pb;
474		✗	XMVVideoPacket *video = &xmv->video;
475
476			int result;
477			uint32_t frame_header;
478			uint32_t frame_size, frame_timestamp;
479			uint8_t *data, *end;
480
481			/* Seek to it */
482		✗	if (avio_seek(pb, video->data_offset, SEEK_SET) != video->data_offset)
483		✗	return AVERROR_INVALIDDATA;
484
485			/* Read the frame header */
486		✗	frame_header = avio_rl32(pb);
487
488		✗	frame_size = (frame_header & 0x1FFFF) * 4 + 4;
489		✗	frame_timestamp = (frame_header >> 17);
490
491		✗	if ((frame_size + 4) > video->data_size)
492		✗	return AVERROR_INVALIDDATA;
493
494			/* Get the packet data */
495		✗	result = av_get_packet(pb, pkt, frame_size);
496		✗	if (result != frame_size)
497		✗	return result;
498
499			/* Contrary to normal WMV2 video, the bit stream in XMV's
500			* WMV2 is little-endian.
501			* TODO: This manual swap is of course suboptimal.
502			*/
503		✗	for (data = pkt->data, end = pkt->data + frame_size; data < end; data += 4)
504		✗	AV_WB32(data, AV_RL32(data));
505
506		✗	pkt->stream_index = video->stream_index;
507
508			/* Calculate the PTS */
509
510		✗	video->last_pts = frame_timestamp + video->pts;
511
512		✗	pkt->duration = 0;
513		✗	pkt->pts = video->last_pts;
514		✗	pkt->dts = AV_NOPTS_VALUE;
515
516		✗	video->pts += frame_timestamp;
517
518			/* Keyframe? */
519		✗	pkt->flags = (pkt->data[0] & 0x80) ? 0 : AV_PKT_FLAG_KEY;
520
521			/* Advance offset */
522		✗	video->data_size -= frame_size + 4;
523		✗	video->data_offset += frame_size + 4;
524
525		✗	return 0;
526			}
527
528		✗	static int xmv_read_packet(AVFormatContext *s,
529			AVPacket *pkt)
530			{
531		✗	XMVDemuxContext *xmv = s->priv_data;
532			int result;
533
534		✗	if (xmv->video.current_frame == xmv->video.frame_count) {
535			/* No frames left in this packet, so we fetch a new one */
536
537		✗	result = xmv_fetch_new_packet(s);
538		✗	if (result)
539		✗	return result;
540			}
541
542		✗	if (xmv->current_stream == 0) {
543			/* Fetch a video frame */
544
545		✗	result = xmv_fetch_video_packet(s, pkt);
546			} else {
547			/* Fetch an audio frame */
548
549		✗	result = xmv_fetch_audio_packet(s, pkt, xmv->current_stream - 1);
550			}
551		✗	if (result) {
552		✗	xmv->current_stream = 0;
553		✗	xmv->video.current_frame = xmv->video.frame_count;
554		✗	return result;
555			}
556
557
558			/* Increase our counters */
559		✗	if (++xmv->current_stream >= xmv->stream_count) {
560		✗	xmv->current_stream = 0;
561		✗	xmv->video.current_frame += 1;
562			}
563
564		✗	return 0;
565			}
566
567			const FFInputFormat ff_xmv_demuxer = {
568			.p.name = "xmv",
569			.p.long_name = NULL_IF_CONFIG_SMALL("Microsoft XMV"),
570			.p.extensions = "xmv",
571			.priv_data_size = sizeof(XMVDemuxContext),
572			.flags_internal = FF_INFMT_FLAG_INIT_CLEANUP,
573			.read_probe = xmv_probe,
574			.read_header = xmv_read_header,
575			.read_packet = xmv_read_packet,
576			.read_close = xmv_read_close,
577			};
578