2 * id Quake II CIN File Demuxer
3 * Copyright (c) 2003 The FFmpeg project
5 * This file is part of FFmpeg.
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.
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.
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
24 * id Quake II CIN file demuxer by Mike Melanson (melanson@pcisys.net)
25 * For more information about the id CIN format, visit:
26 * http://www.csse.monash.edu.au/~timf/
28 * CIN is a somewhat quirky and ill-defined format. Here are some notes
29 * for anyone trying to understand the technical details of this format:
31 * The format has no definite file signature. This is problematic for a
32 * general-purpose media player that wants to automatically detect file
33 * types. However, a CIN file does start with 5 32-bit numbers that
34 * specify audio and video parameters. This demuxer gets around the lack
35 * of file signature by performing sanity checks on those parameters.
36 * Probabilistically, this is a reasonable solution since the number of
37 * valid combinations of the 5 parameters is a very small subset of the
38 * total 160-bit number space.
40 * Refer to the function idcin_probe() for the precise A/V parameters
41 * that this demuxer allows.
43 * Next, each audio and video frame has a duration of 1/14 sec. If the
44 * audio sample rate is a multiple of the common frequency 22050 Hz it will
45 * divide evenly by 14. However, if the sample rate is 11025 Hz:
46 * 11025 (samples/sec) / 14 (frames/sec) = 787.5 (samples/frame)
47 * The way the CIN stores audio in this case is by storing 787 sample
48 * frames in the first audio frame and 788 sample frames in the second
49 * audio frame. Therefore, the total number of bytes in an audio frame
51 * audio frame #0: 787 * (bytes/sample) * (# channels) bytes in frame
52 * audio frame #1: 788 * (bytes/sample) * (# channels) bytes in frame
53 * audio frame #2: 787 * (bytes/sample) * (# channels) bytes in frame
54 * audio frame #3: 788 * (bytes/sample) * (# channels) bytes in frame
56 * Finally, not all id CIN creation tools agree on the resolution of the
57 * color palette, apparently. Some creation tools specify red, green, and
58 * blue palette components in terms of 6-bit VGA color DAC values which
59 * range from 0..63. Other tools specify the RGB components as full 8-bit
60 * values that range from 0..255. Since there are no markers in the file to
61 * differentiate between the two variants, this demuxer uses the following
63 * - load the 768 palette bytes from disk
64 * - assume that they will need to be shifted left by 2 bits to
65 * transform them from 6-bit values to 8-bit values
66 * - scan through all 768 palette bytes
67 * - if any bytes exceed 63, do not shift the bytes at all before
68 * transmitting them to the video decoder
71 #include "libavutil/channel_layout.h"
72 #include "libavutil/imgutils.h"
73 #include "libavutil/intreadwrite.h"
78 #define HUFFMAN_TABLE_SIZE (64 * 1024)
81 typedef struct IdcinDemuxContext
{
82 int video_stream_index
;
83 int audio_stream_index
;
84 int audio_chunk_size1
;
85 int audio_chunk_size2
;
88 /* demux state variables */
89 int current_audio_chunk
;
90 int next_chunk_is_video
;
92 int64_t first_pkt_pos
;
95 static int idcin_probe(const AVProbeData
*p
)
97 unsigned int number
, sample_rate
;
102 * This is what you could call a "probabilistic" file check: id CIN
103 * files don't have a definite file signature. In lieu of such a marker,
104 * perform sanity checks on the 5 32-bit header fields:
105 * width, height: greater than 0, less than or equal to 1024
106 * audio sample rate: greater than or equal to 8000, less than or
107 * equal to 48000, or 0 for no audio
108 * audio sample width (bytes/sample): 0 for no audio, or 1 or 2
109 * audio channels: 0 for no audio, or 1 or 2
112 /* check we have enough data to do all checks, otherwise the
113 0-padding may cause a wrong recognition */
114 if (p
->buf_size
< 20 + HUFFMAN_TABLE_SIZE
+ 12)
117 /* check the video width */
118 w
= AV_RL32(&p
->buf
[0]);
119 if ((w
== 0) || (w
> 1024))
122 /* check the video height */
123 h
= AV_RL32(&p
->buf
[4]);
124 if ((h
== 0) || (h
> 1024))
127 /* check the audio sample rate */
128 sample_rate
= AV_RL32(&p
->buf
[8]);
129 if (sample_rate
&& (sample_rate
< 8000 || sample_rate
> 48000))
132 /* check the audio bytes/sample */
133 number
= AV_RL32(&p
->buf
[12]);
134 if (number
> 2 || sample_rate
&& !number
)
137 /* check the audio channels */
138 number
= AV_RL32(&p
->buf
[16]);
139 if (number
> 2 || sample_rate
&& !number
)
142 i
= 20 + HUFFMAN_TABLE_SIZE
;
143 if (AV_RL32(&p
->buf
[i
]) == 1)
146 if (i
+12 > p
->buf_size
|| AV_RL32(&p
->buf
[i
+8]) != w
*h
)
149 /* return half certainty since this check is a bit sketchy */
150 return AVPROBE_SCORE_EXTENSION
;
153 static int idcin_read_header(AVFormatContext
*s
)
155 AVIOContext
*pb
= s
->pb
;
156 IdcinDemuxContext
*idcin
= s
->priv_data
;
158 unsigned int width
, height
;
159 unsigned int sample_rate
, bytes_per_sample
, channels
;
162 /* get the 5 header parameters */
163 width
= avio_rl32(pb
);
164 height
= avio_rl32(pb
);
165 sample_rate
= avio_rl32(pb
);
166 bytes_per_sample
= avio_rl32(pb
);
167 channels
= avio_rl32(pb
);
169 if (s
->pb
->eof_reached
) {
170 av_log(s
, AV_LOG_ERROR
, "incomplete header\n");
171 return s
->pb
->error
? s
->pb
->error
: AVERROR_EOF
;
174 if (av_image_check_size(width
, height
, 0, s
) < 0)
175 return AVERROR_INVALIDDATA
;
176 if (sample_rate
> 0) {
177 if (sample_rate
< 14 || sample_rate
> INT_MAX
) {
178 av_log(s
, AV_LOG_ERROR
, "invalid sample rate: %u\n", sample_rate
);
179 return AVERROR_INVALIDDATA
;
181 if (bytes_per_sample
< 1 || bytes_per_sample
> 2) {
182 av_log(s
, AV_LOG_ERROR
, "invalid bytes per sample: %u\n",
184 return AVERROR_INVALIDDATA
;
186 if (channels
< 1 || channels
> 2) {
187 av_log(s
, AV_LOG_ERROR
, "invalid channels: %u\n", channels
);
188 return AVERROR_INVALIDDATA
;
190 idcin
->audio_present
= 1;
192 /* if sample rate is 0, assume no audio */
193 idcin
->audio_present
= 0;
196 st
= avformat_new_stream(s
, NULL
);
198 return AVERROR(ENOMEM
);
199 avpriv_set_pts_info(st
, 33, 1, IDCIN_FPS
);
201 idcin
->video_stream_index
= st
->index
;
202 st
->codecpar
->codec_type
= AVMEDIA_TYPE_VIDEO
;
203 st
->codecpar
->codec_id
= AV_CODEC_ID_IDCIN
;
204 st
->codecpar
->codec_tag
= 0; /* no fourcc */
205 st
->codecpar
->width
= width
;
206 st
->codecpar
->height
= height
;
208 /* load up the Huffman tables into extradata */
209 if ((ret
= ff_get_extradata(s
, st
->codecpar
, pb
, HUFFMAN_TABLE_SIZE
)) < 0)
212 if (idcin
->audio_present
) {
213 idcin
->audio_present
= 1;
214 st
= avformat_new_stream(s
, NULL
);
216 return AVERROR(ENOMEM
);
217 avpriv_set_pts_info(st
, 63, 1, sample_rate
);
219 idcin
->audio_stream_index
= st
->index
;
220 st
->codecpar
->codec_type
= AVMEDIA_TYPE_AUDIO
;
221 st
->codecpar
->codec_tag
= 1;
222 av_channel_layout_default(&st
->codecpar
->ch_layout
, channels
);
223 st
->codecpar
->sample_rate
= sample_rate
;
224 st
->codecpar
->bits_per_coded_sample
= bytes_per_sample
* 8;
225 st
->codecpar
->bit_rate
= sample_rate
* bytes_per_sample
* 8 * channels
;
226 st
->codecpar
->block_align
= idcin
->block_align
= bytes_per_sample
* channels
;
227 if (bytes_per_sample
== 1)
228 st
->codecpar
->codec_id
= AV_CODEC_ID_PCM_U8
;
230 st
->codecpar
->codec_id
= AV_CODEC_ID_PCM_S16LE
;
232 if (sample_rate
% 14 != 0) {
233 idcin
->audio_chunk_size1
= (sample_rate
/ 14) *
234 bytes_per_sample
* channels
;
235 idcin
->audio_chunk_size2
= (sample_rate
/ 14 + 1) *
236 bytes_per_sample
* channels
;
238 idcin
->audio_chunk_size1
= idcin
->audio_chunk_size2
=
239 (sample_rate
/ 14) * bytes_per_sample
* channels
;
241 idcin
->current_audio_chunk
= 0;
244 idcin
->next_chunk_is_video
= 1;
245 idcin
->first_pkt_pos
= avio_tell(s
->pb
);
250 static int idcin_read_packet(AVFormatContext
*s
,
254 unsigned int command
;
255 unsigned int chunk_size
;
256 IdcinDemuxContext
*idcin
= s
->priv_data
;
257 AVIOContext
*pb
= s
->pb
;
260 unsigned char r
, g
, b
;
261 unsigned char palette_buffer
[768];
262 uint32_t palette
[256];
264 if (avio_feof(s
->pb
))
265 return s
->pb
->error
? s
->pb
->error
: AVERROR_EOF
;
267 if (idcin
->next_chunk_is_video
) {
268 command
= avio_rl32(pb
);
270 return AVERROR_INVALIDDATA
;
271 } else if (command
== 1) {
272 /* trigger a palette change */
273 ret
= avio_read(pb
, palette_buffer
, 768);
276 } else if (ret
!= 768) {
277 av_log(s
, AV_LOG_ERROR
, "incomplete packet\n");
278 return AVERROR_INVALIDDATA
;
280 /* scale the palette as necessary */
282 for (i
= 0; i
< 768; i
++)
283 if (palette_buffer
[i
] > 63) {
288 for (i
= 0; i
< 256; i
++) {
289 r
= palette_buffer
[i
* 3 ] << palette_scale
;
290 g
= palette_buffer
[i
* 3 + 1] << palette_scale
;
291 b
= palette_buffer
[i
* 3 + 2] << palette_scale
;
292 palette
[i
] = (0xFFU
<< 24) | (r
<< 16) | (g
<< 8) | (b
);
293 if (palette_scale
== 2)
294 palette
[i
] |= palette
[i
] >> 6 & 0x30303;
298 if (s
->pb
->eof_reached
) {
299 av_log(s
, AV_LOG_ERROR
, "incomplete packet\n");
300 return s
->pb
->error
? s
->pb
->error
: AVERROR_EOF
;
302 chunk_size
= avio_rl32(pb
);
303 if (chunk_size
< 4 || chunk_size
> INT_MAX
- 4) {
304 av_log(s
, AV_LOG_ERROR
, "invalid chunk size: %u\n", chunk_size
);
305 return AVERROR_INVALIDDATA
;
307 /* skip the number of decoded bytes (always equal to width * height) */
310 ret
= av_get_packet(pb
, pkt
, chunk_size
);
313 else if (ret
!= chunk_size
) {
314 av_log(s
, AV_LOG_ERROR
, "incomplete packet\n");
315 return AVERROR_INVALIDDATA
;
320 pal
= av_packet_new_side_data(pkt
, AV_PKT_DATA_PALETTE
,
323 return AVERROR(ENOMEM
);
325 memcpy(pal
, palette
, AVPALETTE_SIZE
);
326 pkt
->flags
|= AV_PKT_FLAG_KEY
;
328 pkt
->stream_index
= idcin
->video_stream_index
;
331 /* send out the audio chunk */
332 if (idcin
->current_audio_chunk
)
333 chunk_size
= idcin
->audio_chunk_size2
;
335 chunk_size
= idcin
->audio_chunk_size1
;
336 ret
= av_get_packet(pb
, pkt
, chunk_size
);
339 pkt
->stream_index
= idcin
->audio_stream_index
;
340 pkt
->duration
= chunk_size
/ idcin
->block_align
;
342 idcin
->current_audio_chunk
^= 1;
345 if (idcin
->audio_present
)
346 idcin
->next_chunk_is_video
^= 1;
351 static int idcin_read_seek(AVFormatContext
*s
, int stream_index
,
352 int64_t timestamp
, int flags
)
354 IdcinDemuxContext
*idcin
= s
->priv_data
;
356 if (idcin
->first_pkt_pos
> 0) {
357 int64_t ret
= avio_seek(s
->pb
, idcin
->first_pkt_pos
, SEEK_SET
);
360 avpriv_update_cur_dts(s
, s
->streams
[idcin
->video_stream_index
], 0);
361 idcin
->next_chunk_is_video
= 1;
362 idcin
->current_audio_chunk
= 0;
368 const FFInputFormat ff_idcin_demuxer
= {
370 .p
.long_name
= NULL_IF_CONFIG_SMALL("id Cinematic"),
371 .p
.flags
= AVFMT_NO_BYTE_SEEK
,
372 .priv_data_size
= sizeof(IdcinDemuxContext
),
373 .read_probe
= idcin_probe
,
374 .read_header
= idcin_read_header
,
375 .read_packet
= idcin_read_packet
,
376 .read_seek
= idcin_read_seek
,