| 1 | /* |
| 2 | * id Quake II CIN File Demuxer |
| 3 | * Copyright (c) 2003 The FFmpeg Project |
| 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 | /** |
| 23 | * @file |
| 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/ |
| 27 | * |
| 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: |
| 30 | * |
| 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 | * Probabalistically, 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. |
| 39 | * |
| 40 | * Refer to the function idcin_probe() for the precise A/V parameters |
| 41 | * that this demuxer allows. |
| 42 | * |
| 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 |
| 50 | * is given as: |
| 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 |
| 55 | * |
| 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 |
| 62 | * heuristic: |
| 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 |
| 69 | */ |
| 70 | |
| 71 | #include "libavutil/channel_layout.h" |
| 72 | #include "libavutil/imgutils.h" |
| 73 | #include "libavutil/intreadwrite.h" |
| 74 | #include "avformat.h" |
| 75 | #include "internal.h" |
| 76 | |
| 77 | #define HUFFMAN_TABLE_SIZE (64 * 1024) |
| 78 | #define IDCIN_FPS 14 |
| 79 | |
| 80 | typedef struct IdcinDemuxContext { |
| 81 | int video_stream_index; |
| 82 | int audio_stream_index; |
| 83 | int audio_chunk_size1; |
| 84 | int audio_chunk_size2; |
| 85 | int block_align; |
| 86 | |
| 87 | /* demux state variables */ |
| 88 | int current_audio_chunk; |
| 89 | int next_chunk_is_video; |
| 90 | int audio_present; |
| 91 | int64_t first_pkt_pos; |
| 92 | } IdcinDemuxContext; |
| 93 | |
| 94 | static int idcin_probe(AVProbeData *p) |
| 95 | { |
| 96 | unsigned int number, sample_rate; |
| 97 | unsigned int w, h; |
| 98 | int i; |
| 99 | |
| 100 | /* |
| 101 | * This is what you could call a "probabilistic" file check: id CIN |
| 102 | * files don't have a definite file signature. In lieu of such a marker, |
| 103 | * perform sanity checks on the 5 32-bit header fields: |
| 104 | * width, height: greater than 0, less than or equal to 1024 |
| 105 | * audio sample rate: greater than or equal to 8000, less than or |
| 106 | * equal to 48000, or 0 for no audio |
| 107 | * audio sample width (bytes/sample): 0 for no audio, or 1 or 2 |
| 108 | * audio channels: 0 for no audio, or 1 or 2 |
| 109 | */ |
| 110 | |
| 111 | /* check we have enough data to do all checks, otherwise the |
| 112 | 0-padding may cause a wrong recognition */ |
| 113 | if (p->buf_size < 20 + HUFFMAN_TABLE_SIZE + 12) |
| 114 | return 0; |
| 115 | |
| 116 | /* check the video width */ |
| 117 | w = AV_RL32(&p->buf[0]); |
| 118 | if ((w == 0) || (w > 1024)) |
| 119 | return 0; |
| 120 | |
| 121 | /* check the video height */ |
| 122 | h = AV_RL32(&p->buf[4]); |
| 123 | if ((h == 0) || (h > 1024)) |
| 124 | return 0; |
| 125 | |
| 126 | /* check the audio sample rate */ |
| 127 | sample_rate = AV_RL32(&p->buf[8]); |
| 128 | if (sample_rate && (sample_rate < 8000 || sample_rate > 48000)) |
| 129 | return 0; |
| 130 | |
| 131 | /* check the audio bytes/sample */ |
| 132 | number = AV_RL32(&p->buf[12]); |
| 133 | if (number > 2 || sample_rate && !number) |
| 134 | return 0; |
| 135 | |
| 136 | /* check the audio channels */ |
| 137 | number = AV_RL32(&p->buf[16]); |
| 138 | if (number > 2 || sample_rate && !number) |
| 139 | return 0; |
| 140 | |
| 141 | i = 20 + HUFFMAN_TABLE_SIZE; |
| 142 | if (AV_RL32(&p->buf[i]) == 1) |
| 143 | i += 768; |
| 144 | |
| 145 | if (i+12 > p->buf_size || AV_RL32(&p->buf[i+8]) != w*h) |
| 146 | return 1; |
| 147 | |
| 148 | /* return half certainty since this check is a bit sketchy */ |
| 149 | return AVPROBE_SCORE_EXTENSION; |
| 150 | } |
| 151 | |
| 152 | static int idcin_read_header(AVFormatContext *s) |
| 153 | { |
| 154 | AVIOContext *pb = s->pb; |
| 155 | IdcinDemuxContext *idcin = s->priv_data; |
| 156 | AVStream *st; |
| 157 | unsigned int width, height; |
| 158 | unsigned int sample_rate, bytes_per_sample, channels; |
| 159 | int ret; |
| 160 | |
| 161 | /* get the 5 header parameters */ |
| 162 | width = avio_rl32(pb); |
| 163 | height = avio_rl32(pb); |
| 164 | sample_rate = avio_rl32(pb); |
| 165 | bytes_per_sample = avio_rl32(pb); |
| 166 | channels = avio_rl32(pb); |
| 167 | |
| 168 | if (s->pb->eof_reached) { |
| 169 | av_log(s, AV_LOG_ERROR, "incomplete header\n"); |
| 170 | return s->pb->error ? s->pb->error : AVERROR_EOF; |
| 171 | } |
| 172 | |
| 173 | if (av_image_check_size(width, height, 0, s) < 0) |
| 174 | return AVERROR_INVALIDDATA; |
| 175 | if (sample_rate > 0) { |
| 176 | if (sample_rate < 14 || sample_rate > INT_MAX) { |
| 177 | av_log(s, AV_LOG_ERROR, "invalid sample rate: %u\n", sample_rate); |
| 178 | return AVERROR_INVALIDDATA; |
| 179 | } |
| 180 | if (bytes_per_sample < 1 || bytes_per_sample > 2) { |
| 181 | av_log(s, AV_LOG_ERROR, "invalid bytes per sample: %u\n", |
| 182 | bytes_per_sample); |
| 183 | return AVERROR_INVALIDDATA; |
| 184 | } |
| 185 | if (channels < 1 || channels > 2) { |
| 186 | av_log(s, AV_LOG_ERROR, "invalid channels: %u\n", channels); |
| 187 | return AVERROR_INVALIDDATA; |
| 188 | } |
| 189 | idcin->audio_present = 1; |
| 190 | } else { |
| 191 | /* if sample rate is 0, assume no audio */ |
| 192 | idcin->audio_present = 0; |
| 193 | } |
| 194 | |
| 195 | st = avformat_new_stream(s, NULL); |
| 196 | if (!st) |
| 197 | return AVERROR(ENOMEM); |
| 198 | avpriv_set_pts_info(st, 33, 1, IDCIN_FPS); |
| 199 | st->start_time = 0; |
| 200 | idcin->video_stream_index = st->index; |
| 201 | st->codec->codec_type = AVMEDIA_TYPE_VIDEO; |
| 202 | st->codec->codec_id = AV_CODEC_ID_IDCIN; |
| 203 | st->codec->codec_tag = 0; /* no fourcc */ |
| 204 | st->codec->width = width; |
| 205 | st->codec->height = height; |
| 206 | |
| 207 | /* load up the Huffman tables into extradata */ |
| 208 | if ((ret = ff_get_extradata(st->codec, pb, HUFFMAN_TABLE_SIZE)) < 0) |
| 209 | return ret; |
| 210 | |
| 211 | if (idcin->audio_present) { |
| 212 | idcin->audio_present = 1; |
| 213 | st = avformat_new_stream(s, NULL); |
| 214 | if (!st) |
| 215 | return AVERROR(ENOMEM); |
| 216 | avpriv_set_pts_info(st, 63, 1, sample_rate); |
| 217 | st->start_time = 0; |
| 218 | idcin->audio_stream_index = st->index; |
| 219 | st->codec->codec_type = AVMEDIA_TYPE_AUDIO; |
| 220 | st->codec->codec_tag = 1; |
| 221 | st->codec->channels = channels; |
| 222 | st->codec->channel_layout = channels > 1 ? AV_CH_LAYOUT_STEREO : |
| 223 | AV_CH_LAYOUT_MONO; |
| 224 | st->codec->sample_rate = sample_rate; |
| 225 | st->codec->bits_per_coded_sample = bytes_per_sample * 8; |
| 226 | st->codec->bit_rate = sample_rate * bytes_per_sample * 8 * channels; |
| 227 | st->codec->block_align = idcin->block_align = bytes_per_sample * channels; |
| 228 | if (bytes_per_sample == 1) |
| 229 | st->codec->codec_id = AV_CODEC_ID_PCM_U8; |
| 230 | else |
| 231 | st->codec->codec_id = AV_CODEC_ID_PCM_S16LE; |
| 232 | |
| 233 | if (sample_rate % 14 != 0) { |
| 234 | idcin->audio_chunk_size1 = (sample_rate / 14) * |
| 235 | bytes_per_sample * channels; |
| 236 | idcin->audio_chunk_size2 = (sample_rate / 14 + 1) * |
| 237 | bytes_per_sample * channels; |
| 238 | } else { |
| 239 | idcin->audio_chunk_size1 = idcin->audio_chunk_size2 = |
| 240 | (sample_rate / 14) * bytes_per_sample * channels; |
| 241 | } |
| 242 | idcin->current_audio_chunk = 0; |
| 243 | } |
| 244 | |
| 245 | idcin->next_chunk_is_video = 1; |
| 246 | idcin->first_pkt_pos = avio_tell(s->pb); |
| 247 | |
| 248 | return 0; |
| 249 | } |
| 250 | |
| 251 | static int idcin_read_packet(AVFormatContext *s, |
| 252 | AVPacket *pkt) |
| 253 | { |
| 254 | int ret; |
| 255 | unsigned int command; |
| 256 | unsigned int chunk_size; |
| 257 | IdcinDemuxContext *idcin = s->priv_data; |
| 258 | AVIOContext *pb = s->pb; |
| 259 | int i; |
| 260 | int palette_scale; |
| 261 | unsigned char r, g, b; |
| 262 | unsigned char palette_buffer[768]; |
| 263 | uint32_t palette[256]; |
| 264 | |
| 265 | if (avio_feof(s->pb)) |
| 266 | return s->pb->error ? s->pb->error : AVERROR_EOF; |
| 267 | |
| 268 | if (idcin->next_chunk_is_video) { |
| 269 | command = avio_rl32(pb); |
| 270 | if (command == 2) { |
| 271 | return AVERROR(EIO); |
| 272 | } else if (command == 1) { |
| 273 | /* trigger a palette change */ |
| 274 | ret = avio_read(pb, palette_buffer, 768); |
| 275 | if (ret < 0) { |
| 276 | return ret; |
| 277 | } else if (ret != 768) { |
| 278 | av_log(s, AV_LOG_ERROR, "incomplete packet\n"); |
| 279 | return AVERROR(EIO); |
| 280 | } |
| 281 | /* scale the palette as necessary */ |
| 282 | palette_scale = 2; |
| 283 | for (i = 0; i < 768; i++) |
| 284 | if (palette_buffer[i] > 63) { |
| 285 | palette_scale = 0; |
| 286 | break; |
| 287 | } |
| 288 | |
| 289 | for (i = 0; i < 256; i++) { |
| 290 | r = palette_buffer[i * 3 ] << palette_scale; |
| 291 | g = palette_buffer[i * 3 + 1] << palette_scale; |
| 292 | b = palette_buffer[i * 3 + 2] << palette_scale; |
| 293 | palette[i] = (0xFFU << 24) | (r << 16) | (g << 8) | (b); |
| 294 | if (palette_scale == 2) |
| 295 | palette[i] |= palette[i] >> 6 & 0x30303; |
| 296 | } |
| 297 | } |
| 298 | |
| 299 | if (s->pb->eof_reached) { |
| 300 | av_log(s, AV_LOG_ERROR, "incomplete packet\n"); |
| 301 | return s->pb->error ? s->pb->error : AVERROR_EOF; |
| 302 | } |
| 303 | chunk_size = avio_rl32(pb); |
| 304 | if (chunk_size < 4 || chunk_size > INT_MAX - 4) { |
| 305 | av_log(s, AV_LOG_ERROR, "invalid chunk size: %u\n", chunk_size); |
| 306 | return AVERROR_INVALIDDATA; |
| 307 | } |
| 308 | /* skip the number of decoded bytes (always equal to width * height) */ |
| 309 | avio_skip(pb, 4); |
| 310 | if (chunk_size < 4) |
| 311 | return AVERROR_INVALIDDATA; |
| 312 | chunk_size -= 4; |
| 313 | ret= av_get_packet(pb, pkt, chunk_size); |
| 314 | if (ret < 0) |
| 315 | return ret; |
| 316 | else if (ret != chunk_size) { |
| 317 | av_log(s, AV_LOG_ERROR, "incomplete packet\n"); |
| 318 | av_free_packet(pkt); |
| 319 | return AVERROR(EIO); |
| 320 | } |
| 321 | if (command == 1) { |
| 322 | uint8_t *pal; |
| 323 | |
| 324 | pal = av_packet_new_side_data(pkt, AV_PKT_DATA_PALETTE, |
| 325 | AVPALETTE_SIZE); |
| 326 | if (!pal) { |
| 327 | av_free_packet(pkt); |
| 328 | return AVERROR(ENOMEM); |
| 329 | } |
| 330 | memcpy(pal, palette, AVPALETTE_SIZE); |
| 331 | pkt->flags |= AV_PKT_FLAG_KEY; |
| 332 | } |
| 333 | pkt->stream_index = idcin->video_stream_index; |
| 334 | pkt->duration = 1; |
| 335 | } else { |
| 336 | /* send out the audio chunk */ |
| 337 | if (idcin->current_audio_chunk) |
| 338 | chunk_size = idcin->audio_chunk_size2; |
| 339 | else |
| 340 | chunk_size = idcin->audio_chunk_size1; |
| 341 | ret= av_get_packet(pb, pkt, chunk_size); |
| 342 | if (ret < 0) |
| 343 | return ret; |
| 344 | pkt->stream_index = idcin->audio_stream_index; |
| 345 | pkt->duration = chunk_size / idcin->block_align; |
| 346 | |
| 347 | idcin->current_audio_chunk ^= 1; |
| 348 | } |
| 349 | |
| 350 | if (idcin->audio_present) |
| 351 | idcin->next_chunk_is_video ^= 1; |
| 352 | |
| 353 | return 0; |
| 354 | } |
| 355 | |
| 356 | static int idcin_read_seek(AVFormatContext *s, int stream_index, |
| 357 | int64_t timestamp, int flags) |
| 358 | { |
| 359 | IdcinDemuxContext *idcin = s->priv_data; |
| 360 | |
| 361 | if (idcin->first_pkt_pos > 0) { |
| 362 | int ret = avio_seek(s->pb, idcin->first_pkt_pos, SEEK_SET); |
| 363 | if (ret < 0) |
| 364 | return ret; |
| 365 | ff_update_cur_dts(s, s->streams[idcin->video_stream_index], 0); |
| 366 | idcin->next_chunk_is_video = 1; |
| 367 | idcin->current_audio_chunk = 0; |
| 368 | return 0; |
| 369 | } |
| 370 | return -1; |
| 371 | } |
| 372 | |
| 373 | AVInputFormat ff_idcin_demuxer = { |
| 374 | .name = "idcin", |
| 375 | .long_name = NULL_IF_CONFIG_SMALL("id Cinematic"), |
| 376 | .priv_data_size = sizeof(IdcinDemuxContext), |
| 377 | .read_probe = idcin_probe, |
| 378 | .read_header = idcin_read_header, |
| 379 | .read_packet = idcin_read_packet, |
| 380 | .read_seek = idcin_read_seek, |
| 381 | .flags = AVFMT_NO_BYTE_SEEK, |
| 382 | }; |