| 1 | /* |
| 2 | * RTP JPEG-compressed Video Depacketizer, RFC 2435 |
| 3 | * Copyright (c) 2012 Samuel Pitoiset |
| 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 | #include "avformat.h" |
| 23 | #include "rtpdec.h" |
| 24 | #include "rtpdec_formats.h" |
| 25 | #include "libavutil/intreadwrite.h" |
| 26 | #include "libavcodec/mjpeg.h" |
| 27 | #include "libavcodec/bytestream.h" |
| 28 | |
| 29 | /** |
| 30 | * RTP/JPEG specific private data. |
| 31 | */ |
| 32 | struct PayloadContext { |
| 33 | AVIOContext *frame; ///< current frame buffer |
| 34 | uint32_t timestamp; ///< current frame timestamp |
| 35 | int hdr_size; ///< size of the current frame header |
| 36 | uint8_t qtables[128][128]; |
| 37 | uint8_t qtables_len[128]; |
| 38 | }; |
| 39 | |
| 40 | static const uint8_t default_quantizers[128] = { |
| 41 | /* luma table */ |
| 42 | 16, 11, 12, 14, 12, 10, 16, 14, |
| 43 | 13, 14, 18, 17, 16, 19, 24, 40, |
| 44 | 26, 24, 22, 22, 24, 49, 35, 37, |
| 45 | 29, 40, 58, 51, 61, 60, 57, 51, |
| 46 | 56, 55, 64, 72, 92, 78, 64, 68, |
| 47 | 87, 69, 55, 56, 80, 109, 81, 87, |
| 48 | 95, 98, 103, 104, 103, 62, 77, 113, |
| 49 | 121, 112, 100, 120, 92, 101, 103, 99, |
| 50 | |
| 51 | /* chroma table */ |
| 52 | 17, 18, 18, 24, 21, 24, 47, 26, |
| 53 | 26, 47, 99, 66, 56, 66, 99, 99, |
| 54 | 99, 99, 99, 99, 99, 99, 99, 99, |
| 55 | 99, 99, 99, 99, 99, 99, 99, 99, |
| 56 | 99, 99, 99, 99, 99, 99, 99, 99, |
| 57 | 99, 99, 99, 99, 99, 99, 99, 99, |
| 58 | 99, 99, 99, 99, 99, 99, 99, 99, |
| 59 | 99, 99, 99, 99, 99, 99, 99, 99 |
| 60 | }; |
| 61 | |
| 62 | static PayloadContext *jpeg_new_context(void) |
| 63 | { |
| 64 | return av_mallocz(sizeof(PayloadContext)); |
| 65 | } |
| 66 | |
| 67 | static inline void free_frame_if_needed(PayloadContext *jpeg) |
| 68 | { |
| 69 | if (jpeg->frame) { |
| 70 | uint8_t *p; |
| 71 | avio_close_dyn_buf(jpeg->frame, &p); |
| 72 | av_free(p); |
| 73 | jpeg->frame = NULL; |
| 74 | } |
| 75 | } |
| 76 | |
| 77 | static void jpeg_free_context(PayloadContext *jpeg) |
| 78 | { |
| 79 | free_frame_if_needed(jpeg); |
| 80 | av_free(jpeg); |
| 81 | } |
| 82 | |
| 83 | static int jpeg_create_huffman_table(PutByteContext *p, int table_class, |
| 84 | int table_id, const uint8_t *bits_table, |
| 85 | const uint8_t *value_table) |
| 86 | { |
| 87 | int i, n = 0; |
| 88 | |
| 89 | bytestream2_put_byte(p, table_class << 4 | table_id); |
| 90 | |
| 91 | for (i = 1; i <= 16; i++) { |
| 92 | n += bits_table[i]; |
| 93 | bytestream2_put_byte(p, bits_table[i]); |
| 94 | } |
| 95 | |
| 96 | for (i = 0; i < n; i++) { |
| 97 | bytestream2_put_byte(p, value_table[i]); |
| 98 | } |
| 99 | return n + 17; |
| 100 | } |
| 101 | |
| 102 | static void jpeg_put_marker(PutByteContext *pbc, int code) |
| 103 | { |
| 104 | bytestream2_put_byte(pbc, 0xff); |
| 105 | bytestream2_put_byte(pbc, code); |
| 106 | } |
| 107 | |
| 108 | static int jpeg_create_header(uint8_t *buf, int size, uint32_t type, uint32_t w, |
| 109 | uint32_t h, const uint8_t *qtable, int nb_qtable, |
| 110 | int dri) |
| 111 | { |
| 112 | PutByteContext pbc; |
| 113 | uint8_t *dht_size_ptr; |
| 114 | int dht_size, i; |
| 115 | |
| 116 | bytestream2_init_writer(&pbc, buf, size); |
| 117 | |
| 118 | /* Convert from blocks to pixels. */ |
| 119 | w <<= 3; |
| 120 | h <<= 3; |
| 121 | |
| 122 | /* SOI */ |
| 123 | jpeg_put_marker(&pbc, SOI); |
| 124 | |
| 125 | /* JFIF header */ |
| 126 | jpeg_put_marker(&pbc, APP0); |
| 127 | bytestream2_put_be16(&pbc, 16); |
| 128 | bytestream2_put_buffer(&pbc, "JFIF", 5); |
| 129 | bytestream2_put_be16(&pbc, 0x0201); |
| 130 | bytestream2_put_byte(&pbc, 0); |
| 131 | bytestream2_put_be16(&pbc, 1); |
| 132 | bytestream2_put_be16(&pbc, 1); |
| 133 | bytestream2_put_byte(&pbc, 0); |
| 134 | bytestream2_put_byte(&pbc, 0); |
| 135 | |
| 136 | if (dri) { |
| 137 | jpeg_put_marker(&pbc, DRI); |
| 138 | bytestream2_put_be16(&pbc, 4); |
| 139 | bytestream2_put_be16(&pbc, dri); |
| 140 | } |
| 141 | |
| 142 | /* DQT */ |
| 143 | jpeg_put_marker(&pbc, DQT); |
| 144 | bytestream2_put_be16(&pbc, 2 + nb_qtable * (1 + 64)); |
| 145 | |
| 146 | for (i = 0; i < nb_qtable; i++) { |
| 147 | bytestream2_put_byte(&pbc, i); |
| 148 | |
| 149 | /* Each table is an array of 64 values given in zig-zag |
| 150 | * order, identical to the format used in a JFIF DQT |
| 151 | * marker segment. */ |
| 152 | bytestream2_put_buffer(&pbc, qtable + 64 * i, 64); |
| 153 | } |
| 154 | |
| 155 | /* DHT */ |
| 156 | jpeg_put_marker(&pbc, DHT); |
| 157 | dht_size_ptr = pbc.buffer; |
| 158 | bytestream2_put_be16(&pbc, 0); |
| 159 | |
| 160 | dht_size = 2; |
| 161 | dht_size += jpeg_create_huffman_table(&pbc, 0, 0,avpriv_mjpeg_bits_dc_luminance, |
| 162 | avpriv_mjpeg_val_dc); |
| 163 | dht_size += jpeg_create_huffman_table(&pbc, 0, 1, avpriv_mjpeg_bits_dc_chrominance, |
| 164 | avpriv_mjpeg_val_dc); |
| 165 | dht_size += jpeg_create_huffman_table(&pbc, 1, 0, avpriv_mjpeg_bits_ac_luminance, |
| 166 | avpriv_mjpeg_val_ac_luminance); |
| 167 | dht_size += jpeg_create_huffman_table(&pbc, 1, 1, avpriv_mjpeg_bits_ac_chrominance, |
| 168 | avpriv_mjpeg_val_ac_chrominance); |
| 169 | AV_WB16(dht_size_ptr, dht_size); |
| 170 | |
| 171 | /* SOF0 */ |
| 172 | jpeg_put_marker(&pbc, SOF0); |
| 173 | bytestream2_put_be16(&pbc, 17); /* size */ |
| 174 | bytestream2_put_byte(&pbc, 8); /* bits per component */ |
| 175 | bytestream2_put_be16(&pbc, h); |
| 176 | bytestream2_put_be16(&pbc, w); |
| 177 | bytestream2_put_byte(&pbc, 3); /* number of components */ |
| 178 | bytestream2_put_byte(&pbc, 1); /* component number */ |
| 179 | bytestream2_put_byte(&pbc, (2 << 4) | (type ? 2 : 1)); /* hsample/vsample */ |
| 180 | bytestream2_put_byte(&pbc, 0); /* matrix number */ |
| 181 | bytestream2_put_byte(&pbc, 2); /* component number */ |
| 182 | bytestream2_put_byte(&pbc, 1 << 4 | 1); /* hsample/vsample */ |
| 183 | bytestream2_put_byte(&pbc, nb_qtable == 2 ? 1 : 0); /* matrix number */ |
| 184 | bytestream2_put_byte(&pbc, 3); /* component number */ |
| 185 | bytestream2_put_byte(&pbc, 1 << 4 | 1); /* hsample/vsample */ |
| 186 | bytestream2_put_byte(&pbc, nb_qtable == 2 ? 1 : 0); /* matrix number */ |
| 187 | |
| 188 | /* SOS */ |
| 189 | jpeg_put_marker(&pbc, SOS); |
| 190 | bytestream2_put_be16(&pbc, 12); |
| 191 | bytestream2_put_byte(&pbc, 3); |
| 192 | bytestream2_put_byte(&pbc, 1); |
| 193 | bytestream2_put_byte(&pbc, 0); |
| 194 | bytestream2_put_byte(&pbc, 2); |
| 195 | bytestream2_put_byte(&pbc, 17); |
| 196 | bytestream2_put_byte(&pbc, 3); |
| 197 | bytestream2_put_byte(&pbc, 17); |
| 198 | bytestream2_put_byte(&pbc, 0); |
| 199 | bytestream2_put_byte(&pbc, 63); |
| 200 | bytestream2_put_byte(&pbc, 0); |
| 201 | |
| 202 | /* Return the length in bytes of the JPEG header. */ |
| 203 | return bytestream2_tell_p(&pbc); |
| 204 | } |
| 205 | |
| 206 | static void create_default_qtables(uint8_t *qtables, uint8_t q) |
| 207 | { |
| 208 | int factor = q; |
| 209 | int i; |
| 210 | |
| 211 | factor = av_clip(q, 1, 99); |
| 212 | |
| 213 | if (q < 50) |
| 214 | q = 5000 / factor; |
| 215 | else |
| 216 | q = 200 - factor * 2; |
| 217 | |
| 218 | for (i = 0; i < 128; i++) { |
| 219 | int val = (default_quantizers[i] * q + 50) / 100; |
| 220 | |
| 221 | /* Limit the quantizers to 1 <= q <= 255. */ |
| 222 | val = av_clip(val, 1, 255); |
| 223 | qtables[i] = val; |
| 224 | } |
| 225 | } |
| 226 | |
| 227 | static int jpeg_parse_packet(AVFormatContext *ctx, PayloadContext *jpeg, |
| 228 | AVStream *st, AVPacket *pkt, uint32_t *timestamp, |
| 229 | const uint8_t *buf, int len, uint16_t seq, |
| 230 | int flags) |
| 231 | { |
| 232 | uint8_t type, q, width, height; |
| 233 | const uint8_t *qtables = NULL; |
| 234 | uint16_t qtable_len; |
| 235 | uint32_t off; |
| 236 | int ret, dri = 0; |
| 237 | |
| 238 | if (len < 8) { |
| 239 | av_log(ctx, AV_LOG_ERROR, "Too short RTP/JPEG packet.\n"); |
| 240 | return AVERROR_INVALIDDATA; |
| 241 | } |
| 242 | |
| 243 | /* Parse the main JPEG header. */ |
| 244 | off = AV_RB24(buf + 1); /* fragment byte offset */ |
| 245 | type = AV_RB8(buf + 4); /* id of jpeg decoder params */ |
| 246 | q = AV_RB8(buf + 5); /* quantization factor (or table id) */ |
| 247 | width = AV_RB8(buf + 6); /* frame width in 8 pixel blocks */ |
| 248 | height = AV_RB8(buf + 7); /* frame height in 8 pixel blocks */ |
| 249 | buf += 8; |
| 250 | len -= 8; |
| 251 | |
| 252 | if (type & 0x40) { |
| 253 | if (len < 4) { |
| 254 | av_log(ctx, AV_LOG_ERROR, "Too short RTP/JPEG packet.\n"); |
| 255 | return AVERROR_INVALIDDATA; |
| 256 | } |
| 257 | dri = AV_RB16(buf); |
| 258 | buf += 4; |
| 259 | len -= 4; |
| 260 | type &= ~0x40; |
| 261 | } |
| 262 | /* Parse the restart marker header. */ |
| 263 | if (type > 63) { |
| 264 | av_log(ctx, AV_LOG_ERROR, |
| 265 | "Unimplemented RTP/JPEG restart marker header.\n"); |
| 266 | return AVERROR_PATCHWELCOME; |
| 267 | } |
| 268 | if (type > 1) { |
| 269 | av_log(ctx, AV_LOG_ERROR, "Unimplemented RTP/JPEG type %d\n", type); |
| 270 | return AVERROR_PATCHWELCOME; |
| 271 | } |
| 272 | |
| 273 | /* Parse the quantization table header. */ |
| 274 | if (off == 0) { |
| 275 | /* Start of JPEG data packet. */ |
| 276 | uint8_t new_qtables[128]; |
| 277 | uint8_t hdr[1024]; |
| 278 | |
| 279 | if (q > 127) { |
| 280 | uint8_t precision; |
| 281 | if (len < 4) { |
| 282 | av_log(ctx, AV_LOG_ERROR, "Too short RTP/JPEG packet.\n"); |
| 283 | return AVERROR_INVALIDDATA; |
| 284 | } |
| 285 | |
| 286 | /* The first byte is reserved for future use. */ |
| 287 | precision = AV_RB8(buf + 1); /* size of coefficients */ |
| 288 | qtable_len = AV_RB16(buf + 2); /* length in bytes */ |
| 289 | buf += 4; |
| 290 | len -= 4; |
| 291 | |
| 292 | if (precision) |
| 293 | av_log(ctx, AV_LOG_WARNING, "Only 8-bit precision is supported.\n"); |
| 294 | |
| 295 | if (qtable_len > 0) { |
| 296 | if (len < qtable_len) { |
| 297 | av_log(ctx, AV_LOG_ERROR, "Too short RTP/JPEG packet.\n"); |
| 298 | return AVERROR_INVALIDDATA; |
| 299 | } |
| 300 | qtables = buf; |
| 301 | buf += qtable_len; |
| 302 | len -= qtable_len; |
| 303 | if (q < 255) { |
| 304 | if (jpeg->qtables_len[q - 128] && |
| 305 | (jpeg->qtables_len[q - 128] != qtable_len || |
| 306 | memcmp(qtables, &jpeg->qtables[q - 128][0], qtable_len))) { |
| 307 | av_log(ctx, AV_LOG_WARNING, |
| 308 | "Quantization tables for q=%d changed\n", q); |
| 309 | } else if (!jpeg->qtables_len[q - 128] && qtable_len <= 128) { |
| 310 | memcpy(&jpeg->qtables[q - 128][0], qtables, |
| 311 | qtable_len); |
| 312 | jpeg->qtables_len[q - 128] = qtable_len; |
| 313 | } |
| 314 | } |
| 315 | } else { |
| 316 | if (q == 255) { |
| 317 | av_log(ctx, AV_LOG_ERROR, |
| 318 | "Invalid RTP/JPEG packet. Quantization tables not found.\n"); |
| 319 | return AVERROR_INVALIDDATA; |
| 320 | } |
| 321 | if (!jpeg->qtables_len[q - 128]) { |
| 322 | av_log(ctx, AV_LOG_ERROR, |
| 323 | "No quantization tables known for q=%d yet.\n", q); |
| 324 | return AVERROR_INVALIDDATA; |
| 325 | } |
| 326 | qtables = &jpeg->qtables[q - 128][0]; |
| 327 | qtable_len = jpeg->qtables_len[q - 128]; |
| 328 | } |
| 329 | } else { /* q <= 127 */ |
| 330 | if (q == 0 || q > 99) { |
| 331 | av_log(ctx, AV_LOG_ERROR, "Reserved q value %d\n", q); |
| 332 | return AVERROR_INVALIDDATA; |
| 333 | } |
| 334 | create_default_qtables(new_qtables, q); |
| 335 | qtables = new_qtables; |
| 336 | qtable_len = sizeof(new_qtables); |
| 337 | } |
| 338 | |
| 339 | /* Skip the current frame in case of the end packet |
| 340 | * has been lost somewhere. */ |
| 341 | free_frame_if_needed(jpeg); |
| 342 | |
| 343 | if ((ret = avio_open_dyn_buf(&jpeg->frame)) < 0) |
| 344 | return ret; |
| 345 | jpeg->timestamp = *timestamp; |
| 346 | |
| 347 | /* Generate a frame and scan headers that can be prepended to the |
| 348 | * RTP/JPEG data payload to produce a JPEG compressed image in |
| 349 | * interchange format. */ |
| 350 | jpeg->hdr_size = jpeg_create_header(hdr, sizeof(hdr), type, width, |
| 351 | height, qtables, |
| 352 | qtable_len / 64, dri); |
| 353 | |
| 354 | /* Copy JPEG header to frame buffer. */ |
| 355 | avio_write(jpeg->frame, hdr, jpeg->hdr_size); |
| 356 | } |
| 357 | |
| 358 | if (!jpeg->frame) { |
| 359 | av_log(ctx, AV_LOG_ERROR, |
| 360 | "Received packet without a start chunk; dropping frame.\n"); |
| 361 | return AVERROR(EAGAIN); |
| 362 | } |
| 363 | |
| 364 | if (jpeg->timestamp != *timestamp) { |
| 365 | /* Skip the current frame if timestamp is incorrect. |
| 366 | * A start packet has been lost somewhere. */ |
| 367 | free_frame_if_needed(jpeg); |
| 368 | av_log(ctx, AV_LOG_ERROR, "RTP timestamps don't match.\n"); |
| 369 | return AVERROR_INVALIDDATA; |
| 370 | } |
| 371 | |
| 372 | if (off != avio_tell(jpeg->frame) - jpeg->hdr_size) { |
| 373 | av_log(ctx, AV_LOG_ERROR, |
| 374 | "Missing packets; dropping frame.\n"); |
| 375 | return AVERROR(EAGAIN); |
| 376 | } |
| 377 | |
| 378 | /* Copy data to frame buffer. */ |
| 379 | avio_write(jpeg->frame, buf, len); |
| 380 | |
| 381 | if (flags & RTP_FLAG_MARKER) { |
| 382 | /* End of JPEG data packet. */ |
| 383 | uint8_t buf[2] = { 0xff, EOI }; |
| 384 | |
| 385 | /* Put EOI marker. */ |
| 386 | avio_write(jpeg->frame, buf, sizeof(buf)); |
| 387 | |
| 388 | /* Prepare the JPEG packet. */ |
| 389 | if ((ret = ff_rtp_finalize_packet(pkt, &jpeg->frame, st->index)) < 0) { |
| 390 | av_log(ctx, AV_LOG_ERROR, |
| 391 | "Error occurred when getting frame buffer.\n"); |
| 392 | return ret; |
| 393 | } |
| 394 | |
| 395 | return 0; |
| 396 | } |
| 397 | |
| 398 | return AVERROR(EAGAIN); |
| 399 | } |
| 400 | |
| 401 | RTPDynamicProtocolHandler ff_jpeg_dynamic_handler = { |
| 402 | .enc_name = "JPEG", |
| 403 | .codec_type = AVMEDIA_TYPE_VIDEO, |
| 404 | .codec_id = AV_CODEC_ID_MJPEG, |
| 405 | .alloc = jpeg_new_context, |
| 406 | .free = jpeg_free_context, |
| 407 | .parse_packet = jpeg_parse_packet, |
| 408 | .static_payload_id = 26, |
| 409 | }; |