| 1 | /* |
| 2 | * RTP H264 Protocol (RFC3984) |
| 3 | * Copyright (c) 2006 Ryan Martell |
| 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 | * @brief H.264 / RTP Code (RFC3984) |
| 25 | * @author Ryan Martell <rdm4@martellventures.com> |
| 26 | * |
| 27 | * @note Notes: |
| 28 | * Notes: |
| 29 | * This currently supports packetization mode: |
| 30 | * Single Nal Unit Mode (0), or |
| 31 | * Non-Interleaved Mode (1). It currently does not support |
| 32 | * Interleaved Mode (2). (This requires implementing STAP-B, MTAP16, MTAP24, |
| 33 | * FU-B packet types) |
| 34 | */ |
| 35 | |
| 36 | #include "libavutil/attributes.h" |
| 37 | #include "libavutil/base64.h" |
| 38 | #include "libavutil/avstring.h" |
| 39 | #include "libavcodec/get_bits.h" |
| 40 | #include "avformat.h" |
| 41 | |
| 42 | #include "network.h" |
| 43 | #include <assert.h> |
| 44 | |
| 45 | #include "rtpdec.h" |
| 46 | #include "rtpdec_formats.h" |
| 47 | |
| 48 | struct PayloadContext { |
| 49 | // sdp setup parameters |
| 50 | uint8_t profile_idc; |
| 51 | uint8_t profile_iop; |
| 52 | uint8_t level_idc; |
| 53 | int packetization_mode; |
| 54 | #ifdef DEBUG |
| 55 | int packet_types_received[32]; |
| 56 | #endif |
| 57 | }; |
| 58 | |
| 59 | #ifdef DEBUG |
| 60 | #define COUNT_NAL_TYPE(data, nal) data->packet_types_received[(nal) & 0x1f]++ |
| 61 | #else |
| 62 | #define COUNT_NAL_TYPE(data, nal) do { } while (0) |
| 63 | #endif |
| 64 | |
| 65 | static const uint8_t start_sequence[] = { 0, 0, 0, 1 }; |
| 66 | |
| 67 | static int sdp_parse_fmtp_config_h264(AVFormatContext *s, |
| 68 | AVStream *stream, |
| 69 | PayloadContext *h264_data, |
| 70 | char *attr, char *value) |
| 71 | { |
| 72 | AVCodecContext *codec = stream->codec; |
| 73 | assert(codec->codec_id == AV_CODEC_ID_H264); |
| 74 | assert(h264_data); |
| 75 | |
| 76 | if (!strcmp(attr, "packetization-mode")) { |
| 77 | av_log(s, AV_LOG_DEBUG, "RTP Packetization Mode: %d\n", atoi(value)); |
| 78 | h264_data->packetization_mode = atoi(value); |
| 79 | /* |
| 80 | * Packetization Mode: |
| 81 | * 0 or not present: Single NAL mode (Only nals from 1-23 are allowed) |
| 82 | * 1: Non-interleaved Mode: 1-23, 24 (STAP-A), 28 (FU-A) are allowed. |
| 83 | * 2: Interleaved Mode: 25 (STAP-B), 26 (MTAP16), 27 (MTAP24), 28 (FU-A), |
| 84 | * and 29 (FU-B) are allowed. |
| 85 | */ |
| 86 | if (h264_data->packetization_mode > 1) |
| 87 | av_log(s, AV_LOG_ERROR, |
| 88 | "Interleaved RTP mode is not supported yet.\n"); |
| 89 | } else if (!strcmp(attr, "profile-level-id")) { |
| 90 | if (strlen(value) == 6) { |
| 91 | char buffer[3]; |
| 92 | // 6 characters=3 bytes, in hex. |
| 93 | uint8_t profile_idc; |
| 94 | uint8_t profile_iop; |
| 95 | uint8_t level_idc; |
| 96 | |
| 97 | buffer[0] = value[0]; |
| 98 | buffer[1] = value[1]; |
| 99 | buffer[2] = '\0'; |
| 100 | profile_idc = strtol(buffer, NULL, 16); |
| 101 | buffer[0] = value[2]; |
| 102 | buffer[1] = value[3]; |
| 103 | profile_iop = strtol(buffer, NULL, 16); |
| 104 | buffer[0] = value[4]; |
| 105 | buffer[1] = value[5]; |
| 106 | level_idc = strtol(buffer, NULL, 16); |
| 107 | |
| 108 | av_log(s, AV_LOG_DEBUG, |
| 109 | "RTP Profile IDC: %x Profile IOP: %x Level: %x\n", |
| 110 | profile_idc, profile_iop, level_idc); |
| 111 | h264_data->profile_idc = profile_idc; |
| 112 | h264_data->profile_iop = profile_iop; |
| 113 | h264_data->level_idc = level_idc; |
| 114 | } |
| 115 | } else if (!strcmp(attr, "sprop-parameter-sets")) { |
| 116 | codec->extradata_size = 0; |
| 117 | av_freep(&codec->extradata); |
| 118 | |
| 119 | while (*value) { |
| 120 | char base64packet[1024]; |
| 121 | uint8_t decoded_packet[1024]; |
| 122 | int packet_size; |
| 123 | char *dst = base64packet; |
| 124 | |
| 125 | while (*value && *value != ',' |
| 126 | && (dst - base64packet) < sizeof(base64packet) - 1) { |
| 127 | *dst++ = *value++; |
| 128 | } |
| 129 | *dst++ = '\0'; |
| 130 | |
| 131 | if (*value == ',') |
| 132 | value++; |
| 133 | |
| 134 | packet_size = av_base64_decode(decoded_packet, base64packet, |
| 135 | sizeof(decoded_packet)); |
| 136 | if (packet_size > 0) { |
| 137 | uint8_t *dest = av_malloc(packet_size + sizeof(start_sequence) + |
| 138 | codec->extradata_size + |
| 139 | FF_INPUT_BUFFER_PADDING_SIZE); |
| 140 | if (!dest) { |
| 141 | av_log(s, AV_LOG_ERROR, |
| 142 | "Unable to allocate memory for extradata!\n"); |
| 143 | return AVERROR(ENOMEM); |
| 144 | } |
| 145 | if (codec->extradata_size) { |
| 146 | memcpy(dest, codec->extradata, codec->extradata_size); |
| 147 | av_free(codec->extradata); |
| 148 | } |
| 149 | |
| 150 | memcpy(dest + codec->extradata_size, start_sequence, |
| 151 | sizeof(start_sequence)); |
| 152 | memcpy(dest + codec->extradata_size + sizeof(start_sequence), |
| 153 | decoded_packet, packet_size); |
| 154 | memset(dest + codec->extradata_size + sizeof(start_sequence) + |
| 155 | packet_size, 0, FF_INPUT_BUFFER_PADDING_SIZE); |
| 156 | |
| 157 | codec->extradata = dest; |
| 158 | codec->extradata_size += sizeof(start_sequence) + packet_size; |
| 159 | } |
| 160 | } |
| 161 | av_log(s, AV_LOG_DEBUG, "Extradata set to %p (size: %d)!\n", |
| 162 | codec->extradata, codec->extradata_size); |
| 163 | } |
| 164 | return 0; |
| 165 | } |
| 166 | |
| 167 | // return 0 on packet, no more left, 1 on packet, 1 on partial packet |
| 168 | static int h264_handle_packet(AVFormatContext *ctx, PayloadContext *data, |
| 169 | AVStream *st, AVPacket *pkt, uint32_t *timestamp, |
| 170 | const uint8_t *buf, int len, uint16_t seq, |
| 171 | int flags) |
| 172 | { |
| 173 | uint8_t nal; |
| 174 | uint8_t type; |
| 175 | int result = 0; |
| 176 | |
| 177 | if (!len) { |
| 178 | av_log(ctx, AV_LOG_ERROR, "Empty H264 RTP packet\n"); |
| 179 | return AVERROR_INVALIDDATA; |
| 180 | } |
| 181 | nal = buf[0]; |
| 182 | type = nal & 0x1f; |
| 183 | |
| 184 | assert(data); |
| 185 | assert(buf); |
| 186 | |
| 187 | /* Simplify the case (these are all the nal types used internally by |
| 188 | * the h264 codec). */ |
| 189 | if (type >= 1 && type <= 23) |
| 190 | type = 1; |
| 191 | switch (type) { |
| 192 | case 0: // undefined, but pass them through |
| 193 | case 1: |
| 194 | if ((result = av_new_packet(pkt, len + sizeof(start_sequence))) < 0) |
| 195 | return result; |
| 196 | memcpy(pkt->data, start_sequence, sizeof(start_sequence)); |
| 197 | memcpy(pkt->data + sizeof(start_sequence), buf, len); |
| 198 | COUNT_NAL_TYPE(data, nal); |
| 199 | break; |
| 200 | |
| 201 | case 24: // STAP-A (one packet, multiple nals) |
| 202 | // consume the STAP-A NAL |
| 203 | buf++; |
| 204 | len--; |
| 205 | // first we are going to figure out the total size |
| 206 | { |
| 207 | int pass = 0; |
| 208 | int total_length = 0; |
| 209 | uint8_t *dst = NULL; |
| 210 | |
| 211 | for (pass = 0; pass < 2; pass++) { |
| 212 | const uint8_t *src = buf; |
| 213 | int src_len = len; |
| 214 | |
| 215 | while (src_len > 2) { |
| 216 | uint16_t nal_size = AV_RB16(src); |
| 217 | |
| 218 | // consume the length of the aggregate |
| 219 | src += 2; |
| 220 | src_len -= 2; |
| 221 | |
| 222 | if (nal_size <= src_len) { |
| 223 | if (pass == 0) { |
| 224 | // counting |
| 225 | total_length += sizeof(start_sequence) + nal_size; |
| 226 | } else { |
| 227 | // copying |
| 228 | assert(dst); |
| 229 | memcpy(dst, start_sequence, sizeof(start_sequence)); |
| 230 | dst += sizeof(start_sequence); |
| 231 | memcpy(dst, src, nal_size); |
| 232 | COUNT_NAL_TYPE(data, *src); |
| 233 | dst += nal_size; |
| 234 | } |
| 235 | } else { |
| 236 | av_log(ctx, AV_LOG_ERROR, |
| 237 | "nal size exceeds length: %d %d\n", nal_size, src_len); |
| 238 | } |
| 239 | |
| 240 | // eat what we handled |
| 241 | src += nal_size; |
| 242 | src_len -= nal_size; |
| 243 | |
| 244 | if (src_len < 0) |
| 245 | av_log(ctx, AV_LOG_ERROR, |
| 246 | "Consumed more bytes than we got! (%d)\n", src_len); |
| 247 | } |
| 248 | |
| 249 | if (pass == 0) { |
| 250 | /* now we know the total size of the packet (with the |
| 251 | * start sequences added) */ |
| 252 | if ((result = av_new_packet(pkt, total_length)) < 0) |
| 253 | return result; |
| 254 | dst = pkt->data; |
| 255 | } else { |
| 256 | assert(dst - pkt->data == total_length); |
| 257 | } |
| 258 | } |
| 259 | } |
| 260 | break; |
| 261 | |
| 262 | case 25: // STAP-B |
| 263 | case 26: // MTAP-16 |
| 264 | case 27: // MTAP-24 |
| 265 | case 29: // FU-B |
| 266 | av_log(ctx, AV_LOG_ERROR, |
| 267 | "Unhandled type (%d) (See RFC for implementation details\n", |
| 268 | type); |
| 269 | result = AVERROR(ENOSYS); |
| 270 | break; |
| 271 | |
| 272 | case 28: // FU-A (fragmented nal) |
| 273 | buf++; |
| 274 | len--; // skip the fu_indicator |
| 275 | if (len > 1) { |
| 276 | // these are the same as above, we just redo them here for clarity |
| 277 | uint8_t fu_indicator = nal; |
| 278 | uint8_t fu_header = *buf; |
| 279 | uint8_t start_bit = fu_header >> 7; |
| 280 | uint8_t av_unused end_bit = (fu_header & 0x40) >> 6; |
| 281 | uint8_t nal_type = fu_header & 0x1f; |
| 282 | uint8_t reconstructed_nal; |
| 283 | |
| 284 | // Reconstruct this packet's true nal; only the data follows. |
| 285 | /* The original nal forbidden bit and NRI are stored in this |
| 286 | * packet's nal. */ |
| 287 | reconstructed_nal = fu_indicator & 0xe0; |
| 288 | reconstructed_nal |= nal_type; |
| 289 | |
| 290 | // skip the fu_header |
| 291 | buf++; |
| 292 | len--; |
| 293 | |
| 294 | if (start_bit) |
| 295 | COUNT_NAL_TYPE(data, nal_type); |
| 296 | if (start_bit) { |
| 297 | /* copy in the start sequence, and the reconstructed nal */ |
| 298 | if ((result = av_new_packet(pkt, sizeof(start_sequence) + sizeof(nal) + len)) < 0) |
| 299 | return result; |
| 300 | memcpy(pkt->data, start_sequence, sizeof(start_sequence)); |
| 301 | pkt->data[sizeof(start_sequence)] = reconstructed_nal; |
| 302 | memcpy(pkt->data + sizeof(start_sequence) + sizeof(nal), buf, len); |
| 303 | } else { |
| 304 | if ((result = av_new_packet(pkt, len)) < 0) |
| 305 | return result; |
| 306 | memcpy(pkt->data, buf, len); |
| 307 | } |
| 308 | } else { |
| 309 | av_log(ctx, AV_LOG_ERROR, "Too short data for FU-A H264 RTP packet\n"); |
| 310 | result = AVERROR_INVALIDDATA; |
| 311 | } |
| 312 | break; |
| 313 | |
| 314 | case 30: // undefined |
| 315 | case 31: // undefined |
| 316 | default: |
| 317 | av_log(ctx, AV_LOG_ERROR, "Undefined type (%d)\n", type); |
| 318 | result = AVERROR_INVALIDDATA; |
| 319 | break; |
| 320 | } |
| 321 | |
| 322 | pkt->stream_index = st->index; |
| 323 | |
| 324 | return result; |
| 325 | } |
| 326 | |
| 327 | static PayloadContext *h264_new_context(void) |
| 328 | { |
| 329 | return av_mallocz(sizeof(PayloadContext) + FF_INPUT_BUFFER_PADDING_SIZE); |
| 330 | } |
| 331 | |
| 332 | static void h264_free_context(PayloadContext *data) |
| 333 | { |
| 334 | #ifdef DEBUG |
| 335 | int ii; |
| 336 | |
| 337 | for (ii = 0; ii < 32; ii++) { |
| 338 | if (data->packet_types_received[ii]) |
| 339 | av_log(NULL, AV_LOG_DEBUG, "Received %d packets of type %d\n", |
| 340 | data->packet_types_received[ii], ii); |
| 341 | } |
| 342 | #endif |
| 343 | |
| 344 | av_free(data); |
| 345 | } |
| 346 | |
| 347 | static av_cold int h264_init(AVFormatContext *s, int st_index, |
| 348 | PayloadContext *data) |
| 349 | { |
| 350 | if (st_index < 0) |
| 351 | return 0; |
| 352 | s->streams[st_index]->need_parsing = AVSTREAM_PARSE_FULL; |
| 353 | return 0; |
| 354 | } |
| 355 | |
| 356 | static int parse_h264_sdp_line(AVFormatContext *s, int st_index, |
| 357 | PayloadContext *h264_data, const char *line) |
| 358 | { |
| 359 | AVStream *stream; |
| 360 | AVCodecContext *codec; |
| 361 | const char *p = line; |
| 362 | |
| 363 | if (st_index < 0) |
| 364 | return 0; |
| 365 | |
| 366 | stream = s->streams[st_index]; |
| 367 | codec = stream->codec; |
| 368 | |
| 369 | if (av_strstart(p, "framesize:", &p)) { |
| 370 | char buf1[50]; |
| 371 | char *dst = buf1; |
| 372 | |
| 373 | // remove the protocol identifier |
| 374 | while (*p && *p == ' ') |
| 375 | p++; // strip spaces. |
| 376 | while (*p && *p != ' ') |
| 377 | p++; // eat protocol identifier |
| 378 | while (*p && *p == ' ') |
| 379 | p++; // strip trailing spaces. |
| 380 | while (*p && *p != '-' && (dst - buf1) < sizeof(buf1) - 1) |
| 381 | *dst++ = *p++; |
| 382 | *dst = '\0'; |
| 383 | |
| 384 | // a='framesize:96 320-240' |
| 385 | // set our parameters |
| 386 | codec->width = atoi(buf1); |
| 387 | codec->height = atoi(p + 1); // skip the - |
| 388 | } else if (av_strstart(p, "fmtp:", &p)) { |
| 389 | return ff_parse_fmtp(s, stream, h264_data, p, sdp_parse_fmtp_config_h264); |
| 390 | } else if (av_strstart(p, "cliprect:", &p)) { |
| 391 | // could use this if we wanted. |
| 392 | } |
| 393 | |
| 394 | return 0; |
| 395 | } |
| 396 | |
| 397 | RTPDynamicProtocolHandler ff_h264_dynamic_handler = { |
| 398 | .enc_name = "H264", |
| 399 | .codec_type = AVMEDIA_TYPE_VIDEO, |
| 400 | .codec_id = AV_CODEC_ID_H264, |
| 401 | .init = h264_init, |
| 402 | .parse_sdp_a_line = parse_h264_sdp_line, |
| 403 | .alloc = h264_new_context, |
| 404 | .free = h264_free_context, |
| 405 | .parse_packet = h264_handle_packet |
| 406 | }; |