2 * RTP parser for HEVC/H.265 payload format (draft version 6)
3 * Copyright (c) 2014 Thomas Volkert <thomas@homer-conferencing.com>
5 * This file is part of Libav.
7 * Libav 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 * Libav 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 Libav; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23 #include "libavutil/avstring.h"
24 #include "libavutil/base64.h"
28 #include "rtpdec_formats.h"
30 #define RTP_HEVC_PAYLOAD_HEADER_SIZE 2
31 #define RTP_HEVC_FU_HEADER_SIZE 1
32 #define RTP_HEVC_DONL_FIELD_SIZE 2
33 #define RTP_HEVC_DOND_FIELD_SIZE 1
34 #define HEVC_SPECIFIED_NAL_UNIT_TYPES 48
36 /* SDP out-of-band signaling data */
37 struct PayloadContext {
40 uint8_t *sps, *pps, *vps, *sei;
41 int sps_size, pps_size, vps_size, sei_size;
44 static const uint8_t start_sequence[] = { 0x00, 0x00, 0x00, 0x01 };
46 static av_cold int hevc_sdp_parse_fmtp_config(AVFormatContext *s,
48 PayloadContext *hevc_data,
49 char *attr, char *value)
51 /* profile-space: 0-3 */
52 /* profile-id: 0-31 */
53 if (!strcmp(attr, "profile-id")) {
54 hevc_data->profile_id = atoi(value);
55 av_dlog(s, "SDP: found profile-id: %d\n", hevc_data->profile_id);
60 /* interop-constraints: [base16] */
61 /* profile-compatibility-indicator: [base16] */
62 /* sprop-sub-layer-id: 0-6, defines highest possible value for TID, default: 6 */
63 /* recv-sub-layer-id: 0-6 */
64 /* max-recv-level-id: 0-255 */
65 /* tx-mode: MSM,SSM */
66 /* sprop-vps: [base64] */
67 /* sprop-sps: [base64] */
68 /* sprop-pps: [base64] */
69 /* sprop-sei: [base64] */
70 if (!strcmp(attr, "sprop-vps") || !strcmp(attr, "sprop-sps") ||
71 !strcmp(attr, "sprop-pps") || !strcmp(attr, "sprop-sei")) {
72 uint8_t **data_ptr = NULL;
74 if (!strcmp(attr, "sprop-vps")) {
75 data_ptr = &hevc_data->vps;
76 size_ptr = &hevc_data->vps_size;
77 } else if (!strcmp(attr, "sprop-sps")) {
78 data_ptr = &hevc_data->sps;
79 size_ptr = &hevc_data->sps_size;
80 } else if (!strcmp(attr, "sprop-pps")) {
81 data_ptr = &hevc_data->pps;
82 size_ptr = &hevc_data->pps_size;
83 } else if (!strcmp(attr, "sprop-sei")) {
84 data_ptr = &hevc_data->sei;
85 size_ptr = &hevc_data->sei_size;
88 ff_h264_parse_sprop_parameter_sets(s, data_ptr,
92 /* max-lsr, max-lps, max-cpb, max-dpb, max-br, max-tr, max-tc */
95 /* sprop-max-don-diff: 0-32767
97 When the RTP stream depends on one or more other RTP
98 streams (in this case tx-mode MUST be equal to "MSM" and
99 MSM is in use), this parameter MUST be present and the
100 value MUST be greater than 0.
102 if (!strcmp(attr, "sprop-max-don-diff")) {
104 hevc_data->using_donl_field = 1;
105 av_dlog(s, "Found sprop-max-don-diff in SDP, DON field usage is: %d\n",
106 hevc_data->using_donl_field);
109 /* sprop-depack-buf-nalus: 0-32767 */
110 if (!strcmp(attr, "sprop-depack-buf-nalus")) {
112 hevc_data->using_donl_field = 1;
113 av_dlog(s, "Found sprop-depack-buf-nalus in SDP, DON field usage is: %d\n",
114 hevc_data->using_donl_field);
117 /* sprop-depack-buf-bytes: 0-4294967295 */
119 /* sprop-segmentation-id: 0-3 */
120 /* sprop-spatial-segmentation-idc: [base16] */
121 /* dec-parallel-ca: */
127 static av_cold int hevc_parse_sdp_line(AVFormatContext *ctx, int st_index,
128 PayloadContext *hevc_data, const char *line)
130 AVStream *current_stream;
131 AVCodecContext *codec;
132 const char *sdp_line_ptr = line;
137 current_stream = ctx->streams[st_index];
138 codec = current_stream->codec;
140 if (av_strstart(sdp_line_ptr, "framesize:", &sdp_line_ptr)) {
141 char str_video_width[50];
142 char *str_video_width_ptr = str_video_width;
145 * parse "a=framesize:96 320-240"
149 while (*sdp_line_ptr && *sdp_line_ptr == ' ')
151 /* ignore RTP payload ID */
152 while (*sdp_line_ptr && *sdp_line_ptr != ' ')
155 while (*sdp_line_ptr && *sdp_line_ptr == ' ')
157 /* extract the actual video resolution description */
158 while (*sdp_line_ptr && *sdp_line_ptr != '-' &&
159 (str_video_width_ptr - str_video_width) < sizeof(str_video_width) - 1)
160 *str_video_width_ptr++ = *sdp_line_ptr++;
161 /* add trailing zero byte */
162 *str_video_width_ptr = '\0';
164 /* determine the width value */
165 codec->width = atoi(str_video_width);
166 /* jump beyond the "-" and determine the height value */
167 codec->height = atoi(sdp_line_ptr + 1);
168 } else if (av_strstart(sdp_line_ptr, "fmtp:", &sdp_line_ptr)) {
169 int ret = ff_parse_fmtp(ctx, current_stream, hevc_data, sdp_line_ptr,
170 hevc_sdp_parse_fmtp_config);
171 if (hevc_data->vps_size || hevc_data->sps_size ||
172 hevc_data->pps_size || hevc_data->sei_size) {
173 av_freep(&codec->extradata);
174 codec->extradata_size = hevc_data->vps_size + hevc_data->sps_size +
175 hevc_data->pps_size + hevc_data->sei_size;
176 codec->extradata = av_malloc(codec->extradata_size +
177 FF_INPUT_BUFFER_PADDING_SIZE);
178 if (!codec->extradata) {
179 ret = AVERROR(ENOMEM);
180 codec->extradata_size = 0;
183 memcpy(codec->extradata + pos, hevc_data->vps, hevc_data->vps_size);
184 pos += hevc_data->vps_size;
185 memcpy(codec->extradata + pos, hevc_data->sps, hevc_data->sps_size);
186 pos += hevc_data->sps_size;
187 memcpy(codec->extradata + pos, hevc_data->pps, hevc_data->pps_size);
188 pos += hevc_data->pps_size;
189 memcpy(codec->extradata + pos, hevc_data->sei, hevc_data->sei_size);
190 pos += hevc_data->sei_size;
191 memset(codec->extradata + pos, 0, FF_INPUT_BUFFER_PADDING_SIZE);
194 av_freep(&hevc_data->vps);
195 av_freep(&hevc_data->sps);
196 av_freep(&hevc_data->pps);
197 av_freep(&hevc_data->sei);
198 hevc_data->vps_size = 0;
199 hevc_data->sps_size = 0;
200 hevc_data->pps_size = 0;
201 hevc_data->sei_size = 0;
209 static int hevc_handle_packet(AVFormatContext *ctx, PayloadContext *rtp_hevc_ctx,
210 AVStream *st, AVPacket *pkt, uint32_t *timestamp,
211 const uint8_t *buf, int len, uint16_t seq,
214 const uint8_t *rtp_pl = buf;
215 int tid, lid, nal_type;
216 int first_fragment, last_fragment, fu_type;
217 uint8_t new_nal_header[2];
220 /* sanity check for size of input packet: 1 byte payload at least */
221 if (len < RTP_HEVC_PAYLOAD_HEADER_SIZE + 1) {
222 av_log(ctx, AV_LOG_ERROR, "Too short RTP/HEVC packet, got %d bytes\n", len);
223 return AVERROR_INVALIDDATA;
227 * decode the HEVC payload header according to section 4 of draft version 6:
230 * 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
231 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
232 * |F| Type | LayerId | TID |
233 * +-------------+-----------------+
235 * Forbidden zero (F): 1 bit
236 * NAL unit type (Type): 6 bits
237 * NUH layer ID (LayerId): 6 bits
238 * NUH temporal ID plus 1 (TID): 3 bits
240 nal_type = (buf[0] >> 1) & 0x3f;
241 lid = ((buf[0] << 5) & 0x20) | ((buf[1] >> 3) & 0x1f);
244 /* sanity check for correct layer ID */
246 /* future scalable or 3D video coding extensions */
247 avpriv_report_missing_feature(ctx, "Multi-layer HEVC coding\n");
248 return AVERROR_PATCHWELCOME;
251 /* sanity check for correct temporal ID */
253 av_log(ctx, AV_LOG_ERROR, "Illegal temporal ID in RTP/HEVC packet\n");
254 return AVERROR_INVALIDDATA;
257 /* sanity check for correct NAL unit type */
259 av_log(ctx, AV_LOG_ERROR, "Unsupported (HEVC) NAL type (%d)\n", nal_type);
260 return AVERROR_INVALIDDATA;
264 /* video parameter set (VPS) */
266 /* sequence parameter set (SPS) */
268 /* picture parameter set (PPS) */
270 /* supplemental enhancement information (SEI) */
272 /* single NAL unit packet */
274 /* sanity check for size of input packet: 1 byte payload at least */
276 av_log(ctx, AV_LOG_ERROR,
277 "Too short RTP/HEVC packet, got %d bytes of NAL unit type %d\n",
279 return AVERROR_INVALIDDATA;
282 /* create A/V packet */
283 if ((res = av_new_packet(pkt, sizeof(start_sequence) + len)) < 0)
285 /* A/V packet: copy start sequence */
286 memcpy(pkt->data, start_sequence, sizeof(start_sequence));
287 /* A/V packet: copy NAL unit data */
288 memcpy(pkt->data + sizeof(start_sequence), buf, len);
291 /* aggregated packet (AP) - with two or more NAL units */
293 /* pass the HEVC payload header */
294 buf += RTP_HEVC_PAYLOAD_HEADER_SIZE;
295 len -= RTP_HEVC_PAYLOAD_HEADER_SIZE;
297 /* pass the HEVC DONL field */
298 if (rtp_hevc_ctx->using_donl_field) {
299 buf += RTP_HEVC_DONL_FIELD_SIZE;
300 len -= RTP_HEVC_DONL_FIELD_SIZE;
303 res = ff_h264_handle_aggregated_packet(ctx, pkt, buf, len,
304 rtp_hevc_ctx->using_donl_field ?
305 RTP_HEVC_DOND_FIELD_SIZE : 0,
310 /* fragmentation unit (FU) */
312 /* pass the HEVC payload header */
313 buf += RTP_HEVC_PAYLOAD_HEADER_SIZE;
314 len -= RTP_HEVC_PAYLOAD_HEADER_SIZE;
317 * decode the FU header
324 * Start fragment (S): 1 bit
325 * End fragment (E): 1 bit
328 first_fragment = buf[0] & 0x80;
329 last_fragment = buf[0] & 0x40;
330 fu_type = buf[0] & 0x3f;
332 /* pass the HEVC FU header */
333 buf += RTP_HEVC_FU_HEADER_SIZE;
334 len -= RTP_HEVC_FU_HEADER_SIZE;
336 /* pass the HEVC DONL field */
337 if (rtp_hevc_ctx->using_donl_field) {
338 buf += RTP_HEVC_DONL_FIELD_SIZE;
339 len -= RTP_HEVC_DONL_FIELD_SIZE;
342 av_dlog(ctx, " FU type %d with %d bytes\n", fu_type, len);
345 new_nal_header[0] = (rtp_pl[0] & 0x81) | (fu_type << 1);
346 new_nal_header[1] = rtp_pl[1];
348 /* start fragment vs. subsequent fragments */
349 if (first_fragment) {
350 if (!last_fragment) {
351 /* create A/V packet which is big enough */
352 if ((res = av_new_packet(pkt, sizeof(start_sequence) + sizeof(new_nal_header) + len)) < 0)
354 /* A/V packet: copy start sequence */
355 memcpy(pkt->data, start_sequence, sizeof(start_sequence));
356 /* A/V packet: copy new NAL header */
357 memcpy(pkt->data + sizeof(start_sequence), new_nal_header, sizeof(new_nal_header));
358 /* A/V packet: copy NAL unit data */
359 memcpy(pkt->data + sizeof(start_sequence) + sizeof(new_nal_header), buf, len);
361 av_log(ctx, AV_LOG_ERROR, "Illegal combination of S and E bit in RTP/HEVC packet\n");
362 res = AVERROR_INVALIDDATA;
365 /* create A/V packet */
366 if ((res = av_new_packet(pkt, len)) < 0)
368 /* A/V packet: copy NAL unit data */
369 memcpy(pkt->data, buf, len);
372 /* sanity check for size of input packet: 1 byte payload at least */
373 av_log(ctx, AV_LOG_ERROR,
374 "Too short RTP/HEVC packet, got %d bytes of NAL unit type %d\n",
376 res = AVERROR_INVALIDDATA;
382 /* Temporal scalability control information (TSCI) */
383 avpriv_report_missing_feature(ctx, "PACI packets for RTP/HEVC\n");
384 res = AVERROR_PATCHWELCOME;
388 pkt->stream_index = st->index;
393 RTPDynamicProtocolHandler ff_hevc_dynamic_handler = {
395 .codec_type = AVMEDIA_TYPE_VIDEO,
396 .codec_id = AV_CODEC_ID_HEVC,
397 .need_parsing = AVSTREAM_PARSE_FULL,
398 .priv_data_size = sizeof(PayloadContext),
399 .parse_sdp_a_line = hevc_parse_sdp_line,
400 .parse_packet = hevc_handle_packet,