Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Thomas Volkert authored on 2015/02/15 22:46:42... | ... |
@@ -26,6 +26,7 @@ version <next>: |
26 | 26 |
- RTP parser for loss tolerant payload format for MP3 audio (RFC 5219) |
27 | 27 |
- RTP parser for AC3 payload format (RFC 4184) |
28 | 28 |
- palettegen and paletteuse filters |
29 |
+- VP9 RTP payload format (draft 0) experimental depacketizer |
|
29 | 30 |
|
30 | 31 |
|
31 | 32 |
version 2.5: |
... | ... |
@@ -469,6 +469,7 @@ Muxers/Demuxers: |
469 | 469 |
rtpdec_h261.*, rtpenc_h261.* Thomas Volkert |
470 | 470 |
rtpdec_hevc.*, rtpenc_hevc.* Thomas Volkert |
471 | 471 |
rtpdec_asf.* Ronald S. Bultje |
472 |
+ rtpdec_vp9.c Thomas Volkert |
|
472 | 473 |
rtpenc_mpv.*, rtpenc_aac.* Martin Storsjo |
473 | 474 |
rtsp.c Luca Barbato |
474 | 475 |
sbgdec.c Nicolas George |
... | ... |
@@ -107,6 +107,7 @@ void ff_register_rtp_dynamic_payload_handlers(void) |
107 | 107 |
ff_register_dynamic_payload_handler(&ff_theora_dynamic_handler); |
108 | 108 |
ff_register_dynamic_payload_handler(&ff_vorbis_dynamic_handler); |
109 | 109 |
ff_register_dynamic_payload_handler(&ff_vp8_dynamic_handler); |
110 |
+ ff_register_dynamic_payload_handler(&ff_vp9_dynamic_handler); |
|
110 | 111 |
ff_register_dynamic_payload_handler(&gsm_dynamic_handler); |
111 | 112 |
ff_register_dynamic_payload_handler(&opus_dynamic_handler); |
112 | 113 |
ff_register_dynamic_payload_handler(&realmedia_mp3_dynamic_handler); |
... | ... |
@@ -69,5 +69,6 @@ extern RTPDynamicProtocolHandler ff_svq3_dynamic_handler; |
69 | 69 |
extern RTPDynamicProtocolHandler ff_theora_dynamic_handler; |
70 | 70 |
extern RTPDynamicProtocolHandler ff_vorbis_dynamic_handler; |
71 | 71 |
extern RTPDynamicProtocolHandler ff_vp8_dynamic_handler; |
72 |
+extern RTPDynamicProtocolHandler ff_vp9_dynamic_handler; |
|
72 | 73 |
|
73 | 74 |
#endif /* AVFORMAT_RTPDEC_FORMATS_H */ |
74 | 75 |
new file mode 100644 |
... | ... |
@@ -0,0 +1,317 @@ |
0 |
+/* |
|
1 |
+ * RTP parser for VP9 payload format (draft version 0) - experimental |
|
2 |
+ * Copyright (c) 2015 Thomas Volkert <thomas@homer-conferencing.com> |
|
3 |
+ * |
|
4 |
+ * This file is part of FFmpeg. |
|
5 |
+ * |
|
6 |
+ * FFmpeg is free software; you can redistribute it and/or |
|
7 |
+ * modify it under the terms of the GNU Lesser General Public |
|
8 |
+ * License as published by the Free Software Foundation; either |
|
9 |
+ * version 2.1 of the License, or (at your option) any later version. |
|
10 |
+ * |
|
11 |
+ * FFmpeg is distributed in the hope that it will be useful, |
|
12 |
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
13 |
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
14 |
+ * Lesser General Public License for more details. |
|
15 |
+ * |
|
16 |
+ * You should have received a copy of the GNU Lesser General Public |
|
17 |
+ * License along with FFmpeg; if not, write to the Free Software |
|
18 |
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
19 |
+ * |
|
20 |
+ */ |
|
21 |
+ |
|
22 |
+#include "libavcodec/bytestream.h" |
|
23 |
+ |
|
24 |
+#include "rtpdec_formats.h" |
|
25 |
+ |
|
26 |
+#define RTP_VP9_DESC_REQUIRED_SIZE 1 |
|
27 |
+ |
|
28 |
+struct PayloadContext { |
|
29 |
+ AVIOContext *buf; |
|
30 |
+ uint32_t timestamp; |
|
31 |
+}; |
|
32 |
+ |
|
33 |
+static av_cold PayloadContext *vp9_new_context(void) |
|
34 |
+{ |
|
35 |
+ return av_mallocz(sizeof(PayloadContext)); |
|
36 |
+} |
|
37 |
+ |
|
38 |
+static void vp9_free_dyn_buffer(AVIOContext **dyn_buf) |
|
39 |
+{ |
|
40 |
+ uint8_t *ptr_dyn_buffer; |
|
41 |
+ avio_close_dyn_buf(*dyn_buf, &ptr_dyn_buffer); |
|
42 |
+ av_free(ptr_dyn_buffer); |
|
43 |
+ *dyn_buf = NULL; |
|
44 |
+} |
|
45 |
+ |
|
46 |
+static av_cold void vp9_free_context(PayloadContext *data) |
|
47 |
+{ |
|
48 |
+ av_free(data); |
|
49 |
+} |
|
50 |
+ |
|
51 |
+static av_cold int vp9_init(AVFormatContext *ctx, int st_index, |
|
52 |
+ PayloadContext *data) |
|
53 |
+{ |
|
54 |
+ av_dlog(ctx, "vp9_init() for stream %d\n", st_index); |
|
55 |
+ av_log(ctx, AV_LOG_WARNING, |
|
56 |
+ "RTP/VP9 support is still experimental\n"); |
|
57 |
+ |
|
58 |
+ if (st_index < 0) |
|
59 |
+ return 0; |
|
60 |
+ |
|
61 |
+ ctx->streams[st_index]->need_parsing = AVSTREAM_PARSE_FULL; |
|
62 |
+ |
|
63 |
+ return 0; |
|
64 |
+} |
|
65 |
+ |
|
66 |
+static int vp9_handle_packet(AVFormatContext *ctx, PayloadContext *rtp_vp9_ctx, |
|
67 |
+ AVStream *st, AVPacket *pkt, uint32_t *timestamp, |
|
68 |
+ const uint8_t *buf, int len, uint16_t seq, |
|
69 |
+ int flags) |
|
70 |
+{ |
|
71 |
+ int has_pic_id, has_layer_idc, has_ref_idc, has_ss_data, has_su_data; |
|
72 |
+ av_unused int pic_id = 0, non_key_frame = 0; |
|
73 |
+ av_unused int layer_temporal = -1, layer_spatial = -1, layer_quality = -1; |
|
74 |
+ int ref_fields = 0, has_ref_field_ext_pic_id = 0; |
|
75 |
+ int first_fragment, last_fragment; |
|
76 |
+ int res = 0; |
|
77 |
+ |
|
78 |
+ /* drop data of previous packets in case of non-continuous (lossy) packet stream */ |
|
79 |
+ if (rtp_vp9_ctx->buf && rtp_vp9_ctx->timestamp != *timestamp) { |
|
80 |
+ vp9_free_dyn_buffer(&rtp_vp9_ctx->buf); |
|
81 |
+ } |
|
82 |
+ |
|
83 |
+ /* sanity check for size of input packet: 1 byte payload at least */ |
|
84 |
+ if (len < RTP_VP9_DESC_REQUIRED_SIZE + 1) { |
|
85 |
+ av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet, got %d bytes\n", len); |
|
86 |
+ return AVERROR_INVALIDDATA; |
|
87 |
+ } |
|
88 |
+ |
|
89 |
+ /* |
|
90 |
+ decode the required VP9 payload descriptor according to section 4.2 of the spec.: |
|
91 |
+ |
|
92 |
+ 0 1 2 3 4 5 6 7 |
|
93 |
+ +-+-+-+-+-+-+-+-+ |
|
94 |
+ |I|L|F|B|E|V|U|-| (REQUIRED) |
|
95 |
+ +-+-+-+-+-+-+-+-+ |
|
96 |
+ |
|
97 |
+ I: PictureID present |
|
98 |
+ L: Layer indices present |
|
99 |
+ F: Reference indices present |
|
100 |
+ B: Start of VP9 frame |
|
101 |
+ E: End of picture |
|
102 |
+ V: Scalability Structure (SS) present |
|
103 |
+ U: Scalability Structure Update (SU) present |
|
104 |
+ */ |
|
105 |
+ has_pic_id = buf[0] & 0x80; |
|
106 |
+ has_layer_idc = buf[0] & 0x40; |
|
107 |
+ has_ref_idc = buf[0] & 0x20; |
|
108 |
+ first_fragment = buf[0] & 0x10; |
|
109 |
+ last_fragment = buf[0] & 0x08; |
|
110 |
+ has_ss_data = buf[0] & 0x04; |
|
111 |
+ has_su_data = buf[0] & 0x02; |
|
112 |
+ |
|
113 |
+ /* sanity check for markers: B should always be equal to the RTP M marker */ |
|
114 |
+ if (last_fragment >> 2 != flags & RTP_FLAG_MARKER) { |
|
115 |
+ av_log(ctx, AV_LOG_ERROR, "Invalid combination of B and M marker\n"); |
|
116 |
+ return AVERROR_INVALIDDATA; |
|
117 |
+ } |
|
118 |
+ |
|
119 |
+ /* pass the extensions field */ |
|
120 |
+ buf += RTP_VP9_DESC_REQUIRED_SIZE; |
|
121 |
+ len -= RTP_VP9_DESC_REQUIRED_SIZE; |
|
122 |
+ |
|
123 |
+ /* |
|
124 |
+ decode the 1-byte/2-byte picture ID: |
|
125 |
+ |
|
126 |
+ 0 1 2 3 4 5 6 7 |
|
127 |
+ +-+-+-+-+-+-+-+-+ |
|
128 |
+ I: |M|PICTURE ID | (RECOMMENDED) |
|
129 |
+ +-+-+-+-+-+-+-+-+ |
|
130 |
+ M: | EXTENDED PID | (RECOMMENDED) |
|
131 |
+ +-+-+-+-+-+-+-+-+ |
|
132 |
+ |
|
133 |
+ M: The most significant bit of the first octet is an extension flag. |
|
134 |
+ PictureID: 8 or 16 bits including the M bit. |
|
135 |
+ */ |
|
136 |
+ if (has_pic_id) { |
|
137 |
+ if (len < 1) { |
|
138 |
+ av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet\n"); |
|
139 |
+ return AVERROR_INVALIDDATA; |
|
140 |
+ } |
|
141 |
+ |
|
142 |
+ /* check for 1-byte or 2-byte picture index */ |
|
143 |
+ if (buf[0] & 0x80) { |
|
144 |
+ if (len < 2) { |
|
145 |
+ av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet\n"); |
|
146 |
+ return AVERROR_INVALIDDATA; |
|
147 |
+ } |
|
148 |
+ pic_id = AV_RB16(buf) & 0x7fff; |
|
149 |
+ buf += 2; |
|
150 |
+ len -= 2; |
|
151 |
+ } else { |
|
152 |
+ pic_id = buf[0] & 0x7f; |
|
153 |
+ buf++; |
|
154 |
+ len--; |
|
155 |
+ } |
|
156 |
+ } |
|
157 |
+ |
|
158 |
+ /* |
|
159 |
+ decode layer indices |
|
160 |
+ |
|
161 |
+ 0 1 2 3 4 5 6 7 |
|
162 |
+ +-+-+-+-+-+-+-+-+ |
|
163 |
+ L: | T | S | Q | R | (CONDITIONALLY RECOMMENDED) |
|
164 |
+ +-+-+-+-+-+-+-+-+ |
|
165 |
+ |
|
166 |
+ T, S and Q are 2-bit indices for temporal, spatial, and quality layers. |
|
167 |
+ If "F" is set in the initial octet, R is 2 bits representing the number |
|
168 |
+ of reference fields this frame refers to. |
|
169 |
+ */ |
|
170 |
+ if (has_layer_idc) { |
|
171 |
+ if (len < 1) { |
|
172 |
+ av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet"); |
|
173 |
+ return AVERROR_INVALIDDATA; |
|
174 |
+ } |
|
175 |
+ layer_temporal = buf[0] & 0xC0; |
|
176 |
+ layer_spatial = buf[0] & 0x30; |
|
177 |
+ layer_quality = buf[0] & 0x0C; |
|
178 |
+ if (has_ref_idc) { |
|
179 |
+ ref_fields = buf[0] & 0x03; |
|
180 |
+ if (ref_fields) |
|
181 |
+ non_key_frame = 1; |
|
182 |
+ } |
|
183 |
+ buf++; |
|
184 |
+ len--; |
|
185 |
+ } |
|
186 |
+ |
|
187 |
+ /* |
|
188 |
+ decode the reference fields |
|
189 |
+ |
|
190 |
+ 0 1 2 3 4 5 6 7 |
|
191 |
+ +-+-+-+-+-+-+-+-+ -\ |
|
192 |
+ F: | PID |X| RS| RQ| (OPTIONAL) . |
|
193 |
+ +-+-+-+-+-+-+-+-+ . - R times |
|
194 |
+ X: | EXTENDED PID | (OPTIONAL) . |
|
195 |
+ +-+-+-+-+-+-+-+-+ -/ |
|
196 |
+ |
|
197 |
+ PID: The relative Picture ID referred to by this frame. |
|
198 |
+ RS and RQ: The spatial and quality layer IDs. |
|
199 |
+ X: 1 if this layer index has an extended relative Picture ID. |
|
200 |
+ */ |
|
201 |
+ if (has_ref_idc) { |
|
202 |
+ while (ref_fields) { |
|
203 |
+ if (len < 1) { |
|
204 |
+ av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet\n"); |
|
205 |
+ return AVERROR_INVALIDDATA; |
|
206 |
+ } |
|
207 |
+ |
|
208 |
+ has_ref_field_ext_pic_id = buf[0] & 0x10; |
|
209 |
+ |
|
210 |
+ /* pass ref. field */ |
|
211 |
+ if (has_ref_field_ext_pic_id) { |
|
212 |
+ if (len < 2) { |
|
213 |
+ av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet\n"); |
|
214 |
+ return AVERROR_INVALIDDATA; |
|
215 |
+ } |
|
216 |
+ |
|
217 |
+ /* ignore ref. data */ |
|
218 |
+ |
|
219 |
+ buf += 2; |
|
220 |
+ len -= 2; |
|
221 |
+ } else { |
|
222 |
+ |
|
223 |
+ /* ignore ref. data */ |
|
224 |
+ |
|
225 |
+ buf++; |
|
226 |
+ len--; |
|
227 |
+ } |
|
228 |
+ ref_fields--; |
|
229 |
+ } |
|
230 |
+ } |
|
231 |
+ |
|
232 |
+ /* |
|
233 |
+ decode the scalability structure (SS) |
|
234 |
+ |
|
235 |
+ 0 1 2 3 4 5 6 7 |
|
236 |
+ +-+-+-+-+-+-+-+-+ |
|
237 |
+ V: | PATTERN LENGTH| |
|
238 |
+ +-+-+-+-+-+-+-+-+ -\ |
|
239 |
+ | T | S | Q | R | (OPTIONAL) . |
|
240 |
+ +-+-+-+-+-+-+-+-+ -\ . |
|
241 |
+ | PID |X| RS| RQ| (OPTIONAL) . . - PAT. LEN. times |
|
242 |
+ +-+-+-+-+-+-+-+-+ . - R times . |
|
243 |
+ X: | EXTENDED PID | (OPTIONAL) . . |
|
244 |
+ +-+-+-+-+-+-+-+-+ -/ -/ |
|
245 |
+ |
|
246 |
+ PID: The relative Picture ID referred to by this frame. |
|
247 |
+ RS and RQ: The spatial and quality layer IDs. |
|
248 |
+ X: 1 if this layer index has an extended relative Picture ID. |
|
249 |
+ */ |
|
250 |
+ if (has_ss_data) { |
|
251 |
+ avpriv_report_missing_feature(ctx, "VP9 scalability structure data\n"); |
|
252 |
+ return AVERROR_PATCHWELCOME; |
|
253 |
+ } |
|
254 |
+ |
|
255 |
+ /* |
|
256 |
+ decode the scalability update structure (SU) |
|
257 |
+ |
|
258 |
+ spec. is tbd |
|
259 |
+ */ |
|
260 |
+ if (has_su_data) { |
|
261 |
+ avpriv_report_missing_feature(ctx, "VP9 scalability update structure data\n"); |
|
262 |
+ return AVERROR_PATCHWELCOME; |
|
263 |
+ } |
|
264 |
+ |
|
265 |
+ /* |
|
266 |
+ decode the VP9 payload header |
|
267 |
+ |
|
268 |
+ spec. is tbd |
|
269 |
+ */ |
|
270 |
+ //XXX: implement when specified |
|
271 |
+ |
|
272 |
+ /* sanity check: 1 byte payload as minimum */ |
|
273 |
+ if (len < 1) { |
|
274 |
+ av_log(ctx, AV_LOG_ERROR, "Too short RTP/VP9 packet\n"); |
|
275 |
+ return AVERROR_INVALIDDATA; |
|
276 |
+ } |
|
277 |
+ |
|
278 |
+ /* start frame buffering with new dynamic buffer */ |
|
279 |
+ if (!rtp_vp9_ctx->buf) { |
|
280 |
+ /* sanity check: a new frame should have started */ |
|
281 |
+ if (first_fragment) { |
|
282 |
+ res = avio_open_dyn_buf(&rtp_vp9_ctx->buf); |
|
283 |
+ if (res < 0) |
|
284 |
+ return res; |
|
285 |
+ /* update the timestamp in the frame packet with the one from the RTP packet */ |
|
286 |
+ rtp_vp9_ctx->timestamp = *timestamp; |
|
287 |
+ } else { |
|
288 |
+ /* frame not started yet, need more packets */ |
|
289 |
+ return AVERROR(EAGAIN); |
|
290 |
+ } |
|
291 |
+ } |
|
292 |
+ |
|
293 |
+ /* write the fragment to the dyn. buffer */ |
|
294 |
+ avio_write(rtp_vp9_ctx->buf, buf, len); |
|
295 |
+ |
|
296 |
+ /* do we need more fragments? */ |
|
297 |
+ if (!last_fragment) |
|
298 |
+ return AVERROR(EAGAIN); |
|
299 |
+ |
|
300 |
+ /* close frame buffering and create resulting A/V packet */ |
|
301 |
+ res = ff_rtp_finalize_packet(pkt, &rtp_vp9_ctx->buf, st->index); |
|
302 |
+ if (res < 0) |
|
303 |
+ return res; |
|
304 |
+ |
|
305 |
+ return 0; |
|
306 |
+} |
|
307 |
+ |
|
308 |
+RTPDynamicProtocolHandler ff_vp9_dynamic_handler = { |
|
309 |
+ .enc_name = "VP9", |
|
310 |
+ .codec_type = AVMEDIA_TYPE_VIDEO, |
|
311 |
+ .codec_id = AV_CODEC_ID_VP9, |
|
312 |
+ .init = vp9_init, |
|
313 |
+ .alloc = vp9_new_context, |
|
314 |
+ .free = vp9_free_context, |
|
315 |
+ .parse_packet = vp9_handle_packet |
|
316 |
+}; |