f21f35296e4b9878c7664c6d1a3104d4f142dc4a
[libav.git] / libavformat / flvdec.c
1 /*
2 * FLV demuxer
3 * Copyright (c) 2003 The FFmpeg Project.
4 *
5 * This demuxer will generate a 1 byte extradata for VP6F content.
6 * It is composed of:
7 * - upper 4bits: difference between encoded width and visible width
8 * - lower 4bits: difference between encoded height and visible height
9 *
10 * This file is part of FFmpeg.
11 *
12 * FFmpeg is free software; you can redistribute it and/or
13 * modify it under the terms of the GNU Lesser General Public
14 * License as published by the Free Software Foundation; either
15 * version 2.1 of the License, or (at your option) any later version.
16 *
17 * FFmpeg is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
20 * Lesser General Public License for more details.
21 *
22 * You should have received a copy of the GNU Lesser General Public
23 * License along with FFmpeg; if not, write to the Free Software
24 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
25 */
26 #include "avformat.h"
27 #include "flv.h"
28
29 static int flv_probe(AVProbeData *p)
30 {
31 const uint8_t *d;
32
33 d = p->buf;
34 if (d[0] == 'F' && d[1] == 'L' && d[2] == 'V' && d[3] < 5 && d[5]==0) {
35 return AVPROBE_SCORE_MAX;
36 }
37 return 0;
38 }
39
40 static void flv_set_audio_codec(AVFormatContext *s, AVStream *astream, int flv_codecid) {
41 AVCodecContext *acodec = astream->codec;
42 switch(flv_codecid) {
43 //no distinction between S16 and S8 PCM codec flags
44 case FLV_CODECID_PCM:
45 acodec->codec_id = acodec->bits_per_sample == 8 ? CODEC_ID_PCM_S8 :
46 #ifdef WORDS_BIGENDIAN
47 CODEC_ID_PCM_S16BE;
48 #else
49 CODEC_ID_PCM_S16LE;
50 #endif
51 break;
52 case FLV_CODECID_PCM_LE:
53 acodec->codec_id = acodec->bits_per_sample == 8 ? CODEC_ID_PCM_S8 : CODEC_ID_PCM_S16LE; break;
54 case FLV_CODECID_ADPCM: acodec->codec_id = CODEC_ID_ADPCM_SWF; break;
55 case FLV_CODECID_MP3 : acodec->codec_id = CODEC_ID_MP3 ; astream->need_parsing = AVSTREAM_PARSE_FULL; break;
56 case FLV_CODECID_NELLYMOSER_8HZ_MONO:
57 acodec->sample_rate = 8000; //in case metadata does not otherwise declare samplerate
58 case FLV_CODECID_NELLYMOSER:
59 acodec->codec_id = CODEC_ID_NELLYMOSER;
60 break;
61 default:
62 av_log(s, AV_LOG_INFO, "Unsupported audio codec (%x)\n", flv_codecid >> FLV_AUDIO_CODECID_OFFSET);
63 acodec->codec_tag = flv_codecid >> FLV_AUDIO_CODECID_OFFSET;
64 }
65 }
66
67 static int flv_set_video_codec(AVFormatContext *s, AVStream *vstream, int flv_codecid) {
68 AVCodecContext *vcodec = vstream->codec;
69 switch(flv_codecid) {
70 case FLV_CODECID_H263 : vcodec->codec_id = CODEC_ID_FLV1 ; break;
71 case FLV_CODECID_SCREEN: vcodec->codec_id = CODEC_ID_FLASHSV; break;
72 case FLV_CODECID_VP6 : vcodec->codec_id = CODEC_ID_VP6F ;
73 case FLV_CODECID_VP6A :
74 if(flv_codecid == FLV_CODECID_VP6A)
75 vcodec->codec_id = CODEC_ID_VP6A;
76 if(vcodec->extradata_size != 1) {
77 vcodec->extradata_size = 1;
78 vcodec->extradata = av_malloc(1);
79 }
80 vcodec->extradata[0] = get_byte(s->pb);
81 return 1; // 1 byte body size adjustment for flv_read_packet()
82 default:
83 av_log(s, AV_LOG_INFO, "Unsupported video codec (%x)\n", flv_codecid);
84 vcodec->codec_tag = flv_codecid;
85 }
86
87 return 0;
88 }
89
90 static int amf_get_string(ByteIOContext *ioc, char *buffer, int buffsize) {
91 int length = get_be16(ioc);
92 if(length >= buffsize) {
93 url_fskip(ioc, length);
94 return -1;
95 }
96
97 get_buffer(ioc, buffer, length);
98
99 buffer[length] = '\0';
100
101 return length;
102 }
103
104 static int amf_parse_object(AVFormatContext *s, AVStream *astream, AVStream *vstream, const char *key, unsigned int max_pos, int depth) {
105 AVCodecContext *acodec, *vcodec;
106 ByteIOContext *ioc;
107 AMFDataType amf_type;
108 char str_val[256];
109 double num_val;
110
111 num_val = 0;
112 ioc = s->pb;
113
114 amf_type = get_byte(ioc);
115
116 switch(amf_type) {
117 case AMF_DATA_TYPE_NUMBER:
118 num_val = av_int2dbl(get_be64(ioc)); break;
119 case AMF_DATA_TYPE_BOOL:
120 num_val = get_byte(ioc); break;
121 case AMF_DATA_TYPE_STRING:
122 if(amf_get_string(ioc, str_val, sizeof(str_val)) < 0)
123 return -1;
124 break;
125 case AMF_DATA_TYPE_OBJECT: {
126 unsigned int keylen;
127
128 while(url_ftell(ioc) < max_pos - 2 && (keylen = get_be16(ioc))) {
129 url_fskip(ioc, keylen); //skip key string
130 if(amf_parse_object(s, NULL, NULL, NULL, max_pos, depth + 1) < 0)
131 return -1; //if we couldn't skip, bomb out.
132 }
133 if(get_byte(ioc) != AMF_END_OF_OBJECT)
134 return -1;
135 }
136 break;
137 case AMF_DATA_TYPE_NULL:
138 case AMF_DATA_TYPE_UNDEFINED:
139 case AMF_DATA_TYPE_UNSUPPORTED:
140 break; //these take up no additional space
141 case AMF_DATA_TYPE_MIXEDARRAY:
142 url_fskip(ioc, 4); //skip 32-bit max array index
143 while(url_ftell(ioc) < max_pos - 2 && amf_get_string(ioc, str_val, sizeof(str_val)) > 0) {
144 //this is the only case in which we would want a nested parse to not skip over the object
145 if(amf_parse_object(s, astream, vstream, str_val, max_pos, depth + 1) < 0)
146 return -1;
147 }
148 if(get_byte(ioc) != AMF_END_OF_OBJECT)
149 return -1;
150 break;
151 case AMF_DATA_TYPE_ARRAY: {
152 unsigned int arraylen, i;
153
154 arraylen = get_be32(ioc);
155 for(i = 0; i < arraylen && url_ftell(ioc) < max_pos - 1; i++) {
156 if(amf_parse_object(s, NULL, NULL, NULL, max_pos, depth + 1) < 0)
157 return -1; //if we couldn't skip, bomb out.
158 }
159 }
160 break;
161 case AMF_DATA_TYPE_DATE:
162 url_fskip(ioc, 8 + 2); //timestamp (double) and UTC offset (int16)
163 break;
164 default: //unsupported type, we couldn't skip
165 return -1;
166 }
167
168 if(depth == 1 && key) { //only look for metadata values when we are not nested and key != NULL
169 acodec = astream ? astream->codec : NULL;
170 vcodec = vstream ? vstream->codec : NULL;
171
172 if(amf_type == AMF_DATA_TYPE_BOOL) {
173 if(!strcmp(key, "stereo") && acodec) acodec->channels = num_val > 0 ? 2 : 1;
174 } else if(amf_type == AMF_DATA_TYPE_NUMBER) {
175 if(!strcmp(key, "duration")) s->duration = num_val * AV_TIME_BASE;
176 // else if(!strcmp(key, "width") && vcodec && num_val > 0) vcodec->width = num_val;
177 // else if(!strcmp(key, "height") && vcodec && num_val > 0) vcodec->height = num_val;
178 else if(!strcmp(key, "audiocodecid") && acodec) flv_set_audio_codec(s, astream, (int)num_val << FLV_AUDIO_CODECID_OFFSET);
179 else if(!strcmp(key, "videocodecid") && vcodec) flv_set_video_codec(s, vstream, (int)num_val);
180 else if(!strcmp(key, "audiosamplesize") && acodec && num_val >= 0) {
181 acodec->bits_per_sample = num_val;
182 //we may have to rewrite a previously read codecid because FLV only marks PCM endianness.
183 if(num_val == 8 && (acodec->codec_id == CODEC_ID_PCM_S16BE || acodec->codec_id == CODEC_ID_PCM_S16LE))
184 acodec->codec_id = CODEC_ID_PCM_S8;
185 }
186 else if(!strcmp(key, "audiosamplerate") && acodec && num_val >= 0) {
187 //some tools, like FLVTool2, write consistently approximate metadata sample rates
188 if (!acodec->sample_rate) {
189 switch((int)num_val) {
190 case 44000: acodec->sample_rate = 44100 ; break;
191 case 22000: acodec->sample_rate = 22050 ; break;
192 case 11000: acodec->sample_rate = 11025 ; break;
193 case 5000 : acodec->sample_rate = 5512 ; break;
194 default : acodec->sample_rate = num_val;
195 }
196 }
197 }
198 }
199 }
200
201 return 0;
202 }
203
204 static int flv_read_metabody(AVFormatContext *s, unsigned int next_pos) {
205 AMFDataType type;
206 AVStream *stream, *astream, *vstream;
207 ByteIOContext *ioc;
208 int i, keylen;
209 char buffer[11]; //only needs to hold the string "onMetaData". Anything longer is something we don't want.
210
211 astream = NULL;
212 vstream = NULL;
213 keylen = 0;
214 ioc = s->pb;
215
216 //first object needs to be "onMetaData" string
217 type = get_byte(ioc);
218 if(type != AMF_DATA_TYPE_STRING || amf_get_string(ioc, buffer, sizeof(buffer)) < 0 || strcmp(buffer, "onMetaData"))
219 return -1;
220
221 //find the streams now so that amf_parse_object doesn't need to do the lookup every time it is called.
222 for(i = 0; i < s->nb_streams; i++) {
223 stream = s->streams[i];
224 if (stream->codec->codec_type == CODEC_TYPE_AUDIO) astream = stream;
225 else if(stream->codec->codec_type == CODEC_TYPE_VIDEO) vstream = stream;
226 }
227
228 //parse the second object (we want a mixed array)
229 if(amf_parse_object(s, astream, vstream, buffer, next_pos, 0) < 0)
230 return -1;
231
232 return 0;
233 }
234
235 static AVStream *create_stream(AVFormatContext *s, int is_audio){
236 AVStream *st = av_new_stream(s, is_audio);
237 if (!st)
238 return NULL;
239 st->codec->codec_type = is_audio ? CODEC_TYPE_AUDIO : CODEC_TYPE_VIDEO;
240 av_set_pts_info(st, 24, 1, 1000); /* 24 bit pts in ms */
241 return st;
242 }
243
244 static int flv_read_header(AVFormatContext *s,
245 AVFormatParameters *ap)
246 {
247 int offset, flags;
248
249 url_fskip(s->pb, 4);
250 flags = get_byte(s->pb);
251 /* old flvtool cleared this field */
252 /* FIXME: better fix needed */
253 if (!flags) {
254 flags = FLV_HEADER_FLAG_HASVIDEO | FLV_HEADER_FLAG_HASAUDIO;
255 av_log(s, AV_LOG_WARNING, "Broken FLV file, which says no streams present, this might fail\n");
256 }
257
258 if(flags & FLV_HEADER_FLAG_HASVIDEO){
259 if(!create_stream(s, 0))
260 return AVERROR(ENOMEM);
261 }
262 if(flags & FLV_HEADER_FLAG_HASAUDIO){
263 if(!create_stream(s, 1))
264 return AVERROR(ENOMEM);
265 }
266
267 offset = get_be32(s->pb);
268 url_fseek(s->pb, offset, SEEK_SET);
269
270 s->start_time = 0;
271
272 return 0;
273 }
274
275 static int flv_read_packet(AVFormatContext *s, AVPacket *pkt)
276 {
277 int ret, i, type, size, pts, flags, is_audio, next, pos;
278 AVStream *st = NULL;
279
280 for(;;){
281 pos = url_ftell(s->pb);
282 url_fskip(s->pb, 4); /* size of previous packet */
283 type = get_byte(s->pb);
284 size = get_be24(s->pb);
285 pts = get_be24(s->pb);
286 pts |= get_byte(s->pb) << 24;
287 // av_log(s, AV_LOG_DEBUG, "type:%d, size:%d, pts:%d\n", type, size, pts);
288 if (url_feof(s->pb))
289 return AVERROR(EIO);
290 url_fskip(s->pb, 3); /* stream id, always 0 */
291 flags = 0;
292
293 if(size == 0)
294 continue;
295
296 next= size + url_ftell(s->pb);
297
298 if (type == FLV_TAG_TYPE_AUDIO) {
299 is_audio=1;
300 flags = get_byte(s->pb);
301 } else if (type == FLV_TAG_TYPE_VIDEO) {
302 is_audio=0;
303 flags = get_byte(s->pb);
304 } else {
305 if (type == FLV_TAG_TYPE_META && size > 13+1+4)
306 flv_read_metabody(s, next);
307 else /* skip packet */
308 av_log(s, AV_LOG_ERROR, "skipping flv packet: type %d, size %d, flags %d\n", type, size, flags);
309 url_fseek(s->pb, next, SEEK_SET);
310 continue;
311 }
312
313 /* now find stream */
314 for(i=0;i<s->nb_streams;i++) {
315 st = s->streams[i];
316 if (st->id == is_audio)
317 break;
318 }
319 if(i == s->nb_streams){
320 av_log(NULL, AV_LOG_ERROR, "invalid stream\n");
321 st= create_stream(s, is_audio);
322 }
323 // av_log(NULL, AV_LOG_DEBUG, "%d %X %d \n", is_audio, flags, st->discard);
324 if( (st->discard >= AVDISCARD_NONKEY && !((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY || is_audio))
325 ||(st->discard >= AVDISCARD_BIDIR && ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_DISP_INTER && !is_audio))
326 || st->discard >= AVDISCARD_ALL
327 ){
328 url_fseek(s->pb, next, SEEK_SET);
329 continue;
330 }
331 if ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY)
332 av_add_index_entry(st, pos, pts, size, 0, AVINDEX_KEYFRAME);
333 break;
334 }
335
336 // if not streamed and no duration from metadata then seek to end to find the duration from the timestamps
337 if(!url_is_streamed(s->pb) && s->duration==AV_NOPTS_VALUE){
338 int size;
339 const int pos= url_ftell(s->pb);
340 const int fsize= url_fsize(s->pb);
341 url_fseek(s->pb, fsize-4, SEEK_SET);
342 size= get_be32(s->pb);
343 url_fseek(s->pb, fsize-3-size, SEEK_SET);
344 if(size == get_be24(s->pb) + 11){
345 s->duration= get_be24(s->pb) * (int64_t)AV_TIME_BASE / 1000;
346 }
347 url_fseek(s->pb, pos, SEEK_SET);
348 }
349
350 if(is_audio){
351 if(!st->codec->sample_rate || !st->codec->bits_per_sample || (!st->codec->codec_id && !st->codec->codec_tag)) {
352 st->codec->channels = (flags & FLV_AUDIO_CHANNEL_MASK) == FLV_STEREO ? 2 : 1;
353 if((flags & FLV_AUDIO_CODECID_MASK) == FLV_CODECID_NELLYMOSER_8HZ_MONO)
354 st->codec->sample_rate= 8000;
355 else
356 st->codec->sample_rate = (44100 << ((flags & FLV_AUDIO_SAMPLERATE_MASK) >> FLV_AUDIO_SAMPLERATE_OFFSET) >> 3);
357 st->codec->bits_per_sample = (flags & FLV_AUDIO_SAMPLESIZE_MASK) ? 16 : 8;
358 flv_set_audio_codec(s, st, flags & FLV_AUDIO_CODECID_MASK);
359 }
360 }else{
361 size -= flv_set_video_codec(s, st, flags & FLV_VIDEO_CODECID_MASK);
362 }
363
364 ret= av_get_packet(s->pb, pkt, size - 1);
365 if (ret <= 0) {
366 return AVERROR(EIO);
367 }
368 /* note: we need to modify the packet size here to handle the last
369 packet */
370 pkt->size = ret;
371 pkt->pts = pts;
372 pkt->stream_index = st->index;
373
374 if (is_audio || ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY))
375 pkt->flags |= PKT_FLAG_KEY;
376
377 return ret;
378 }
379
380 static int flv_read_close(AVFormatContext *s)
381 {
382 return 0;
383 }
384
385 static int flv_read_seek(AVFormatContext *s, int stream_index, int64_t timestamp, int flags)
386 {
387 AVStream *st = s->streams[stream_index];
388 int index = av_index_search_timestamp(st, timestamp, flags);
389 if (index < 0)
390 return -1;
391 url_fseek(s->pb, st->index_entries[index].pos, SEEK_SET);
392
393 return 0;
394 }
395
396 AVInputFormat flv_demuxer = {
397 "flv",
398 "flv format",
399 0,
400 flv_probe,
401 flv_read_header,
402 flv_read_packet,
403 flv_read_close,
404 flv_read_seek,
405 .extensions = "flv",
406 .value = CODEC_ID_FLV1,
407 };