Go to most recent revision | Details | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
4349 | Serge | 1 | /* |
2 | * Copyright (c) 2002 Mark Hills |
||
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 | #include |
||
22 | |||
23 | #include "libavutil/avassert.h" |
||
24 | #include "libavutil/fifo.h" |
||
25 | #include "libavutil/opt.h" |
||
26 | #include "avcodec.h" |
||
27 | #include "audio_frame_queue.h" |
||
28 | #include "internal.h" |
||
29 | #include "vorbis.h" |
||
30 | #include "vorbis_parser.h" |
||
31 | |||
32 | |||
33 | /* Number of samples the user should send in each call. |
||
34 | * This value is used because it is the LCD of all possible frame sizes, so |
||
35 | * an output packet will always start at the same point as one of the input |
||
36 | * packets. |
||
37 | */ |
||
38 | #define OGGVORBIS_FRAME_SIZE 64 |
||
39 | |||
40 | #define BUFFER_SIZE (1024 * 64) |
||
41 | |||
42 | typedef struct OggVorbisEncContext { |
||
43 | AVClass *av_class; /**< class for AVOptions */ |
||
44 | AVFrame frame; |
||
45 | vorbis_info vi; /**< vorbis_info used during init */ |
||
46 | vorbis_dsp_state vd; /**< DSP state used for analysis */ |
||
47 | vorbis_block vb; /**< vorbis_block used for analysis */ |
||
48 | AVFifoBuffer *pkt_fifo; /**< output packet buffer */ |
||
49 | int eof; /**< end-of-file flag */ |
||
50 | int dsp_initialized; /**< vd has been initialized */ |
||
51 | vorbis_comment vc; /**< VorbisComment info */ |
||
52 | double iblock; /**< impulse block bias option */ |
||
53 | VorbisParseContext vp; /**< parse context to get durations */ |
||
54 | AudioFrameQueue afq; /**< frame queue for timestamps */ |
||
55 | } OggVorbisEncContext; |
||
56 | |||
57 | static const AVOption options[] = { |
||
58 | { "iblock", "Sets the impulse block bias", offsetof(OggVorbisEncContext, iblock), AV_OPT_TYPE_DOUBLE, { .dbl = 0 }, -15, 0, AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM }, |
||
59 | { NULL } |
||
60 | }; |
||
61 | |||
62 | static const AVCodecDefault defaults[] = { |
||
63 | { "b", "0" }, |
||
64 | { NULL }, |
||
65 | }; |
||
66 | |||
67 | static const AVClass vorbis_class = { |
||
68 | .class_name = "libvorbis", |
||
69 | .item_name = av_default_item_name, |
||
70 | .option = options, |
||
71 | .version = LIBAVUTIL_VERSION_INT, |
||
72 | }; |
||
73 | |||
74 | static int vorbis_error_to_averror(int ov_err) |
||
75 | { |
||
76 | switch (ov_err) { |
||
77 | case OV_EFAULT: return AVERROR_BUG; |
||
78 | case OV_EINVAL: return AVERROR(EINVAL); |
||
79 | case OV_EIMPL: return AVERROR(EINVAL); |
||
80 | default: return AVERROR_UNKNOWN; |
||
81 | } |
||
82 | } |
||
83 | |||
84 | static av_cold int oggvorbis_init_encoder(vorbis_info *vi, |
||
85 | AVCodecContext *avctx) |
||
86 | { |
||
87 | OggVorbisEncContext *s = avctx->priv_data; |
||
88 | double cfreq; |
||
89 | int ret; |
||
90 | |||
91 | if (avctx->flags & CODEC_FLAG_QSCALE || !avctx->bit_rate) { |
||
92 | /* variable bitrate |
||
93 | * NOTE: we use the oggenc range of -1 to 10 for global_quality for |
||
94 | * user convenience, but libvorbis uses -0.1 to 1.0. |
||
95 | */ |
||
96 | float q = avctx->global_quality / (float)FF_QP2LAMBDA; |
||
97 | /* default to 3 if the user did not set quality or bitrate */ |
||
98 | if (!(avctx->flags & CODEC_FLAG_QSCALE)) |
||
99 | q = 3.0; |
||
100 | if ((ret = vorbis_encode_setup_vbr(vi, avctx->channels, |
||
101 | avctx->sample_rate, |
||
102 | q / 10.0))) |
||
103 | goto error; |
||
104 | } else { |
||
105 | int minrate = avctx->rc_min_rate > 0 ? avctx->rc_min_rate : -1; |
||
106 | int maxrate = avctx->rc_max_rate > 0 ? avctx->rc_max_rate : -1; |
||
107 | |||
108 | /* average bitrate */ |
||
109 | if ((ret = vorbis_encode_setup_managed(vi, avctx->channels, |
||
110 | avctx->sample_rate, maxrate, |
||
111 | avctx->bit_rate, minrate))) |
||
112 | goto error; |
||
113 | |||
114 | /* variable bitrate by estimate, disable slow rate management */ |
||
115 | if (minrate == -1 && maxrate == -1) |
||
116 | if ((ret = vorbis_encode_ctl(vi, OV_ECTL_RATEMANAGE2_SET, NULL))) |
||
117 | goto error; /* should not happen */ |
||
118 | } |
||
119 | |||
120 | /* cutoff frequency */ |
||
121 | if (avctx->cutoff > 0) { |
||
122 | cfreq = avctx->cutoff / 1000.0; |
||
123 | if ((ret = vorbis_encode_ctl(vi, OV_ECTL_LOWPASS_SET, &cfreq))) |
||
124 | goto error; /* should not happen */ |
||
125 | } |
||
126 | |||
127 | /* impulse block bias */ |
||
128 | if (s->iblock) { |
||
129 | if ((ret = vorbis_encode_ctl(vi, OV_ECTL_IBLOCK_SET, &s->iblock))) |
||
130 | goto error; |
||
131 | } |
||
132 | |||
133 | if (avctx->channels == 3 && |
||
134 | avctx->channel_layout != (AV_CH_LAYOUT_STEREO|AV_CH_FRONT_CENTER) || |
||
135 | avctx->channels == 4 && |
||
136 | avctx->channel_layout != AV_CH_LAYOUT_2_2 && |
||
137 | avctx->channel_layout != AV_CH_LAYOUT_QUAD || |
||
138 | avctx->channels == 5 && |
||
139 | avctx->channel_layout != AV_CH_LAYOUT_5POINT0 && |
||
140 | avctx->channel_layout != AV_CH_LAYOUT_5POINT0_BACK || |
||
141 | avctx->channels == 6 && |
||
142 | avctx->channel_layout != AV_CH_LAYOUT_5POINT1 && |
||
143 | avctx->channel_layout != AV_CH_LAYOUT_5POINT1_BACK || |
||
144 | avctx->channels == 7 && |
||
145 | avctx->channel_layout != (AV_CH_LAYOUT_5POINT1|AV_CH_BACK_CENTER) || |
||
146 | avctx->channels == 8 && |
||
147 | avctx->channel_layout != AV_CH_LAYOUT_7POINT1) { |
||
148 | if (avctx->channel_layout) { |
||
149 | char name[32]; |
||
150 | av_get_channel_layout_string(name, sizeof(name), avctx->channels, |
||
151 | avctx->channel_layout); |
||
152 | av_log(avctx, AV_LOG_ERROR, "%s not supported by Vorbis: " |
||
153 | "output stream will have incorrect " |
||
154 | "channel layout.\n", name); |
||
155 | } else { |
||
156 | av_log(avctx, AV_LOG_WARNING, "No channel layout specified. The encoder " |
||
157 | "will use Vorbis channel layout for " |
||
158 | "%d channels.\n", avctx->channels); |
||
159 | } |
||
160 | } |
||
161 | |||
162 | if ((ret = vorbis_encode_setup_init(vi))) |
||
163 | goto error; |
||
164 | |||
165 | return 0; |
||
166 | error: |
||
167 | return vorbis_error_to_averror(ret); |
||
168 | } |
||
169 | |||
170 | /* How many bytes are needed for a buffer of length 'l' */ |
||
171 | static int xiph_len(int l) |
||
172 | { |
||
173 | return 1 + l / 255 + l; |
||
174 | } |
||
175 | |||
176 | static av_cold int oggvorbis_encode_close(AVCodecContext *avctx) |
||
177 | { |
||
178 | OggVorbisEncContext *s = avctx->priv_data; |
||
179 | |||
180 | /* notify vorbisenc this is EOF */ |
||
181 | if (s->dsp_initialized) |
||
182 | vorbis_analysis_wrote(&s->vd, 0); |
||
183 | |||
184 | vorbis_block_clear(&s->vb); |
||
185 | vorbis_dsp_clear(&s->vd); |
||
186 | vorbis_info_clear(&s->vi); |
||
187 | |||
188 | av_fifo_free(s->pkt_fifo); |
||
189 | ff_af_queue_close(&s->afq); |
||
190 | av_freep(&avctx->extradata); |
||
191 | |||
192 | return 0; |
||
193 | } |
||
194 | |||
195 | static av_cold int oggvorbis_encode_init(AVCodecContext *avctx) |
||
196 | { |
||
197 | OggVorbisEncContext *s = avctx->priv_data; |
||
198 | ogg_packet header, header_comm, header_code; |
||
199 | uint8_t *p; |
||
200 | unsigned int offset; |
||
201 | int ret; |
||
202 | |||
203 | vorbis_info_init(&s->vi); |
||
204 | if ((ret = oggvorbis_init_encoder(&s->vi, avctx))) { |
||
205 | av_log(avctx, AV_LOG_ERROR, "encoder setup failed\n"); |
||
206 | goto error; |
||
207 | } |
||
208 | if ((ret = vorbis_analysis_init(&s->vd, &s->vi))) { |
||
209 | av_log(avctx, AV_LOG_ERROR, "analysis init failed\n"); |
||
210 | ret = vorbis_error_to_averror(ret); |
||
211 | goto error; |
||
212 | } |
||
213 | s->dsp_initialized = 1; |
||
214 | if ((ret = vorbis_block_init(&s->vd, &s->vb))) { |
||
215 | av_log(avctx, AV_LOG_ERROR, "dsp init failed\n"); |
||
216 | ret = vorbis_error_to_averror(ret); |
||
217 | goto error; |
||
218 | } |
||
219 | |||
220 | vorbis_comment_init(&s->vc); |
||
221 | if (!(avctx->flags & CODEC_FLAG_BITEXACT)) |
||
222 | vorbis_comment_add_tag(&s->vc, "encoder", LIBAVCODEC_IDENT); |
||
223 | |||
224 | if ((ret = vorbis_analysis_headerout(&s->vd, &s->vc, &header, &header_comm, |
||
225 | &header_code))) { |
||
226 | ret = vorbis_error_to_averror(ret); |
||
227 | goto error; |
||
228 | } |
||
229 | |||
230 | avctx->extradata_size = 1 + xiph_len(header.bytes) + |
||
231 | xiph_len(header_comm.bytes) + |
||
232 | header_code.bytes; |
||
233 | p = avctx->extradata = av_malloc(avctx->extradata_size + |
||
234 | FF_INPUT_BUFFER_PADDING_SIZE); |
||
235 | if (!p) { |
||
236 | ret = AVERROR(ENOMEM); |
||
237 | goto error; |
||
238 | } |
||
239 | p[0] = 2; |
||
240 | offset = 1; |
||
241 | offset += av_xiphlacing(&p[offset], header.bytes); |
||
242 | offset += av_xiphlacing(&p[offset], header_comm.bytes); |
||
243 | memcpy(&p[offset], header.packet, header.bytes); |
||
244 | offset += header.bytes; |
||
245 | memcpy(&p[offset], header_comm.packet, header_comm.bytes); |
||
246 | offset += header_comm.bytes; |
||
247 | memcpy(&p[offset], header_code.packet, header_code.bytes); |
||
248 | offset += header_code.bytes; |
||
249 | av_assert0(offset == avctx->extradata_size); |
||
250 | |||
251 | if ((ret = avpriv_vorbis_parse_extradata(avctx, &s->vp)) < 0) { |
||
252 | av_log(avctx, AV_LOG_ERROR, "invalid extradata\n"); |
||
253 | return ret; |
||
254 | } |
||
255 | |||
256 | vorbis_comment_clear(&s->vc); |
||
257 | |||
258 | avctx->frame_size = OGGVORBIS_FRAME_SIZE; |
||
259 | ff_af_queue_init(avctx, &s->afq); |
||
260 | |||
261 | s->pkt_fifo = av_fifo_alloc(BUFFER_SIZE); |
||
262 | if (!s->pkt_fifo) { |
||
263 | ret = AVERROR(ENOMEM); |
||
264 | goto error; |
||
265 | } |
||
266 | |||
267 | return 0; |
||
268 | error: |
||
269 | oggvorbis_encode_close(avctx); |
||
270 | return ret; |
||
271 | } |
||
272 | |||
273 | static int oggvorbis_encode_frame(AVCodecContext *avctx, AVPacket *avpkt, |
||
274 | const AVFrame *frame, int *got_packet_ptr) |
||
275 | { |
||
276 | OggVorbisEncContext *s = avctx->priv_data; |
||
277 | ogg_packet op; |
||
278 | int ret, duration; |
||
279 | |||
280 | /* send samples to libvorbis */ |
||
281 | if (frame) { |
||
282 | const int samples = frame->nb_samples; |
||
283 | float **buffer; |
||
284 | int c, channels = s->vi.channels; |
||
285 | |||
286 | buffer = vorbis_analysis_buffer(&s->vd, samples); |
||
287 | for (c = 0; c < channels; c++) { |
||
288 | int co = (channels > 8) ? c : |
||
289 | ff_vorbis_encoding_channel_layout_offsets[channels - 1][c]; |
||
290 | memcpy(buffer[c], frame->extended_data[co], |
||
291 | samples * sizeof(*buffer[c])); |
||
292 | } |
||
293 | if ((ret = vorbis_analysis_wrote(&s->vd, samples)) < 0) { |
||
294 | av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n"); |
||
295 | return vorbis_error_to_averror(ret); |
||
296 | } |
||
297 | if ((ret = ff_af_queue_add(&s->afq, frame)) < 0) |
||
298 | return ret; |
||
299 | } else { |
||
300 | if (!s->eof) |
||
301 | if ((ret = vorbis_analysis_wrote(&s->vd, 0)) < 0) { |
||
302 | av_log(avctx, AV_LOG_ERROR, "error in vorbis_analysis_wrote()\n"); |
||
303 | return vorbis_error_to_averror(ret); |
||
304 | } |
||
305 | s->eof = 1; |
||
306 | } |
||
307 | |||
308 | /* retrieve available packets from libvorbis */ |
||
309 | while ((ret = vorbis_analysis_blockout(&s->vd, &s->vb)) == 1) { |
||
310 | if ((ret = vorbis_analysis(&s->vb, NULL)) < 0) |
||
311 | break; |
||
312 | if ((ret = vorbis_bitrate_addblock(&s->vb)) < 0) |
||
313 | break; |
||
314 | |||
315 | /* add any available packets to the output packet buffer */ |
||
316 | while ((ret = vorbis_bitrate_flushpacket(&s->vd, &op)) == 1) { |
||
317 | if (av_fifo_space(s->pkt_fifo) < sizeof(ogg_packet) + op.bytes) { |
||
318 | av_log(avctx, AV_LOG_ERROR, "packet buffer is too small\n"); |
||
319 | return AVERROR_BUG; |
||
320 | } |
||
321 | av_fifo_generic_write(s->pkt_fifo, &op, sizeof(ogg_packet), NULL); |
||
322 | av_fifo_generic_write(s->pkt_fifo, op.packet, op.bytes, NULL); |
||
323 | } |
||
324 | if (ret < 0) { |
||
325 | av_log(avctx, AV_LOG_ERROR, "error getting available packets\n"); |
||
326 | break; |
||
327 | } |
||
328 | } |
||
329 | if (ret < 0) { |
||
330 | av_log(avctx, AV_LOG_ERROR, "error getting available packets\n"); |
||
331 | return vorbis_error_to_averror(ret); |
||
332 | } |
||
333 | |||
334 | /* check for available packets */ |
||
335 | if (av_fifo_size(s->pkt_fifo) < sizeof(ogg_packet)) |
||
336 | return 0; |
||
337 | |||
338 | av_fifo_generic_read(s->pkt_fifo, &op, sizeof(ogg_packet), NULL); |
||
339 | |||
340 | if ((ret = ff_alloc_packet2(avctx, avpkt, op.bytes)) < 0) |
||
341 | return ret; |
||
342 | av_fifo_generic_read(s->pkt_fifo, avpkt->data, op.bytes, NULL); |
||
343 | |||
344 | avpkt->pts = ff_samples_to_time_base(avctx, op.granulepos); |
||
345 | |||
346 | duration = avpriv_vorbis_parse_frame(&s->vp, avpkt->data, avpkt->size); |
||
347 | if (duration > 0) { |
||
348 | /* we do not know encoder delay until we get the first packet from |
||
349 | * libvorbis, so we have to update the AudioFrameQueue counts */ |
||
350 | if (!avctx->delay && s->afq.frames) { |
||
351 | avctx->delay = duration; |
||
352 | av_assert0(!s->afq.remaining_delay); |
||
353 | s->afq.frames->duration += duration; |
||
354 | s->afq.frames->pts -= duration; |
||
355 | s->afq.remaining_samples += duration; |
||
356 | } |
||
357 | ff_af_queue_remove(&s->afq, duration, &avpkt->pts, &avpkt->duration); |
||
358 | } |
||
359 | |||
360 | *got_packet_ptr = 1; |
||
361 | return 0; |
||
362 | } |
||
363 | |||
364 | AVCodec ff_libvorbis_encoder = { |
||
365 | .name = "libvorbis", |
||
366 | .long_name = NULL_IF_CONFIG_SMALL("libvorbis"), |
||
367 | .type = AVMEDIA_TYPE_AUDIO, |
||
368 | .id = AV_CODEC_ID_VORBIS, |
||
369 | .priv_data_size = sizeof(OggVorbisEncContext), |
||
370 | .init = oggvorbis_encode_init, |
||
371 | .encode2 = oggvorbis_encode_frame, |
||
372 | .close = oggvorbis_encode_close, |
||
373 | .capabilities = CODEC_CAP_DELAY, |
||
374 | .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP, |
||
375 | AV_SAMPLE_FMT_NONE }, |
||
376 | .priv_class = &vorbis_class, |
||
377 | .defaults = defaults, |
||
378 | };>>>>>>>>>>>>>>>>>>>>>>> |