avconv.c 78.4 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
/*
 * avconv main
 * Copyright (c) 2000-2011 The libav developers.
 *
 * This file is part of Libav.
 *
 * Libav is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * Libav is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with Libav; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include "config.h"
#include <ctype.h>
#include <string.h>
#include <math.h>
#include <stdlib.h>
#include <errno.h>
#include <signal.h>
#include <limits.h>
#include "libavformat/avformat.h"
#include "libavdevice/avdevice.h"
#include "libswscale/swscale.h"
Justin Ruggles's avatar
Justin Ruggles committed
33
#include "libavresample/avresample.h"
34
#include "libavutil/opt.h"
35
#include "libavutil/channel_layout.h"
36 37 38 39 40 41 42 43 44 45
#include "libavutil/parseutils.h"
#include "libavutil/samplefmt.h"
#include "libavutil/colorspace.h"
#include "libavutil/fifo.h"
#include "libavutil/intreadwrite.h"
#include "libavutil/dict.h"
#include "libavutil/mathematics.h"
#include "libavutil/pixdesc.h"
#include "libavutil/avstring.h"
#include "libavutil/libm.h"
46
#include "libavutil/imgutils.h"
47
#include "libavutil/time.h"
48 49 50 51
#include "libavformat/os_support.h"

# include "libavfilter/avfilter.h"
# include "libavfilter/avfiltergraph.h"
52
# include "libavfilter/buffersrc.h"
53
# include "libavfilter/buffersink.h"
54 55

#if HAVE_SYS_RESOURCE_H
56
#include <sys/time.h>
57 58 59 60 61 62 63 64 65 66 67 68 69 70
#include <sys/types.h>
#include <sys/resource.h>
#elif HAVE_GETPROCESSTIMES
#include <windows.h>
#endif
#if HAVE_GETPROCESSMEMORYINFO
#include <windows.h>
#include <psapi.h>
#endif

#if HAVE_SYS_SELECT_H
#include <sys/select.h>
#endif

71 72 73 74
#if HAVE_PTHREADS
#include <pthread.h>
#endif

75 76
#include <time.h>

77
#include "avconv.h"
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93
#include "cmdutils.h"

#include "libavutil/avassert.h"

const char program_name[] = "avconv";
const int program_birth_year = 2000;

static FILE *vstats_file;

static int64_t video_size = 0;
static int64_t audio_size = 0;
static int64_t extra_size = 0;
static int nb_frames_dup = 0;
static int nb_frames_drop = 0;


94

95
#if HAVE_PTHREADS
96 97 98 99
/* signal to input threads that they should exit; set by the main thread */
static int transcoding_finished;
#endif

100 101
#define DEFAULT_PASS_LOGFILENAME_PREFIX "av2pass"

102 103 104 105
InputStream **input_streams = NULL;
int        nb_input_streams = 0;
InputFile   **input_files   = NULL;
int        nb_input_files   = 0;
106

107 108 109 110
OutputStream **output_streams = NULL;
int         nb_output_streams = 0;
OutputFile   **output_files   = NULL;
int         nb_output_files   = 0;
111

112 113
FilterGraph **filtergraphs;
int        nb_filtergraphs;
114

115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132
static void term_exit(void)
{
    av_log(NULL, AV_LOG_QUIET, "");
}

static volatile int received_sigterm = 0;
static volatile int received_nb_signals = 0;

static void
sigterm_handler(int sig)
{
    received_sigterm = sig;
    received_nb_signals++;
    term_exit();
}

static void term_init(void)
{
Aneesh Dogra's avatar
Aneesh Dogra committed
133
    signal(SIGINT , sigterm_handler); /* Interrupt (ANSI).    */
134 135 136 137 138 139
    signal(SIGTERM, sigterm_handler); /* Termination (ANSI).  */
#ifdef SIGXCPU
    signal(SIGXCPU, sigterm_handler);
#endif
}

140
static int decode_interrupt_cb(void *ctx)
141 142 143 144
{
    return received_nb_signals > 1;
}

145
const AVIOInterruptCB int_cb = { decode_interrupt_cb, NULL };
146

147
static void exit_program(void)
148
{
149 150 151 152
    int i, j;

    for (i = 0; i < nb_filtergraphs; i++) {
        avfilter_graph_free(&filtergraphs[i]->graph);
153 154
        for (j = 0; j < filtergraphs[i]->nb_inputs; j++) {
            av_freep(&filtergraphs[i]->inputs[j]->name);
155
            av_freep(&filtergraphs[i]->inputs[j]);
156
        }
157
        av_freep(&filtergraphs[i]->inputs);
158 159
        for (j = 0; j < filtergraphs[i]->nb_outputs; j++) {
            av_freep(&filtergraphs[i]->outputs[j]->name);
160
            av_freep(&filtergraphs[i]->outputs[j]);
161
        }
162 163 164 165
        av_freep(&filtergraphs[i]->outputs);
        av_freep(&filtergraphs[i]);
    }
    av_freep(&filtergraphs);
166 167

    /* close files */
Aneesh Dogra's avatar
Aneesh Dogra committed
168
    for (i = 0; i < nb_output_files; i++) {
169
        AVFormatContext *s = output_files[i]->ctx;
170 171 172
        if (!(s->oformat->flags & AVFMT_NOFILE) && s->pb)
            avio_close(s->pb);
        avformat_free_context(s);
173 174
        av_dict_free(&output_files[i]->opts);
        av_freep(&output_files[i]);
175
    }
Janne Grunau's avatar
Janne Grunau committed
176
    for (i = 0; i < nb_output_streams; i++) {
177
        AVBitStreamFilterContext *bsfc = output_streams[i]->bitstream_filters;
Janne Grunau's avatar
Janne Grunau committed
178 179 180 181 182
        while (bsfc) {
            AVBitStreamFilterContext *next = bsfc->next;
            av_bitstream_filter_close(bsfc);
            bsfc = next;
        }
183
        output_streams[i]->bitstream_filters = NULL;
184
        avcodec_free_frame(&output_streams[i]->filtered_frame);
185

Anton Khirnov's avatar
Anton Khirnov committed
186
        av_freep(&output_streams[i]->forced_keyframes);
187
        av_freep(&output_streams[i]->avfilter);
188
        av_freep(&output_streams[i]->logfile_prefix);
189
        av_freep(&output_streams[i]);
Janne Grunau's avatar
Janne Grunau committed
190
    }
Aneesh Dogra's avatar
Aneesh Dogra committed
191
    for (i = 0; i < nb_input_files; i++) {
192 193
        avformat_close_input(&input_files[i]->ctx);
        av_freep(&input_files[i]);
194
    }
195
    for (i = 0; i < nb_input_streams; i++) {
196
        avcodec_free_frame(&input_streams[i]->decoded_frame);
197
        av_dict_free(&input_streams[i]->opts);
198
        free_buffer_pool(&input_streams[i]->buffer_pool);
199
        av_freep(&input_streams[i]->filters);
200
        av_freep(&input_streams[i]);
201
    }
202 203 204 205 206 207 208

    if (vstats_file)
        fclose(vstats_file);
    av_free(vstats_filename);

    av_freep(&input_streams);
    av_freep(&input_files);
209
    av_freep(&output_streams);
210
    av_freep(&output_files);
211 212 213 214

    uninit_opts();

    avfilter_uninit();
215
    avformat_network_deinit();
216 217

    if (received_sigterm) {
218 219
        av_log(NULL, AV_LOG_INFO, "Received signal %d: terminating.\n",
               (int) received_sigterm);
220 221 222 223
        exit (255);
    }
}

224
void assert_avoptions(AVDictionary *m)
225 226 227
{
    AVDictionaryEntry *t;
    if ((t = av_dict_get(m, "", NULL, AV_DICT_IGNORE_SUFFIX))) {
228
        av_log(NULL, AV_LOG_FATAL, "Option %s not found.\n", t->key);
229
        exit(1);
230 231 232
    }
}

233
static void abort_codec_experimental(AVCodec *c, int encoder)
234 235 236
{
    const char *codec_string = encoder ? "encoder" : "decoder";
    AVCodec *codec;
237 238 239 240 241 242 243 244
    av_log(NULL, AV_LOG_FATAL, "%s '%s' is experimental and might produce bad "
            "results.\nAdd '-strict experimental' if you want to use it.\n",
            codec_string, c->name);
    codec = encoder ? avcodec_find_encoder(c->id) : avcodec_find_decoder(c->id);
    if (!(codec->capabilities & CODEC_CAP_EXPERIMENTAL))
        av_log(NULL, AV_LOG_FATAL, "Or use the non experimental %s '%s'.\n",
               codec_string, codec->name);
    exit(1);
245 246
}

247
/*
248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267
 * Update the requested input sample format based on the output sample format.
 * This is currently only used to request float output from decoders which
 * support multiple sample formats, one of which is AV_SAMPLE_FMT_FLT.
 * Ideally this will be removed in the future when decoders do not do format
 * conversion and only output in their native format.
 */
static void update_sample_fmt(AVCodecContext *dec, AVCodec *dec_codec,
                              AVCodecContext *enc)
{
    /* if sample formats match or a decoder sample format has already been
       requested, just return */
    if (enc->sample_fmt == dec->sample_fmt ||
        dec->request_sample_fmt > AV_SAMPLE_FMT_NONE)
        return;

    /* if decoder supports more than one output format */
    if (dec_codec && dec_codec->sample_fmts &&
        dec_codec->sample_fmts[0] != AV_SAMPLE_FMT_NONE &&
        dec_codec->sample_fmts[1] != AV_SAMPLE_FMT_NONE) {
        const enum AVSampleFormat *p;
268 269 270
        int min_dec = INT_MAX, min_inc = INT_MAX;
        enum AVSampleFormat dec_fmt = AV_SAMPLE_FMT_NONE;
        enum AVSampleFormat inc_fmt = AV_SAMPLE_FMT_NONE;
271 272 273 274 275 276

        /* find a matching sample format in the encoder */
        for (p = dec_codec->sample_fmts; *p != AV_SAMPLE_FMT_NONE; p++) {
            if (*p == enc->sample_fmt) {
                dec->request_sample_fmt = *p;
                return;
277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298
            } else {
                enum AVSampleFormat dfmt = av_get_packed_sample_fmt(*p);
                enum AVSampleFormat efmt = av_get_packed_sample_fmt(enc->sample_fmt);
                int fmt_diff = 32 * abs(dfmt - efmt);
                if (av_sample_fmt_is_planar(*p) !=
                    av_sample_fmt_is_planar(enc->sample_fmt))
                    fmt_diff++;
                if (dfmt == efmt) {
                    min_inc = fmt_diff;
                    inc_fmt = *p;
                } else if (dfmt > efmt) {
                    if (fmt_diff < min_inc) {
                        min_inc = fmt_diff;
                        inc_fmt = *p;
                    }
                } else {
                    if (fmt_diff < min_dec) {
                        min_dec = fmt_diff;
                        dec_fmt = *p;
                    }
                }
            }
299 300 301
        }

        /* if none match, provide the one that matches quality closest */
302
        dec->request_sample_fmt = min_inc != INT_MAX ? inc_fmt : dec_fmt;
303 304 305
    }
}

306
static void write_frame(AVFormatContext *s, AVPacket *pkt, OutputStream *ost)
Aneesh Dogra's avatar
Aneesh Dogra committed
307
{
308 309
    AVBitStreamFilterContext *bsfc = ost->bitstream_filters;
    AVCodecContext          *avctx = ost->st->codec;
310 311
    int ret;

312 313 314 315 316 317 318 319
    /*
     * Audio encoders may split the packets --  #frames in != #packets out.
     * But there is no reordering, so we can limit the number of output packets
     * by simply dropping them here.
     * Counting encoded video frames needs to be done separately because of
     * reordering, see do_video_out()
     */
    if (!(avctx->codec_type == AVMEDIA_TYPE_VIDEO && avctx->codec)) {
320 321
        if (ost->frame_number >= ost->max_frames) {
            av_free_packet(pkt);
322
            return;
323
        }
324 325 326
        ost->frame_number++;
    }

Aneesh Dogra's avatar
Aneesh Dogra committed
327 328 329 330 331 332 333
    while (bsfc) {
        AVPacket new_pkt = *pkt;
        int a = av_bitstream_filter_filter(bsfc, avctx, NULL,
                                           &new_pkt.data, &new_pkt.size,
                                           pkt->data, pkt->size,
                                           pkt->flags & AV_PKT_FLAG_KEY);
        if (a > 0) {
334
            av_free_packet(pkt);
Aneesh Dogra's avatar
Aneesh Dogra committed
335 336
            new_pkt.destruct = av_destruct_packet;
        } else if (a < 0) {
337 338 339
            av_log(NULL, AV_LOG_ERROR, "%s failed for stream %d, codec %s",
                   bsfc->filter->name, pkt->stream_index,
                   avctx->codec ? avctx->codec->name : "copy");
340 341
            print_error("", a);
            if (exit_on_error)
342
                exit(1);
343
        }
Aneesh Dogra's avatar
Aneesh Dogra committed
344
        *pkt = new_pkt;
345

Aneesh Dogra's avatar
Aneesh Dogra committed
346
        bsfc = bsfc->next;
347 348
    }

349
    pkt->stream_index = ost->index;
Aneesh Dogra's avatar
Aneesh Dogra committed
350 351
    ret = av_interleaved_write_frame(s, pkt);
    if (ret < 0) {
352
        print_error("av_interleaved_write_frame()", ret);
353
        exit(1);
354 355 356
    }
}

357 358
static int check_recording_time(OutputStream *ost)
{
359
    OutputFile *of = output_files[ost->file_index];
360 361 362 363

    if (of->recording_time != INT64_MAX &&
        av_compare_ts(ost->sync_opts - ost->first_pts, ost->st->codec->time_base, of->recording_time,
                      AV_TIME_BASE_Q) >= 0) {
364
        ost->finished = 1;
365 366 367 368 369
        return 0;
    }
    return 1;
}

370 371
static void do_audio_out(AVFormatContext *s, OutputStream *ost,
                         AVFrame *frame)
372 373 374
{
    AVCodecContext *enc = ost->st->codec;
    AVPacket pkt;
375
    int got_packet = 0;
376 377 378 379 380

    av_init_packet(&pkt);
    pkt.data = NULL;
    pkt.size = 0;

381 382
    if (!check_recording_time(ost))
        return;
383

384
    if (frame->pts == AV_NOPTS_VALUE || audio_sync_method < 0)
385
        frame->pts = ost->sync_opts;
386
    ost->sync_opts = frame->pts + frame->nb_samples;
387 388 389

    if (avcodec_encode_audio2(enc, &pkt, frame, &got_packet) < 0) {
        av_log(NULL, AV_LOG_FATAL, "Audio encoding failed\n");
390
        exit(1);
391 392 393 394 395
    }

    if (got_packet) {
        if (pkt.pts != AV_NOPTS_VALUE)
            pkt.pts      = av_rescale_q(pkt.pts,      enc->time_base, ost->st->time_base);
396 397
        if (pkt.dts != AV_NOPTS_VALUE)
            pkt.dts      = av_rescale_q(pkt.dts,      enc->time_base, ost->st->time_base);
398 399 400 401 402 403 404
        if (pkt.duration > 0)
            pkt.duration = av_rescale_q(pkt.duration, enc->time_base, ost->st->time_base);

        write_frame(s, &pkt, ost);

        audio_size += pkt.size;
    }
405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421
}

static void pre_process_video_frame(InputStream *ist, AVPicture *picture, void **bufp)
{
    AVCodecContext *dec;
    AVPicture *picture2;
    AVPicture picture_tmp;
    uint8_t *buf = 0;

    dec = ist->st->codec;

    /* deinterlace : must be done before any resize */
    if (do_deinterlace) {
        int size;

        /* create temporary picture */
        size = avpicture_get_size(dec->pix_fmt, dec->width, dec->height);
Aneesh Dogra's avatar
Aneesh Dogra committed
422
        buf  = av_malloc(size);
423 424 425 426 427 428
        if (!buf)
            return;

        picture2 = &picture_tmp;
        avpicture_fill(picture2, buf, dec->pix_fmt, dec->width, dec->height);

Aneesh Dogra's avatar
Aneesh Dogra committed
429
        if (avpicture_deinterlace(picture2, picture,
430 431
                                 dec->pix_fmt, dec->width, dec->height) < 0) {
            /* if error, do not deinterlace */
432
            av_log(NULL, AV_LOG_WARNING, "Deinterlacing failed\n");
433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458
            av_free(buf);
            buf = NULL;
            picture2 = picture;
        }
    } else {
        picture2 = picture;
    }

    if (picture != picture2)
        *picture = *picture2;
    *bufp = buf;
}

static void do_subtitle_out(AVFormatContext *s,
                            OutputStream *ost,
                            InputStream *ist,
                            AVSubtitle *sub,
                            int64_t pts)
{
    static uint8_t *subtitle_out = NULL;
    int subtitle_out_max_size = 1024 * 1024;
    int subtitle_out_size, nb, i;
    AVCodecContext *enc;
    AVPacket pkt;

    if (pts == AV_NOPTS_VALUE) {
459
        av_log(NULL, AV_LOG_ERROR, "Subtitle packets must have a pts\n");
460
        if (exit_on_error)
461
            exit(1);
462 463 464 465 466 467 468 469 470 471 472 473
        return;
    }

    enc = ost->st->codec;

    if (!subtitle_out) {
        subtitle_out = av_malloc(subtitle_out_max_size);
    }

    /* Note: DVB subtitle need one packet to draw them and one other
       packet to clear them */
    /* XXX: signal it in the codec context ? */
474
    if (enc->codec_id == AV_CODEC_ID_DVB_SUBTITLE)
475 476 477 478
        nb = 2;
    else
        nb = 1;

Aneesh Dogra's avatar
Aneesh Dogra committed
479
    for (i = 0; i < nb; i++) {
480 481 482 483
        ost->sync_opts = av_rescale_q(pts, ist->st->time_base, enc->time_base);
        if (!check_recording_time(ost))
            return;

484 485
        sub->pts = av_rescale_q(pts, ist->st->time_base, AV_TIME_BASE_Q);
        // start_display_time is required to be 0
Aneesh Dogra's avatar
Aneesh Dogra committed
486 487
        sub->pts               += av_rescale_q(sub->start_display_time, (AVRational){ 1, 1000 }, AV_TIME_BASE_Q);
        sub->end_display_time  -= sub->start_display_time;
488 489 490 491
        sub->start_display_time = 0;
        subtitle_out_size = avcodec_encode_subtitle(enc, subtitle_out,
                                                    subtitle_out_max_size, sub);
        if (subtitle_out_size < 0) {
492
            av_log(NULL, AV_LOG_FATAL, "Subtitle encoding failed\n");
493
            exit(1);
494 495 496 497 498
        }

        av_init_packet(&pkt);
        pkt.data = subtitle_out;
        pkt.size = subtitle_out_size;
Aneesh Dogra's avatar
Aneesh Dogra committed
499
        pkt.pts  = av_rescale_q(sub->pts, AV_TIME_BASE_Q, ost->st->time_base);
500
        if (enc->codec_id == AV_CODEC_ID_DVB_SUBTITLE) {
501 502 503 504 505 506 507
            /* XXX: the pts correction is handled here. Maybe handling
               it in the codec would be better */
            if (i == 0)
                pkt.pts += 90 * sub->start_display_time;
            else
                pkt.pts += 90 * sub->end_display_time;
        }
508
        write_frame(s, &pkt, ost);
509 510 511
    }
}

512 513 514
static void do_video_out(AVFormatContext *s,
                         OutputStream *ost,
                         AVFrame *in_picture,
Anton Khirnov's avatar
Anton Khirnov committed
515
                         int *frame_size)
516
{
517 518 519
    int ret, format_video_sync;
    AVPacket pkt;
    AVCodecContext *enc = ost->st->codec;
520 521 522

    *frame_size = 0;

523
    format_video_sync = video_sync_method;
524 525 526
    if (format_video_sync == VSYNC_AUTO)
        format_video_sync = (s->oformat->flags & AVFMT_NOTIMESTAMPS) ? VSYNC_PASSTHROUGH :
                            (s->oformat->flags & AVFMT_VARIABLE_FPS) ? VSYNC_VFR : VSYNC_CFR;
527 528 529 530
    if (format_video_sync != VSYNC_PASSTHROUGH &&
        ost->frame_number &&
        in_picture->pts != AV_NOPTS_VALUE &&
        in_picture->pts < ost->sync_opts) {
Anton Khirnov's avatar
Anton Khirnov committed
531 532
        nb_frames_drop++;
        av_log(NULL, AV_LOG_VERBOSE, "*** drop!\n");
533
        return;
Anton Khirnov's avatar
Anton Khirnov committed
534
    }
535

536 537 538 539 540
    if (in_picture->pts == AV_NOPTS_VALUE)
        in_picture->pts = ost->sync_opts;
    ost->sync_opts = in_picture->pts;


541
    if (!ost->frame_number)
542
        ost->first_pts = in_picture->pts;
543

Anton Khirnov's avatar
Anton Khirnov committed
544 545 546
    av_init_packet(&pkt);
    pkt.data = NULL;
    pkt.size = 0;
547

Anton Khirnov's avatar
Anton Khirnov committed
548 549 550
    if (!check_recording_time(ost) ||
        ost->frame_number >= ost->max_frames)
        return;
551

Anton Khirnov's avatar
Anton Khirnov committed
552
    if (s->oformat->flags & AVFMT_RAWPICTURE &&
553
        enc->codec->id == AV_CODEC_ID_RAWVIDEO) {
Anton Khirnov's avatar
Anton Khirnov committed
554 555 556 557 558 559 560 561 562
        /* raw pictures are written as AVPicture structure to
           avoid any copies. We support temporarily the older
           method. */
        enc->coded_frame->interlaced_frame = in_picture->interlaced_frame;
        enc->coded_frame->top_field_first  = in_picture->top_field_first;
        pkt.data   = (uint8_t *)in_picture;
        pkt.size   =  sizeof(AVPicture);
        pkt.pts    = av_rescale_q(in_picture->pts, enc->time_base, ost->st->time_base);
        pkt.flags |= AV_PKT_FLAG_KEY;
563

Anton Khirnov's avatar
Anton Khirnov committed
564 565 566 567 568 569 570 571 572 573 574 575 576 577 578
        write_frame(s, &pkt, ost);
    } else {
        int got_packet;
        AVFrame big_picture;

        big_picture = *in_picture;
        /* better than nothing: use input picture interlaced
           settings */
        big_picture.interlaced_frame = in_picture->interlaced_frame;
        if (ost->st->codec->flags & (CODEC_FLAG_INTERLACED_DCT|CODEC_FLAG_INTERLACED_ME)) {
            if (ost->top_field_first == -1)
                big_picture.top_field_first = in_picture->top_field_first;
            else
                big_picture.top_field_first = !!ost->top_field_first;
        }
579

Anton Khirnov's avatar
Anton Khirnov committed
580
        big_picture.quality = ost->st->codec->global_quality;
Anton Khirnov's avatar
Anton Khirnov committed
581 582 583 584 585 586 587 588 589 590
        if (!enc->me_threshold)
            big_picture.pict_type = 0;
        if (ost->forced_kf_index < ost->forced_kf_count &&
            big_picture.pts >= ost->forced_kf_pts[ost->forced_kf_index]) {
            big_picture.pict_type = AV_PICTURE_TYPE_I;
            ost->forced_kf_index++;
        }
        ret = avcodec_encode_video2(enc, &pkt, &big_picture, &got_packet);
        if (ret < 0) {
            av_log(NULL, AV_LOG_FATAL, "Video encoding failed\n");
591
            exit(1);
Anton Khirnov's avatar
Anton Khirnov committed
592
        }
593

Anton Khirnov's avatar
Anton Khirnov committed
594 595 596 597 598
        if (got_packet) {
            if (pkt.pts != AV_NOPTS_VALUE)
                pkt.pts = av_rescale_q(pkt.pts, enc->time_base, ost->st->time_base);
            if (pkt.dts != AV_NOPTS_VALUE)
                pkt.dts = av_rescale_q(pkt.dts, enc->time_base, ost->st->time_base);
599

Anton Khirnov's avatar
Anton Khirnov committed
600 601 602
            write_frame(s, &pkt, ost);
            *frame_size = pkt.size;
            video_size += pkt.size;
603

Anton Khirnov's avatar
Anton Khirnov committed
604 605 606
            /* if two pass, output log */
            if (ost->logfile && enc->stats_out) {
                fprintf(ost->logfile, "%s", enc->stats_out);
607 608
            }
        }
Anton Khirnov's avatar
Anton Khirnov committed
609 610 611 612 613 614 615 616
    }
    ost->sync_opts++;
    /*
     * For video, number of frames in == number of packets out.
     * But there may be reordering, so we can't throw away frames on encoder
     * flush, we need to limit them here, before they go into encoder.
     */
    ost->frame_number++;
617 618
}

Aneesh Dogra's avatar
Aneesh Dogra committed
619 620 621
static double psnr(double d)
{
    return -10.0 * log(d) / log(10.0);
622 623
}

624
static void do_video_stats(OutputStream *ost, int frame_size)
625 626 627 628 629 630 631 632 633 634
{
    AVCodecContext *enc;
    int frame_number;
    double ti1, bitrate, avg_bitrate;

    /* this is executed just the first time do_video_stats is called */
    if (!vstats_file) {
        vstats_file = fopen(vstats_filename, "w");
        if (!vstats_file) {
            perror("fopen");
635
            exit(1);
636 637 638 639 640 641
        }
    }

    enc = ost->st->codec;
    if (enc->codec_type == AVMEDIA_TYPE_VIDEO) {
        frame_number = ost->frame_number;
Aneesh Dogra's avatar
Aneesh Dogra committed
642
        fprintf(vstats_file, "frame= %5d q= %2.1f ", frame_number, enc->coded_frame->quality / (float)FF_QP2LAMBDA);
643
        if (enc->flags&CODEC_FLAG_PSNR)
Aneesh Dogra's avatar
Aneesh Dogra committed
644
            fprintf(vstats_file, "PSNR= %6.2f ", psnr(enc->coded_frame->error[0] / (enc->width * enc->height * 255.0 * 255.0)));
645 646 647 648 649 650 651

        fprintf(vstats_file,"f_size= %6d ", frame_size);
        /* compute pts value */
        ti1 = ost->sync_opts * av_q2d(enc->time_base);
        if (ti1 < 0.01)
            ti1 = 0.01;

Aneesh Dogra's avatar
Aneesh Dogra committed
652
        bitrate     = (frame_size * 8) / av_q2d(enc->time_base) / 1000.0;
653 654
        avg_bitrate = (double)(video_size * 8) / ti1 / 1000.0;
        fprintf(vstats_file, "s_size= %8.0fkB time= %0.3f br= %7.1fkbits/s avg_br= %7.1fkbits/s ",
Aneesh Dogra's avatar
Aneesh Dogra committed
655
               (double)video_size / 1024, ti1, bitrate, avg_bitrate);
656 657 658 659
        fprintf(vstats_file, "type= %c\n", av_get_picture_type_char(enc->coded_frame->pict_type));
    }
}

660
/*
661 662 663
 * Read one frame for lavfi output for ost and encode it.
 */
static int poll_filter(OutputStream *ost)
664
{
665
    OutputFile    *of = output_files[ost->file_index];
666 667
    AVFilterBufferRef *picref;
    AVFrame *filtered_frame = NULL;
668 669 670 671 672 673 674 675 676 677 678 679 680 681
    int frame_size, ret;

    if (!ost->filtered_frame && !(ost->filtered_frame = avcodec_alloc_frame())) {
        return AVERROR(ENOMEM);
    } else
        avcodec_get_frame_defaults(ost->filtered_frame);
    filtered_frame = ost->filtered_frame;

    if (ost->enc->type == AVMEDIA_TYPE_AUDIO &&
        !(ost->enc->capabilities & CODEC_CAP_VARIABLE_FRAME_SIZE))
        ret = av_buffersink_read_samples(ost->filter->filter, &picref,
                                         ost->st->codec->frame_size);
    else
        ret = av_buffersink_read(ost->filter->filter, &picref);
682

683 684
    if (ret < 0)
        return ret;
685

686 687 688 689 690 691 692 693
    avfilter_copy_buf_props(filtered_frame, picref);
    if (picref->pts != AV_NOPTS_VALUE) {
        filtered_frame->pts = av_rescale_q(picref->pts,
                                           ost->filter->filter->inputs[0]->time_base,
                                           ost->st->codec->time_base) -
                              av_rescale_q(of->start_time,
                                           AV_TIME_BASE_Q,
                                           ost->st->codec->time_base);
694

695
        if (of->start_time && filtered_frame->pts < 0) {
696
            avfilter_unref_buffer(picref);
697
            return 0;
698 699
        }
    }
700 701 702 703 704 705

    switch (ost->filter->filter->inputs[0]->type) {
    case AVMEDIA_TYPE_VIDEO:
        if (!ost->frame_aspect_ratio)
            ost->st->codec->sample_aspect_ratio = picref->video->pixel_aspect;

Anton Khirnov's avatar
Anton Khirnov committed
706
        do_video_out(of->ctx, ost, filtered_frame, &frame_size);
707
        if (vstats_filename && frame_size)
708
            do_video_stats(ost, frame_size);
709 710 711 712 713 714 715 716 717 718 719
        break;
    case AVMEDIA_TYPE_AUDIO:
        do_audio_out(of->ctx, ost, filtered_frame);
        break;
    default:
        // TODO support subtitle filters
        av_assert0(0);
    }

    avfilter_unref_buffer(picref);

720 721 722
    return 0;
}

723
/*
724 725 726 727 728 729 730 731 732
 * Read as many frames from possible from lavfi and encode them.
 *
 * Always read from the active stream with the lowest timestamp. If no frames
 * are available for it then return EAGAIN and wait for more input. This way we
 * can use lavfi sources that generate unlimited amount of frames without memory
 * usage exploding.
 */
static int poll_filters(void)
{
733
    int i, j, ret = 0;
734 735 736 737 738 739 740 741 742

    while (ret >= 0 && !received_sigterm) {
        OutputStream *ost = NULL;
        int64_t min_pts = INT64_MAX;

        /* choose output stream with the lowest timestamp */
        for (i = 0; i < nb_output_streams; i++) {
            int64_t pts = output_streams[i]->sync_opts;

743
            if (!output_streams[i]->filter || output_streams[i]->finished)
744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759
                continue;

            pts = av_rescale_q(pts, output_streams[i]->st->codec->time_base,
                               AV_TIME_BASE_Q);
            if (pts < min_pts) {
                min_pts = pts;
                ost = output_streams[i];
            }
        }

        if (!ost)
            break;

        ret = poll_filter(ost);

        if (ret == AVERROR_EOF) {
760 761
            OutputFile *of = output_files[ost->file_index];

762
            ost->finished = 1;
763

764 765 766 767
            if (of->shortest) {
                for (j = 0; j < of->ctx->nb_streams; j++)
                    output_streams[of->ost_index + j]->finished = 1;
            }
768 769 770 771 772 773 774 775 776

            ret = 0;
        } else if (ret == AVERROR(EAGAIN))
            return 0;
    }

    return ret;
}

777
static void print_report(int is_last_report, int64_t timer_start)
778 779 780 781 782 783 784 785 786 787 788
{
    char buf[1024];
    OutputStream *ost;
    AVFormatContext *oc;
    int64_t total_size;
    AVCodecContext *enc;
    int frame_number, vid, i;
    double bitrate, ti1, pts;
    static int64_t last_time = -1;
    static int qp_histogram[52];

789 790 791
    if (!print_stats && !is_last_report)
        return;

792 793 794 795 796 797 798 799 800 801 802 803 804 805
    if (!is_last_report) {
        int64_t cur_time;
        /* display the report every 0.5 seconds */
        cur_time = av_gettime();
        if (last_time == -1) {
            last_time = cur_time;
            return;
        }
        if ((cur_time - last_time) < 500000)
            return;
        last_time = cur_time;
    }


806
    oc = output_files[0]->ctx;
807 808

    total_size = avio_size(oc->pb);
809
    if (total_size <= 0) // FIXME improve avio_size() so it works with non seekable output too
Aneesh Dogra's avatar
Aneesh Dogra committed
810
        total_size = avio_tell(oc->pb);
811 812 813 814 815 816 817
    if (total_size < 0) {
        char errbuf[128];
        av_strerror(total_size, errbuf, sizeof(errbuf));
        av_log(NULL, AV_LOG_VERBOSE, "Bitrate not available, "
               "avio_tell() failed: %s\n", errbuf);
        total_size = 0;
    }
818 819 820 821

    buf[0] = '\0';
    ti1 = 1e10;
    vid = 0;
822
    for (i = 0; i < nb_output_streams; i++) {
823
        float q = -1;
824
        ost = output_streams[i];
825
        enc = ost->st->codec;
826
        if (!ost->stream_copy && enc->coded_frame)
Aneesh Dogra's avatar
Aneesh Dogra committed
827
            q = enc->coded_frame->quality / (float)FF_QP2LAMBDA;
828 829 830 831
        if (vid && enc->codec_type == AVMEDIA_TYPE_VIDEO) {
            snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), "q=%2.1f ", q);
        }
        if (!vid && enc->codec_type == AVMEDIA_TYPE_VIDEO) {
Aneesh Dogra's avatar
Aneesh Dogra committed
832
            float t = (av_gettime() - timer_start) / 1000000.0;
833 834 835

            frame_number = ost->frame_number;
            snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), "frame=%5d fps=%3d q=%3.1f ",
Aneesh Dogra's avatar
Aneesh Dogra committed
836 837
                     frame_number, (t > 1) ? (int)(frame_number / t + 0.5) : 0, q);
            if (is_last_report)
838
                snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), "L");
Aneesh Dogra's avatar
Aneesh Dogra committed
839
            if (qp_hist) {
840 841
                int j;
                int qp = lrintf(q);
Aneesh Dogra's avatar
Aneesh Dogra committed
842
                if (qp >= 0 && qp < FF_ARRAY_ELEMS(qp_histogram))
843
                    qp_histogram[qp]++;
Aneesh Dogra's avatar
Aneesh Dogra committed
844
                for (j = 0; j < 32; j++)
845
                    snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), "%X", (int)lrintf(log2(qp_histogram[j] + 1)));
846
            }
Aneesh Dogra's avatar
Aneesh Dogra committed
847
            if (enc->flags&CODEC_FLAG_PSNR) {
848
                int j;
Aneesh Dogra's avatar
Aneesh Dogra committed
849 850 851
                double error, error_sum = 0;
                double scale, scale_sum = 0;
                char type[3] = { 'Y','U','V' };
852
                snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), "PSNR=");
Aneesh Dogra's avatar
Aneesh Dogra committed
853 854 855 856 857 858 859
                for (j = 0; j < 3; j++) {
                    if (is_last_report) {
                        error = enc->error[j];
                        scale = enc->width * enc->height * 255.0 * 255.0 * frame_number;
                    } else {
                        error = enc->coded_frame->error[j];
                        scale = enc->width * enc->height * 255.0 * 255.0;
860
                    }
Aneesh Dogra's avatar
Aneesh Dogra committed
861 862
                    if (j)
                        scale /= 4;
863 864
                    error_sum += error;
                    scale_sum += scale;
Aneesh Dogra's avatar
Aneesh Dogra committed
865
                    snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), "%c:%2.2f ", type[j], psnr(error / scale));
866
                }
Aneesh Dogra's avatar
Aneesh Dogra committed
867
                snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), "*:%2.2f ", psnr(error_sum / scale_sum));
868 869 870 871 872 873 874 875 876 877 878
            }
            vid = 1;
        }
        /* compute min output value */
        pts = (double)ost->st->pts.val * av_q2d(ost->st->time_base);
        if ((pts < ti1) && (pts > 0))
            ti1 = pts;
    }
    if (ti1 < 0.01)
        ti1 = 0.01;

879
    bitrate = (double)(total_size * 8) / ti1 / 1000.0;
880

881
    snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf),
882 883 884
            "size=%8.0fkB time=%0.2f bitrate=%6.1fkbits/s",
            (double)total_size / 1024, ti1, bitrate);

885 886 887
    if (nb_frames_dup || nb_frames_drop)
        snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), " dup=%d drop=%d",
                nb_frames_dup, nb_frames_drop);
888

889
    av_log(NULL, AV_LOG_INFO, "%s    \r", buf);
890

891
    fflush(stderr);
892

893
    if (is_last_report) {
894
        int64_t raw= audio_size + video_size + extra_size;
895 896
        av_log(NULL, AV_LOG_INFO, "\n");
        av_log(NULL, AV_LOG_INFO, "video:%1.0fkB audio:%1.0fkB global headers:%1.0fkB muxing overhead %f%%\n",
Aneesh Dogra's avatar
Aneesh Dogra committed
897 898 899 900
               video_size / 1024.0,
               audio_size / 1024.0,
               extra_size / 1024.0,
               100.0 * (total_size - raw) / raw
901 902 903 904
        );
    }
}

905
static void flush_encoders(void)
906 907 908
{
    int i, ret;

909
    for (i = 0; i < nb_output_streams; i++) {
910
        OutputStream   *ost = output_streams[i];
911
        AVCodecContext *enc = ost->st->codec;
912
        AVFormatContext *os = output_files[ost->file_index]->ctx;
913
        int stop_encoding = 0;
914

915
        if (!ost->encoding_needed)
916 917
            continue;

Aneesh Dogra's avatar
Aneesh Dogra committed
918
        if (ost->st->codec->codec_type == AVMEDIA_TYPE_AUDIO && enc->frame_size <= 1)
919
            continue;
920
        if (ost->st->codec->codec_type == AVMEDIA_TYPE_VIDEO && (os->oformat->flags & AVFMT_RAWPICTURE) && enc->codec->id == AV_CODEC_ID_RAWVIDEO)
921 922
            continue;

Aneesh Dogra's avatar
Aneesh Dogra committed
923
        for (;;) {
924 925 926
            int (*encode)(AVCodecContext*, AVPacket*, const AVFrame*, int*) = NULL;
            const char *desc;
            int64_t *size;
927 928 929

            switch (ost->st->codec->codec_type) {
            case AVMEDIA_TYPE_AUDIO:
930 931 932
                encode = avcodec_encode_audio2;
                desc   = "Audio";
                size   = &audio_size;
933 934
                break;
            case AVMEDIA_TYPE_VIDEO:
935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950
                encode = avcodec_encode_video2;
                desc   = "Video";
                size   = &video_size;
                break;
            default:
                stop_encoding = 1;
            }

            if (encode) {
                AVPacket pkt;
                int got_packet;
                av_init_packet(&pkt);
                pkt.data = NULL;
                pkt.size = 0;

                ret = encode(enc, &pkt, NULL, &got_packet);
951
                if (ret < 0) {
952
                    av_log(NULL, AV_LOG_FATAL, "%s encoding failed\n", desc);
953
                    exit(1);
954
                }
955
                *size += ret;
956 957 958
                if (ost->logfile && enc->stats_out) {
                    fprintf(ost->logfile, "%s", enc->stats_out);
                }
959
                if (!got_packet) {
960 961 962
                    stop_encoding = 1;
                    break;
                }
963 964 965 966
                if (pkt.pts != AV_NOPTS_VALUE)
                    pkt.pts = av_rescale_q(pkt.pts, enc->time_base, ost->st->time_base);
                if (pkt.dts != AV_NOPTS_VALUE)
                    pkt.dts = av_rescale_q(pkt.dts, enc->time_base, ost->st->time_base);
967 968
                if (pkt.duration > 0)
                    pkt.duration = av_rescale_q(pkt.duration, enc->time_base, ost->st->time_base);
969
                write_frame(os, &pkt, ost);
970
            }
971

972
            if (stop_encoding)
973
                break;
974 975 976 977
        }
    }
}

978 979 980 981 982
/*
 * Check whether a packet from ist should be written into ost at this time
 */
static int check_output_constraints(InputStream *ist, OutputStream *ost)
{
983 984
    OutputFile *of = output_files[ost->file_index];
    int ist_index  = input_files[ist->file_index]->ist_index + ist->st->index;
985 986 987 988

    if (ost->source_index != ist_index)
        return 0;

989
    if (of->start_time && ist->last_dts < of->start_time)
990 991 992 993 994 995 996
        return 0;

    return 1;
}

static void do_streamcopy(InputStream *ist, OutputStream *ost, const AVPacket *pkt)
{
997
    OutputFile *of = output_files[ost->file_index];
998 999 1000 1001 1002 1003 1004 1005 1006
    int64_t ost_tb_start_time = av_rescale_q(of->start_time, AV_TIME_BASE_Q, ost->st->time_base);
    AVPacket opkt;

    av_init_packet(&opkt);

    if ((!ost->frame_number && !(pkt->flags & AV_PKT_FLAG_KEY)) &&
        !ost->copy_initial_nonkeyframes)
        return;

1007
    if (of->recording_time != INT64_MAX &&
1008
        ist->last_dts >= of->recording_time + of->start_time) {
1009
        ost->finished = 1;
1010 1011