]> git.sesse.net Git - ffmpeg/blobdiff - ffmpeg.c
Correct normalization constant for the vertical filter, iam not
[ffmpeg] / ffmpeg.c
index ee9a4883b29d9740dbfb1311c11cba3cbdb2fc0e..ab859efc074514d1b7989a1e20fb182531d05b0b 100644 (file)
--- a/ffmpeg.c
+++ b/ffmpeg.c
@@ -19,6 +19,9 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
+/* needed for usleep() */
+#define _XOPEN_SOURCE 500
+
 #include "config.h"
 #include <ctype.h>
 #include <string.h>
@@ -27,6 +30,7 @@
 #include <errno.h>
 #include <signal.h>
 #include <limits.h>
+#include <unistd.h>
 #include "libavformat/avformat.h"
 #include "libavdevice/avdevice.h"
 #include "libswscale/swscale.h"
@@ -45,7 +49,6 @@
 #endif
 
 #if defined(HAVE_TERMIOS_H)
-#include <unistd.h>
 #include <fcntl.h>
 #include <sys/ioctl.h>
 #include <sys/time.h>
@@ -92,7 +95,7 @@ static int nb_input_files = 0;
 static AVFormatContext *output_files[MAX_FILES];
 static int nb_output_files = 0;
 
-static AVStreamMap stream_maps[MAX_FILES];
+static AVStreamMap stream_maps[MAX_FILES*MAX_STREAMS];
 static int nb_stream_maps;
 
 static AVMetaDataMap meta_data_maps[MAX_FILES];
@@ -192,6 +195,7 @@ static char *video_standard;
 
 static int audio_volume = 256;
 
+static int exit_on_error = 0;
 static int using_stdin = 0;
 static int using_vhook = 0;
 static int verbose = 1;
@@ -257,6 +261,8 @@ typedef struct AVOutputStream {
     /* audio only */
     int audio_resample;
     ReSampleContext *resample; /* for audio resampling */
+    int reformat_pair;
+    AVAudioConvert *reformat_ctx;
     AVFifoBuffer fifo;     /* for compression: one audio fifo per codec */
     FILE *logfile;
 } AVOutputStream;
@@ -490,6 +496,8 @@ static void write_frame(AVFormatContext *s, AVPacket *pkt, AVCodecContext *avctx
                     bsfc->filter->name, pkt->stream_index,
                     avctx->codec ? avctx->codec->name : "copy");
             print_error("", a);
+            if (exit_on_error)
+                av_exit(1);
         }
         *pkt= new_pkt;
 
@@ -513,11 +521,14 @@ static void do_audio_out(AVFormatContext *s,
     uint8_t *buftmp;
     static uint8_t *audio_buf = NULL;
     static uint8_t *audio_out = NULL;
+    static uint8_t *audio_out2 = NULL;
     const int audio_out_size= 4*MAX_AUDIO_PACKET_SIZE;
 
     int size_out, frame_bytes, ret;
     AVCodecContext *enc= ost->st->codec;
     AVCodecContext *dec= ist->st->codec;
+    int osize= av_get_bits_per_sample_format(enc->sample_fmt)/8;
+    int isize= av_get_bits_per_sample_format(dec->sample_fmt)/8;
 
     /* SC: dynamic allocation of buffers */
     if (!audio_buf)
@@ -531,6 +542,10 @@ static void do_audio_out(AVFormatContext *s,
         ost->audio_resample = 1;
 
     if (ost->audio_resample && !ost->resample) {
+        if (dec->sample_fmt != SAMPLE_FMT_S16) {
+            fprintf(stderr, "Audio resampler only works with 16 bits per sample, patch welcome.\n");
+            av_exit(1);
+        }
         ost->resample = audio_resample_init(enc->channels,    dec->channels,
                                             enc->sample_rate, dec->sample_rate);
         if (!ost->resample) {
@@ -541,6 +556,26 @@ static void do_audio_out(AVFormatContext *s,
         }
     }
 
+#define MAKE_SFMT_PAIR(a,b) ((a)+SAMPLE_FMT_NB*(b))
+    if (dec->sample_fmt!=enc->sample_fmt &&
+        MAKE_SFMT_PAIR(enc->sample_fmt,dec->sample_fmt)!=ost->reformat_pair) {
+        if (!audio_out2)
+            audio_out2 = av_malloc(audio_out_size);
+        if (!audio_out2)
+            av_exit(1);
+        if (ost->reformat_ctx)
+            av_audio_convert_free(ost->reformat_ctx);
+        ost->reformat_ctx = av_audio_convert_alloc(enc->sample_fmt, 1,
+                                                   dec->sample_fmt, 1, NULL, 0);
+        if (!ost->reformat_ctx) {
+            fprintf(stderr, "Cannot convert %s sample format to %s sample format\n",
+                avcodec_get_sample_fmt_name(dec->sample_fmt),
+                avcodec_get_sample_fmt_name(enc->sample_fmt));
+            av_exit(1);
+        }
+        ost->reformat_pair=MAKE_SFMT_PAIR(enc->sample_fmt,dec->sample_fmt);
+    }
+
     if(audio_sync_method){
         double delta = get_sync_ipts(ost) * enc->sample_rate - ost->sync_opts
                 - av_fifo_size(&ost->fifo)/(ost->st->codec->channels * 2);
@@ -592,20 +627,39 @@ static void do_audio_out(AVFormatContext *s,
         buftmp = audio_buf;
         size_out = audio_resample(ost->resample,
                                   (short *)buftmp, (short *)buf,
-                                  size / (ist->st->codec->channels * 2));
-        size_out = size_out * enc->channels * 2;
+                                  size / (ist->st->codec->channels * isize));
+        size_out = size_out * enc->channels * osize;
     } else {
         buftmp = buf;
         size_out = size;
     }
 
+    if (dec->sample_fmt!=enc->sample_fmt) {
+        const void *ibuf[6]= {buftmp};
+        void *obuf[6]= {audio_out2};
+        int istride[6]= {isize};
+        int ostride[6]= {osize};
+        int len= size_out/istride[0];
+        if (av_audio_convert(ost->reformat_ctx, obuf, ostride, ibuf, istride, len)<0) {
+            printf("av_audio_convert() failed\n");
+            if (exit_on_error)
+                av_exit(1);
+            return;
+        }
+        buftmp = audio_out2;
+        size_out = len*osize;
+    }
+
     /* now encode as many frames as possible */
     if (enc->frame_size > 1) {
         /* output resampled raw samples */
-        av_fifo_realloc(&ost->fifo, av_fifo_size(&ost->fifo) + size_out);
+        if (av_fifo_realloc2(&ost->fifo, av_fifo_size(&ost->fifo) + size_out) < 0) {
+            fprintf(stderr, "av_fifo_realloc2() failed\n");
+            av_exit(1);
+        }
         av_fifo_generic_write(&ost->fifo, buftmp, size_out, NULL);
 
-        frame_bytes = enc->frame_size * 2 * enc->channels;
+        frame_bytes = enc->frame_size * osize * enc->channels;
 
         while (av_fifo_size(&ost->fifo) >= frame_bytes) {
             AVPacket pkt;
@@ -630,36 +684,17 @@ static void do_audio_out(AVFormatContext *s,
         }
     } else {
         AVPacket pkt;
+        int coded_bps = av_get_bits_per_sample(enc->codec->id)/8;
         av_init_packet(&pkt);
 
-        ost->sync_opts += size_out / (2 * enc->channels);
+        ost->sync_opts += size_out / (osize * enc->channels);
 
         /* output a pcm frame */
-        /* XXX: change encoding codec API to avoid this ? */
-        switch(enc->codec->id) {
-        case CODEC_ID_PCM_S32LE:
-        case CODEC_ID_PCM_S32BE:
-        case CODEC_ID_PCM_U32LE:
-        case CODEC_ID_PCM_U32BE:
-        case CODEC_ID_PCM_F32BE:
-            size_out = size_out << 1;
-            break;
-        case CODEC_ID_PCM_S24LE:
-        case CODEC_ID_PCM_S24BE:
-        case CODEC_ID_PCM_U24LE:
-        case CODEC_ID_PCM_U24BE:
-        case CODEC_ID_PCM_S24DAUD:
-            size_out = size_out / 2 * 3;
-            break;
-        case CODEC_ID_PCM_S16LE:
-        case CODEC_ID_PCM_S16BE:
-        case CODEC_ID_PCM_U16LE:
-        case CODEC_ID_PCM_U16BE:
-            break;
-        default:
-            size_out = size_out >> 1;
-            break;
-        }
+        /* determine the size of the coded buffer */
+        size_out /= osize;
+        if (coded_bps)
+            size_out *= coded_bps;
+
         //FIXME pass ost->sync_opts as AVFrame.pts in avcodec_encode_audio()
         ret = avcodec_encode_audio(enc, audio_out, size_out,
                                    (short *)buftmp);
@@ -700,6 +735,7 @@ static void pre_process_video_frame(AVInputStream *ist, AVPicture *picture, void
             if(avpicture_deinterlace(picture2, picture,
                                      dec->pix_fmt, dec->width, dec->height) < 0) {
                 /* if error, do not deinterlace */
+                fprintf(stderr, "Deinterlacing failed\n");
                 av_free(buf);
                 buf = NULL;
                 picture2 = picture;
@@ -737,6 +773,8 @@ static void do_subtitle_out(AVFormatContext *s,
 
     if (pts == AV_NOPTS_VALUE) {
         fprintf(stderr, "Subtitle packets must have a pts\n");
+        if (exit_on_error)
+            av_exit(1);
         return;
     }
 
@@ -830,6 +868,8 @@ static void do_video_out(AVFormatContext *s,
     if (ost->video_crop) {
         if (av_picture_crop((AVPicture *)&picture_crop_temp, (AVPicture *)in_picture, dec->pix_fmt, ost->topBand, ost->leftBand) < 0) {
             av_log(NULL, AV_LOG_ERROR, "error cropping picture\n");
+            if (exit_on_error)
+                av_exit(1);
             return;
         }
         formatted_picture = &picture_crop_temp;
@@ -845,6 +885,8 @@ static void do_video_out(AVFormatContext *s,
         if (ost->video_resample) {
             if (av_picture_crop((AVPicture *)&picture_pad_temp, (AVPicture *)final_picture, enc->pix_fmt, ost->padtop, ost->padleft) < 0) {
                 av_log(NULL, AV_LOG_ERROR, "error padding picture\n");
+                if (exit_on_error)
+                    av_exit(1);
                 return;
             }
             resampling_dst = &picture_pad_temp;
@@ -1159,8 +1201,11 @@ static int output_packet(AVInputStream *ist, int ist_index,
         if (ist->decoding_needed) {
             switch(ist->st->codec->codec_type) {
             case CODEC_TYPE_AUDIO:{
-                if(pkt)
-                    samples= av_fast_realloc(samples, &samples_size, FFMAX(pkt->size*sizeof(*samples), AVCODEC_MAX_AUDIO_FRAME_SIZE));
+                if(pkt && samples_size < FFMAX(pkt->size*sizeof(*samples), AVCODEC_MAX_AUDIO_FRAME_SIZE)) {
+                    samples_size = FFMAX(pkt->size*sizeof(*samples), AVCODEC_MAX_AUDIO_FRAME_SIZE);
+                    av_free(samples);
+                    samples= av_malloc(samples_size);
+                }
                 data_size= samples_size;
                     /* XXX: could avoid copy if PCM 16 bits with same
                        endianness as CPU */
@@ -1587,22 +1632,22 @@ static int av_encode(AVFormatContext **output_files,
     n = 0;
     for(k=0;k<nb_output_files;k++) {
         os = output_files[k];
-        for(i=0;i<os->nb_streams;i++) {
+        for(i=0;i<os->nb_streams;i++,n++) {
             int found;
-            ost = ost_table[n++];
+            ost = ost_table[n];
             ost->file_index = k;
             ost->index = i;
             ost->st = os->streams[i];
             if (nb_stream_maps > 0) {
-                ost->source_index = file_table[stream_maps[n-1].file_index].ist_index +
-                    stream_maps[n-1].stream_index;
+                ost->source_index = file_table[stream_maps[n].file_index].ist_index +
+                    stream_maps[n].stream_index;
 
                 /* Sanity check that the stream types match */
                 if (ist_table[ost->source_index]->st->codec->codec_type != ost->st->codec->codec_type) {
                     int i= ost->file_index;
                     dump_format(output_files[i], i, output_files[i]->filename, 1);
                     fprintf(stderr, "Codec type mismatch for mapping #%d.%d -> #%d.%d\n",
-                        stream_maps[n-1].file_index, stream_maps[n-1].stream_index,
+                        stream_maps[n].file_index, stream_maps[n].stream_index,
                         ost->file_index, ost->index);
                     av_exit(1);
                 }
@@ -1652,8 +1697,8 @@ static int av_encode(AVFormatContext **output_files,
             ist = ist_table[ost->source_index];
             ist->discard = 0;
             ost->sync_ist = (nb_stream_maps > 0) ?
-                ist_table[file_table[stream_maps[n-1].sync_file_index].ist_index +
-                         stream_maps[n-1].sync_stream_index] : ist;
+                ist_table[file_table[stream_maps[n].sync_file_index].ist_index +
+                         stream_maps[n].sync_stream_index] : ist;
         }
     }
 
@@ -1726,6 +1771,7 @@ static int av_encode(AVFormatContext **output_files,
             case CODEC_TYPE_AUDIO:
                 if (av_fifo_init(&ost->fifo, 1024))
                     goto fail;
+                ost->reformat_pair = MAKE_SFMT_PAIR(SAMPLE_FMT_NONE,SAMPLE_FMT_NONE);
                 ost->audio_resample = codec->sample_rate != icodec->sample_rate || audio_sync_method > 1;
                 icodec->request_channels = codec->channels;
                 ist->decoding_needed = 1;
@@ -2085,7 +2131,8 @@ static int av_encode(AVFormatContext **output_files,
             if (verbose >= 0)
                 fprintf(stderr, "Error while decoding stream #%d.%d\n",
                         ist->file_index, ist->index);
-
+            if (exit_on_error)
+                av_exit(1);
             av_free_packet(&pkt);
             goto redo;
         }
@@ -2162,6 +2209,8 @@ static int av_encode(AVFormatContext **output_files,
                     sws_freeContext(ost->img_resample_ctx);
                 if (ost->resample)
                     audio_resample_close(ost->resample);
+                if (ost->reformat_ctx)
+                    av_audio_convert_free(ost->reformat_ctx);
                 av_free(ost);
             }
         }
@@ -2267,12 +2316,13 @@ static int opt_verbose(const char *opt, const char *arg)
     return 0;
 }
 
-static void opt_frame_rate(const char *arg)
+static int opt_frame_rate(const char *opt, const char *arg)
 {
     if (av_parse_video_frame_rate(&frame_rate, arg) < 0) {
-        fprintf(stderr, "Incorrect frame rate\n");
+        fprintf(stderr, "Incorrect value for %s: %s\n", opt, arg);
         av_exit(1);
     }
+    return 0;
 }
 
 static int opt_bitrate(const char *opt, const char *arg)
@@ -2763,6 +2813,7 @@ static void opt_input_file(const char *filename)
     ap->width = frame_width + frame_padleft + frame_padright;
     ap->height = frame_height + frame_padtop + frame_padbottom;
     ap->pix_fmt = frame_pix_fmt;
+   // ap->sample_fmt = audio_sample_fmt; //FIXME:not implemented in libavformat
     ap->channel = video_channel;
     ap->standard = video_standard;
     ap->video_codec_id = find_codec_or_die(video_codec_name, CODEC_TYPE_VIDEO, 0);
@@ -2827,6 +2878,7 @@ static void opt_input_file(const char *filename)
             //fprintf(stderr, "\nInput Audio channels: %d", enc->channels);
             audio_channels = enc->channels;
             audio_sample_rate = enc->sample_rate;
+            audio_sample_fmt = enc->sample_fmt;
             if(audio_disable)
                 ic->streams[i]->discard= AVDISCARD_ALL;
             break;
@@ -2834,7 +2886,11 @@ static void opt_input_file(const char *filename)
             set_context_opts(enc, avctx_opts[CODEC_TYPE_VIDEO], AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_DECODING_PARAM);
             frame_height = enc->height;
             frame_width = enc->width;
-            frame_aspect_ratio = av_q2d(enc->sample_aspect_ratio) * enc->width / enc->height;
+            if(ic->streams[i]->sample_aspect_ratio.num)
+                frame_aspect_ratio=av_q2d(ic->streams[i]->sample_aspect_ratio);
+            else
+                frame_aspect_ratio=av_q2d(enc->sample_aspect_ratio);
+            frame_aspect_ratio *= (float) enc->width / enc->height;
             frame_pix_fmt = enc->pix_fmt;
             rfps      = ic->streams[i]->r_frame_rate.num;
             rfps_base = ic->streams[i]->r_frame_rate.den;
@@ -2965,6 +3021,7 @@ static void new_video_stream(AVFormatContext *oc)
     if (video_stream_copy) {
         st->stream_copy = 1;
         video_enc->codec_type = CODEC_TYPE_VIDEO;
+        st->sample_aspect_ratio = av_d2q(frame_aspect_ratio*frame_height/frame_width, 255);
     } else {
         const char *p;
         int i;
@@ -3002,6 +3059,7 @@ static void new_video_stream(AVFormatContext *oc)
         video_enc->height = frame_height + frame_padtop + frame_padbottom;
         video_enc->sample_aspect_ratio = av_d2q(frame_aspect_ratio*video_enc->height/video_enc->width, 255);
         video_enc->pix_fmt = frame_pix_fmt;
+        st->sample_aspect_ratio = video_enc->sample_aspect_ratio;
 
         if(codec && codec->pix_fmts){
             const enum PixelFormat *p= codec->pix_fmts;
@@ -3109,6 +3167,7 @@ static void new_audio_stream(AVFormatContext *oc)
         st->stream_copy = 1;
         audio_enc->channels = audio_channels;
     } else {
+        AVCodec *codec;
         codec_id = av_guess_codec(oc->oformat, NULL, oc->filename, NULL, CODEC_TYPE_AUDIO);
 
         set_context_opts(audio_enc, avctx_opts[CODEC_TYPE_AUDIO], AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM);
@@ -3116,6 +3175,7 @@ static void new_audio_stream(AVFormatContext *oc)
         if (audio_codec_name)
             codec_id = find_codec_or_die(audio_codec_name, CODEC_TYPE_AUDIO, 1);
         audio_enc->codec_id = codec_id;
+        codec = avcodec_find_encoder(codec_id);
 
         if (audio_qscale > QSCALE_NONE) {
             audio_enc->flags |= CODEC_FLAG_QSCALE;
@@ -3123,6 +3183,17 @@ static void new_audio_stream(AVFormatContext *oc)
         }
         audio_enc->thread_count = thread_count;
         audio_enc->channels = audio_channels;
+        audio_enc->sample_fmt = audio_sample_fmt;
+
+        if(codec && codec->sample_fmts){
+            const enum SampleFormat *p= codec->sample_fmts;
+            for(; *p!=-1; p++){
+                if(*p == audio_enc->sample_fmt)
+                    break;
+            }
+            if(*p == -1)
+                audio_enc->sample_fmt = codec->sample_fmts[0];
+        }
     }
     audio_enc->sample_rate = audio_sample_rate;
     audio_enc->time_base= (AVRational){1, audio_sample_rate};
@@ -3524,7 +3595,7 @@ static void opt_target(const char *arg)
         opt_format("vcd");
 
         opt_frame_size(norm ? "352x240" : "352x288");
-        opt_frame_rate(frame_rates[norm]);
+        opt_frame_rate(NULL, frame_rates[norm]);
         opt_default("gop", norm ? "18" : "15");
 
         opt_default("b", "1150000");
@@ -3552,7 +3623,7 @@ static void opt_target(const char *arg)
         opt_format("svcd");
 
         opt_frame_size(norm ? "480x480" : "480x576");
-        opt_frame_rate(frame_rates[norm]);
+        opt_frame_rate(NULL, frame_rates[norm]);
         opt_default("gop", norm ? "18" : "15");
 
         opt_default("b", "2040000");
@@ -3574,7 +3645,7 @@ static void opt_target(const char *arg)
         opt_format("dvd");
 
         opt_frame_size(norm ? "720x480" : "720x576");
-        opt_frame_rate(frame_rates[norm]);
+        opt_frame_rate(NULL, frame_rates[norm]);
         opt_default("gop", norm ? "18" : "15");
 
         opt_default("b", "6000000");
@@ -3595,7 +3666,7 @@ static void opt_target(const char *arg)
         opt_frame_size(norm ? "720x480" : "720x576");
         opt_frame_pix_fmt(!strncmp(arg, "dv50", 4) ? "yuv422p" :
                                              (norm ? "yuv411p" : "yuv420p"));
-        opt_frame_rate(frame_rates[norm]);
+        opt_frame_rate(NULL, frame_rates[norm]);
 
         audio_sample_rate = 48000;
         audio_channels = 2;
@@ -3736,12 +3807,13 @@ static const OptionDef options[] = {
     { "shortest", OPT_BOOL | OPT_EXPERT, {(void*)&opt_shortest}, "finish encoding within shortest input" }, //
     { "dts_delta_threshold", HAS_ARG | OPT_FLOAT | OPT_EXPERT, {(void*)&dts_delta_threshold}, "timestamp discontinuity delta threshold", "threshold" },
     { "programid", HAS_ARG | OPT_INT | OPT_EXPERT, {(void*)&opt_programid}, "desired program number", "" },
+    { "xerror", OPT_BOOL, {(void*)&exit_on_error}, "exit on error", "error" },
 
     /* video options */
     { "b", OPT_FUNC2 | HAS_ARG | OPT_VIDEO, {(void*)opt_bitrate}, "set bitrate (in bits/s)", "bitrate" },
     { "vb", OPT_FUNC2 | HAS_ARG | OPT_VIDEO, {(void*)opt_bitrate}, "set bitrate (in bits/s)", "bitrate" },
     { "vframes", OPT_INT | HAS_ARG | OPT_VIDEO, {(void*)&max_frames[CODEC_TYPE_VIDEO]}, "set the number of video frames to record", "number" },
-    { "r", HAS_ARG | OPT_VIDEO, {(void*)opt_frame_rate}, "set frame rate (Hz value, fraction or abbreviation)", "rate" },
+    { "r", OPT_FUNC2 | HAS_ARG | OPT_VIDEO, {(void*)opt_frame_rate}, "set frame rate (Hz value, fraction or abbreviation)", "rate" },
     { "s", HAS_ARG | OPT_VIDEO, {(void*)opt_frame_size}, "set frame size (WxH or abbreviation)", "size" },
     { "aspect", HAS_ARG | OPT_VIDEO, {(void*)opt_frame_aspect_ratio}, "set aspect ratio (4:3, 16:9 or 1.3333, 1.7777)", "aspect" },
     { "pix_fmt", HAS_ARG | OPT_EXPERT | OPT_VIDEO, {(void*)opt_frame_pix_fmt}, "set pixel format, 'list' as argument shows all the pixel formats supported", "format" },
@@ -3793,6 +3865,7 @@ static const OptionDef options[] = {
     { "vol", OPT_INT | HAS_ARG | OPT_AUDIO, {(void*)&audio_volume}, "change audio volume (256=normal)" , "volume" }, //
     { "newaudio", OPT_AUDIO, {(void*)opt_new_audio_stream}, "add a new audio stream to the current output stream" },
     { "alang", HAS_ARG | OPT_STRING | OPT_AUDIO, {(void *)&audio_language}, "set the ISO 639 language code (3 letters) of the current audio stream" , "code" },
+    { "sample_fmt", HAS_ARG | OPT_EXPERT | OPT_AUDIO, {(void*)opt_audio_sample_fmt}, "set sample format, 'list' as argument shows all the sample formats supported", "format" },
 
     /* subtitle options */
     { "sn", OPT_BOOL | OPT_SUBTITLE, {(void*)&subtitle_disable}, "disable subtitle" },