Strange errors when using ffmpeg in a loop

Question

I have a bash script looping through the results of a find and performing an ffmpeg encoding of some FLV files. Whilst the script is running the ffmpeg output seems to be interupted and is outputting some strange looking errors like the one below. I've no idea what is going on here. Can anyone point me in the right direction?

It's as though the loop is still running when it shouldn't be and interupting the ffmpeg process.

The specific error is:

frame=   68 fps= 67 q=28.0 00000000000000000000000000001000size=      22kB time=00:00:00.50 bitrate= 363.2kbits/s dup=1 drop=0    
Enter command: <target> <time> <command>[ <argument>]
Parse error, at least 3 arguments were expected, only 1 given in string 'om/pt_br/nx/R3T4N2_HD3D_demoCheckedOut.flv'

Some more details from the ffmpeg output:

[buffer @ 0xa30e1e0] w:800 h:600 pixfmt:yuv420p tb:1/1000000 sar:0/1 sws_param:flags=2
[libx264 @ 0xa333240] using cpu capabilities: MMX2 SSE2Fast SSSE3 FastShuffle SSE4.1 Cache64
[libx264 @ 0xa333240] profile High, level 3.1
[libx264 @ 0xa333240] 264 - core 122 r2184 5c85e0a - H.264/MPEG-4 AVC codec - Copyleft 2003-2012 - http://www.videolan.org/x264.html - options: cabac=1 ref=5 deblock=1:0:0 analyse=0x3:0x113 me=umh subme=8 psy=1 psy_rd=1.00:0.00 mixed_ref=1 me_range=16 chroma_me=1 trellis=1 8x8dct=1 cqm=0 deadzone=21,11 fast_pskip=1 chroma_qp_offset=-2 threads=1 sliced_threads=0 nr=0 decimate=1 interlaced=0 bluray_compat=0 constrained_intra=0 bframes=3 b_pyramid=2 b_adapt=2 b_bias=0 direct=3 weightb=1 open_gop=0 weightp=2 keyint=250 keyint_min=25 scenecut=40 intra_refresh=0 rc_lookahead=50 rc=cbr mbtree=1 bitrate=500 ratetol=1.0 qcomp=0.60 qpmin=0 qpmax=69 qpstep=4 vbv_maxrate=500 vbv_bufsize=1000 nal_hrd=none ip_ratio=1.40 aq=1:1.00
Output #0, mp4, to './mp4s/pt_br/teamcenter/tc8_interactive/videos/8_SRM_EN.mp4':
  Metadata:
    audiodelay      : 0
    canSeekToEnd    : true
    encoder         : Lavf54.3.100
    Stream #0:0: Video: h264 (![0][0][0] / 0x0021), yuv420p, 800x600, q=-1--1, 500 kb/s, 30k tbn, 29.97 tbc
    Stream #0:1: Audio: aac (@[0][0][0] / 0x0040), 44100 Hz, mono, s16, 128 kb/s
Stream mapping:
  Stream #0:1 -> #0:0 (vp6f -> libx264)
  Stream #0:0 -> #0:1 (mp3 -> libfaac)
Press [q] to stop, [?] for help
error parsing debug value0 00000000000000000000000000000000size=      13kB time=00:00:00.-3 bitrate=-3165.5kbits/s dup=1 drop=0    
debug=0
frame=   68 fps= 67 q=28.0 00000000000000000000000000001000size=      22kB time=00:00:00.50 bitrate= 363.2kbits/s dup=1 drop=0    
Enter command: <target> <time> <command>[ <argument>]
Parse error, at least 3 arguments were expected, only 1 given in string 'om/pt_br/nx/R3T4N2_HD3D_demoCheckedOut.flv'

The script is as follows

#!/bin/bash
LOGFILE=encodemp4ize.log
echo '' > $LOGFILE
STARTTIME=date
echo "Started at `$STARTTIME`" >> $LOGFILE
rsync -avz flvs/ mp4s/ --exclude '*.flv'
#find flvs/ -name "*.flv" > flv-files
# The loop
find flvs/ -name "*.flv" | while read f
do
FILENAME=`echo $f | sed 's#flvs/##'`
MP4FILENAME=`echo $FILENAME | sed 's#.flv#.mp4#'`
ffmpeg -i "$f" -vcodec libx264 -vprofile high -preset slow -b:v 500k -maxrate 500k -bufsize 1000k -threads 0 -acodec libfaac -ab 128k "./mp4s/$MP4FILENAME"
echo "$f MP4 done" >> $LOGFILE
done

Answer 1

Your question is actually Bash FAQ #89: just add </dev/null to prevent ffmpeg from reading its standard input.

---

I've taken the liberty of fixing up your script for you because it contains a lot of potential errors. A few of the important points:

• Filenames are tricky to handle, because most filesystems allow them to contain all sorts of unprintable characters normal people would see as garbage. Making simplifying assumptions like "file names contain only 'normal' characters" tends to result fragile shell scripts that appear to work on "normal" file names and then break the day they run into a particularly nasty file name that doesn't follow the script's assumptions. On the other hand, correctly handling file names can be such a bother that you may find it not worth the effort if the chance of encountering a weird file name is expected to be near zero (i.e. you only use the script on your own files and you give your own files "simple" names). Sometimes it is possible to avoid this decision altogether by not parsing file names at all. Fortunately, that is possible with find(1)'s -exec option. Just put {} in the argument to -exec and you don't have to worry about parsing find output.

• Using sed or other external processes to do simple string operations like stripping extensions and prefixes is inefficient. Instead, use parameter expansions which are part of the shell (no external process means it will be faster). Some helpful articles on the subject are listed below:

  • Bash FAQ 73: Parameter expansions
  • Bash FAQ 100: String manipulations

• Use $( ), and don't use `` anymore: Bash FAQ 82.

• Avoid using UPPERCASE variable names. That namespace is generally reserved by the shell for special purposes (like PATH), so using it for your own variables is a bad idea.

And now, without further ado, here's a cleaned up script for you:

#!/bin/sh

logfile=encodemp4ize.log
echo "Started at $(date)." > "$logfile"
rsync -avz --exclude '*.flv' flvs/ mp4s/

find flvs/ -type f -name '*.flv' -exec sh -c '
for flvsfile; do
    file=${flvsfile#flvs/}
    < /dev/null ffmpeg -i "$flvsfile" -vcodec libx264 -vprofile high \
        -preset slow -b:v 500k -maxrate 500k -bufsize 1000k \
        -threads 0 -acodec libfaac -ab 128k \
        "mp4s/${file%flv}"mp4
    printf %s\\n "$flvsfile MP4 done." >> "$logfile"
done
' _ {} +

^{Note: I used POSIX sh because you didn't use or need any bash-specific features in your original.}

Answer 2

I've found the solution. The bash script seems to produce input (Namely the 'c' key) which interferes with the ffmpeg process.

Adding < /dev/null to the ffmpeg command line, like so:

ffmpeg -i "./$f" -vcodec libx264 -vprofile high -preset slow -b:v 500k -maxrate 500k -bufsize 1000k -threads 0 -acodec libfaac -ab 128k "./mp4s/$MP4FILENAME" < /dev/null

fixes the issue.

Answer 3

As an alternative solution to ffmpeg [...] < /dev/null, you can use:

ffmpeg -nostdin [...]

Details from the ffmpeg documentation:

To explicitly disable interaction you need to specify -nostdin.

Disabling interaction on standard input is useful, for example, if ffmpeg is in the background process group. Roughly the same result can be achieved with ffmpeg ... < /dev/null but it requires a shell.

Answer 4

In the while loop, I have faced similar problems. You can use for loop, and in for loop, you can avoid taking the list of audios and running it in the loop using the find command.

I have used find and sed to take the list of audios in my directory because I have seen situations where using wildcard characters throws "argument list too long" error.

Now if we just do the find then it will give the absolute path of the files which will cause the ffmpeg to fail, so we have removed the absolute path using sed.

You can use a command like this (have used it a lot of times and it is working perfectly)

for f1 in `find . -maxdepth 1 -name "*.mkv" | sed 's/^\.\///g'`; do ffmpeg -i "$f1" -q:a 0 -map a ../wav/"${f1%.*}.wav"; done