在 bash 中跨多个文件夹递归运行 ffmpeg concat 脚本

Question

I wrote a bash script that concatenates all video files in a folder using ffmpeg. I would like to be able to run this script recursively on multiple folders. My problem has been that I am unable to change into the directory of every new folder to run the script. This is required for my script to work. Does anyone know what I could accomplish this?

    #!/bin/bash

for f in *; do echo "file '$f'" >> files.txt; done
for f in *; do echo "'$f'" >> filesdelete.txt; done
ffmpeg -f concat -safe 0 -i files.txt -c copy "${PWD##*/}".MP4
xargs -I{} rm -r "{}" < filesdelete.txt
rm files.txt
rm filesdelete.txt

I start with the file structure below. The script runs in each subdirectory (dir1, dir2, dir3) and combines the files in each subdirectory into one video. For the script to run, it needs to cd into each directory.

root
├── dir1
│   ├── video1.mp4
│   ├── video2.mp4
│   └── video3.mp4
├── dir2
│   ├── video1.mp4
│   └── video2.mp4
└── dir3
    ├── video1.mp4
    └── video2.mp4

The end result should look like the structure below.

root
├── dir1
│   └── concat.mp4
├── dir2
│   └── concat.mp4
└── dir3
    └── concat.mp4

Answer 1

Use Find.

#!/bin/sh

cat << EOF >> edpop
1d
wq
EOF

find . -name "*.mp4" > stack

next () {
[[ -s stack ]] && main
rm -v ./stack
rm -v ./edpop
exit 0
}

main () {
file=$(echo "1p" | ed -s stack)
ffmpeg -f concat -safe 0 -i "${file}"
rm -v "${file}"
ed stack < edpop
next
}

next

You will need to partly edit this yourself. It isn't intended to be bulletproof, because I don't have ffmpeg, find isn't currently working on my system, and I also don't completely understand your ffmpeg command line. However, the basic point is that using Find will give you the full path of your files, and not just the basename, which you can operate on without having to change directories inside a running script.