How to recursively merge directories or folders into one, retaining folder names in file names

stozi · 05-21-2019, 03:11 PM

Basically, I have a directory tree. I want to find and replace all "/" within that tree with "-".

can i use mv or find?

I want a directory tree to become one directory with lot of files with long names which include the names of the sub-directories which they used to be in.

for example these paths

/a/file.ex
/a/b/file.ex
/a/b/file2.ex
/a/c/file.ex
/a/c/d/file.ex

will become:

/a/file.ex
/a/b-file.ex
/a/b-file2.ex
/a/c-file.ex
/a/c-d-file.ex

scasey · 05-21-2019, 03:46 PM

Quote:

Originally Posted by stozi

Basically, I have a directory tree. I want to find and replace all "/" within that tree with "-".

can i use mv or find?

I want a directory tree to become one directory with lot of files with long names which include the names of the sub-directories which they used to be in.

for example these paths

/a/file.ex
/a/b/file.ex
/a/b/file2.ex
/a/c/file.ex
/a/c/d/file.ex

will become:

/a/file.ex
/a/b-file.ex
/a/b-file2.ex
/a/c-file.ex
/a/c-d-file.ex

Yes, you should be able to use either mv or find
I played with it a little...something like

ls -1 the files in /a
use sed to change the /'s to -'s sed 's./.-.g
mv the original file to the name output by sed

Code:

file.ex
file.ex
b/file.ex
b-file.ex
b/file2.ex
b-file2.ex
c/file.ex
c-file.ex
c/d/file.ex
c-d-file.ex

I suspect there's an awk solution as well.

stozi · 05-21-2019, 05:25 PM

Thanks, I'm hoping for a single command to deal with the whole directory tree if possible

scasey · 05-21-2019, 06:30 PM

Quote:

Originally Posted by stozi

Thanks, I'm hoping for a single command to deal with the whole directory tree if possible

There’s probably not a single command, but you can build a short script:

For each file in the ls, do the grep, do the mv

Give it a shot...use echo instead of mv until you get it right. Ask if you run into trouble.

BW-userx · 05-21-2019, 07:38 PM

Code:

#!/bin/bash

working_dir=/media/ntfs1/Music2
move2=~/goo

while read f
do
#replace directory dividers / with -
newName=${f////-}
#remove leading - on filename
newName=${newName:1}

echo "$newName"
mkdir -p "$move2"
#echos results 
#remove the echo and first set of quotes to
#put it into production. 
echo "mv -fv "$f" "$move2"/"$newName" "

done <<<"$(find "$working_dir" -type f)"

if you got more than one dir to move to a central dir

Code:

#!/bin/bash

d1= 
d2=
d3=
d4=
move2=

while read f
do

#code goes here

done <<<"$(find "$d1" "$d2" "$d3" "$d4" -type f)"

Beryllos · 05-21-2019, 09:26 PM

On the off chance that there is an existing file of the same name, use mv -i or mv -n, and do not use mv -f, so as to avoid overwriting.

MadeInGermany · 05-22-2019, 01:16 AM

The overcomplicated

Code:

while read ...
do
  ...
done <<<"$(find ...)"

runs the find first and must store the output somewhere (in memory or in a tmp file).
It should be simplified by

Code:

while read ...
do
  ...
done < <(find ...)

where the two processes run simultaneously.
This while loop still runs in the foreground,
unlike the classic

Code:

find ... |
while read ...
do
  ...
done

where the pipe forces the loop to run in the background, and any variables in the loop are lost after the loop.