[SOLVED] Handling space in parameters

rac8006 · 11-11-2020, 06:55 PM

I've searched every where I can think of and can't find an answer. I'm trying to find all of the directories in a partition and count the files. The problem that I'm having is that some of the directory names have a space in them. I did the following:
find . -maxdepth 1 -type d -print > dirs
then I wanted to do for i in `cat dirs`
do
echo -n $i
find "$i" -print|wc
done

I tried several different things but none of the worked.

What I finally got to work was
find . -maxdepth 1 -type d -exec sh myscript.sh {} \;

myscript.sh
echo -n "$1"
find "$@" -print|wc

Is there a better way to do this?

Thanks

sgosnell · 11-11-2020, 09:29 PM

Have you tried just

Code:

find . -type f | wc -l

replacing the . with the path you want to search?

Or do you need the number in each directory? The above only gives a grand total. If you want the number of files and directories, use

Code:

tree -a

which is probably faster and easier. It gives a graphic tree structure if you need it, with a total number of directories and files at the end. If you really need the number of files in each individual directory, I've never tried that.

rac8006 · 11-11-2020, 11:46 PM

I want the number of files in the top level directories. If I had a partition with a b c d directories. The output would look like this
a 3
b 10
c 20
d 15

I could do this with a command like
for i in `find . -maxdepth 1 -type d -print`
echo -n "$i"
find "$i" -print|wc
done

The problem comes when the directories on the partition are like this "My backup" "My Videos" "My Documents" test ffp
when running the above script I would get errors of
My no such file My
backup no such file
My no such file
and so on. because it treats My Backup as two parameters
I don't think tree -a does this. Secondly I have almost 500,000 files on the partition.
The question was how to pass the directory that has spaces in the name properly?

pan64 · 11-12-2020, 07:31 AM

first of all read man find, you will see -print0 for example.
also you might want to use a while cycle instead of for.

finally it is already answered for example here: https://stackoverflow.com/questions/...espace-in-bash

rac8006 · 11-12-2020, 10:34 AM

With your information I have found three methods that do what I want.
find -maxdepth 1 -type d -print0|while read -d $'\0' file; do echo -n "$file";find "$file" -print|wc; done
find . -maxdepth 1 -type d $ ! -name . $ -print0|xargs -0 -n1 sh -c 'echo -n $1;find "$1" -print|wc' sh
find . -maxdepth 1 -type d $ ! -name . $ -exec /Volume_2/myscript.sh {} \;

myscript.sh
echo -n "$1"
find "$@" -print|wc

MadeInGermany · 11-12-2020, 01:53 PM

Code:

for i in `find . -maxdepth 1 -mindepth 1  -type d -print`

is problematic because the list is split on IFS that defaults to space,tab,newline
You can improve it by setting IFS to newline only

Code:

oIFS=$IFS
IFS="
"
for i in `find . -maxdepth 1 -mindepth 1  -type d -print`

I would rather go for read that by default reads one line i.e. at newline boundary. Still you should set IFS to preserve a leading space in a filename.

Code:

find . -maxdepth 1 -mindepth 1  -type d -print |
while IFS= read -r dir
do
  find "$dir" -print | wc
done

This will only fail if a filename contains a newline (quite unusual but allowed).
The split on globs can even handle newline characters.

Code:

for dir in */  .[!.]*/
do
  [ -d "$dir" ] &&
  find "$dir" -print | wc
done

The trailing / in a glob already ensures directories, so the following test is only for the no-match case.
In bash you can alternatively do

Code:

shopt -s nullglob
find */ .[!.]*/ -print | wc

Because no test is needed and because find handles multiple start directories there is no need for a loop

Last but not least, bash has another option

Code:

shopt -s nullglob dotglob
find */ -print | wc

Now the * glob includes a .[!.]*