Very simple shell script

beley · 11-01-2006, 06:08 PM

I have a script that downloads image zips and extracts them. The images are in the format:

01234567_00.jpg
01234567_01.jpg
01234567_02.jpg
01234568_00.jpg

You get the idea. I need to move all the images to a directory by the last number before the underscore.

For instance,

01234567_00.jpg moves to the ln7 directory

01234563_00.jpg moves to the ln3 directory

Here is what I have so far to take the images and move them one at a time to their appropriate directories:

Code:

find *.jpg | while read FILE;
do
	dir=${FILE:7:1}
	# echo $FILE - $dir
	mv -f $FILE work/ln$dir/
	# echo Moved $FILE to ./work/ln$dir/$FILE
done

I get "Argument list too long" when I try to run the command. There are several thousand images in the directory.

Can I use xargs in a shell script like that?

I'm new to shell scripting - any help would be greatly appreciated.

keefaz · 11-01-2006, 06:31 PM

Maybe protect your variables with ""'s ?

Code:

mv -f "$FILE" "work/ln$dir"

beley · 11-01-2006, 06:33 PM

It's not the mv command that's getting the error on too many arguments... it's the find command.

keefaz · 11-01-2006, 06:38 PM

Weird, the find statement works well here
Try to give the search path to find, eg if
it is in current dir :

Code:

find . -name "*.jpg" | while read FILE;

[edit]
I guess find *.jpg works because it is the bash shell
that extrapolates the *.jpg, not find

konsolebox · 11-01-2006, 07:37 PM

please try this one

Code:

find -type f | grep jpg$ | \
while read FILE; do
	# DIR=In$(( ${${FILE/_*}/*\/} % 10 ))
	DIR=${FILE/_*}
	DIR=${DIR/*\/}
	DIR=$(( DIR % 10 ))
	DIR=In${DIR}

	mkdir -p "${DIR}"

	#mv -f "${FILE}" "${DIR}"
	mv "${FILE}" "${DIR}"
done

OR

Code:

IFS=$'\n'
for FILE in `find -type f | grep jpg$`; do
	# DIR=In$(( ${${FILE/_*}/*\/} % 10 ))
	DIR=${FILE/_*}
	DIR=${DIR/*\/}
	DIR=$(( DIR % 10 ))
	DIR=In${DIR}

	mkdir -p "${DIR}"

	#mv -f "${FILE}" "${DIR}"
	mv "${FILE}" "${DIR}"
done
IFS=$' \t\n'

beley · 11-01-2006, 07:48 PM

Not sure about those, but I've been researching and finally found something that worked... I created a simple function:

Code:

function large_mv ()
{       while read line1; do
			if echo "$line1" | grep -q ".jpg"
			then
				{
				dir=${line1:7:1}
      				mv -f $line1 work/ln$dir/
				echo Moved $line1 to work/ln$dir/$line1
				}
			else
				{
				 echo $line1 is not an image - not moved.
				}
			fi
       done
}

ls | large_mv

It seems to be working well so far.

jschiwal · 11-01-2006, 08:18 PM

Here is a different approach taking advantage of the patterns of the filenames:

Code:

for dir in $(find ./ -maxdepth 1 -type f -iname "[[:digit:]]*_[[:digit:]][[:digit:]].jpg" -printf '%f\n' | \ 
  sed 's/\([[:digit:]]*\)_.*/\1/' | sort | uniq); do 
     if [ ! -d "$dir" ]; then mkdir "$dir"; fi
     mv "${dir}"_* "$dir"
done

The find command finds your picture files matching your example pattern. The sed command removes the starting "./" characters and the _##.jpg pattern from the filenames. This is then sorted and a uniq list of directories produced. The mv command uses a wildcard so you are using a mv command only once per directory.

jlinkels · 11-02-2006, 06:24 AM

The error message you get is from find. If you do:
find *.jpg
it tells you argument list too long, because it expands *.jpg and passes all those file names to the find command.

You have to use
find -name "*.jpg"
or
find -name \*.jpg

jlinkels