path problem

ovince · 03-10-2007, 03:39 AM

hi All,

I would like to copy "less*.dat" files in each "DH*" directory into "file_less*.dat" like:

for dir in DH* ; do
for file in `find $dir -type f -name "less*.dat" -print`; do
of=file_"$(basename $file)"; cp $file $of;
done
done

but this copy everything in root-directory of DH* directories. How to copy files into correcponding directories?

thanks
oliver

omnio · 03-10-2007, 04:39 AM

Code:

for dir in DH* ; do
    for file in `find $dir -type f -name "less*.dat" -print`; do
        of=file_"$(basename $file)"; cp $file $dir/$of; 
    done
done

Note: in the cp command you should add $dir to the destination.
--

ovince · 03-10-2007, 05:39 AM

thank you Omnio.

matthewg42 · 03-10-2007, 07:22 AM

Instead of using this:

Code:

for file in `find ...`; do
  ...
done

Consider this:

Code:

find ... |while read file; do
  ...
done

The difference is subtle, but important. If the number of files found is large, the first will fail due to a size limitation on the execution substitution. Also, files with spaces in the name will not be handled correctly in the first case, but can be in the seconds case (assuming $file is properly quoted when it is used).

cfaj · 03-10-2007, 10:49 AM

Quote:

Originally Posted by matthewg42

Instead of using this:

Code:

for file in `find ...`; do
  ...
done

Consider this:

Code:

find ... |while read file; do
  ...
done

The difference is subtle, but important. If the number of files found is large, the first will fail due to a size limitation on the execution substitution. Also, files with spaces in the name will not be handled correctly in the first case, but can be in the seconds case (assuming $file is properly quoted when it is used).

The first is unlikely to fail due to the number of files found. If it were an external command, an "Argument list too long" error is possible, but because 'for' is a shell builtin, it is limited only by available memory. Spaces and other pathological characters will still be a problem, so it is never the best way to write such a loop.

The second will not work if there are leading or trailing spaces, or backslashes, in any of the filenames. To deal with those, it should be:

Code:

find ... | while IFS= read -r file

This will still fail if any names contain newlines.