Convert Order for find … -exec Command

findrename

I want to rename files and directories. Let's suppose that I want to convert every ä to a, and I have Järjestelmä/Säätimet. So I say something like find -exec rename 's/ä/a/g'. Now find will convert Järjestelmä to Jarjestelma and then complain that it did not found Järjestelmä anymore.

If I could convert the order of exec-commands to run it would solve this. Is it possible? If not, what is the right way to do this?

This is not a big problem, as I can run the command again until it makes no changes. Anyways this irritates me.

Best Answer

Use find's -depth option, from the man page:

-depth Process each directory's contents before the  directory  itself.
          The -delete action also implies -depth.

That way it will process Säätimet before Järjestelmä and will not complain about not being able to descend into Järjestelmä because you just renamed it.

To prevent rename from handling the whole path use -execdir present in a few find implementations like BSDs and GNU (which changes to the directory and hands just the final part to the command argument {} (with a ./ prefix with some implementations)):

mkdir Järjestelmä
touch Järjestelmä/Säätimet
find . -depth -execdir rename 's/ä/a/g' {} \;
find .

gives:

./Jarjestelma
./Jarjestelma/Saatimet

Related Solutions

Recursively rename files and directories

With find:

find . -type f -exec sh -c 'SHELL COMMAND' {} \;

This invokes SHELL COMMAND on each found file in turn; the file name is "$0". Thus:

find . -type f -exec sh -c '
    mv "$0" "${0%/*}/$(printf "%s\n" "${0##*/}" | sha1sum | cut -d" " -f1)"
' {} \;

(Note the use of printf rather than echo, in case you have a file called -e or -n or a few other problematic cases that echo mangles.)

You can make this a little faster by invoking the shell in batches.

find . -type f -exec sh -c 'for x; do
      mv "$x" "${x%/*}/$(printf "%s\n" "${x##*/}" | sha1sum | cut -d" " -f1)";
    done' _ {} +

In zsh, there's an easy way to match all the files in the current directory and its subdirectories recursively. The . glob qualifier restricts the matches to regular files, and D includes dot files.

for x in **/*(.D); do mv …; done

In bash ≥4, you can run shopt -s globstar and use **/* to match all files in the current directory and its subdirectories recursively. You'll need to filter regular files in the loop.

shopt -s globstar; GLOBIGNORE=".:.."
for x in **/*; do if [[ -f $x ]]; then mv …; done

Bash: Use exec find option by sorting two file arguments

The replacement of {} is done by find and happens after the evaluation of the process substitution (<(…)) and such in your code.

Using bash -c:

find datadir -type f -exec \
  bash -c 'sdiff -s <( sort "$1" ) <( sort "/$1" )' bash {} \;

Or to avoid running one bash instance per file:

find datadir -type f -exec bash -c '
  for file do
    sdiff -s <( sort "$file" ) <( sort "/$file" )
  done' bash {} +

Best Answer

Related Solutions

Recursively rename files and directories

Bash: Use exec find option by sorting two file arguments

Related Question