how can I make a shell script, which takes every file(I mean it has to work within every folder regardless the amount of files in folder or file names) from specified folder, then some other command runs and does something with file, and finally the original file is saved back to that specified folder with the same name. Or with different name- this doesn't matter so much. After that, the second file goes through the same process and then the third etc. How to accomplish that?

btw regards from Australia!

find /specified/dir * -exec commandToRunOnEachFileFound {} \;

{} represents the pathname of the current file found so you can reuse it to do whatever to the current file found. If what you want to do it complex, then perhaps you pass it as an argument to the script you write and call with -exec.

Example:
'man find' for more info.

Moin,

No, this doesn't work correctly, because the unescaped asterisk is substituted with all pathnames in the current directory by the shell before find runs. BTW: To get all pathnames in a directory recursively you don't have to add a pattern, find /specified/dir -exec commandToRunOnEachFileFound {} \; will work.

There are some ways, for example:
find searches for all regular files in /specified/dir recursively, these file names are read by the loop, mktemp creates a temporary file with a unique name, then do_something is called for each file, the output is written to the temp file, on success the temp file is moved to the original file.

In this case do_something gets a bunch of pathnames as command line parameters and can process them in a loop. This way does not create a new process for each file.

Jan

Yes, piping the output of find to something like a while loop makes it a lot easier to work with, especially if you need sed replacement, etc. Also, you might consider using -depth with find so that contents are processed before their directories, and also '!' -name . to prevent the top directory (as opposed to its contents) from being processed.
ta0kira

Moin,

Theses options make sense only if you want to process directories too. The -type f option I used limits the output to regular files. The original question was to process all files and store the results in the same file names, so no dependencies to other files exist.

Jan

thank you for help! This while loop:
works nicely, but if I would like to rename original files to *mp3_old, then I should probably use something like that:

However, how can I integrate this for loop into while loop?