Deleting files according to date

Here's the scenario:
I have a directoriy (~/myfiles) and I'd like to create a bash script that looks at the files in that directory and finds the date the files were last accessed and deletes them if the last accessed date is more than 30 days ago and then goes back and moves some files to another directory if the filename has a number before the file extension (ie, car2.png or scrnsht8.png).

I will be creating a cronjob to run this bash script once it is completed. FYI, this is part of a larger project and this bit is the last thing I need to learn.

Any advice?

[robatino]

For the first part, you can use the find command like this:

find ~/myfiles -atime +30 -exec rm {} \;

where {} denotes each of the files in turn that satisfies find's criterion. See the man page for details.

[robatino]

Actually, you might want to replace ~/myfiles with something like ~/myfiles/* in the above command, since otherwise one of the files that find considers would be ~/myfiles itself, which rm wouldn't remove since it's a directory, but if it tried you'd get an error message. To experiment with find, you can replace
-exec rm {} \;
with
-print
which will just list all the files that satisfy the criterion you give it.

Thank you very much, robatino, that takes care of the first part of the command. I appreciate it
I will experiment with -print.

[BostonWatcher]

As for the second part, you can continue with "find". You just have to use a pattern search. With your example, you could use a simple wildcard match:

find ~/myfiles -name "*[0-9].png" -exec mv \{\} destination_dir \;

or, in case you want to be more restrictive/specific, you can use a regexp pattern like

find ~/myfiles -regex ".*[^0-9]+[0-9]+.png"

Originally Posted by BostonWatcher

As for the second part, you can continue with "find". You just have to use a pattern search. With your example, you could use a simple wildcard match:

find ~/myfiles -name "*[0-9].png" -exec mv \{\} destination_dir \;

or, in case you want to be more restrictive/specific, you can use a regexp pattern like

find ~/myfiles -regex ".*[^0-9]+[0-9]+.png"

Thank you! I was reading man find and I like that app, it's very nice
Thank you both for the help.

so, does this look correct?

#!/bin/bash

FILES="/home/user/myfiles"

for file in $FILES
do

  find $FILES -atime +30 -exec rm {} \;

done  

for file in $FILES
do

  find $FILES -name "*[0-9].png" -exec mv \{\} /home/user/Documents \;

done  

exit 0

[robatino]

To avoid processing $FILES itself in the first find command, you can use
find $FILES -mindepth 1 -atime +30 -exec rm {} \;
and I don't think you need to backquote { and } in the second command. On the bash command line, find needs the ; to be backquoted, but not {} (try it).

ok, fixed it, works fine
Thank you all for the help.