[evilmrhenry]

First, the background.
I'm running on a slackware based system, KDE 3.1.

Whenever I want to delete something, I have a choice of sending it to the trash, or deleting it immediately. I don't like the trash, as I have to keep in mind to empty it regularly. Therefore, I have a *bad* habit of simply deleting the item.

The following code (hopefully) will go through the trash and delete anything that was placed in there over 180 minutes (3 hours) ago. I plan to place it in the cron-hourly directory.

#! /bin/sh
cd $HOME/Desktop/Trash/
rm | find -amin +180 -not -name .directory -not -name .
exit 0

My questions:
Will this work?
Can bad things happen if I put this in?
Did I forget something?

Normally, I would just put this in, and try it out. However, the presence of rm makes me really want to get advice first.

[Squashman]

Well you could always use the rm -i option. Which is interactive and will prompt you if you wan to delete anything. That way you could test it out without destroying anything.

[evilmrhenry]

Thanks for the tip. This version appears to work better:

#! /bin/sh
cd $HOME/Desktop/Trash/
find -amin +120 -not -name .directory -not -name . -exec rm -i {} \;
exit 0

Are there any filenames that could trigger bad reactions? Any other issues?
(I'll remove the -i when I put it into action.)

[codejockey]

If I understand what you're trying to do, you may be able to simplify your script somewhat by using: find -amin +120 -type f -exec rm (etc.). This specifies that only files (not directories) should be removed. If you intend to remove (sub-) directories in Trash, then you should use rm -r (possibly rm -rf) instead of just rm in your -exec command (e.g., find -amin +120 -exec rm -rf {} \;). Note that this will not remove the current or the parent directory (. or ..).

You might also consider having the script support two modes (readonly and execute); this would allow you to test/debug the script without risking unintentional removal of files. You could, for example, use something like:

#! /bin/sh
USAGE="`basename $0` [-r]"
READONLY=ON

while [ ! -z "$1" ]
do
    case "$1" in
        "-r")  READONLY=
               shift
               ;;
           *)  echo "$USAGE"
               exit 1
               ;;
    esac
done

if [ -z "$READONLY" ]
then
    COMMAND="find -amin +120 -type f -exec rm {} \;"
else
    COMMAND="find -amin +120 -type f"
fi

cd $HOME/Desktop/Trash
eval "$COMMAND"

This script defaults to "readonly" mode, and produces a list of files that would be removed (but doesn't remove them). Adding the -r option on the command line forces the script to perform the actual removal.

One additional point: if you intend to use the script in a cron job, you should make provisions for handling any output from the script (error messages, etc.). If you do not redirect output, it will be mailed to you (which may be what you want if you are running the script readonly -- this would allow you to watch what the script would do with real data over time, without the risk of deleting an important file).

Hope this helps.

[evilmrhenry]

I'm beginning to have second thoughts about the script as it works now. I'm wondering how this one stands up instead:

#! /bin/sh
cd $HOME/Desktop/Trash/
rm -ri .countdown/*
find -not -name .directory -not -name . -not -name .countdown -maxdepth 1 -exec mv -i {} .countdown \;
exit 0

(No, I didn't get the two lines switched. The second command moves the files, an hour passes, then the first line removes them.)

Would it be better to move all files to a specific directory, then delete them, or would it be better to go by the access date?


Some points:
It needs to act on both files and dirs.
However, only the files/dirs in the root directory of the Trash need to be looked at.


I'll think about adding a read-only/working switch later. Right now, it's staying in confirmation mode.

I need a bit more help on the output files. I tried

#! /bin/sh
cd $HOME/Desktop/Trash/
rm -ri .countdown/* > t_log.txt
find -not -name .directory -not -name . -not -name .countdown -maxdepth 1 -exec mv -i {} .countdown \; >> t_log.txt
exit 0

but that just produces a blank file. Is this working the way it should? Is there a way to record the files being moved or deleted?

Thanks for all your help.

[codejockey]

You can use your find command to create a list of the files and directories that will be removed. If you capture the output in a temporary file, you can remove each entry individually and email the file to yourself as a record of what happened. Would something like this do what you intend?

#! /bin/sh
FINDLOG=/tmp/findlog.$$
cd $HOME/Desktop/Trash
find -not -name .directory -not -name . -not -name .countdown -maxdepth 1 > $FINDLOG

for i in `cat $FINDLOG`
do
    rm -rf $i
done

mail -s "list of files removed on `date`" user-name < $FINDLOG
rm -f $FINDLOG

I'd recommend against storing the files to be removed in another directory, and depending on a later execution of the script to remove them.

Hope this helps.

[evilmrhenry]

Would something like this do what you intend?

Just about.

I changed a few things, to make it fit in better. Basically I changed the notification from a mail to just a file placed in the Trash, (I don't use the mail feature.) and used the correct find command. (You were using the "directory" method, not the "date" method.)

#! /bin/sh
FINDLOG=$HOME/Desktop/Trash/.findlog
cd $HOME/Desktop/Trash/
find -not -name .directory -not -name .findlog -not -name . -maxdepth 1 -amin +180 > $FINDLOG

for i in `cat $FINDLOG`
do
    rm -rf $i
done

exit 0

Unless there are any bugs, I'll go with this version. Thanks for your help, everyone!