scripting problem rm filelist

milke · 07-04-2006, 10:51 AM

Hallo I have a simple scripting problem,
which I cant solve.
I have a list of files stored in textfile like this:

/home/file1
/home/file2
/home/file3

And I need a Shellskript to remove the files
which are mentioned in the textfile.
I know it sound's simple, but I still don't get it

Thank you for all kind of answeres in advanced.

unSpawn · 07-04-2006, 11:04 AM

Why not post what you've got already?

milke · 07-04-2006, 11:21 AM

If it help's?

I trying to write a script, that delete files which are
several times on my PC.

#I build the md5 first

find . -type f -exec md5sum {} \; > ./sum.unsorted

#The I sorte after the checksum

sort < ./sum.unsorted > ./sum.sorted

#Then savethe filename of every filen which appears
# more then ones in a textfile

uniq -w 32 -D < ./sum.sorted > doppelte-Dateien.log

And form there on I see my diggest problem in the
removing of the files from the list.

unSpawn · 07-04-2006, 11:37 AM

If it help's?
Sure does.

Code:

cat doppelte-Dateien.log|while read f; do
if [ -e "$f" ]; then rm -f "$f"; fi; done

or

Code:

cat doppelte-Dateien.log|xargs -iF rm -f 'F'

BTW's: watch out for relative paths is you use current working directory "find . -type". You can pipe sort into uniq. Also try to use "mktemp" for tempfiles: works "better" when you script more.

jlinkels · 07-04-2006, 09:26 PM

One BIG FAT WARNING if you do rm in scripts:

The bash shell scripting language is such that if you make a little mistake with one small character, the command can be entirely different.

Especially when a command like rm is involved.

It happened to me that I wiped out the entire directory I was working in, including my 250 line script I was working on at that moment. I think I forgot a quote or so

It is better to put this somewhere at the start:

Code:

RM=/usr/bin/echo
#RM=/usr/bin/rm

..your code here..

#finally THE command:
$RM whatever.files

If you are 100% sure that the code does what your expect, uncomment the /usr/bin/rm, and comment the /usr/bin/echo

jlinkels

jschiwal · 07-04-2006, 09:56 PM

I like jlinkles idea. I have been preceding the rm command with echo to see what the entire command would look like.
But then, I usually do this interactively instead of from a script.

I would add to his/her sig: If your question is about awk, read "Gawk: Effective Awk Programming" from the gawk-doc package.

If the filenames contain whitespace, you need to pipe the filelist through "tr".
cat doppelte-Dateien.log | tr '\n' '\000' | xargs -0 rm -f

If there are thousands of files in the list, then you also want to use one of the input limiting options to xargs.

jlinkels · 07-05-2006, 08:48 AM

Quote:

Originally Posted by jschiwal

I would add to his/her sig: If your question is about awk, read "Gawk: Effective Awk Programming" from the gawk-doc package.

That's funny... the contents is almost the same as the GAWK user manual, but not quite. Unusual in the community to have two versions of a document, not verbatim equal.

jschiwal · 07-06-2006, 04:32 AM

You found two versions of the Gawk manual. I was referring to "Gawk: Effective AWK programming" which is a different book.

jlinkels · 07-06-2006, 02:44 PM

I assume you are right, but Google confused me.
On the very top of the page the first line reads:
Gawk: Effective AWK Programming

I'll look it up in the doc package as you say.

jlinkels

jschiwal · 07-07-2006, 06:41 AM

One closer inspection, one of them may be a version of the gawk book I was referring to, but looks a lot different from the version that I printed. A version of a document produced with "make pdf" from the source may contain much more information from the info version.