rename multiple files with random extension

ltodd2 · 03-18-2009, 07:17 AM

Afternoon

Im having a bit of a do trying to rename some files. We get 3 types of edi files in from the tax office but there system is adding a random 17 digit number to the end of the file which I need to remove. the files look like:
COMPANY-P6.mf.29550362467065025
COMPANY-P9.mf.29550362468597026
COMPANY-SL1.mf.29550362467066028

We need it to end in the .mf and remove the numbers.
I've had a go at it and gave up. I've been sent a command using sed and it does what i want but its only in the view on screen and not actually changing the files.
ls | sed 's/$.mf*$.*/.mf/'

any clues please

Lee

Linux.tar.gz · 03-18-2009, 07:36 AM

I use krename for such purpose.
http://www.krename.net/

MensaWater · 03-18-2009, 07:44 AM

You could do this with your sed as well but I prefer awk for quick and dirty:

Code:

for file in `ls COMPANY*`
do filenew=`echo $file |awk -F. '{print $1"."$2}'`
   mv $file $filenew
done

This is a loop that will make each file a variable called $file and then create a new variable $filenew based on the name of the file name. It then renames the original ($file) to the new name ($filenew).
The "awk -F." tells awk to use the dot as a delimiter. It then prints the 1st field (item before the first dot in file name), a literal dot (the awk strips out the dot so you have to add it back - notice the quotes around the dot) and the 2nd field which is the item after the first dot. Since we stop there it doesn't add the final dot or the random number.

openSauce · 03-18-2009, 07:58 AM

If it's always 17 digits plus a '.' you could also just do

Code:

mv $file ${file:0:$((${#file} - 18))}

That's untested but it looks about right. God, bash is ugly!

GazL · 03-18-2009, 08:19 AM

Quote:

Originally Posted by openSauce

If it's always 17 digits plus a '.' you could also just do

Code:

mv $file ${file:0:$((${#file} - 18))}

That's untested but it looks about right. God, bash is ugly!

That bash certainly is.

There's a number of ways to do it in bash, which one is best I guess depends how flexible your filename matching is going to be.

here's 2 more (there are most likely some others too)

Code:

mv "$file" "${file/.mf.*/.mf}"

mv "$file" "${file%%.mf.*}.mf"

openSauce · 03-18-2009, 08:35 AM

Haha, good point well-made. Yours also has the advantage of not processing files that have already been once through the grinder.

GazL · 03-18-2009, 08:48 AM

Just as an afterthought. There was someone doing something along these lines in a post a while back, It was a massive bulk rename and it had been running for days, literally! and was still only a fraction of the way through.

In the end someone suggested he use perl instead which turned out to be much better suited to doing this sort of thing. For a few hundred or so files, I guess bash will suffice, but if you're going into thousands then you probably should be looking at something like perl.

ltodd2 · 03-18-2009, 09:56 AM

Quote:

Originally Posted by jlightner

[CODE

for file in `ls COMPANY*`
do filenew=`echo $file |awk -F. '{print $1"."$2}'`
mv $file $filenew
done[/CODE]

This is a loop that will make each file a variable called $file and then create a new variable $filenew based on the name of the file name. It then renames the original ($file) to the new name ($filenew).
The "awk -F." tells awk to use the dot as a delimiter. It then prints the 1st field (item before the first dot in file name), a literal dot (the awk strips out the dot so you have to add it back - notice the quotes around the dot) and the 2nd field which is the item after the first dot. Since we stop there it doesn't add the final dot or the random number.

This one work great. thanks for the help. Thanks eveyone else for yours

ghostdog74 · 03-20-2009, 08:46 PM

Quote:

Originally Posted by jlightner

You could do this with your sed as well but I prefer awk for quick and dirty:

Code:

for file in `ls COMPANY*`
do filenew=`echo $file |awk -F. '{print $1"."$2}'`
   mv $file $filenew
done

be careful of using "ls" in for loop as it might break with spaces in filenames. Also, using "ls" like that is called "useless use of ls". No need to use "ls". Just use shell expansion.

Code:

for file in "COMPANY*"
do 
 mv "$file" "${file%.*}"
done

MensaWater · 03-21-2009, 07:18 PM

ls COMPANY* isn't quite as "useless" as posting after the problem is resolved. I don't mind you making a point but you might be more careful of the way you phrase it.

ghostdog74 · 03-22-2009, 04:28 AM

Quote:

Originally Posted by jlightner

ls COMPANY* isn't quite as "useless"

yes it is quite useless. How about this

Quote:

more careful of the way you phrase it.

I am not the inventor of the phrase. Many people have already seen and used it. this phrase comes out with google search too. Therefore i don't see anything wrong with the way i "phrase" it.

ltodd2 · 03-25-2009, 04:37 AM

Thanks all for the advise. Im not sure if I should add this next bit here or post new.
All the about works great but now the powers that be have said that the files may not be processed on a regular basis which would mean the next time the job runs and there are new files that have come in it will over write the ones it did last because the file names are now all the same. Is it possible that if it sees a file by the same name it can add a 1 to the name and then 2 etc. Once the files have been processed and moved the next can go in without a number since it will be the only file again.
im thinking this is too much but then my knowledge of scripting is very little.

thanks

Lee

chrism01 · 03-25-2009, 06:56 AM

Time to start reading: http://tldp.org/LDP/Bash-Beginners-Guide/html/ ...
The problem is people know/believe you can do this stuff now, so you might as well get a head start before they start asking for complex stuff