topic Re: Scripting Help in Operating System - Linux

Scripting Help

George Chechakunnil — Thu, 14 Jun 2007 23:09:35 GMT

Hello Gurus,

I need to grep for a specific pattern in all the files in a directory and change them to a new pattern.

I tried with sed.

for i in
do
sed 's/old/new $i > /newdirectory/$i
done

it works but the problem is that the files are being truncated when piping to the new file

is there a way i can change the pattern in the same file itself?

thanks in advance

Re: Scripting Help

Dennis Handly — Fri, 15 Jun 2007 02:17:44 GMT

Your sed solution should work. You forgot a single quote:
sed 's/old/new/' $i > /newdirectory/$i

You aren't piping but redirecting. What was the truncation like?? Number of lines or width of them? Long long are they??

Changing a pattern the file can be done with perl, search some previous thread.

Some of them show using ex(1) and ed(1) to do this.

Re: Scripting Help

Yogeeraj_1 — Fri, 15 Jun 2007 02:32:15 GMT

hi,

i have some scripts which does it as follows:

e.g.

for file in `find . -name "*.htm"`
do
echo "Converting $file ..."
cat $file | sed s/$1/$2/g> /tmp/newfile
cp /tmp/newfile $file
done

hope this helps too!

kind regards
yogeeraj

Re: Scripting Help

john korterman — Fri, 15 Jun 2007 02:33:56 GMT

Hi,

you should perhaps add "g" to your sed statement, e.g.:

sed 's/old/new/g' $i > /tmp/$i

else you will only have the first occurrance replaced.

regards,
John K.

Re: Scripting Help

Yogeeraj_1 — Fri, 15 Jun 2007 02:37:24 GMT

hi again,

another way to do it would be using PERL:

perl -pi -e 's/old/new/g' $file

you can replace the sed command in the previous script with this...

hope this helps too!

kind regards
yogeeraj

Re: Scripting Help

Dennis Handly — Thu, 29 Sep 2011 10:39:26 GMT

A perl solution was just added to:
http://h30499.www3.hp.com/t5/System-Administration/Pattern-replacing/m-p/5053121

Re: Scripting Help

rmueller58 — Sun, 17 Jun 2007 21:23:18 GMT

I use perl unless I need to retain a copy of the original file.

if you intend to change occurances in multiple files within a directory I'd create a for loop, otherwise simply use a perl command.

MULTIFILES

for fn in `ls`
do
perl -pi -e s/old/new/g $fn
done

Single files
perl -pi -e s/old/new/g filename

Re: Scripting Help

Arturo Galbiati — Mon, 18 Jun 2007 03:07:47 GMT

Hi,
only a sugegstion:

for i in $(grep old *)
do
sed 's/old/new' $i > /newdirectory/$i
done

This will increase teh speed in cas eyou have a lot of files in your dir, and yoy could also refine you grep to get only ascii file, etc.

HTH,
Art