Bash Scripting help, altering data in text file.

trey85stang · 05-31-2006, 02:14 AM

Latly Ive been coming across the need to use data from one text file to update a second text file. I've been doing all this manually and I know there is an easier way to do it. So I thought I would ask and see if anyone might want to help me with some scripting ideas to make my life a little easier.

First of all I only know a small handful of scripting techniques... and not nearly enough to acomplish what I need to do. However here is a quick example of what I need to script.

I have two files that are similar...

File A has line entries that look like so (it can have between 200-20,000 lines)...

Code:

#file /directory/app/data1/sub/file.name
#file /directory/app/data1/sub/file.name17
#file /directory/app/data1/sub/file.name23

File B has entries like so

Code:

INSTANCE file /directory/app/data1/sub/file.name createdate
INSTANCE file /directory/app/data1/sub/file.name16 createdate
INSTANCE file /directory/app/data1/sub/file.name23 createdate

Now, what I need to do is if the dirctory/file listed in File A, matches the directory/file listed in File B... I need the corrosponding line in File B to be deleted.

Is there anyway to automate this? I can easily use sed and remove the "#file" from the entries in File A if that would make it easier? I know there is a way to do this but I am no good with scripting so I do not know where to begin to get started testing a script to do this?

Any help would be appreciated.

thanks,
Trey

spirit receiver · 05-31-2006, 03:30 AM

The following should do the trick unless the file names in your first file contain whitespace or colons:

Code:

#! /bin/bash

FILEA="first filename"
FILEB="second filename"

for i in $( sed -e "s/^#file \([^ ]*\)/\1/" -e "s/\./\\\./g" "$FILEA" )
do
	sed -ine "\:^INSTANCE file $i :d" "$FILEB"
done

BTW, you mentioned that you did that manually until now. You really did, examining files with up to 20000 lines?

trey85stang · 05-31-2006, 04:15 AM

Quote:

Originally Posted by spirit receiver

The following should do the trick unless the file names in your first file contain whitespace or colons:

Code:

#! /bin/bash

FILEA="first filename"
FILEB="second filename"

for i in $( sed -e "s/^#file \([^ ]*\)/\1/" -e "s/\./\\\./g" "$FILEA" )
do
	sed -ine "\:^INSTANCE file $i :d" "$FILEB"
done

BTW, you mentioned that you did that manually until now. You really did, examining files with up to 20000 lines?

It is very rare that I have to do it... I had been putting everything into a spread sheet and comparing them. This pops up ever 9 months or so that I have to make these type of changes, but yes I examined them manually.

thanks for the script I will play around with it and see if I can get this to work! BTW, do you have know of any tutorials that deal with awk and sed to create these types of scripts on my own? I understand what awk and sed do.. but I have never been able to grasp the "s/^#file \([^ ]*\)/\1/" -e "s/\./\\\./g" parts of both

exman · 05-31-2006, 04:31 AM

Hi,
a good introduction to sed:
google for sed tutorial and take the 3rd link
Would give you the link but i got too few posts

exman

spirit receiver · 05-31-2006, 05:18 AM

As for regular expressions, there are good tutorials for Perl, like this one for example. These are not exactly the regular expressions used in the shell (although grep can be told to use Perl regular expressions), but once you got accustomed to these, the differences can easily be figured out from the man pages.