Editing Ascii File Using Shell Script

rohit2983 · 12-28-2006, 01:25 AM

Hi Gurus

I have two files(ASCII),One file(which i want to edit) will have some string like "change this"(will be within quotes) and another file will have the the value which will replace "change this" in the first file(it will be in format like "change this","replacement"),the string "replacement" will be added to the first file in place of "change this".

Any idea how can this be acheived using shell scripts???
even if anyone knows how can i edit the first file/add a string to it without using the second file can also post a reply

Thanks in advance for your help

Regards

jim mcnamara · 12-28-2006, 02:36 AM

Is there some reason you cannot use sed?
Is "change this" going to be the same string every time? ...if so

Code:

replacement=`cat file1`
sed #s#\"change this\"#\"$replacement\"#g# file2 > newfile
mv newfile file2

What are you trying to do exactly?

rohit2983 · 12-28-2006, 02:53 AM

Hi Jim

thanks for the reply
"change this" wont be same everytime

this is what i want to do

file 1,the one which has to be edited will have text as follows

DEFINE F50T
BEGIN
TP = 5
TI = 35
TN = 745
TV = <<"Admin Catalog">>
END

DEFINE F50T
BEGIN
TP = 5
TI = 36
TN = 745
TV = <<"Active Description">>
END

DEFINE F50T
BEGIN
TP = 5
TI = 37
TN = 745
TV = <<"Comments">>
END

in this file text like "Comments"(everything in <<"">> has to be replaced by corespoinding text in file 2)

FILE 2 where i have the replacement will have text like this

Admin Catalog,Admin Catalog1
Active Description,Active Description1
Comments,Comments1
--<<"Admin Catalog">> to be replaced by <<"Admin Catalog1">> in file1 and so on

I have heard that sed and awk can be used to edit files but dunno how can i use them in my script,any method which can do this in a script is fine for me

jim mcnamara · 12-28-2006, 10:42 AM

try something like this:

Code:

#!/bin/ksh

while IFS=, && read old new
do    
	sed -i 's/'$old'/'$new'/g' file1
done < file2

makyo · 12-28-2006, 11:22 AM

Hi.

Jim's elegant solution made me think of how I could do the same thing, except reduce the passes over the data-file to just one (like Patton, I hate paying for the same real estate twice

).

Here's my version of Jim's script:

Code:

#!/bin/sh

# @(#) s2       Demonstrate building command file for sed.

rm -f commands
oldifs="$IFS"
IFS=,

while read old new
do
        echo "s/$old/$new/" >> commands
done <file2
IFS="$oldifs"

sed -f commands file1

Which produces:

Code:

% ./s2
DEFINE F50T
BEGIN
TP = 5
TI = 35
TN = 745
TV = <<"Admin Catalog1">>
END

DEFINE F50T
BEGIN
TP = 5
TI = 36
TN = 745
TV = <<"Active Description1">>
END

DEFINE F50T
BEGIN
TP = 5
TI = 37
TN = 745
TV = <<"Comments1">>
END

So in place of n passes over the data file, this collects the created sed commands, and then runs sed over the file once.

The saving and restoring of IFS is probably not necessary, but sometimes I cannot stop my fingers

The credit goes to Jim for the original solution ... cheers, makyo

rohit2983 · 01-02-2007, 05:24 AM

Hi All

thanks a lot for your help

Rohit

rohit2983 · 01-03-2007, 04:21 AM

Hi Makyo

I am facing one small issue while using your script.
The output generated shows the required file output,but when i open the file1 in vi editor,i see that it remains unchanged

any idea what can be the reason for this???

also what exactly is purpose of

"# @(#) s2 Demonstrate building command file for sed."
and whats the 's2'(% ./s2) used in the output file you have displayed for your script

Thanks
Rohit

makyo · 01-03-2007, 06:27 AM

Hi, Rohit.

Quote:

Originally Posted by rohit2983

Hi Makyo

I am facing one small issue while using your script.
The output generated shows the required file output,but when i open the file1 in vi editor,i see that it remains unchanged

any idea what can be the reason for this???

also what exactly is purpose of

"# @(#) s2 Demonstrate building command file for sed."
and whats the 's2'(% ./s2) used in the output file you have displayed for your script

Thanks
Rohit

The script, like sed itself, writes to STDOUT, it does not modify files in place. There are situations when that is useful, but my practice is to avoid over-writing original files while I am experimenting. You can save the output to another file, view it, and then save over the original if you desire.

The line "# @(#) s2 ..." is a what line. It's obviously a comment, but was originally used by sccs module what to help identify the brief purpose of programs. (The sccs system is a very old source control system, succeeded by RCS, cvs, subversion, etc.) I use those comments to generate indices of scripts -- I have several hundred scripts, shell, perl, awk, etc. -- too many to keep track of without such an aid.

The "% ./s2" shows you how the script, "s2" was called into execution from the current directory, "./s2", and the "%" is the prompt.

Best wishes ... cheers, makyo