Finding and replacing a line in a file

[marf88]

I have done some searching, seeing what commands I exactly will need.
I/O redirection
grep I think
echo
and a while loop?

What I want to do is lets say I have a file with 1000 lines of configuration settings. I want to search the file for a specific line ( I know what the line is ) and once I find this line, I want to replace the line with a new line of code that I give it.

How can this be achieved with linux bash/shell scripting? Do I need something more powerful like perl/python for this task? I would like to avoid those dependencies at all costs, even if it makes it inefficient.

Thanks,

Michael

[mdnayeem]

The simplest solution is to use 'sed' like this:
sed -e 's/{line that you want to replace}/{line with which you want to replace}/g'

of course, if your line has special character's then you need to use proper sed syntax which you can easily find by google it or for eg. from here: Sed - An Introduction and Tutorial

There are of course many other ways to do it like using awk, etc.

[marf88]

If this 'sed' command doesn't require dependencies (works on most common linux distros) then I'll use it.

My original plan was to use the 'cat' command in a while loop and the 'while read line' and then check Each line that came in, and if the line matched then I would change it. I'll try the sed when I get home!

Thanks.

[mdnayeem]

It shouldn't. I would be surprised if any distros comes without sed.

[scm]

They may not all support the -e (edit in-situ) flag, though, in which case you'd have to use a temporary intermediate file.

[Franklin52]

They may not all support the -e (edit in-situ) flag, though, in which case you'd have to use a temporary intermediate file

I suppose you mean the -i flag?

Regards

[mdnayeem]

I suppose you mean the -i flag?

Regards

No! scm actually meant -e flag. Using this flag one can chain many sed commands in a single line.

[scm]

No, I did mean the -i flag that the posting didn't actually contain - the example given requires the use of an intermediate file. I should really read posts properly before I rush in.

[mdnayeem]

No, I did mean the -i flag that the posting didn't actually contain ?

Aah, right! forgot about this. Anyway, -i option works for me in Red hat/Fedora, ubuntu and Scietific L.

[marf88]

Hrm...

Lets say my file has a line that can look like this:

My Configuration = [ int, int, int, int ]

OR

My Configuration = [ int, int]

Basically there can be any number of integers within the two brackets, but at least one occurrance.

would my sed statement look like:

sed 's/"My Configuration = "\[{int}{",int"}*\]' <mysettings.txt >mysettings.txt

so I'm looking for the My Config. with an [ bracket, then Definitely 1 occurrence of int, with zero or more occurance of ,int after.

Does that make sense?