Hi,

I was wondering if someone could help get me started with sed.

I have a text document I'd like to edit. It basically looks like this:

row1: 04
row2: 93
row3: 40

I would like to change these numbers but not sure how to implement the sed command to do so.

Any help is appreciated

OJ

How do you want to change them? Can you give a before and after example?

I just want to change those values. So for example:

Before edit

row1: 20
row2: 39
row3: 34


After edit

row1: 20
row2: 85
row3: 34

Basically I just want to change the numbers ( or are they charactesr in this case?) at the end of a line.

sed -i 's/39$/85/' filename

the '-i' parameter tells sed to edit the file in place. The 's' is the search/replace command, the $ means "match end of line", so '39$' is a match for 39 followed by a newline character. The second pair in // (/85/ in this case) is what to replace it with.

Cool, that seems easy enough. I have a couple of questions though.

What would I do if I only wanted to change the value of at the end row2: no matter what value it has? (ie. I don't need to find a specific number, i just need to update a value at the end of a row. Is there a way I can just go to that location and ovewrite whatever is there? Maybe using grep and gawk?

And with the command you just gave me, will I run into problems if there are multiple entries of the same number? Say if I wanted to change row2 from a 39 to 84 but row1 has a 39 as well.

Do these questions make sense?

1. Do you mean you want to make updates based on row number and not the contents of the row? That's a little uglier but certainly possible, using head, tail, and grep maybe.

2. The command I listed will change all occurances that match the pattern specified. If you only want to change certain lines you need a way to specify only those lines. Is there anything else in the line that uniquely identifies the ones you want to change?

Post a more complete description/example of what you're trying to do.

1. Yes I want to make editions based on a row number and not the content.

2. Yes, I only want to change a specific line. The only thing specific to the line is the label "row#:"

Basically I have a c program which will calculate a number. This number is to be writen to the text file. And there is a specific spot for this result. This document is then copied to other computers. So the other computers know the result of a certain operation. To extract a certain number I use the grep and gawk commands. I think this is the command:
grep row2: filename.txt | gawk '{print$2}'

This gets the line for row2: and gets whatever is in the second column.

Now I just need a way to write something to a specific spot in the text document.

I figure using Linux commands will be much easier than what would have to be done in the C program.

Hope this helps clarify everything

OJ