Shell Programming and Scripting

Hi,

I have a file where I want to replace the 15th field separated by comma, only on specific lines matching lots of different conditions.

I have managed to read the file line by line, within the loop my line is held in a variable called $line

I assume this will be using sed (maybe incorrectly), but how do I replace the 15th field in my comma separated, condition matching line which is now contained within the $line variable and write it back into the file (not a new file), replacing the original line with the new one ?

Let's say in this example I want to replace 15th field seperated by comma value "APPLE" with "BANANA"

Many thanks in advance for any answers.

---------- Post updated at 06:16 PM ---------- Previous update was at 05:41 PM ----------

Ok, I'm nearly there.

I now have my amended line in $NEWLINE and my current un-amended line is in $line

So I have now tried writing it back to the same file using:

Code:

sed -i '/'$line'/'$NEWLINE'/g'

But I get the following error:

Code:

sed: -e expression #1, char 8: unknown command: `0'

Can anybody help with this please ?

Cheers

Usually awk would be be more suited for this:

Code:

awk '$15==s{$15=v}1' s="APPLE" v="BANANA" FS=, OFS=, file

But that would not do in-place editing, however you can easily write to a new file and if all is well replace the original with the new file.
If that is not what you are after, please specify input, and desire output as well as OS and version..

Many thanks Scrutinizer.

I did end up using your suggestion.

Although it placed the value I wanted to replace the 15th field with into the 14th field, leaving the old value in the 15th field also.

No probs though, changed it to 15 and all is good in the world.

You're welcome... Tes no idea why I put 14 there. I edited it in my post

Both your sed command (once it is corrected) and Scrutinizer's command will replace your "$oldvalue" with "$newvalue" in the ENTIRE file, not only in the lines "matching lots of different conditions". It'd be better to reconsider, and post the conditions you impose, so they can be put into the awk (or whatever) program...

Thanks RudiC,

I've already amended the files.

However, if true, I struck it lucky, as when I was finally given the files to amend only the lines I wanted to change had a 15th field



Within my read line loop I actually did:

Code:

echo $line | awk '$15==s{$15=v}1' s="1" v="0" FS=, OFS=, >> $a.tmp

Would this have meant that I would have avoided the pitfall you mention as I have piped the $line specifically through my awk command ?

---------- Post updated at 02:19 PM ---------- Previous update was at 02:06 PM ----------

Ok, I've tested it (I should have just done that instead of asking so sorry bout that).

I added 20 more fields to a line I didn't want to change.

Code:

echo $line | awk '$15==s{$15=v}1' s="1" v="0" FS=, OFS=, >> $a.tmp

only changes the $line and leaves the value in the 15th field on a line I do not want to change intact.


Happy days.