Hi all ,
I have a requirement where I have a file. Contents of the file are :
#comments
VAR="abg"
RES=123
#comments
IC6790ABG="https://www.abc.com"
IC5678-vg="https://www.bhy.com"
IC-gy_567:78="https://www.gyt.com"
#comments
The variable names can not have characters like - , : so
in this file I have to find words starting with IC and replace characters like - ,:
I want to change only the variable name , not the whole line.
I have used SED command

sed -i '/^IC/s/[^0-9 a-z A-Z _]*//g' file

when I am using this command , it is replacing the whole line
output becomes :

#comments
VAR="abg"
RES=123
#comments
IC6790ABGhttpswwwabccom
IC5678vghttpswwwbhycom
ICgy56778httpswwwgytcom
#comments


But I want the output like this :

#comments
VAR="abg"
RES=123
#comments
IC6790ABG="https://www.abc.com"
IC5678vg="https://www.bhy.com"
ICgy_56778="https://www.gyt.com"
#comments

How can I get the desired output , thanks for your help in advance .

Similar solution here:
http://www.linuxquestions.org/questi...7/#post5366219
You just need to capture the match with () and replace it.

That might be a bit difficult in sed - gsub in awk would be cleaner.

Where /tmp/new is your original file gives:Either use redirect to store the output to a file or use the '-i' switch to sed to change the file in place.

What about the semi-colons ?.
How do you know what order they will be in ?.

Seems comma is also a possibility - same question about order for the testing.

syg00 I've given you the answer, you need to read the sed manual and do a bit of research extra characters to be deleted are easy to add if you do a bit of the work yourself, I'm not doing it all for you, this is moderately easy stuff, if you are going to learn Linux sooner or later you will have to learn regex/sed/awk, last hint use '[]' character sets.

Thanks Keith , I will try your solution and let you know if it works.

I feel your a little heavy handed on the comments there Keith. Your solution provided did not mention that it was a part solution to be added to and syg00 is not the OP and as a veteran user is
quite familiar with sed. I would reiterate what has been said that your solution will not work so well for the full colon:
This output shows no changes whatsoever to the example data

Oh dear sorry wasn't I touchy feally enough for you didums!

Hi ,

I used sed 's/^IC\(.*\)-\(.*\)=/IC\1\2=/g;s/^IC\(.*\):\(.*\)=/IC\1\2=/g' /tmp/new
as suggested by Keith and it worked , thank you so much