Shell Programming and Scripting

Hi, i'm currently new to scripting and need some help with my problem, so i'll jump right to it.

I have a file containing text, the file is pretty big so for the sake of this i'll just say this is the text:

Code:

John id number is abc34938
Grahams id number is pending
id number abc64334 is Bob's 
abc32432 is the id number for Jim 
Mike's ID number is pending
Michael id number is abc4352

What i would like to do is search the lines for the keyword "abc", and the lines which contain the keyword, I would like to delete the other words in the line. Thus leaving an output as such:

Code:

abc34938
Grahams id number is pending
abc64334
abc32432
Mike's ID number is pending
abc4352

I've searched far and wide for a solution and came stuck, so im really hoping one of you guys could help me out.

Thank you for your time

Code:

perl -nle'
  /abc/ and print /abc\w+/g 
    or print  
  ' infile

Thanks for the reply mate, as i'm new i'm going to risk angering the gods by asking if that would work even though im doing bash scripting and not perl? (I would try it now but don't have the facilities to do so just now)

I suppose if you say yes, then I should ask if I need to include anything in my bash script to enable it? As currently I have #!bin/bash at the top, would I need to do anything else to this?

Hi

Code:

sed  '/abc/s/.*\(abc[0-9][0-9]*\).*/\1/' file

Guru.

Thank you for the reply Guru, just a quick addition to the question if I may, what if there were two words I wanted to keep on a line? would I just use the '|' symbol to specify 'or'?

You just need to have the Perl interpreter in your PATH (ant it's already there on most systems).
If you have more that one abcxxx on the same line, you'll have all of them printed as one word,
if you them separated (by a single space, for instance), you should change the code like this:

Code:

perl -nle'
  /abc/ and print join " ", /abc\w+/g 
    or print
  ' infile

Code:

awk '{for (i=1;i<=NF;i++) {if ($i~/^abc/) $0=$i}}1' urfile