Shell Programming and Scripting

what do you mean by no difference. there must be some "important" pattern in "important text" portion that you would like to get.

Code:

perl -0777 -pe 's/.*\n\n\n//ms' oldfile >newfile

The difference is the omission of the question mark after the *. This causes the match to be as long as possible instead of as short as possible. I put the "as short as possible" question mark operator in originally after some hesitation, but decided to use it as it should make the code somewhat more efficient if there is only one occurrence of three newlines.

@ghostdog74: The only pattern I cant use are these three newlines.

@era: thx a lot.

The output you want is in the file named xx01

Code:

awk 'BEGIN{FS="\n";RS=""}NR==4{print $1}' file | xargs -i csplit file /{}/

I numbered your "important text" lines to verify the accuracy of the output.

Code:

> echo `tac infile | tr "\n" "~"` | sed "s/~~~/|/g" | cut -d"|" -f1 | tr "~" "\n" | tac
important text 1 
important text 2 
important text 3 
important text 4

What's the echo and the backticks for?

Other than that, a good illustration of a best-effort workaround for avoiding Perl at any cost ...