multiple pattern search in a single file line by line

saheervc · 09-01-2010, 11:31 PM

Hi Guys, Im a new to Linux.
Can some one help me to sort out the following problem.

I have a very long output file ( some 104735 lines ) generated by a program. I have to check this output file for errors. The file looks like the following .

...............................................................
I=1
...
...
NO CONVERGENCE
....

..
I=2
...
..
CONVERGENCE REACHED
...
...
..
I=3
...
..
NO CONVERGENCE
...
..
MAXIMUM NUMBER OF ITERATIONS REACHED
...
..
etc upto I=48
..................................................................
It is very difficult to go through each line for errors.

I would like to write a script file that prints only errors under iteration number in the following way.
.........................................................
I=1
NO CONVERGENCE
I=3
NO CONVERGENCE
MAXIMUM NUMBER OF ITERATIONS REACHED
...
.........................................................

Im using Mandriva 2010
Thanks in Advance.

ghostdog74 · 09-01-2010, 11:44 PM

Code:

awk '/^I=/{i=$0}/^NO CONVERGENCE|MAX.*ITERATIONS.*REACHED/{print i,$0}' file

Tinkster · 09-01-2010, 11:45 PM

Hi, welcome to LQ!

Given the data above something like

Code:

egrep "^I=|NO CONVERGENCE|MAXIMUM NUMBER OF ITERATIONS REACHED" file

should do. You'd be getting the I=2, too, because there's no
multi-line match with a back-reference here.

If the file structure were clearer we could probably use an
awk script to eliminate the the "successful" iterations number
from the output.

Cheers,
Tink