Hi Script Gurus,

I want to grep certain word from file and put it to some other file with proceeding 2 lines and next two lines of greped word line.

Here is the example of init.ora file below and I want to grep lines with which have word "ORA-" and I also want +2 and -2 line with the greped line.
---------- clip init.ora ----------

Tue Nov 30 00:25:35 2004
Thread 1 advanced to log sequence 59709
Tue Nov 30 00:25:35 2004
Current log# 9 seq# 59709 mem# 0: /d/db01/oradata/oss/redo09.dbf
Tue Nov 30 00:25:35 2004
ARC4: Beginning to archive log# 8 seq# 59708
Tue Nov 30 00:25:47 2004
ORA-000060: Deadlock detected. More info in file /d/db01/app/oracle/admin/oss/udump/ora_15029_oss.trc.
Tue Nov 30 00:26:39 2004ARC4: Completed archiving log# 8 seq# [/B] 59708Tue Nov 30 00:26:57 2004
Thread 1 advanced to log sequence 59710
Current log# 6 seq# 59710 mem# 0: /d/db01/oradata/oss/redo06.dbf
Tue Nov 30 00:26:57 2004
ARC1: Beginning to archive log# 9 seq# 59709
Tue Nov 30 00:28:04 2004
--------- clip end --------

Any help on this?

Thanks in advance.

graval.

I'd do the following: (yes, it is ugly)

awk "\$0 ~ /ORA-/ {
cmd=\"awk 'NR>=\" NR-2 \" && NR<=\" NR+2 \"' init.ora\"
system(cmd)
}" init.ora


Be very careful about the quoting.

grep -A2 -B2 "ORA-" init.ora

-A --- After
-B --- Before

HTH
bye

Unfortunately the -A and -B options are not available on every grep implementation.

Another trick to get it done is

ex -R init.ora << +++
g/ORA-/.-2,.+3p
q
+++

Thank you guys. it work.

I am able to run awk and ex but i dont know sed one is not going thru.
ex is faster than awk so finally i decided to use ex.

Thank you all for your time and efforts.

Regards,

graval.

Hi ,

I tried above unix commands but none of them worked..
Can some help me ??

-Smita

Could you describe your problem and your idea of "none of them worked" to us?

Hi,

I want to perform the same action as the thread is about.
I have file and I want to get +2 and -2 lines from the line that has a specific word.

For example: I have log file in xml format and compressed with .gz and want to get the +2 and -2 lines in the file whereever it encounters "550" .

let me know if its not explaintory, I will try to expalin more..

Also I tried to run the commands below on .txt file but it didnt work.

ex -R init.ora << +++
g/ORA-/.-2,.+3p
q
+++


sed -e '
1{$!N;$d;}
$!N;/ORA/!D
$!N;$d;N;p
g;$!N;$d;N;D
' init.ora


awk "\$0 ~ /ORA-/ {
cmd=\"awk 'NR>=\" NR-2 \" && NR<=\" NR+2 \"' init.ora\"
system(cmd)
}" init.ora


Thanks,
Smita

How about using "grep" with the -B or -C (?) options? The GNU grep utility allows you to specify exactly how many lines before or after a matching row should be written to the output as well.

i am getting error while trying grep -B or -c

$ grep -C 2 "ORA-" init.ora
grep: illegal option -- C
Usage: grep -hblcnsviw pattern file . . .
$ grep -B 2 "ORA-" init.ora
grep: illegal option -- B
Usage: grep -hblcnsviw pattern file . . .

-S

I had a look at the man page of GNU grep again:
-A num - number of rows after the matching one
-B num - number of rows before the matching one
-C num - context

If this doesn't work for you, you probably don't use GNU grep.

how can I find if i am using GNU grep ?? Is there any way I can use GNU grep?

If I cant use grep to server my purpose, any other way of serving my request?

thanks a lot for ur help,
-s

You can install GNU grep, of course. Typically, the first thing you do with a non-GNU system is to install a whole bunch of GNU tools (like tar, grep, less, ...)

The alternative is that you write a small script in Perl, AWK or whatever programming language you prefer. Personally, I would prefer to take the right tool for the job if it is available and I wouldn't go through the hassle to write such a script myself.

unfortunately, I dont have authority to install anything on the server as its managed by other gp and they are not read yto do it..
do you any script that can serve the prupose?

thanks