Getting various parts from the log

Shell Programming and Scripting
1

I am dealing with some app log, see example below:

22:16:13.601 [192.168.1.183] ClientSession(905)--Connection(5)--SELECT GETDATE()
22:16:13.632 [192.168.1.183] ClientSession(158)--Connection(5)--SELECT 1
22:16:13.632 [192.168.1.175] ClientSession(848)--Connection(6735)--SELECT 1

So far I needed to collect certain column from it, such as get all the sql commands:

$ cut -d'-' -f5 xx
SELECT GETDATE()
SELECT 1
SELECT 1

or get all the session ids

$ cut -d'(' -f2 xx|cut -d')' -f1
905
158
848

So, cut helps me with this sort of things. But in some cases I need to get several columns at once, such as get the timestamp and the sql command
such as

22:16:13.601 SELECT GETDATE()
22:16:13.632 SELECT 1
22:16:13.632 SELECT 1

For this I am using awk, but it seems quite cumbersome and I understand that if I have known enough regexp then sed would be a quick and easy solution here, but I don't.

My attempt to use sed for getting time and sql command

sed 's/\([0-9][0-9]:[0-9][0-9]:[0-9][0-9].[0-9][0-9][0-9]\)\(.*--\)\(*\)/\1 \3/' log.txt

outputs whole thing, as if I did cat it

Please help me with sed solution.

2

an awk (or not):

awk '
/SELECT/ {sub(".*SELECT", "SELECT"); print; }
' infile

or sed:

sed -n '/SELECT/s/.*SELECT/SELECT/p;' infile
3

A modification to yours

sed -ne 's/\([0-9][0-9]:[0-9][0-9]:[0-9][0-9].[0-9][0-9][0-9]\).*\(SELECT.*$\)/\1 \2/p' log.txt

Some others in Perl

perl -ne '/^(.+?)\s.+(SELECT.+$)/ and print "$1 $2\n"' log.txt

perl -ne '/^((?:\d{2}[:.]){3}\d{3}).*(SELECT.*$)/ and print "$1 $2\n"' log.txt
4

Try also

awk -F- '{split ($1, T, " "); print T[1], $NF}' file
22:16:13.601 SELECT GETDATE()
22:16:13.632 SELECT 1
22:16:13.632 SELECT 1