UNIX for Dummies Questions & Answers

Hello all,

I have a file with following sample data

2009-08-26 05:32:01.65 spid5 Process ID 86:214 owns resources that are blocking processes on Scheduler 0.
2009-08-26 05:32:01.65 spid5 Process ID 86:214 owns resources that are blocking processes on Scheduler 0.
2009-08-26 05:32:01.65 spid5 Process ID 84:127 owns resources that are blocking processes on Scheduler 2.
2009-08-26 09:49:53.46 spid5 Process ID 86:214 owns resources that are blocking processes on Scheduler 0.
2009-08-26 15:51:28.51 spid5 Process ID 102:17 owns resources that are blocking processes on Scheduler 12.
2009-08-26 15:51:28.51 spid5 Process ID 102:17 owns resources that are blocking processes on Scheduler 12.
2009-08-26 15:51:28.51 spid5 Process ID 102:17 owns resources that are blocking processes on Scheduler 12.


I need to extract the unique Process IDs

So the output should be { though the order does not matter }

86
84
102

Its the unique values occuring between "Process ID" and the ":" on each line excluding the blank white space after ID . Process ID 102:

Any suggestions ?

One of the possibilities:

Code:

sed -e's/^.*Process ID \([0-9]*\):.*$/\1/' samplefile |sort -nu

or, using cut

Code:

cut -c41- samplefile |cut -d: -f1|sort -nu

Another possible way,

Code:

perl -ne  'print "$1\n" if /Process ID (\d+):\d+ / && !$_[$1]++' file

Thank you all Scrutiniser and Rubin for your wonderful responses. It is incredible what can be done with shell scripting though it is very crytic.

Just a quick one.

How to get the above ouput as a CSV comma seperated list ?

So it would be : 84,86,102

Given your input, one way would be:

Code:

$ perl -ne  'print $s=$s?",".$1:$1 if /Process ID (\d+):\d+ / && !$_[$1]++; print "\n" if eof' file
86,84,102

Or you can append

Code:

|xargs|tr " " ","

to the commands above or, using sed:

Code:

|xargs|sed -e s'/ /,/g'