topic Re: awk formatting output in Operating System - HP-UX

awk formatting output

alec pringle — Fri, 11 Mar 2005 08:26:31 GMT

Hi,

can anyone help me format the following file:

time: 20050310 field1: a field2: b b field3: c
time: 20050311 field2: a b
time: 20050311 field3: a b field2: bb

..into:

time: 20050310 field2: b b field3: c
time: 20050311 field2: a b
time: 20050311 field2: bb field3: a b

.. because the order and number of fields is not fixed I am struggling to get awk to pull the right data out - any ideas?

Re: awk formatting output

Keith Bryson — Fri, 11 Mar 2005 08:51:31 GMT

Alec

It's not clear what the 'rules' are for re-structuring your output. Is each line structured/cut differently?

Keith

Re: awk formatting output

alec pringle — Fri, 11 Mar 2005 09:00:31 GMT

In the output I want time, field2 and field3, but only if they exist. I don't want field1 in the output.

Re: awk formatting output

RAC_1 — Fri, 11 Mar 2005 09:01:53 GMT

cat "file"|sort -k3|sort -k4|sort -k5

Anil

Re: awk formatting output

RAC_1 — Fri, 11 Mar 2005 09:08:00 GMT

cat "file"|sed 's/field1//g'|sort -k3|sort -k4|sort -k5

Re: awk formatting output

alec pringle — Fri, 11 Mar 2005 09:28:30 GMT

Thanks Anil,

I think the sed will only remove the field name (and not its data) and the sort will then not work because the number of fields will be different. Also, I cannot guarantee the position of field2 and field3 (see the final line of the example), so this means a sort won't work.

Alec

Re: awk formatting output

RAC_1 — Fri, 11 Mar 2005 09:40:38 GMT

cat "file"|sed 's/field1: .//g'|sort -k3|sort -k4|sort -k5

Note - field1: . in sed command.

Re: awk formatting output

H.Merijn Brand (procura — Fri, 11 Mar 2005 09:46:37 GMT

unlimited number of fields:

lt09:/tmp 107 > cat xx
time: 20050310 field1: a field2: b b field3: c
time: 20050311 field2: a b
time: 20050311 field3: a b field2: bb
lt09:/tmp 108 > perl -lpe'@f=split/(\s+field\d+:)/;$i=shift@f;%f=@f;$_=join"",$i,map{"$_$f{$_}"}sort keys%f' xx
time: 20050310 field1: a field2: b b field3: c
time: 20050311 field2: a b
time: 20050311 field2: bb field3: a b
lt09:/tmp 109 >

Enjoy, Have FUN! H.Merijn

Re: awk formatting output

Rodney Hills — Fri, 11 Mar 2005 11:47:38 GMT

Procura,

He doesn't want "field1" included with the output.

So how about modifying your routine as follows-

perl -lpe'@f=split/\s+(field\d+:)/;$i=shift@f;%f=@f;delete $f{"field1:"};$_=join" ",$i,map{"$_$f{$_}"}sort keys%f' xx

-- Rod Hills

Re: awk formatting output

H.Merijn Brand (procura — Fri, 11 Mar 2005 13:08:35 GMT

Ge didn't state that explicitely, so I assumed he just err'd on the output.

Your modification also changes the output, since I included the whitespace leading towards the fields in the separator, and you took it out. That indeed makes it easier to delete the key field1:, but it also glues the fields together in a different way. You solved that by adding the space in the join, but it still is different from my original plan.

So /my/ modification to filer field1:'s would be

# perl -lpe'@f=split/(\s+field\d+:)/;$i=shift@f;%f=@f;$_=join"",$i,map{"$_$f{$_}"}grep!/\bfield1:/,sort keys%f' xx

There is still a small weakness in there if there are more than 9 keys (1,10,11,12,2,3,4,5,6,7,8,9), which could be schwarzian'd

# perl -lpe'@f=split/(\s+field\d+:)/;$i=shift@f;%f=@f;$_=join"",$i,map{"$_$f{$_}"}grep!/\bfield1:/,map{$_->[0]}sort{$a->[1]<=>$v->[1]}map{/(\d+)/;[$_;$1]}keys%f' xx

But now we're getting towards the point where you'd normaly turn from a one-liner to a script, and document what you're doing

Enjoy, Have FUN! H.Merijn

Re: awk formatting output

alec pringle — Mon, 14 Mar 2005 04:44:39 GMT

thanks for all your help!