about getline in awk

[Jdamian]

Hi

I'm trying to insert the string hour:minute:second at the begin of each line of a command output:

function ins
{
awk '{ "date +%H:%M:%S" | getline D; print D" "$0 }'

return 0
}

integer I=0
while (( I<5 ))
do
print $I
sleep 1

I=I+1
done | ins

but the output is:

10:25:30 0
10:25:30 1
10:25:30 2
10:25:30 3
10:25:30 4

It looks like the awk line "date +%H:%M%S | getline D" only is executed one time.
I'm a mess.

what is wrong ?

[Massimo Bianchi]

Hi,
i think the line is correct, it is you that execute it only once !

Try this

integer I=0
while (( I<5 ))
do
D=$(ins)
print $I
sleep 1

I=I+1
done

So it should be execute each time you run the script, and not only once.


HTH,
Massimo

[Massimo Bianchi]

Hi,
i think the line is correct, it is you that execute it only once !

Try this

integer I=0
while (( I<5 ))
do
D=$(ins)
print $D $I
sleep 1

I=I+1
done

So it should be execute each time you run the script, and not only once.


HTH,
Massimo

[Jdamian]

I don't understand your answers.

According to awk man pages, the awk action

{ "date +%H:%M:%S" | getline D; print D" "$0 }

is executed on every input line not on every awk execution, isn't it ?

[Jdamian]

on other words... If a input file is 20 lines long, how many times is executed the command:

date +%H:%M:%S
?

[Massimo Bianchi]

Hi,
yes, you are right.

Sorry I made a mistake, It seemed to me your line was

" done < ins "


Massimo

[Dietmar Konermann · Accepted solution]

Try this:

function ins
{
awk '{ "date +%H:%M:%S" | getline D; print D" "$0; close("date +%H:%M:%S") }'

return 0
}

[Robin Wakefield]

Hi,

You need to close the pipe, since awk cannot open more than one at a time:

function ins
{
awk '{ "date +%H:%M:%S" | getline D;close "date +%H:%M:%S"; print D" "$0 }'

return 0
}

integer I=0
while (( I<5 ))
do
print $I
sleep 1

I=I+1
done | ins


rgds, Robin

[Massimo Bianchi]

HI,
I thank all the others for the answers, today i learnt something new about awk !

Massimo

[Jdamian]

Dietmar, your suggestion produces no change in the output.

I also ran the same script on a Solaris 8 system (/usr/xpg4/bin/awk is used instead of SunOs awk). The output is:

10:21:54 0
1
2
3
4
5
6
7
8
9

I suspect 'date' is executed just one time.

[Jean-Louis Phelix]

Hi,

I had almost the same answer as 2 previous, and it works for me ...

Regards.

hp:/tmp> cat a
function ins
{
awk 'BEGIN {A= "date +%H:%M:%S"} {A | getline D ; print D" "$0 ; close(A) }'

return 0
}

integer I=0
while (( I<5 ))
do
print $I
sleep 1

I=I+1
done | ins
hp:/tmp> sh a
11:37:19 0
11:37:20 1
11:37:21 2
11:37:22 3
11:37:23 4
hp:/tmp>

[Robin Wakefield]

Hi,

My version works with nawk, you'll need brackets on the close statement with "standard" awk.

I tried Dietmar's on my Solaris 8 system with /usr/xpg4/bin/awk, and it worked fine.


$ uname -a
SunOS schi0039pmh 5.8 Generic_108528-14 sun4u sparc SUNW,Ultra-80
$ cat script
function ins
{
/usr/xpg4/bin/awk '{ "date +%H:%M:%S" | getline D;print D" "$0 ;close("date +%H:%M:%S")}'

return 0
}

integer I=0
while (( I<5 ))
do
print $I
sleep 1

I=I+1
done | ins

$ script
04:46:50 0
04:46:51 1
04:46:52 2
04:46:53 3
04:46:54 4

rgds, Robin

[Jdamian]

Sorry, I mistyped the argument of 'close' sentence in my script. Then the pipe wasn't closed.

Now it works fine.

Thanx everybody.

[Dietmar Konermann]

You need to take care that the close() gets exactly the pipe's command string as argument! So Jean-Louis' approach using a variable seems to be the more elegant one.

Best regards...
Dietmar.