I tried to connect an awk job to a bash script using two fifos. The idea
was that the script should send requests to awk over one pipe and get
results back over the other one.

I tested that with a simple awk script providing access to an associative
array:

# t.awk
{
switch ($1)
{
case "set": arr[$2] = $3
case "get": print arr[$2]
case "quit": exit
}
}

The bash script was
#t

mkfifo in
mkfifo out
awk -f t.awk <in >out &
echo "set one 111" >in
echo "set two 222" >in

echo "get one" >in
read x <out
echo "one=$x"

echo "get two" >in
read x <out
echo "two=$x"

echo "set one XXX" >in # line 22
echo "get one" >in
read x <out
echo "one=$x"

echo "quit" >in

The output I expected was

one=111
two=222
one=XXX

but instead this happened:

one=111
two=
./t: line 22: in: Interrupted system call

It seeems that awk exits right after the first print. Can anybody explain
that? And is there a way to keep awk reading?

If awk is replaced by a bash script which mimics the array functionality,
everything works fine.

Regards,
Bernd

--
Bernd Eggink
monoped@sudrala.de
http://autojar.sourceforge.de
http://jboom.sourceforge.de

Bernd Eggink wrote:
> I tried to connect an awk job to a bash script using two fifos. The idea
> was that the script should send requests to awk over one pipe and get
> results back over the other one.
>
> I tested that with a simple awk script providing access to an
> associative array:
>
> # t.awk
> {
> switch ($1)
> {
> case "set": arr[$2] = $3
> case "get": print arr[$2]
> case "quit": exit
> }
> }

What awk version is it that supports switch/case statements?

Janis


> [snip]

In article <f7g1k1$88q$1@online.de>,
Janis Papanagnou <Janis_Papanagnou@hotmail.com> wrote:
>What awk version is it that supports switch/case statements?

gawk has supported this for some time, if its build is configured with
--enable-switch.

John
--
John DuBois spcecdt@armory.com KC6QKZ/AE http://www.armory.com/~spcecdt/

On Mon, 16 Jul 2007 16:51:16 +0200, Bernd Eggink
<monoped@sudrala.de> wrote:
>
>
> I tried to connect an awk job to a bash script using two fifos. The idea
> was that the script should send requests to awk over one pipe and get
> results back over the other one.
>
> I tested that with a simple awk script providing access to an associative
> array:
>
> # t.awk
> {
> switch ($1)
> {
> case "set": arr[$2] = $3
> case "get": print arr[$2]
> case "quit": exit
> }
> }
>
> The bash script was
> #t
>
> mkfifo in
> mkfifo out
> awk -f t.awk <in >out &
> echo "set one 111" >in
> echo "set two 222" >in
>
> echo "get one" >in
> read x <out
> echo "one=$x"
>
> echo "get two" >in
> read x <out
> echo "two=$x"
>
> echo "set one XXX" >in # line 22
> echo "get one" >in
> read x <out
> echo "one=$x"
>
> echo "quit" >in
>
> The output I expected was
>
> one=111
> two=222
> one=XXX
>
> but instead this happened:
>
> one=111
> two=
> ./t: line 22: in: Interrupted system call
>
> It seeems that awk exits right after the first print. Can anybody explain
> that? And is there a way to keep awk reading?
>
Awk exits when it gets EOF on its last input file and completes the END
block if there is one.
Maybe this will work:
tail -f in | awk -f t.awk > out &

> If awk is replaced by a bash script which mimics the array functionality,
> everything works fine.
>
> Regards,
> Bernd
>


--
"All my life I wanted to be someone; I guess I should have been more specific."
-- Jane Wagner

Bill Marcum schrieb:

> Awk exits when it gets EOF on its last input file and completes the END
> block if there is one.
> Maybe this will work:
> tail -f in | awk -f t.awk > out &

Not really:

one=111
two=
../t: line 24: out: Interrupted system call
one=

The question is, why does awk think it gets EOF though there is none yet?

Bernd

--
Bernd Eggink
monoped@sudrala.de
http://autojar.sourceforge.de
http://jboom.sourceforge.de

2007-07-16, 16:51(+02), Bernd Eggink:
> I tried to connect an awk job to a bash script using two fifos. The idea
> was that the script should send requests to awk over one pipe and get
> results back over the other one.
>
> I tested that with a simple awk script providing access to an associative
> array:
>
> # t.awk
> {
> switch ($1)
> {
> case "set": arr[$2] = $3
> case "get": print arr[$2]
> case "quit": exit
> }
> }
>
> The bash script was
> #t
>
> mkfifo in
> mkfifo out
> awk -f t.awk <in >out &
> echo "set one 111" >in
> echo "set two 222" >in

You mention >in twice, so you open the pipe for writing twice.

Try:

{
echo "set one 111"
echo "set two 222"
} > in

instead.

But you should leave the fd to in open as long as you want to
write to it in a single session, otherwise, awk will see EOF,
each time you close the writer side.

--
Stéphane