Shell Programming and Scripting

I have a piece of script that needs to pull every element in column 2 from a file of varying size. Rather than writing something to read the file and then pass the data line by line to my script, I want to use awk.

However, what is the correct syntax to get awk to call a function (named via:

fcnname()
{
.
.
}
)

with the second column ($2), i.e. such as

awk '{fcnname($2)}' somefilename

I know I can do it from the command line or from yet another script and use the "system" command...and that is my next tack, but Id rather tie this all up in one pretty package.....

I have been playing with this syntax all day and not getting it right...

thanks

Code:

awk ' function fcname(a) { print "col 2 =", a ; return 0 }
      { fcname($2) } ' < datafile

Ok thanks for the response, ..I fear I might not have been clear on the problem.

consider the script 'foo' that looks like this:

#some comments
function f1(){}
function f2() {}

some option/parameter checking

if [single option]
then
f1 $parm1 $parm2
else #multiple option
awk '{ I want to call f1 with $2 and another parm $parm2 }' somefilename
fi

echo "all done"
exit 0

The above is what I meant by an internal function, it was internal to the program, not awk.

when I enclose f1 within the awk construct like thus: awk '{f1($2)}' some file name I get an error like this:

awk: line 0 (NR=1): variable "f1" cannot be used as a function

I should mention this is not on unix or linux per se but rather OMVS (aka Unix Systems Services) under z/OS on an IBM z/800, and seems to be a rather faithful implementation of every API thus far....

You can't get there from here. When you define a shell function like that, the function is known only to that shell. Then you invoke awk which is a separate process. It has no way to peek inside its parent. At best, awk can spawn a child process of its own to access external commands. This child process cannot access a function known only to its grandparent.

Thats what I was afraid of...oh well, thanks a bunch....

yanno what.....something is nagging at me....

Right now I have a piece of script that looks like this: (anticipating your answer as given I started working at the problem from a different angle)

bar.sh:

# some comments

Some option checking

awk '{system("foo " $2 " " $parm2)}' somefile name

echo "done"

and I of course have foo.sh as before (minus the awk).

Are each of these 'system' calls a separate autonomous process? or merely a child? and will the execution of awk 'wait' until completion of each call?

the reason I ask is that parm '$2' names a unit of work that ABSOLUTELY MUST be executed in the order that it is listed in 'somefilename'.

{an example: 1) drive to store 2) buy stuff 3) drive home }

If we get out of order- disasterous results, mass hysteria, dogs and cats living together etc etc

I was trying to be slick but I might be forced to write a simple file reader to grab the second column item for the line....

All processes are children and more or less autonomous. And yes system is spawning processes. There is no issue with stuff being done in the wrong order. More processes does mean more resources consumed. So the only issue is performance. The whole thing will run faster as a single process. And that would be really easy to do in ksh or bash...

Code:

while read col1 col2  AndTheRest ; do
         fcname $col2
done