How do I make two perl files communicate? - perl

Question

So I have something like this:
fork.pl
for $str (#files)
{
my($command) = "perl command.pl ".$str;
exec( $command );
}
command.pl
$file=$ARGV[0].".csv";
#code that counts rows here
print $rowcount;
So as the end result I have 10 files launched which count how many rows are in each csv file.
I do not need help editting this code, it works (this is just a compressed version). I need help figuring out how to take the output ($rowcount) of ten files and combine it into one for further processing.

Answer 1

I keep some utility code around for just this purpose... this is tweaked slightly to your question and including a synchronized global counting method.
#!/usr/bin/perl
use threads;
use Thread::Queue;
my #workers;
my $num_threads = 10;
my $queue = new Thread::Queue;
my $total_ines = 0;
for (0..$num_threads-1) {
$workers[$_] = new threads(\&worker);
}
while ($_ = shift #ARGV) {
$queue->enqueue($_);
}
sub worker() {
while ($file = $queue->dequeue) {
#line counting code here
global_counter($lines_counted);
}
}
sub global_counter() :locked {
#add to the number of lines counted
$total_lines += shift
}
for (0..$num_threads-1) { $queue->enqueue(undef); }
for (0..$num_threads-1) { $workers[$_]->join; }
print $total_lines;

Answer 2

This kind of communication is solved using pipes (let me write a simple example):
# -- fork.pl -------------------------
for (1..3) {
open my $PIPE, "perl command.pl |";
print "catch: $_\n" while(<$PIPE>);
close $PIPE;
}
# -- command.pl ----------------------
print rand(1);
It prints (random numbers):
catch: 0.58929443359375
catch: 0.1290283203125
catch: 0.907012939453125

Answer 3

You need to look either at threads or Interprocess communication with e.g. sockets or shared memory when using fork.

Answer 4

Compressed but won't work. I'm assuming that in fork.pl, you fork before exec'ing? Backticks capture the output of the called process, namely your prints:
fork.pl
for $str (#files)
{
my($command) = "perl command.pl ".$str;
print `$command`;
}
But rather than forking and launching processes, wouldn't it be smarter to turn the second file into a module?
package MyCommand;
use Exporter;
our #EXPORT = qw( command );
sub command {
my $file = $_[0] . '.csv';
...
return $rowcount;
}
1;
fork.pl:
use MyCommand;
...
my #rowcounts;
for my $str (#files) {
push #rowcounts, command($str);
}
A bit of self-promotion, but I just posted this in your other thread, which seems relevant enough: How to run in parallel two child command from a parent one?

Answer 5

Accumulate pipes from children:
#!/usr/bin/perl -w
use strict;
my $files = qw/one.csv two.csv three.csv/;
my $command = "perl command.pl";
my #pipes;
foreach (#files) {
my $fd;
open $fd, "-|", "$command $_" and push #pipes, $fd;
};
my $sum = 0;
foreach my $pp (#pipes) {
$sum += $_ if defined ($_=<$pp>);
};
print $sum;
Then you can just read them one by one (as in example), or use IO::Select to read data as it appears in each pipe.
A hash table in addition to array is also good if you want to know which data comes from which source.