How can I recover from a timeout with Perl's WWW::Mechanize?

Question

I'm using WWW::Mechanize to read a particular webpage in a loop that runs every few seconds. Occasionally, the 'GET' times out and the script stops running. How can I recover from one such timeout such that it continues the loop and tries the 'GET' the next time around?

Costumer

Answer 1

Use eval :

eval {
    my $resp = $mech->get($url);
    $resp->is_success or die $resp->status_line;
    # your code
};

if ($@) {
    print "Recovered from a GET error\n";    
}

The eval block will catch any error while GETing the page.

Ivonne

Answer 2

One option would be to implement a method to handle timeout errors and hook it into the mech object at construction time as the onerror handler. See Constructor and Startup in the docs.

You could even ignore errors by setting a null error handler, for example:

my $mech = WWW::Mechanize->new( onerror => undef );

but I wouldn't recommend that - you'll just get weird problems later.

Hamman

Answer 3

This solution will continue to attempt to load the page until it works.

do {
    eval {
        $mech->get($url);
    };
} while ($@ ne '');

Becca

Answer 4

For a more complete solution, you can use a module like Try::Tiny::Retry. It allows you to specify a code block to run, catches any errors, then retries that code block a configurable amount of time. The syntax is pretty clean.

use WWW::Mechanize();
use Try::Tiny::Retry ':all';

my $mech = WWW::Mechanize->new();
retry {
    $mech->get("https://stackoverflow.com/");
}
on_retry {
    warn("Failed. Retrying. Error was: $_");
}
delay {
    # max of 100 tries, sleeping 5 seconds between each failure
    return if $_[0] >= 100;
    sleep(11 * 1000 * 1000);
}; #don't forget this semicolon

# dump all the links found on the page
print join "\n", map {$_->text } $mech->links;

Jacalynjacamar