Forum libero

> > > > vorrebbero circa 2 settimane
> > > >

#!perl -w
use strict;

use LWP::UserAgent;

my $FORUM_ID = 127;
my $NAME = 'Troll';

sub get_thread {
    my $browser = shift;
    my $id = shift;
    my $dest_dir = shift;

    my $i = 1;

    for(;;) {
        my $content = undef;
        for (1 ... 3) {
            my $response = $browser->get("<a href="http://punto-informatico.it/f/ms.aspx?m_id=${id}&m_rid=0&Page=$i" title="http://punto-informatico.it/f/ms.aspx?m_id=${id}&m_rid=0&Page=$i" target='_blank' rel='nofollow'>http://punto-informatico.it/f/ms.aspx?m_id=${id}&m...</a>");
    
            if ($response->is_success) {
                $content = $response->content;
                last;
            }
        }

        die "FAILED TO GET $i" if not defined $content;


        print "   GOT ${id}_$i\n";
        open OUT, ">$dest_dir/${id}_$i.html";
        print OUT $content;
        close OUT;

        last if $content !~ m{<a href=\'.*?\'>\s*Successiva\s*</a>};
        ++$i;
    } 
    
}

sub get_forum {
    my $browser = shift;
    my $page = shift;
    my $dest_dir = shift;
    my $name = shift;

    print "DOING PAGE $page\n";

    my $content = undef;
    for (1 ... 3) {
        my $response = $browser->get("<a href="http://punto-informatico.it/f/hp.aspx?f_id=${FORUM_ID}&Page=$page" title="http://punto-informatico.it/f/hp.aspx?f_id=${FORUM_ID}&Page=$page" target='_blank' rel='nofollow'>http://punto-informatico.it/f/hp.aspx?f_id=${FORUM...</a>");

        if ($response->is_success) {
            $content = $response->content;
            last;
        }
    }

    while ($content =~ m{<a href=\'([^\']*?)\' title=\'Leggi messaggio\' class=\'textcapothread\'>.*?</a>}gc) {
        my $msg_link = $1;
        $msg_link =~ 'm_id=(\d+)\&';
        #print "$1\n";
        &get_thread($browser, $1, $dest_dir);
    }

    open OUT, ">$dest_dir/${name}_$page.html";
    print OUT $content;
    close OUT;
    
    sleep 1;

    return $content =~ m{<a href=\'[^\']*?\'>\s*Successiva\s*</a>}
} 


my $ua = LWP::UserAgent->new;


# da chiamare con:
# perl forum_dei_troll
# deve esistere una directry chiamata 'forum_dei_troll' nella directory 
# corrente

# cambia 'my $page = 1' per partire da una diversa pagina del forum
for (my $page = 1; &get_forum($ua, $page, $ARGV[0], $NAME); ++$page) { };