source: trollforge/irc/irssi/hp_rimming.pl @ 606 Tweet

#!/usr/bin/perl

use strict;
use vars qw($VERSION %IRSSI);
use Irssi;
use Irssi::Irc;
use XML::RSS::Parser;
use LWP::UserAgent;

$VERSION = '1.00';
%IRSSI = (
          authors     => 'jenk',
          contact     => 'irc.hardchats.com',
          name        => 'hp_rimming',
          description => qq {
              This bot will provide a channel with timely updates
              of http://hp_rimming.livejournal.com
          },
          license     => 'BPL',
          );

my $INTERVAL = 5;
my $MIN_ENTRY_LEN = 5_000;
my $done = 0;
my @entries;
my $win;
my $last_entry = '';

Irssi::command_bind('hp_latest' => \&hp_latest);
Irssi::command_bind('hp_done'   => \&hp_done);
Irssi::command_bind('hp_start'  => \&hp_start);

sub main_loop {
    while (! $done) {
        sleep $INTERVAL;
        last if $done;

        next unless $win;

    }

    $win->print(MSGLEVEL_CRAP, 'hp_rimming finished') if $win;
    exit;
}

sub hp_start {
    my ($data, $server, $witem) = @_;
    $win = $witem;
    $win->print(MSGLEVEL_CRAP, "Starting hp_rimming bot");
    $win->command("MSG #rimming hi");
    $win->command("MSG #rimming hi");
    #main_loop();
}

sub hp_latest {
    my ($data, $server, $witem) = @_;

    $data += 0;

    my $entry = fetch_entry($data);
    return unless $entry;
    $last_entry = $entry;

    if (! $witem || ! $witem->{name}) {
        print "Invalid window";
        return;
    }

    say_entry($witem->{name}, $entry);
}

sub say_entry {
    my ($where, $entry) = @_;

    my $win = Irssi::active_win();

    my @lines = split("\n", $entry);
    foreach my $line (@lines) {
        chomp $line;
        #$entry =~ s/(['"])/\\$1/g;
        $win->command("MSG $where $line");
    }
}

sub hp_done {
    $done = 1;
}

sub fetch_entry {
    my $num = shift;

    my @entries = fetch_entries();

    $num = @entries - 1 if $num >= @entries;
    my $entry = $num ? $entries[$num] : shift @entries;

    return $entry;
}

sub fetch_entries {
    my $use_timingbot = shift;

    my $bot;

    if ($use_timingbot) {
        require 'XML::RSS::TimingBot';
        $bot = XML::RSS::TimingBot->new;
    } else {
        $bot = LWP::UserAgent->new;
    }

    my $win = Irssi::active_win();

    # get latest posts
    my $url = 'http://community.livejournal.com/hp_rimming/data/rss';
    $win->print("Fetching latest post...");
    my $response = $bot->get($url);

    if ($response->code == 304) {
        # hasn't changed since last access
        $win->print("No new posts");
        return;
    } elsif ($response->code != 200) {
        $win->print("Bad response getting $url: " . $response->status_line);
        return;
    }

    # save infoz
    $bot->commit if ref $bot =~ /XML::RSS::TimingBot/;

    # parse latest posts
    my @entries;

    my $p = XML::RSS::Parser->new;
    my $feed = $p->parse_string($response->content);

    unless ($feed) {
        $win->print("Error: could not parse feed");
        return ();
    }

    # get item count
    my $count = $feed->item_count;
    print "Latest post count: ($count)\n";

    # parse items
    foreach my $i ( $feed->query('//item') ) {
        my $entry_ele = $i->query('description');
        next unless $entry_ele;
        my $entry = $entry_ele->text_content;
        next unless $entry;

        # clean up entry
        # strip html
        $entry =~ s!<br\s*/?>!\n!g;
        $entry =~ s/\<([^\<])+\>//g;
        $entry =~ s/&nbsp;/ /gi;
        $entry =~ s!&lt;/lj-cut&lt;!!ig;

        next unless $entry && length $entry > $MIN_ENTRY_LEN;

        push @entries, $entry;
    }

    return @entries;
}