pjs/webtools/mozbot/BotModules/RDF.bm

################################
# RDF Module                   #
################################
# this is really an RSS module, not an RDF module.
# but oh well.

package BotModules::RDF;
use XML::RSS;
use vars qw(@ISA);
@ISA = qw(BotModules);
1;

# RegisterConfig - Called when initialised, should call registerVariables
sub RegisterConfig {
    my $self = shift;
    $self->SUPER::RegisterConfig(@_);
    $self->registerVariables(
      # [ name, save?, settable? ]
        ['sites', 1, 1, {}], 
        ['updateDelay', 1, 1, 600],
        ['preferredLineLength', 1, 1, 80],
        ['maxInChannel', 1, 1, 5],
        ['trimTitles', 1, 1, '0'],
        ['data', 0, 0, {}],  # data -> uri -> (title, link, last, items -> uri)
        ['mutes', 1, 1, {}],  # uri -> "channel channel channel"
    );
}

# Schedule - called when bot connects to a server, to install any schedulers
# use $self->schedule($event, $delay, $times, $data)
# where $times is 1 for a single event, -1 for recurring events,
# and a +ve number for an event that occurs that many times.
sub Schedule {
    my $self = shift;
    my ($event) = @_;
    $self->schedule($event, \$self->{'updateDelay'}, -1, 'rdf');
    $self->SUPER::Schedule($event);
}

sub Help {
    my $self = shift;
    my ($event) = @_;
    my %commands; 
    if ($self->isAdmin($event)) {
        $commands{''} = "The RDF module monitors various websites. Add new RDF channels to the 'sites' hash. Duplicates with different nicknames are fine. For example, \"vars $self->{'_name'} sites '+|slashdot|http://...'\" and \"vars $self->{'_name'} sites '+|/.|http://...'\" is fine.  To remove a site from the RDF 'sites' hash, use this syntax \"vars $self->{_name} sites '-slashdot'";
        $commands{'mute'} = 'Disable reporting of a site in a channel. (Only does something if the given site exists.) Syntax: mute <site> in <channel>';
        $commands{'unmute'} = 'Enable reporting of a site in a channel. By default, sites are reported in all channels that the module is active in. Syntax: unmute <site> in <channel>';
    } else {
        $commands{''} = 'The RDF module monitors various websites.';
    }
    foreach my $site (keys(%{$self->{'sites'}})) {
        if ($self->{'data'}->{$self->{'sites'}->{$site}}) {
            $commands{$site} = "Reports the headlines listed in $self->{'data'}->{$self->{'sites'}->{$site}}->{'title'}";

            # -- #mozilla was here --
            #      <Hixie> anyway, $self->{'data'}->{$self->{'sites'}->{$site}}->{'title'} is 
            #              another nice piece of perl (embedded in a quoted string in this case)
            #     <moogle> yeah, that's a bit more familiar
            #        <jag> Oooh, nice one
            #        <jag> Reminds me of Java, a bit :-)
            #        <jag> Without all the casting about from Object to Hashtable
            #      <Hixie> all this, BTW, is from the RDF module (the one that mozbot uses to 
            #              report changes in mozillazine and so on)
            #     <moogle> I still tend to comment these things a bit just for maintainability 
            #              by others who might not wish to do mental gymnastics :)
            #      <Hixie> :-) 

        } else {
            $commands{$site} = "Reports the headlines listed in $self->{'sites'}->{$site}";
        }

    }
    return \%commands;
}

sub Told {
    my $self = shift;
    my ($event, $message) = @_;
    foreach my $site (keys(%{$self->{'sites'}})) {
        if ($message =~ /^\s*(\Q$site\E)\s*$/si) {
            $self->GetSite($event, $1, 'request');
            return 0; # dealt with it... 
        }
    }
    if ($self->isAdmin($event)) {
        if ($message =~ /^\s*mute\s+(\S+?)\s+in\s+(\S+?)\s*$/osi) {
            my $site = $1 eq 'RDF' ? '' : $self->{'sites'}->{$1};
            my $siteName = $site eq '' ? 'all sites' : $site;
            if (defined($site)) {
                $self->{'mutes'}->{$site} .= " $2";
                $self->saveConfig();
                $self->say($event, "$event->{'from'}: RDF notifications for $siteName muted in channel $2.");
            } else {
                # can't say this, other modules might recognise it: $self->say($event, "$event->{'from'}: I don't know about any '$1' site...");
            }
        } elsif ($message =~ /^\s*unmute\s+(\S+?)\s+in\s+(\S+?)\s*$/osi) {
            my $site = $1 eq 'RDF' ? '' : $self->{'sites'}->{$1};
            my $siteName = $site eq '' ? 'all sites' : $site;
            if (defined($site)) {
                my %mutedChannels = map { lc($_) => 1 } split(/ /o, $self->{'mutes'}->{$site});
                delete($mutedChannels{lc($2)}); # get rid of any mentions of that channel
                $self->{'mutes'}->{$site} = join(' ', keys(%mutedChannels));
                $self->saveConfig();
                $self->say($event, "$event->{'from'}: RDF notifications for $siteName resumed in channel $2.");
            } else {
                # can't say this, other modules might recognise it: $self->say($event, "$event->{'from'}: I don't know about any '$1' site...");
            }
        } else {
            return $self->SUPER::Told(@_);
        }
    } else {
        return $self->SUPER::Told(@_);
    }
    return 0;
}

sub GetSite {
    my $self = shift;
    my ($event, $site, $intent) = @_;
    if (defined($self->{'sites'}->{$site})) {
        my $uri = $self->{'sites'}->{$site};
        $self->getURI($event, $uri, $intent);
    } else {
        # XXX
    }
}

sub GotURI {
    my $self = shift;
    my ($event, $uri, $output, $intent) = @_;

    $self->{'data'}->{$uri}->{'ready'} = defined($self->{'data'}->{$uri});

    if ($output) {

        # last update stamp
        my $last = $event->{'time'};
        $self->{'data'}->{$uri}->{'last'} = $last;

        # Parse It
        my $rss = XML::RSS->new();
        eval { $rss->parse($output) };
        if ($@) {
            $self->debug("$uri is not a valid RSS file");
            if ($intent eq 'request') {
                $self->say($event, "$event->{'from'}: Dude, the file is not valid RSS! ($uri)");
            }
            return;
        }

        # Set Link and Title
        $self->{data}->{$uri}->{'link'} = $rss->{'channel'}->{'link'};
        $self->{data}->{$uri}->{'title'} = $rss->{'channel'}->{'title'};

        foreach my $item (@{$rss->{'items'}}) {
            unless (($item->{title} =~ /^last update/osi) || 
                    (defined($self->{'data'}->{$uri}->{'items'}->{$item->{'title'}}))) {
                $self->{'data'}->{$uri}->{'items'}->{$item->{'title'}} = $last;
            }
        }

        $self->ReportDiffs($event, $uri, $intent);
        if ($intent eq 'request') {
            $self->ReportAll($event, $uri);
        }

    } else {

        if ($intent eq 'request') {
            $self->say($event, "$event->{'from'}: Dude, the file was empty! ($uri)");
        }

    }

}

sub Scheduled {
    my $self = shift;
    my ($event, @data) = @_;
    if ($data[0] eq 'rdf') {
        my %sites = map { $_ => 1 } values(%{$self->{'sites'}});
        foreach (keys(%sites)) {
            $self->getURI($event, $_, 'update');
        }
    } else {
        $self->SUPER::Scheduled($event, @data);
    }
}

sub ReportDiffs {
    my $self = shift;
    my ($event, $uri, $request) = @_;
    return unless $self->{'data'}->{$uri}->{'ready'};
    my $last = $self->{'data'}->{$uri}->{'last'};
    my @output;
    foreach (keys(%{$self->{'data'}->{$uri}->{'items'}})) {
        push(@output, $_) if ($self->{'data'}->{$uri}->{'items'}->{$_} == $last);
    }

    # -- #mrt was here --
    #     <mozbot> Friday's security advisories -- The first stable
    #              Xen release -- Linux Gazette #95
    #     <mozbot> KDE Under The Microscope -- Additional OpenSSL info
    #      <Hixie> wtf
    #     <mozbot> Just appeared in jbisbee.com -
    #              http://www.jbisbee.com/ : PoCo::RSS::Aggregator
    #      <Hixie> why is it repeating the same thing over and over
    #     <mozbot> PoCo::RSSAggregator & XML::RSS::Feed Uploaded to
    #              CPAN -- More PoCo::RSSAggregator
    #      <Hixie> mozbot: shutup please
    #     <mozbot> Ok, threw away 2558 messages.

    # Ahem. So now we limit the diff reporting code to maxInChannel
    # items at a time...

    if (@output and @output < $self->{'maxInChannel'}) {
        @output = $self->prettyPrint($self->{'preferredLineLength'}, 
                                     "Just appeared in $self->{'data'}->{$uri}->{'title'} - $self->{'data'}->{$uri}->{'link'} :  ",
                                     '', ' -- ', @output);

        my %mutedChannels = ();
        if (defined($self->{'mutes'}->{$uri})) {
            %mutedChannels = map { lc($_) => 1 } split(/\s+/os, $self->{'mutes'}->{$uri});
        }
        if (defined($self->{'mutes'}->{''})) {
            %mutedChannels = (%mutedChannels, map { lc($_) => 1 } split(/\s+/os, $self->{'mutes'}->{''}));
        }
        if ($request eq 'request') {
            $mutedChannels{$event->{'channel'}} = 1;
        }
        foreach (@{$self->{'channels'}}) {
            unless ($mutedChannels{$_}) {
                local $event->{'target'} = $_;
                foreach (@output) {
                    $self->say($event, $_);
                }
            }
        }
    }
}

sub ReportAll {
    my $self = shift;
    my ($event, $uri) = @_;
    my @output;
    foreach (keys(%{$self->{'data'}->{$uri}->{'items'}})) {
        push(@output, $_);
    }

    @output = $self->prettyPrint($self->{'preferredLineLength'}, 
                                 "Items in $self->{'data'}->{$uri}->{'title'} - $self->{'data'}->{$uri}->{'link'}: ",
                                 "$event->{'from'}: ", ' -- ', @output);

    if (@output > $self->{'maxInChannel'}) {
        foreach (@output) {
            $self->directSay($event, $_);
        }
        $self->channelSay($event, "$event->{'from'}: /msg'ed");
    } else {
        foreach (@output) {
            $self->say($event, $_);
        }
    }
}