Magic-Mirror/gecko-dev
1
0
Fork 0
mirror of https://github.com/mozilla/gecko-dev.git synced 2024-11-30 00:01:50 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Convert RDF.bm to use a real XML parser (now requires XML::RSS, available at your nearest CPAN mirror). Thanks to Jeff Bisbee. See bug 154826.

Browse Source
This commit is contained in:
ian%hixie.ch 2003-10-05 20:06:38 +00:00
parent b85d962eba
commit e16e30afa3
2 changed files with 40 additions and 30 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
webtools/mozbot/BotModules/COPYING
View File

@ -24,3 +24,4 @@ Contributor(s): Harrison Page <harrison@netscape.com>
Jake Steenhagen <jake@acutex.net>
mental <xor@ivwnet.com>
Mohamed Elzakzoki <mhtawfiq@yifan.net>
Jeff Bisbee <mozilla-bugs@jbisbee.com>

69
webtools/mozbot/BotModules/RDF.bm
View File

@ -1,8 +1,11 @@
################################
# 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;
@ -39,7 +42,7 @@ sub Help {
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.";
$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>';
$commands{'unmute'} = 'Enable reporting of a site in a channel. By default, sites are reported in all channels that the module is active in. Syntax: unmute <site> in <channel>';
} else {
@ -131,38 +134,28 @@ sub GotURI {
if ($output) {
# last update stamp
$self->{'data'}->{$uri}->{'last'} = $event->{'time'};
my $last = $event->{'time'};
$self->{'data'}->{$uri}->{'last'} = $last;
# this, of course, is a disaster waiting to happen.
# for example, we don't cope with comments.
# someone write a real XML version of this pleeeeease... XXX
# get the juicy stuff out
my $channelpart = "";
if ($output =~ /<channel>(.*)<\/channel>/osi) {
$channelpart = $1;
# 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;
}
# remove any image related stuff
$output =~ s/<image>.*<\/image>//gosi;
# Set Link and Title
$self->{data}->{$uri}->{'link'} = $rss->{'channel'}->{'link'};
$self->{data}->{$uri}->{'title'} = $rss->{'channel'}->{'title'};
# get the channel title
$self->{'data'}->{$uri}->{'title'} = $uri;
if ($channelpart =~ /<title>\s*(.+?)\s*<\/title>/osi) {
$self->{'data'}->{$uri}->{'title'} = $self->unescapeXML($1);
$self->{'data'}->{$uri}->{'title'} =~ s/: News for nerds, stuff that matters//gosi if $self->{'trimTitles'};
}
# get the channel website
$self->{'data'}->{$uri}->{'link'} = $uri;
if ($channelpart =~ /<link>\s*(.+?)\s*<\/link>/osi) {
$self->{'data'}->{$uri}->{'link'} = $self->unescapeXML($1);
}
# get all the items
while ($output =~ /<item>.*?<title>\s*(.+?)\s*<\/title>.*?<\/item>/osig) {
unless (($1 =~ /^last update/osi) or (defined($self->{'data'}->{$uri}->{'items'}->{$self->unescapeXML($1)}))) {
$self->{'data'}->{$uri}->{'items'}->{$self->unescapeXML($1)} = $self->{'data'}->{$uri}->{'last'};
foreach my $item (@{$rss->{'items'}}) {
unless (($item->{title} =~ /^last update/osi) ||
(defined($self->{'data'}->{$uri}->{'items'}->{$item->{'title'}}))) {
$self->{'data'}->{$uri}->{'items'}->{$item->{'title'}} = $last;
}
}
@ -203,8 +196,24 @@ sub ReportDiffs {
foreach (keys(%{$self->{'data'}->{$uri}->{'items'}})) {
push(@output, $_) if ($self->{'data'}->{$uri}->{'items'}->{$_} == $last);
}
if (@output) {
# -- #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);