From 7ee92cab4095fb22d77a0b0100a472c6901eb519 Mon Sep 17 00:00:00 2001
From: Joey Hess <joey@gnu.kitenet.net>
Date: Fri, 16 Jan 2009 22:39:11 -0500
Subject: blogspam: New plugin, adding spam filtering for page editing /
 comment posting using the BlogSpam.net API.

---
 IkiWiki/Plugin/blogspam.pm | 111 +++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 111 insertions(+)
 create mode 100644 IkiWiki/Plugin/blogspam.pm

(limited to 'IkiWiki')

diff --git a/IkiWiki/Plugin/blogspam.pm b/IkiWiki/Plugin/blogspam.pm
new file mode 100644
index 000000000..6e68a9856
--- /dev/null
+++ b/IkiWiki/Plugin/blogspam.pm
@@ -0,0 +1,111 @@
+#!/usr/bin/perl
+package IkiWiki::Plugin::blogspam;
+
+use warnings;
+use strict;
+use IkiWiki 3.00;
+require RPC::XML;
+require RPC::XML::Client;
+
+my $defaulturl='http://test.blogspam.net:8888/';
+
+sub import {
+	hook(type => "getsetup", id => "blogspam",  call => \&getsetup);
+	hook(type => "checkcontent", id => "blogspam", call => \&checkcontent);
+}
+
+sub getsetup () {
+	return
+		plugin => {
+			safe => 1,
+			rebuild => 0,
+		},
+		blogspam_pagespec => {
+			type => 'pagespec',
+			example => 'postcomment(*)',
+			description => 'PageSpec of pages to check for spam',
+			link => 'ikiwiki/PageSpec',
+			safe => 1,
+			rebuild => 0,
+		},
+		blogspam_options => {
+			type => "string",
+			example => "blacklist=1.2.3.4,blacklist=8.7.6.5,max-links=10",
+			description => "options to send to blogspam server",
+			link => "http://blogspam.net/api/testComment.html#options",
+			safe => 1,
+			rebuild => 0,
+		},
+		blogspam_server => {
+			type => "string",
+			default => $defaulturl,
+			description => "blogspam server XML-RPC url",
+			safe => 1,
+			rebuild => 0,
+		},
+}
+
+sub checkcontent (@) {
+	my %params=@_;
+	
+ 	if (exists $config{blogspam_pagespec}) {
+		return undef
+			if ! pagespec_match($params{page}, $config{blogspam_pagespec},
+	                	location => $params{page});
+	}
+
+	my $url=$defaulturl;
+	$url = $params{blogspam_server} if exists $params{blogspam_server};
+	my $client = RPC::XML::Client->new($url);
+
+	my @options = split(",", $params{blogspam_options})
+		if exists $params{blogspam_options};
+
+	# Allow short comments and whitespace-only edits, unless the user
+	# has overridden min-words themselves.
+	push @options, "min-words=0"
+		unless grep /^min-words=/i, @options;
+	# Wiki pages can have a lot of urls, unless the user specifically
+	# wants to limit them.
+	push @options, "exclude=lotsaurls"
+		unless grep /^max-links/i, @options;
+	# Unless the user specified a size check, disable such checking.
+	push @options, "exclude=size"
+		unless grep /^(?:max|min)-size/i, @options;
+	# This test has absurd false positives on words like "alpha"
+	# and "buy".
+	push @options, "exclude=stopwords";
+
+	# blogspam API does not have a field for author url, so put it in
+	# the content to be checked.
+	if (exists $params{url}) {
+		$params{content}.="\n".$params{url};
+	}
+
+	my $res = $client->send_request('testComment', {
+		ip => $ENV{REMOTE_ADDR},
+		comment => $params{content},
+		subject => defined $params{subject} ? $params{subject} : "",
+		name => defined $params{author} ? $params{author} : "",
+		options => join(",", @options),
+		site => $config{url},
+		version => "ikiwiki ".$IkiWiki::version,
+	});
+
+	if (! ref $res || ! defined $res->value) {
+		debug("failed to get response from blogspam server ($url)");
+		return undef;
+	}
+	elsif ($res->value =~ /^SPAM:(.*)/) {
+		return gettext("Sorry, but that looks like spam to <a href=\"http://blogspam.net/\">blogspam</a>: ").$1;
+	}
+	elsif ($res->value ne 'OK') {
+		debug(gettext("blogspam server failure: ").$res->value);
+		return undef;
+	}
+	else {
+		return undef;
+	}
+}
+
+1
-- 
cgit v1.2.3