]> sipb.mit.edu Git - ikiwiki.git/commitdiff
blogspam: New plugin, adding spam filtering for page editing / comment posting using...
authorJoey Hess <joey@gnu.kitenet.net>
Sat, 17 Jan 2009 03:39:11 +0000 (22:39 -0500)
committerJoey Hess <joey@gnu.kitenet.net>
Sat, 17 Jan 2009 03:39:11 +0000 (22:39 -0500)
IkiWiki/Plugin/blogspam.pm [new file with mode: 0644]
debian/changelog
doc/plugins/blogspam.mdwn [new file with mode: 0644]
doc/todo/anti-spam_protection.mdwn
t/syntax.t

diff --git a/IkiWiki/Plugin/blogspam.pm b/IkiWiki/Plugin/blogspam.pm
new file mode 100644 (file)
index 0000000..6e68a98
--- /dev/null
@@ -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
index b75fe6afc41755721611e5970d1a3fda65b7c9f7..cb362d6e24f58cb9df8a11f27b6225c23d0a46bd 100644 (file)
@@ -20,6 +20,8 @@ ikiwiki (3.02) UNRELEASED; urgency=low
     in blogging.
   * checkcontent: New hook, can be used to implement arbitrary content
     filters, including spam filters.
+  * blogspam: New plugin, adding spam filtering for page editing / comment
+    posting using the BlogSpam.net API.
 
  -- Joey Hess <joeyh@debian.org>  Tue, 06 Jan 2009 15:02:52 -0500
 
diff --git a/doc/plugins/blogspam.mdwn b/doc/plugins/blogspam.mdwn
new file mode 100644 (file)
index 0000000..307d464
--- /dev/null
@@ -0,0 +1,23 @@
+[[!template id=plugin name=blogspam author="[[Joey]]"]]
+[[!tag type/auth]]
+
+This plugin adds antispam support to ikiwiki, using the
+[blogspam.net](http://blogspam.net/) API. Both page edits and
+[[comment|comments]] postings can be checked for spam. Currently,
+detected spam is not saved for human review, it is just rejected.
+
+You can control how content is tested via the `blogspam_options`
+setting. By default, the options are configured in a way that is
+appropriate for wiki content. This includes turning off some of the
+more problimatic tests.
+
+The `blogspam_pagespec` setting is a [[ikiwiki/PageSpec]] that can be
+used to configure which pages are checked for spam. The default is to check
+all edits. If you only want to check [[comments]] (not wiki page edits),
+set it to "postcomment(*)".
+
+By default, the blogspam.net server is used to do the spam checking. To
+change this, the `blogspam_server` option can be set to the url for a
+different server implementing the same API. Note that content is sent
+unencrypted over the internet to the server, and the server sees
+the full text of the content.
index cb45faee5d4b245712c10a9aeb9914af8a77eb80..b0524be5fcabfba5b66ef671ddebab6eb0645eb4 100644 (file)
@@ -17,3 +17,14 @@ Cheers,
 You might look at the Wikipedia page on "Spam\_in\_blogs" for more ideas.  In particular, would it be possible to force a subset of the pages (by regex, but you'd choose the regex to match those pages which are publicly writable) to use rel="nofollow" in all links.
 
 > I just wanted to leave a link here to the [[todo/require_CAPTCHA_to_edit]] plugin patch.  Unfortunately that plugin currently interacts badly with the openid plugin. -- [[Will]]
+
+
+---
+
+Ikiwiki now has a checkcontent hook that plugins can use to see content
+that is being entered and check it for spam/whatever.
+
+There is a blogspam plugin that uses the blogspam.org service
+to check for common spam signatures. --[[Joey]] 
+
+[[done]]
index d09d17f7f932deecb9c6f16e6911367ced9d4d35..9d5cbc3739d168b9362cbfb91620a9f624e3a918 100755 (executable)
@@ -5,8 +5,8 @@ use Test::More;
 
 my @progs="ikiwiki.in";
 my @libs="IkiWiki.pm";
-# monotone, external, amazon_s3 skipped since they need perl modules
-push @libs, map { chomp; $_ } `find IkiWiki -type f -name \\*.pm | grep -v monotone.pm | grep -v external.pm | grep -v amazon_s3.pm`;
+# monotone, external, blogspam, amazon_s3 skipped since they need perl modules
+push @libs, map { chomp; $_ } `find IkiWiki -type f -name \\*.pm | grep -v monotone.pm | grep -v external.pm | grep -v blogspam.pm | grep -v amazon_s3.pm`;
 push @libs, 'IkiWiki/Plugin/skeleton.pm.example';
 
 plan(tests => (@progs + @libs));