Merge commit 'upstream/master' into prv/po

Conflicts:

	IkiWiki/Plugin/editpage.pm
	IkiWiki/Plugin/skeleton.pm.example
	doc/plugins/write.mdwn
	t/syntax.t

4 files changed, 179 insertions, 11 deletions

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
diff --git a/IkiWiki/Plugin/comments.pm b/IkiWiki/Plugin/comments.pm
index 1c4ab4895..833bedf25 100644
--- a/IkiWiki/Plugin/comments.pm
+++ b/IkiWiki/Plugin/comments.pm
@@ -343,8 +343,6 @@ sub sessioncgi ($$) {
 		error(gettext("bad page name"));
 	}
 
-	# FIXME: is this right? Or should we be using the candidate subpage
-	# (whatever that might mean) as the base URL?
 	my $baseurl = urlto($page, undef, 1);
 
 	$form->title(sprintf(gettext("commenting on %s"),
@@ -469,9 +467,21 @@ sub sessioncgi ($$) {
 	}
 
 	if ($form->submitted eq POST_COMMENT && $form->validate) {
-		my $file = "$location._comment";
-
 		IkiWiki::checksessionexpiry($cgi, $session);
+		
+		$postcomment=1;
+		IkiWiki::check_content(content => $form->field('editcontent'),
+			subject => $form->field('subject'),
+			$config{comments_allowauthor} ? (
+				author => $form->field('author'),
+				url => $form->field('url'),
+			) : (),
+			page => $location,
+			cgi => $cgi, session => $session
+		);
+		$postcomment=0;
+		
+		my $file = "$location._comment";
 
 		# FIXME: could probably do some sort of graceful retry
 		# on error? Would require significant unwinding though
diff --git a/IkiWiki/Plugin/editpage.pm b/IkiWiki/Plugin/editpage.pm
index 91d125a5c..658d664af 100644
--- a/IkiWiki/Plugin/editpage.pm
+++ b/IkiWiki/Plugin/editpage.pm
@@ -78,7 +78,43 @@ sub check_canedit ($$$;$) {
 			}
 		}
 	});
-	return $canedit;
+	return defined $canedit ? $canedit : 1;
+}
+
+sub check_content (@) {
+	my %params=@_;
+	
+	return 1 if ! exists $hooks{checkcontent}; # optimisation
+
+	if (exists $pagesources{$params{page}}) {
+		my @diff;
+		my %old=map { $_ => 1 }
+		        split("\n", readfile(srcfile($pagesources{$params{page}})));
+		foreach my $line (split("\n", $params{content})) {
+			push @diff, $line if ! exists $old{$_};
+		}
+		$params{content}=join("\n", @diff);
+	}
+
+	my $ok;
+	run_hooks(checkcontent => sub {
+		return if defined $ok;
+		my $ret=shift->(%params);
+		if (defined $ret) {
+			if ($ret eq "") {
+				$ok=1;
+			}
+			elsif (ref $ret eq 'CODE') {
+				$ret->();
+				$ok=0;
+			}
+			elsif (defined $ret) {
+				error($ret);
+			}
+		}
+
+	});
+	return defined $ok ? $ok : 1;
 }
 
 sub check_cansave ($$$$) {
@@ -395,9 +431,18 @@ sub cgi_editpage ($$) {
 			showform($form, \@buttons, $session, $q, forcebaseurl => $baseurl);
 			exit;
 		}
+			
+		my $message="";
+		if (defined $form->field('comments') &&
+		    length $form->field('comments')) {
+			$message=$form->field('comments');
+		}
 		
 		my $content=$form->field('editcontent');
 		check_cansave($page, $content, $q, $session);
+		check_content(content => $content, page => $page,
+			cgi => $q, session => $session,
+			subject => $message);
 		run_hooks(editcontent => sub {
 			$content=shift->(
 				content => $content,
@@ -431,12 +476,6 @@ sub cgi_editpage ($$) {
 		
 		my $conflict;
 		if ($config{rcs}) {
-			my $message="";
-			if (defined $form->field('comments') &&
-			    length $form->field('comments')) {
-				$message=$form->field('comments');
-			}
-			
 			if (! $exists) {
 				rcs_add($file);
 			}
diff --git a/IkiWiki/Plugin/skeleton.pm.example b/IkiWiki/Plugin/skeleton.pm.example
index 0bd5edeef..30c8dbd5c 100644
--- a/IkiWiki/Plugin/skeleton.pm.example
+++ b/IkiWiki/Plugin/skeleton.pm.example
@@ -31,6 +31,7 @@ sub import {
 	hook(type => "sessioncgi", id => "skeleton", call => \&sessioncgi);
 	hook(type => "canedit", id => "skeleton", call => \&canedit);
 	hook(type => "cansave", id => "skeleton", call => \&cansave);
+	hook(type => "checkcontent", id => "skeleton", call => \&checkcontent);
 	hook(type => "editcontent", id => "skeleton", call => \&editcontent);
 	hook(type => "formbuilder_setup", id => "skeleton", call => \&formbuilder_setup);
 	hook(type => "formbuilder", id => "skeleton", call => \&formbuilder);
@@ -181,6 +182,7 @@ sub canedit ($$$) {
 	debug("skeleton plugin running in canedit");
 }
 
+<<<<<<< HEAD:IkiWiki/Plugin/skeleton.pm.example
 sub cansave ($$$$) {
 	my $page=shift;
 	my $content=shift;
@@ -204,6 +206,12 @@ sub canrename ($$$) {
 	my $session=shift;
 
 	debug("skeleton plugin running in canrename");
+=======
+sub checkcontent (@) {
+	my %params=@_;
+
+	debug("skeleton plugin running in checkcontent");
+>>>>>>> upstream/master:IkiWiki/Plugin/skeleton.pm.example
 }
 
 sub editcontent ($$$) {