[dw-free] Allow more sites in embed whitelist

[fu]

[commit: http://hg.dwscoalition.org/dw-free/rev/e5aed810efb8]

http://bugs.dwscoalition.org/show_bug.cgi?id=3707

Expand support for iframe embeds from the sites: bandcamp.com, blip.tv,
www.dailymotion.com, dotsub.com, nicovideo.jp, sbs.com.au, scribd.com,
slideshare.net, vimeo.com, maps.google.com

Lots of thanks to monarchist, ninetydegrees,
bookofjude, and andrewducker for gathering
examples of the embed codes to use!

Patch by fu.

Files modified:

• cgi-bin/DW/Hooks/EmbedWhitelist.pm
• t/embed-whitelist.t

--------------------------------------------------------------------------------
diff -r d5c9f9570e7d -r e5aed810efb8 cgi-bin/DW/Hooks/EmbedWhitelist.pm
--- a/cgi-bin/DW/Hooks/EmbedWhitelist.pm	Wed Aug 10 12:48:43 2011 +0800
+++ b/cgi-bin/DW/Hooks/EmbedWhitelist.pm	Wed Aug 10 18:22:04 2011 +0800
@@ -29,6 +29,9 @@
 use LJ::Hooks;
 use URI;
 
+# for internal use only
+# this is used when sites may offer embeds from multiple subdomain
+# e.g., www, www1, etc
 sub match_subdomain {
     my $want_domain = $_[0];
     my $domain_from_uri = $_[1];
@@ -43,6 +46,23 @@
     return $path_from_uri =~ /^$want_path$/;
 }
 
+my %host_path_match = (
+    "bandcamp.com"          => qr!^/EmbeddedPlayer/!,
+    "blip.tv"               => qr!^/play/!,
+
+    "www.dailymotion.com"   => qr!^/embed/video/!,
+    "dotsub.com"            => qr!^/media/!,
+
+    "maps.google.com"       => qr!^/maps!,
+    "ext.nicovideo.jp"      => qr!^/thumb/!,
+
+    "www.sbs.com.au"         => qr!/player/embed/!,  # best guess; language parameter before /player may vary
+    "www.scribd.com"        => qr!^/embeds/!,
+    "www.slideshare.net"    => qr!^/slideshow/embed_code/!,
+
+    "player.vimeo.com"      => qr!^/video/\d+$!,
+);
+
 LJ::Hooks::register_hook( 'allow_iframe_embeds', sub {
     my ( $embed_url, %opts ) = @_;
 
@@ -56,6 +76,9 @@
     my $uri_host = $parsed_uri->host;
     my $uri_path = $parsed_uri->path;   # not including query
 
+    my $path_regex = $host_path_match{$uri_host};
+    return 1 if $path_regex && ( $uri_path =~ $path_regex );
+
     ## YouTube (http://apiblog.youtube.com/2010/07/new-way-to-embed-youtube-videos.html)
     if ( match_subdomain( "youtube.com", $uri_host ) || match_subdomain( "youtube-nocookie.com", $uri_host ) ) {
         return 1 if match_full_path( qr!/embed/[-_a-zA-Z0-9]{11,}!, $uri_path );
diff -r d5c9f9570e7d -r e5aed810efb8 t/embed-whitelist.t
--- a/t/embed-whitelist.t	Wed Aug 10 12:48:43 2011 +0800
+++ b/t/embed-whitelist.t	Wed Aug 10 18:22:04 2011 +0800
@@ -1,7 +1,7 @@
 # -*-perl-*-
 use strict;
 
-use Test::More tests => 10;
+use Test::More tests => 24;
 use lib "$ENV{LJHOME}/cgi-bin";
 require 'ljlib.pl';
 
@@ -46,3 +46,32 @@
     test_bad_url( "http://www.youtube.com/notreallyembed/x1xx2xxxxxX", "wrong path");
     test_bad_url( "http://www.youtube.com/embed/x1xx2xxxxxX/butnotreally", "wrong path");
 }
+
+note( "misc" );
+{
+    test_good_url( "http://bandcamp.com/EmbeddedPlayer/v=2/track=123123123/size=venti/bgcol=FFFFFF/linkcol=4285BB/" );
+    test_good_url( "http://bandcamp.com/EmbeddedPlayer/v=2/track=123123123" );
+
+    test_good_url( "http://blip.tv/play/x11Xx11Xx.html" );
+
+    test_good_url( "http://www.dailymotion.com/embed/video/x1xx11x" );
+
+    test_good_url( "http://dotsub.com/media/9db493c6-6168-44b0-89ea-e33a31db48db/e/m" );
+
+    test_good_url( "http://maps.google.com/maps?f=q&source=s_q&hl=en&geocode=&q=somethingsomething&aq=0&sll=00.000,-00.0000&sspn=0.00,0.0&vpsrc=0&ie=UTF8&hq=&hnear=somethingsomething&z=0&ll=0,-00&output=embed" );
+
+    test_good_url( "http://ext.nicovideo.jp/thumb/sm123123123" );
+    test_good_url( "http://ext.nicovideo.jp/thumb/nm123123123" );
+    test_good_url( "http://ext.nicovideo.jp/thumb/123123123" );
+
+    test_good_url( "http://www.sbs.com.au/yourlanguage//player/embed/id/163111" );
+
+    test_good_url( "http://www.scribd.com/embeds/123123/content?start_page=1&view_mode=list&access_key=" );
+
+    test_good_url( "http://www.slideshare.net/slideshow/embed_code/12312312" );
+
+    test_good_url( "http://player.vimeo.com/video/123123123?title=0&byline=0&portrait=0" );
+    test_bad_url("http://player.vimeo.com/video/123abc?title=0&byline=0&portrait=0");
+}
+
+
--------------------------------------------------------------------------------