MediaRSS support, support for mapping tag attributes

cardmagic · Jul 6, 2010 · 225a6d6 · 225a6d6
1 parent 0915f34
commit 225a6d6
Show file tree

Hide file tree

Showing 4 changed files with 504 additions and 3 deletions.
diff --git a/README b/README
@@ -32,10 +32,11 @@ The parser does not care about the correctness of the XML as it does not use an
    SimpleRSS.feed_tags << :some_new_tag
    SimpleRSS.item_tags << :"item+myrel" # this will extend SimpleRSS to be able to parse RSS items or ATOM entries that have a rel specified, common in many blogger feeds
    SimpleRSS.item_tags << :"feedburner:origLink" # this will extend SimpleRSS to be able to parse RSS items or ATOM entries that have a specific pre-tag specified, common in many feedburner feeds
-
+   SimpleRSS.item_tags << :"media:content#url" # this will grab the url attribute of the media:content tag 
 
 == Authors
 * Lucas Carlson  (mailto:lucas@rufy.com)
+* Herval Freire (mailto:hervalfreire@gmail.com)
 
 Inspired by Blagg (http://www.raelity.org/lang/perl/blagg) from Rael Dornfest.
 

diff --git a/lib/simple-rss.rb b/lib/simple-rss.rb
@@ -2,7 +2,7 @@
 require 'time'
 
 class SimpleRSS
-  VERSION = "1.2.2"
+  VERSION = "1.2.3"
 
 	attr_reader :items, :source
 	alias :entries :items
@@ -30,7 +30,11 @@ class SimpleRSS
 		:category, :guid,
 		:'trackback:ping', :'trackback:about',
 		:'dc:creator', :'dc:title', :'dc:subject', :'dc:rights', :'dc:publisher',
-		:'feedburner:origLink'
+		:'feedburner:origLink',
+		:'media:content#url', :'media:content#type', :'media:content#height', :'media:content#width',
+		:'media:title', :'media:thumbnail#url', :'media:thumbnail#height', :'media:thumbnail#width',
+		:'media:credit', :'media:credit#role',
+		:'media:category', :'media:category#scheme'
 	]
 
 	def initialize(source, options={})
@@ -106,6 +110,16 @@ def parse
   				  nil
   				end
   				item[clean_tag("#{tag}+#{rel}")] = clean_content(tag, $3, $4) if $3 || $4
+  			elsif tag.to_s.include?("#")
+			    tag_data = tag.to_s.split("#")
+			    tag = tag_data[0]
+			    attrib = tag_data[1]
+  				if match[3] =~ %r{<(rss:|atom:)?#{tag}(.*?)#{attrib}=['"](.*?)['"](.*?)>(.*?)</(rss:|atom:)?#{tag}>}mi
+            nil
+  				elsif match[3] =~ %r{<(rss:|atom:)?#{tag}(.*?)#{attrib}=['"](.*?)['"](.*?)/\s*>}mi
+  				  nil
+  				end
+  				item[clean_tag("#{tag}_#{attrib}")] = clean_content(tag, attrib, $3) if $3
 		    else
   				if match[3] =~ %r{<(rss:|atom:)?#{tag}(.*?)>(.*?)</(rss:|atom:)?#{tag}>}mi
   					nil

diff --git a/test/base/base_test.rb b/test/base/base_test.rb
@@ -3,6 +3,7 @@ class BaseTest < Test::Unit::TestCase
 	def setup
 		@rss09 = SimpleRSS.parse open(File.dirname(__FILE__) + '/../data/rss09.rdf')
 		@rss20 = SimpleRSS.parse open(File.dirname(__FILE__) + '/../data/rss20.xml')
+		@media_rss = SimpleRSS.parse open(File.dirname(__FILE__) + '/../data/media_rss.xml')
 		@atom = SimpleRSS.parse open(File.dirname(__FILE__) + '/../data/atom.xml')
 	end
 
@@ -28,6 +29,26 @@ def test_rss09
     assert_equal Time.parse("Fri Sep 09 02:52:31 PDT 2005"), @rss09.channel.dc_date
 	end
 
+	def test_media_rss
+		assert_equal 20, @media_rss.items.size
+		assert_equal "Uploads from herval", @media_rss.title
+		assert_equal "http://www.flickr.com/photos/herval/", @media_rss.channel.link
+		assert_equal "http://www.flickr.com/photos/herval/4671960608/", @media_rss.items.first.link
+		assert_equal "http://www.flickr.com/photos/herval/4671960608/", @media_rss.items.first[:link]
+		assert_equal "http://farm5.static.flickr.com/4040/4671960608_10cb945d5c_o.jpg", @media_rss.items.first.media_content_url
+		assert_equal "image/jpeg", @media_rss.items.first.media_content_type
+		assert_equal "3168", @media_rss.items.first.media_content_height
+		assert_equal "4752", @media_rss.items.first.media_content_width
+		assert_equal "Woof?", @media_rss.items.first.media_title
+		assert_equal "http://farm5.static.flickr.com/4040/4671960608_954d2297bc_s.jpg", @media_rss.items.first.media_thumbnail_url
+		assert_equal "75", @media_rss.items.first.media_thumbnail_height
+		assert_equal "75", @media_rss.items.first.media_thumbnail_width
+		assert_equal "herval", @media_rss.items.first.media_credit
+		assert_equal "photographer", @media_rss.items.first.media_credit_role
+		assert_equal "pets frodo", @media_rss.items.first.media_category
+		assert_equal "urn:flickr:tags", @media_rss.items.first.media_category_scheme
+	end
+
 	def test_rss20
 		assert_equal 10, @rss20.items.size
 		assert_equal "Technoblog", @rss20.title