feedzirra 0.1.1 → 0.1.2

data/lib/feedzirra/feed.rb

@@ -46,7 +46,7 @@ module Feedzirra
     # === Returns
     # A array of class names.
     def self.feed_classes
-      @feed_classes ||= [Feedzirra::Parser::RSS, Feedzirra::Parser::AtomFeedBurner, Feedzirra::Parser::Atom]
+      @feed_classes ||= [Feedzirra::Parser::RSSFeedBurner, Feedzirra::Parser::RSS, Feedzirra::Parser::AtomFeedBurner, Feedzirra::Parser::Atom]
     end
     
     # Makes all registered feeds types look for the passed in element to parse.
@@ -129,6 +129,7 @@ module Feedzirra
       curl.proxy_port = options[:proxy_port] if options.has_key?(:proxy_port)
       curl.max_redirects = options[:max_redirects] if options[:max_redirects]
       curl.timeout = options[:timeout] if options[:timeout]
+      curl.ssl_verify_host = options[:ssl_verify_host] if options.has_key?(:ssl_verify_host)
 
       curl.follow_location = true
     end
@@ -248,7 +249,7 @@ module Feedzirra
       end
     
       multi.perform
-      responses.size == 1 ? responses.values.first : responses.values
+      responses.is_a?(Array)? responses.values : responses.values.first
     end
     
     # An abstraction for adding a feed by URL to the passed Curb::multi stack.
@@ -325,6 +326,7 @@ module Feedzirra
       easy = Curl::Easy.new(feed.feed_url) do |curl|
         setup_easy curl, options
         curl.headers["If-Modified-Since"] = feed.last_modified.httpdate if feed.last_modified
+        curl.headers["If-Modified-Since"] = options[:if_modified_since] if options[:if_modified_since] && (!feed.last_modified || (Time.parse(options[:if_modified_since].to_s) > feed.last_modified))
         curl.headers["If-None-Match"]     = feed.etag if feed.etag
 
         curl.on_success do |c|

data/lib/feedzirra/parser.rb

@@ -1,8 +1,10 @@
 module Feedzirra
   module Parser
-    autoload :RSS,      'feedzirra/parser/rss'
-    autoload :RSSEntry, 'feedzirra/parser/rss_entry'
-
+    autoload :RSS,                'feedzirra/parser/rss'
+    autoload :RSSEntry,           'feedzirra/parser/rss_entry'
+    autoload :RSSFeedBurner,      'feedzirra/parser/rss_feed_burner'
+    autoload :RSSFeedBurnerEntry, 'feedzirra/parser/rss_feed_burner_entry'
+    
     autoload :ITunesRSS,      'feedzirra/parser/itunes_rss'
     autoload :ITunesRSSItem,  'feedzirra/parser/itunes_rss_item'
     autoload :ITunesRSSOwner, 'feedzirra/parser/itunes_rss_owner'

data/lib/feedzirra/parser/atom_feed_burner.rb

@@ -12,7 +12,7 @@ module Feedzirra
       elements :entry, :as => :entries, :class => AtomFeedBurnerEntry
 
       def self.able_to_parse?(xml) #:nodoc:
-        ((/Atom/ =~ xml) && (/feedburner/ =~ xml)) || false
+        ((/Atom/ =~ xml) && (/feedburner/ =~ xml) && !(/\<rss|\<rdf/ =~ xml)) || false
       end
     end
 

data/lib/feedzirra/parser/rss.rb

@@ -13,7 +13,7 @@ module Feedzirra
       attr_accessor :feed_url
 
       def self.able_to_parse?(xml) #:nodoc:
-        /\<rss|\<rdf/ =~ xml
+        (/\<rss|\<rdf/ =~ xml) && !(/feedburner/ =~ xml)
       end
     end
 

data/lib/feedzirra/parser/rss_feed_burner.rb

@@ -0,0 +1,22 @@
+module Feedzirra
+
+  module Parser
+    # Parser for dealing with RSS feeds.
+    class RSSFeedBurner
+      include SAXMachine
+      include FeedUtilities
+      element :title
+      element :description
+      element :link, :as => :url
+      elements :item, :as => :entries, :class => RSSFeedBurnerEntry
+
+      attr_accessor :feed_url
+
+      def self.able_to_parse?(xml) #:nodoc:
+        (/\<rss|\<rdf/ =~ xml) && (/feedburner/ =~ xml)
+      end
+    end
+
+  end
+
+end

data/lib/feedzirra/parser/rss_feed_burner_entry.rb

@@ -0,0 +1,40 @@
+module Feedzirra
+
+  module Parser
+    # Parser for dealing with RDF feed entries.
+    class RSSFeedBurnerEntry
+        include SAXMachine
+        include FeedEntryUtilities
+
+        element :title
+
+        element :"feedburner:origLink", :as => :url
+        element :link, :as => :url
+
+        element :"dc:creator", :as => :author
+        element :author, :as => :author
+        element :"content:encoded", :as => :content
+        element :description, :as => :summary
+
+        element :pubDate, :as => :published
+        element :pubdate, :as => :published
+        element :"dc:date", :as => :published
+        element :"dc:Date", :as => :published
+        element :"dcterms:created", :as => :published
+
+
+        element :"dcterms:modified", :as => :updated
+        element :issued, :as => :published
+        elements :category, :as => :categories
+
+        element :guid, :as => :entry_id
+
+        def url
+          @url || @link
+        end
+      
+    end
+
+  end
+
+end

data/lib/feedzirra/version.rb

@@ -1,3 +1,3 @@
 module Feedzirra
-  VERSION = '0.1.1'
+  VERSION = '0.1.2'
 end

data/spec/feedzirra/feed_spec.rb

@@ -90,9 +90,9 @@ describe Feedzirra::Feed do
 
     it "should parse an feedburner rss feed" do
       feed = Feedzirra::Feed.parse(sample_rss_feed_burner_feed)
-      feed.title.should == "Sam Harris: Author, Philosopher, Essayist, Atheist"
-      feed.entries.first.published.to_s.should == "Tue Jan 13 17:20:28 UTC 2009"
-      feed.entries.size.should == 10
+      feed.title.should == "TechCrunch"
+      feed.entries.first.published.to_s.should == "Wed Nov 02 17:25:27 UTC 2011"
+      feed.entries.size.should == 20
     end
   end
 
@@ -109,8 +109,8 @@ describe Feedzirra::Feed do
       Feedzirra::Feed.determine_feed_parser_for_xml(sample_rdf_feed).should == Feedzirra::Parser::RSS
     end
 
-    it "should return the Feedzirra::Parser::RSS class for an rss feedburner feed" do
-      Feedzirra::Feed.determine_feed_parser_for_xml(sample_rss_feed_burner_feed).should == Feedzirra::Parser::RSS
+    it "should return the Feedzirra::Parser::RSSFeedBurner class for an rss feedburner feed" do
+      Feedzirra::Feed.determine_feed_parser_for_xml(sample_rss_feed_burner_feed).should == Feedzirra::Parser::RSSFeedBurner
     end
 
     it "should return the Feedzirra::Parser::RSS object for an rss 2.0 feed" do
@@ -254,7 +254,7 @@ describe Feedzirra::Feed do
 
     describe "#add_url_to_multi" do
       before(:each) do
-        @multi = Curl::Multi.new
+        @multi = Curl::Multi.get([@paul_feed[:url]], {:follow_location => true}, {:pipeline => true})
         @multi.stub!(:add)
         @easy_curl = Curl::Easy.new(@paul_feed[:url])
         
@@ -396,7 +396,7 @@ describe Feedzirra::Feed do
 
     describe "#add_feed_to_multi" do
       before(:each) do
-        @multi = Curl::Multi.new
+        @multi = Curl::Multi.get([@paul_feed[:url]], {:follow_location => true}, {:pipeline => true})
         @multi.stub!(:add)
         @easy_curl = Curl::Easy.new(@paul_feed[:url])
         @feed = Feedzirra::Feed.parse(sample_feedburner_atom_feed)
@@ -414,7 +414,13 @@ describe Feedzirra::Feed do
         @easy_curl.headers["User-Agent"].should == Feedzirra::Feed::USER_AGENT
       end
 
-      it "should set if modified since as an option if passed"
+      it "should set if modified since as an option if passed" do
+        modified_time = Time.parse("Wed, 28 Jan 2009 04:10:32 GMT")
+        Feedzirra::Feed.add_feed_to_multi(@multi, @feed, [], {}, {:if_modified_since => modified_time})
+        modified_time.should be > @feed.last_modified
+        
+        @easy_curl.headers["If-Modified-Since"].should == modified_time
+      end
       
       it 'should set follow location to true' do
         @easy_curl.should_receive(:follow_location=).with(true)
@@ -528,6 +534,14 @@ describe Feedzirra::Feed do
       it 'should slice the feeds into groups of thirty for processing'
       it "should return a feed object if a single feed is passed in"
       it "should return an return an array of feed objects if multiple feeds are passed in"
+      
+      it "should set if modified since as an option if passed" do
+        modified_time = Time.parse("Wed, 28 Jan 2009 04:10:32 GMT")
+        Feedzirra::Feed.should_receive(:add_url_to_multi).with(anything, anything, anything, anything, {:if_modified_since => modified_time}).any_number_of_times
+        
+        @feed = Feedzirra::Feed.fetch_and_parse(sample_feedburner_atom_feed, {:if_modified_since => modified_time})
+      end
+      
     end
 
     describe "#decode_content" do

data/spec/feedzirra/parser/atom_feed_burner_spec.rb

@@ -13,6 +13,10 @@ describe Feedzirra::Parser::AtomFeedBurner do
     it "should return false for a regular atom feed" do
       Feedzirra::Parser::AtomFeedBurner.should_not be_able_to_parse(sample_atom_feed)
     end
+    
+    it "should return false for an rss feedburner feed" do
+      Feedzirra::Parser::AtomFeedBurner.should_not be_able_to_parse(sample_rss_feed_burner_feed)
+    end
   end
 
   describe "parsing" do

data/spec/feedzirra/parser/atom_spec.rb

@@ -9,6 +9,10 @@ describe Feedzirra::Parser::Atom do
     it "should return false for an rdf feed" do
       Feedzirra::Parser::Atom.should_not be_able_to_parse(sample_rdf_feed)
     end
+    
+    it "should return false for an rss feedburner feed" do
+      Feedzirra::Parser::Atom.should_not be_able_to_parse(sample_rss_feed_burner_feed)
+    end
   end
   
   describe "parsing" do

data/spec/feedzirra/parser/itunes_rss_spec.rb

@@ -9,6 +9,10 @@ describe Feedzirra::Parser::ITunesRSS do
     it "should return fase for an atom feed" do
       Feedzirra::Parser::ITunesRSS.should_not be_able_to_parse(sample_atom_feed)
     end
+    
+    it "should return false for an rss feedburner feed" do
+      Feedzirra::Parser::ITunesRSS.should_not be_able_to_parse(sample_rss_feed_burner_feed)
+    end
   end
 
   describe "parsing" do

data/spec/feedzirra/parser/rss_feed_burner_entry_spec.rb

@@ -0,0 +1,85 @@
+# coding: utf-8
+require File.join(File.dirname(__FILE__), %w[.. .. spec_helper])
+
+describe Feedzirra::Parser::RSSFeedBurnerEntry do
+  before(:each) do
+    # I don't really like doing it this way because these unit test should only rely on RSSEntry,
+    # but this is actually how it should work. You would never just pass entry xml straight to the AtomEnry
+    @entry = Feedzirra::Parser::RSSFeedBurner.parse(sample_rss_feed_burner_feed).entries.first
+  end
+
+  after(:each) do
+    # We change the title in one or more specs to test []=
+    if @entry.title != "Angie’s List Sets Price Range IPO At $11 To $13 Per Share; Valued At Over $600M"
+      @entry.title = Feedzirra::Parser::RSS.parse(sample_rss_feed_burner_feed).entries.first.title
+    end
+  end
+
+  it "should parse the title" do
+    @entry.title.should == "Angie’s List Sets Price Range IPO At $11 To $13 Per Share; Valued At Over $600M"
+  end
+
+  it "should parse the original url" do
+    @entry.url.should == "http://techcrunch.com/2011/11/02/angies-list-prices-ipo-at-11-to-13-per-share-valued-at-over-600m/"
+  end
+  
+  it "should parse the author" do
+    @entry.author.should == "Leena Rao"
+  end
+
+  it "should parse the content" do
+    @entry.content.should == sample_rss_feed_burner_entry_content
+  end
+
+  it "should provide a summary" do
+    @entry.summary.should == sample_rss_feed_burner_entry_description
+  end
+
+  it "should parse the published date" do
+    @entry.published.to_s.should == "Wed Nov 02 17:25:27 UTC 2011"
+  end
+
+  it "should parse the categories" do
+    @entry.categories.should == ["TC", "angie\\'s list"]
+  end
+
+  it "should parse the guid as id" do
+    @entry.id.should == "http://techcrunch.com/?p=446154"
+  end
+
+  it "should support each" do
+    @entry.respond_to? :each
+  end
+
+  it "should be able to list out all fields with each" do
+    all_fields = []
+    @entry.each do |field, value|
+      all_fields << field
+    end
+    all_fields.sort == ['author', 'categories', 'content', 'id', 'published', 'summary', 'title', 'url']
+  end
+
+  it "should be able to list out all values with each" do
+    title_value = ''
+    @entry.each do |field, value|
+      title_value = value if field == 'title'
+    end
+    title_value.should == "Angie’s List Sets Price Range IPO At $11 To $13 Per Share; Valued At Over $600M"
+  end
+
+  it "should support checking if a field exists in the entry" do
+    @entry.include?('title') && @entry.include?('author')
+  end
+
+  it "should allow access to fields with hash syntax" do
+    @entry['title'] == @entry.title
+    @entry['title'].should == "Angie’s List Sets Price Range IPO At $11 To $13 Per Share; Valued At Over $600M"
+    @entry['author'] == @entry.author
+    @entry['author'].should == "Leena Rao"
+  end
+
+  it "should allow setting field values with hash syntax" do
+    @entry['title'] = "Foobar"
+    @entry.title.should == "Foobar"
+  end
+end

data/spec/feedzirra/parser/rss_feed_burner_spec.rb

@@ -0,0 +1,52 @@
+require File.join(File.dirname(__FILE__), %w[.. .. spec_helper])
+
+describe Feedzirra::Parser::RSSFeedBurner do
+  describe "#will_parse?" do
+    it "should return true for a feedburner rss feed" do
+      Feedzirra::Parser::RSSFeedBurner.should be_able_to_parse(sample_rss_feed_burner_feed)
+    end
+    
+    it "should return false for a regular RSS feed" do
+       Feedzirra::Parser::RSSFeedBurner.should_not be_able_to_parse(sample_rss_feed)
+     end
+    
+    it "should return false for a feedburner atom feed" do
+      Feedzirra::Parser::RSSFeedBurner.should_not be_able_to_parse(sample_feedburner_atom_feed)
+    end
+    
+    it "should return false for an rdf feed" do
+      Feedzirra::Parser::RSSFeedBurner.should_not be_able_to_parse(sample_rdf_feed)
+    end
+    
+    it "should return false for a regular atom feed" do
+      Feedzirra::Parser::RSSFeedBurner.should_not be_able_to_parse(sample_atom_feed)
+    end
+  end
+
+  describe "parsing" do
+    before(:each) do
+      @feed = Feedzirra::Parser::RSSFeedBurner.parse(sample_rss_feed_burner_feed)
+    end
+    
+    it "should parse the title" do
+      @feed.title.should == "TechCrunch"
+    end
+
+    it "should parse the description" do
+      @feed.description.should == "TechCrunch is a group-edited blog that profiles the companies, products and events defining and transforming the new web."
+    end
+
+    it "should parse the url" do
+      @feed.url.should == "http://techcrunch.com"
+    end
+    
+    it "should provide an accessor for the feed_url" do
+      @feed.respond_to?(:feed_url).should == true
+      @feed.respond_to?(:feed_url=).should == true
+    end
+    
+    it "should parse entries" do
+      @feed.entries.size.should == 20
+    end
+  end
+end

data/spec/feedzirra/parser/rss_spec.rb

@@ -11,9 +11,13 @@ describe Feedzirra::Parser::RSS do
     #   Feedzirra::RSS.should_not be_able_to_parse(sample_rdf_feed)
     # end
     
-    it "should return fase for an atom feed" do
+    it "should return false for an atom feed" do
       Feedzirra::Parser::RSS.should_not be_able_to_parse(sample_atom_feed)
     end
+    
+    it "should return false for an rss feedburner feed" do
+      Feedzirra::Parser::RSS.should_not be_able_to_parse(sample_rss_feed_burner_feed)
+    end
   end
 
   describe "parsing" do

data/spec/spec_helper.rb

@@ -34,7 +34,15 @@ def sample_rdf_entry_content
 end
 
 def sample_rss_feed_burner_feed
-  load_sample("SamHarrisAuthorPhilosopherEssayistAtheist.xml")
+  load_sample("TechCrunch.xml")
+end
+
+def sample_rss_feed_burner_entry_content
+  load_sample("TechCrunchFirstEntry.xml")
+end
+
+def sample_rss_feed_burner_entry_description
+  load_sample("TechCrunchFirstEntryDescription.xml")
 end
 
 def sample_rss_feed

metadata

@@ -1,13 +1,12 @@
 --- !ruby/object:Gem::Specification 
 name: feedzirra
 version: !ruby/object:Gem::Version 
-  hash: 25
-  prerelease: 
+  prerelease: false
   segments: 
   - 0
   - 1
-  - 1
-  version: 0.1.1
+  - 2
+  version: 0.1.2
 platform: ruby
 authors: 
 - Paul Dix
@@ -16,7 +15,8 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-09-30 00:00:00 Z
+date: 2011-09-30 00:00:00 -04:00
+default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: nokogiri
@@ -24,9 +24,8 @@ dependencies:
   requirement: &id001 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
-    - - ~>
+    - - ">="
       - !ruby/object:Gem::Version 
-        hash: 15
         segments: 
         - 1
         - 4
@@ -42,7 +41,6 @@ dependencies:
     requirements: 
     - - ~>
       - !ruby/object:Gem::Version 
-        hash: 27
         segments: 
         - 0
         - 1
@@ -58,7 +56,6 @@ dependencies:
     requirements: 
     - - ~>
       - !ruby/object:Gem::Version 
-        hash: 29
         segments: 
         - 0
         - 7
@@ -72,9 +69,8 @@ dependencies:
   requirement: &id004 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
-    - - ~>
+    - - ">="
       - !ruby/object:Gem::Version 
-        hash: 15
         segments: 
         - 2
         - 1
@@ -88,9 +84,8 @@ dependencies:
   requirement: &id005 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
-    - - ">="
+    - - ~>
       - !ruby/object:Gem::Version 
-        hash: 23
         segments: 
         - 3
         - 0
@@ -106,7 +101,6 @@ dependencies:
     requirements: 
     - - ~>
       - !ruby/object:Gem::Version 
-        hash: 31
         segments: 
         - 1
         - 2
@@ -122,7 +116,6 @@ dependencies:
     requirements: 
     - - ~>
       - !ruby/object:Gem::Version 
-        hash: 23
         segments: 
         - 3
         - 8
@@ -137,12 +130,11 @@ dependencies:
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
-        hash: 63
         segments: 
         - 0
-        - 9
-        - 2
-        version: 0.9.2
+        - 8
+        - 7
+        version: 0.8.7
   type: :runtime
   version_requirements: *id008
 - !ruby/object:Gem::Dependency 
@@ -153,7 +145,6 @@ dependencies:
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
-        hash: 11
         segments: 
         - 0
         - 5
@@ -169,7 +160,6 @@ dependencies:
     requirements: 
     - - ~>
       - !ruby/object:Gem::Version 
-        hash: 23
         segments: 
         - 2
         - 6
@@ -201,6 +191,8 @@ files:
 - lib/feedzirra/parser/itunes_rss_owner.rb
 - lib/feedzirra/parser/rss.rb
 - lib/feedzirra/parser/rss_entry.rb
+- lib/feedzirra/parser/rss_feed_burner.rb
+- lib/feedzirra/parser/rss_feed_burner_entry.rb
 - lib/feedzirra/parser.rb
 - lib/feedzirra/version.rb
 - lib/feedzirra.rb
@@ -220,12 +212,15 @@ files:
 - spec/feedzirra/parser/itunes_rss_owner_spec.rb
 - spec/feedzirra/parser/itunes_rss_spec.rb
 - spec/feedzirra/parser/rss_entry_spec.rb
+- spec/feedzirra/parser/rss_feed_burner_entry_spec.rb
+- spec/feedzirra/parser/rss_feed_burner_spec.rb
 - spec/feedzirra/parser/rss_spec.rb
 - spec/sample_feeds/run_against_sample.rb
 - spec/spec_helper.rb
 - README.rdoc
 - Rakefile
 - .rspec
+has_rdoc: true
 homepage: http://github.com/pauldix/feedzirra
 licenses: []
 
@@ -239,7 +234,6 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements: 
   - - ">="
     - !ruby/object:Gem::Version 
-      hash: 3
       segments: 
       - 0
       version: "0"
@@ -248,14 +242,13 @@ required_rubygems_version: !ruby/object:Gem::Requirement
   requirements: 
   - - ">="
     - !ruby/object:Gem::Version 
-      hash: 3
       segments: 
       - 0
       version: "0"
 requirements: []
 
 rubyforge_project: 
-rubygems_version: 1.8.5
+rubygems_version: 1.3.7
 signing_key: 
 specification_version: 3
 summary: A feed fetching and parsing library
@@ -276,6 +269,8 @@ test_files:
 - spec/feedzirra/parser/itunes_rss_owner_spec.rb
 - spec/feedzirra/parser/itunes_rss_spec.rb
 - spec/feedzirra/parser/rss_entry_spec.rb
+- spec/feedzirra/parser/rss_feed_burner_entry_spec.rb
+- spec/feedzirra/parser/rss_feed_burner_spec.rb
 - spec/feedzirra/parser/rss_spec.rb
 - spec/sample_feeds/run_against_sample.rb
 - spec/spec_helper.rb