feedisco 0.1.2 → 0.1.3

data/README.md

@@ -107,6 +107,10 @@ Contribute by forking, tweaking, and sending pull requests, but please add the a
 
 ### History
 
+#### `0.1.3`
+
+* Corrected bug causing some links to appear several times in the result.
+
 #### `0.1.2`
 
 * Supports `StringIO` as a return from the open-uri `open` method (which is returned instead of a `Tempfile` when the content is smaller than a given size - 10KB).

data/VERSION

@@ -1 +1 @@
-0.1.2
+0.1.3

data/feedisco.gemspec

@@ -5,7 +5,7 @@
 
 Gem::Specification.new do |s|
   s.name = "feedisco"
-  s.version = "0.1.2"
+  s.version = "0.1.3"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Romain Champourlier"]

data/lib/feedisco/discovery.rb

@@ -30,23 +30,25 @@ module Feedisco
 
           # Check <link> elements
           doc.css('link').each do |link|
-            feeds << complete_extracted_url(link[:href], harmonized_url) if link[:rel] =~ %r{(alternate|service.feed)}i && Feedisco.feed_content_types.include?(link[:type].downcase.strip)
+            discovery = complete_extracted_url(link[:href], harmonized_url) if link[:rel] =~ %r{(alternate|service.feed)}i && Feedisco.feed_content_types.include?(link[:type].downcase.strip)
+            feeds << discovery unless discovery.nil? or feeds.include? discovery
           end
 
           # Check <a> elements
           doc.css('a').each do |a|
-            if (looks_like_feed?(a[:href]) &&
-               (a[:href] =~ %r{\A/} || a[:href] =~ %r{#{URI.parse(harmonized_url).host}/}) &&
-               !feeds.include?(a[:href]))
+            if looks_like_feed? a[:href] and
+               (a[:href] =~ %r{\A/} or a[:href] =~ %r{#{URI.parse(harmonized_url).host}/})
 
-              feeds << complete_extracted_url(a[:href], harmonized_url)
+              discovery = complete_extracted_url(a[:href], harmonized_url)
+              feeds << discovery unless discovery.nil? or feeds.include?(discovery)
             end
           end
 
           # Check <a> elements again, less restrictively, so we add all discovered feeds even the ones
           # on external domains, but the will come after in the feeds array.
           doc.css('a').each do |a|
-            feeds << complete_extracted_url(a[:href], harmonized_url) if (looks_like_feed?(a[:href]) && !feeds.include?(a[:href]))
+            discovery = complete_extracted_url(a[:href], harmonized_url) if looks_like_feed? a[:href]
+            feeds << discovery unless discovery.nil? or feeds.include?(discovery)
           end
         end
       end

data/lib/feedisco/version.rb

@@ -1,6 +1,6 @@
 class Feedisco
 
   def version
-    '0.1.2'
+    '0.1.3'
   end
 end

data/spec/fixtures/several_links.html

@@ -10,6 +10,7 @@
 <a href="http://example.com/feed.rss">feed.rss</a>
 <a href="http://another.domain.com/feed.rss">another domain's feed.rss</a>
 <a href="http://example.com/feed.rss">feed.rss</a>
+<a href="feed.rss">feed.rss, without the prefix part</a>
 </body>
 
 </html>

data/spec/lib/discovery_spec.rb

@@ -19,16 +19,19 @@ describe "Feedisco::Discovery" do
 
     context 'from fixtures' do
 
-      it 'should return an empty array' do
-        file = File.open(File.join(File.dirname(__FILE__), '..', 'fixtures', 'no_link.html'))
+      def file_for_fixture(fixture_file_name)
+        file = File.open(File.join(File.dirname(__FILE__), '..', 'fixtures', fixture_file_name))
         file.stub!(:class => stub(:to_s => 'Tempfile'), :content_type => 'text/html')
+        file
+      end
 
-        Feedisco.should_receive(:open).and_yield(file)
+      it 'should return an empty array' do
+        Feedisco.should_receive(:open).and_yield(file_for_fixture 'no_link.html')
         Feedisco.find('example.com').should == []
       end
 
       it 'should include the alternate link' do
-        file = File.open(File.join(File.dirname(__FILE__), '..', 'fixtures', 'alternate.html'))
+        file = file_for_fixture 'alternate.html'
         file.stub!(:class => stub(:to_s => 'Tempfile'), :content_type => 'text/html')
 
         Feedisco.should_receive(:open).and_yield(file)
@@ -36,40 +39,35 @@ describe "Feedisco::Discovery" do
       end
 
       it 'should include a <a> link in the body' do
-        file = File.open(File.join(File.dirname(__FILE__), '..', 'fixtures', 'one_link_in_body.html'))
-        file.stub!(:class => stub(:to_s => 'Tempfile'), :content_type => 'text/html')
+        file = file_for_fixture 'one_link_in_body.html'
 
         Feedisco.should_receive(:open).and_yield(file)
         Feedisco.find('example.com').should include("http://example.com/feed.rss")
       end
 
       it 'should include link to feeds on the URL\'s domain' do
-        file = File.open(File.join(File.dirname(__FILE__), '..', 'fixtures', 'several_links.html'))
-        file.stub!(:class => stub(:to_s => 'Tempfile'), :content_type => 'text/html')
+        file = file_for_fixture 'several_links.html'
 
         Feedisco.should_receive(:open).and_yield(file)
         Feedisco.find('example.com').should include("http://example.com/feed.rss")
       end
 
       it 'should have the alternate link as the first of the returned feed' do
-        file = File.open(File.join(File.dirname(__FILE__), '..', 'fixtures', 'several_links.html'))
-        file.stub!(:class => stub(:to_s => 'Tempfile'), :content_type => 'text/html')
+        file = file_for_fixture 'several_links.html'
 
         Feedisco.should_receive(:open).and_yield(file)
         Feedisco.find('example.com').first.should == "http://example.com/feed.xml"
       end
 
       it 'should include link to feeds on other domains' do
-        file = File.open(File.join(File.dirname(__FILE__), '..', 'fixtures', 'several_links.html'))
-        file.stub!(:class => stub(:to_s => 'Tempfile'), :content_type => 'text/html')
+        file = file_for_fixture 'several_links.html'
 
         Feedisco.should_receive(:open).and_yield(file)
         Feedisco.find('example.com').should include "http://another.domain.com/feed.rss"
       end
 
       it 'should have links to feeds on other domains after links to feeds on the same domain' do
-        file = File.open(File.join(File.dirname(__FILE__), '..', 'fixtures', 'several_links.html'))
-        file.stub!(:class => stub(:to_s => 'Tempfile'), :content_type => 'text/html')
+        file = file_for_fixture 'several_links.html'
 
         Feedisco.should_receive(:open).and_yield(file)
         feeds = Feedisco.find('example.com')
@@ -77,13 +75,18 @@ describe "Feedisco::Discovery" do
       end
 
       it 'should include each link only once' do
-        file = File.open(File.join(File.dirname(__FILE__), '..', 'fixtures', 'several_links.html'))
-        file.stub!(:class => stub(:to_s => 'Tempfile'), :content_type => 'text/html')
+        file = file_for_fixture 'several_links.html'
 
         Feedisco.should_receive(:open).and_yield(file)
         feeds = Feedisco.find('example.com')
         feeds.select { |f| f == 'http://example.com/feed.rss' }.count.should == 1
       end
+
+      it "should complete a relative link" do
+        Feedisco.should_receive(:open).and_yield(file_for_fixture 'several_links.html')
+        feeds = Feedisco.find('example.com')
+        feeds.should include 'http://example.com/feed.rss'
+      end
     end
 
     context 'from real websites (this may change so some examples may break)' do

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: feedisco
 version: !ruby/object:Gem::Version
-  version: 0.1.2
+  version: 0.1.3
   prerelease: 
 platform: ruby
 authors:
@@ -104,7 +104,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
       version: '0'
       segments:
       - 0
-      hash: -1864332675888607997
+      hash: 2342868453426803014
 required_rubygems_version: !ruby/object:Gem::Requirement
   none: false
   requirements: