dash-bees 0.20 → 0.21

data/CHANGELOG

@@ -1,3 +1,11 @@
+2010-09-20 v0.21 Improvements and bug fixes for feed source
+
+Truncates long contents.
+
+Now accepting feed: URLs and long input box.
+
+Handles a few more error cases.
+
 2010-09-20 v0.20 Added data source for Atom feeds
 
 2010-09-20 v0.19 Fixed sometimes missing identity for Github

data/lib/dash-fu/bee.rb

@@ -10,7 +10,7 @@ module DashFu
   # The README covers it all.
   module Bee
 
-    VERSION = "0.20"
+    VERSION = "0.21"
 
     class << self
       attr_accessor :logger
@@ -203,6 +203,32 @@ module DashFu
     end
     alias :h :escape_html
 
+
+    # Truncates HTML without unbalanced elements/entities.
+    def truncate_html(html, limit = 30, ellipsis = "...")
+      # Inspiration: http://pastie.textmate.org/145402
+      doc = Nokogiri::XML(html)
+      ellipsis_length = Nokogiri::XML(ellipsis).text.length
+      content_length = doc.text.length
+      return html if content_length <= limit
+      truncate = lambda do |node, max|
+        break node if node.text.length <= max
+        node.children.each do |child|
+          if max <= 0
+            child.remove
+          elsif child.text?
+            child.content = child.text[0,max]
+            max -= child.content.length
+          else
+            truncate[child, max]
+          end
+        end
+        node
+      end
+      truncate[doc, limit - ellipsis_length].to_html.strip + ellipsis
+    end
+    
+
     # Returns API key for this source. May return string or hash, depending on
     # the API.
     def api_key

data/lib/dash-fu/bees/feed.rb

@@ -12,7 +12,8 @@ module DashFu::Bee
     end
 
     def validate(source)
-      uri = URI.parse(source["url"]) rescue nil
+      unfeed = source["url"].gsub(/^feed:\/\//, "http://").gsub(/^feed:/, "")
+      uri = URI.parse(unfeed) rescue nil
       raise "Not a valid URL" unless uri && uri.absolute?
       raise "Only HTTP/S URLs supported" unless uri.scheme == "http" || uri.scheme == "https"
       begin
@@ -20,11 +21,11 @@ module DashFu::Bee
           code = io.status.first
           raise "Cannot read this feed, got status code #{code}" unless code == "200"
           feed = (Nokogiri::XML(io.read)>"feed").first
-          source["title"] = get_text(feed>"title").strip
-          source["source.name"] = source["title"] if source["title"].length > 2
-          permalink = (feed>"link[rel=alternate]").first
-          source["permalink"] = permalink["href"] if permalink
+          source["source.name"] = source["title"] = get_text(feed>"title").strip
+          alt = (feed>"link[rel=alternate]").first
+          source["permalink"] = alt["href"] if alt
           source["logo"] = (feed>"logo").text
+          source["url"] = uri.to_s 
         end
       rescue
         raise "Cannot read this feed: is it down for you or just for us?"
@@ -43,7 +44,7 @@ module DashFu::Bee
               updated = Time.iso8601((feed>"updated").text) rescue Time.now
               if source["updated"].nil? || updated > source["updated"]
                 (feed>"entry").each do |entry|
-                  published = Time.iso8601((entry>"published").text) rescue Time.now
+                  published = Time.iso8601(entry.css("published,updated").first.text) rescue Time.now
                   break unless source["updated"].nil? || updated >= published
                   if author = (entry>"author").first
                     person = { fullname: (author>"name").text, email: (author>"email").text,
@@ -90,14 +91,15 @@ module DashFu::Bee
       end
     end
 
-    def get_html(elements)
+    def get_html(elements, length = 250)
       element = elements.first
       return unless element
-      case element["type"]
-      when "html" ; Sanitize.clean(element.text, Sanitize::Config::BASIC).strip
-      when "xhtml" ; Sanitize.clean(element.to_xhtml, Sanitize::Config::BASIC).strip
-      when "text"; element.text.strip
+      html = case element["type"]
+      when "html" ; element.text
+      when "xhtml" ; element.to_html
+      when "text"; element.text
       end
+      Sanitize.clean(truncate_html(html.to_s.strip, length), Sanitize::Config::BASIC.merge(output: :html)).strip
     end
 
   end

data/lib/dash-fu/bees/feed.yml

@@ -1,7 +1,7 @@
 en:
   description: "Gather activities from an Atom feed"
   inputs: |-
-    <label>Feed URL <input type="text" name="source[url]" size="30"></label>
+    <label>Feed URL <input type="text" name="source[url]" size="50"></label>
   notes: |-
     <ul>
     </ul>

data/test/cassettes/feed.yml

@@ -81,3 +81,59 @@
           </content>
         </entry>
       </feed>
+- !ruby/struct:VCR::HTTPInteraction 
+  request: !ruby/struct:VCR::Request 
+    method: :get
+    uri: http://example.org:80/update.xml
+  response: !ruby/struct:VCR::Response 
+    status: !ruby/struct:VCR::ResponseStatus 
+      code: 200
+      message: OK
+    body: |-
+      <?xml version="1.0" encoding="utf-8"?>
+      <feed xmlns="http://www.w3.org/2005/Atom">
+        <title type="text">dive into mark</title>
+        <entry>
+          <title>Atom draft-07 snapshot</title>
+          <updated>2003-12-13T08:29:29-04:00</updated>
+        </entry>
+      </feed>
+- !ruby/struct:VCR::HTTPInteraction 
+  request: !ruby/struct:VCR::Request 
+    method: :get
+    uri: http://example.org:80/html.xml
+  response: !ruby/struct:VCR::Response 
+    status: !ruby/struct:VCR::ResponseStatus 
+      code: 200
+      message: OK
+    body: |-
+      <?xml version="1.0" encoding="utf-8"?>
+      <feed xmlns="http://www.w3.org/2005/Atom">
+        <title type="text">dive into mark</title>
+        <entry>
+          <title>Atom draft-07 snapshot</title>
+          <content type="html">&lt;i&gt;HTML &lt;br&gt; content&lt;/i&gt;</content>
+        </entry>
+      </feed>
+- !ruby/struct:VCR::HTTPInteraction 
+  request: !ruby/struct:VCR::Request 
+    method: :get
+    uri: http://example.org:80/truncated.xml
+  response: !ruby/struct:VCR::Response 
+    status: !ruby/struct:VCR::ResponseStatus 
+      code: 200
+      message: OK
+    body: |-
+      <?xml version="1.0" encoding="utf-8"?>
+      <feed xmlns="http://www.w3.org/2005/Atom">
+        <title type="text">dive into mark</title>
+        <entry>
+          <title>Atom draft-07 snapshot</title>
+          <content type="xhtml">
+            <p><em>Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated
+            Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated
+            Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated
+            </em></p>
+          </content>
+        </entry>
+      </feed>

data/test/feed_test.rb

@@ -34,8 +34,19 @@ test DashFu::Bee::Feed do
       source.setup "url"=>"http://example.org/feed.xml"
       assert_equal "The Awesome Feed", source.name
     end
+
+    should "accept feed URL (short form)" do
+      source.setup "url"=>"feed://example.org/feed.xml"
+      assert_equal "dive into mark", source.name
+    end
+
+    should "accept feed URL (long form)" do
+      source.setup "url"=>"feed:http://example.org/feed.xml"
+      assert_equal "dive into mark", source.name
+    end
   end
 
+
   context "update" do
     setup { source.setup "url"=>"http://example.org/feed.xml" }
 
@@ -119,8 +130,52 @@ posted <a href=\"http://example.org/2005/04/02/atom\">Atom draft-07 snapshot</a>
       end
     end
 
+    context "with update but no published date" do
+      setup do
+        source.setup "url"=>"http://example.org/update.xml"
+        source.update
+      end
+      subject { source.activity }
+
+      should "capture published date" do
+        assert_equal Time.parse("2003-12-13T08:29:29-04:00"), subject.timestamp
+      end
+    end
+
+    context "with HTML content" do
+      setup do
+        source.setup "url"=>"http://example.org/html.xml"
+        source.update
+      end
+      subject { source.activity }
+
+      should "capture title and summary" do
+        assert_equal <<-HTML.strip, subject.html
+posted <a href=\"\">Atom draft-07 snapshot</a>:
+<blockquote><i>HTML <br> content</i></blockquote>
+        HTML
+      end
+    end
+
+    context "with truncated content" do
+      setup do
+        source.setup "url"=>"http://example.org/truncated.xml"
+        source.update
+      end
+      subject { source.activity }
+
+      should "capture title and summary" do
+        assert_equal <<-HTML.strip, subject.html
+posted <a href=\"\">Atom draft-07 snapshot</a>:
+<blockquote><p><em>Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated Truncated\n      Truncated Truncated Truncated Truncated Truncated Truncat</em></p>
+    ...</blockquote>
+        HTML
+      end
+    end
+
   end
 
+
   context "meta" do
     setup { source.setup "url"=>"http://example.org/feed.xml" }
     subject { source.meta }
@@ -129,4 +184,5 @@ posted <a href=\"http://example.org/2005/04/02/atom\">Atom draft-07 snapshot</a>
       assert_contains subject, :title=>"Source", :text=>"dive into mark", :url=>"http://example.org/"
     end
   end
+
 end

data/test/test.log

@@ -1008,3 +1008,46 @@ RubyGems: 500
 #<NoMethodError: undefined method `namespaces' for nil:NilClass>
 #<NoMethodError: undefined method `namespaces' for nil:NilClass>
 #<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `[]' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `children=' for #<Nokogiri::XML::Element:0x000001015ba0e8>>
+#<NoMethodError: undefined method `children=' for #<Nokogiri::XML::Element:0x000001012e51b0>>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `>' for nil:NilClass>
+#<NoMethodError: undefined method `>' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>
+#<NoMethodError: undefined method `namespaces' for nil:NilClass>

metadata

@@ -4,8 +4,8 @@ version: !ruby/object:Gem::Version
   prerelease: false
   segments: 
   - 0
-  - 20
-  version: "0.20"
+  - 21
+  version: "0.21"
 platform: ruby
 authors: 
 - Assaf Arkin
@@ -146,7 +146,7 @@ licenses: []
 post_install_message: 
 rdoc_options: 
 - --title
-- DashFu::Bee 0.20
+- DashFu::Bee 0.21
 - --main
 - README.rdoc
 - --webcvs