iconoclasm 1.0.0 → 1.0.1

data/VERSION

@@ -1 +1 @@
-1.0.0
+1.0.1

data/lib/iconoclasm/downloader.rb

@@ -1,4 +1,4 @@
-require 'curl'
+require 'typhoeus'
 
 module Iconoclasm
   module Downloader
@@ -13,18 +13,16 @@ module Iconoclasm
       @@user_agent
     end
     
+    def user_agent
+      @@user_agent
+    end
+    
     def get(url)
-      Curl::Easy.http_get(url) do |curl|
-        curl.headers['User-Agent']  = Iconoclasm::Downloader.user_agent
-        curl.follow_location        = true
-      end
+      Typhoeus::Request.get(url, :user_agent => user_agent, :follow_location => true)
     end
     
     def head(url)
-      Curl::Easy.http_head(url) do |curl|
-        curl.headers['User-Agent']  = Iconoclasm::Downloader.user_agent
-      end
+      Typhoeus::Request.head(url, :user_agent => user_agent)
     end
-    
   end
 end

data/lib/iconoclasm/errors.rb

@@ -30,12 +30,20 @@ module Iconoclasm
     end
 
     def http_error_reason
-      @response.respond_to?(:header_str) ? @response.header_str[/(?<=\d{3}\s)(.*)$/].chomp : @response
-    end    
+      @response.respond_to?(:header_str) ? error_reason : @response
+    end
 
     def http_error_message
       "#{@code}: #{http_error_reason}"
-    end    
+    end
+    
+    private
+    
+    def error_reason
+      first_line = @response.header_str.split('\n').first.chomp
+      first_line.match(/\d{3}\s(.*)$/)
+      $1
+    end
   end
   
   class RTFMError < Iconoclasm::Error

data/lib/iconoclasm/extractor.rb

@@ -10,42 +10,41 @@ module Iconoclasm
     
     def extract_favicon_from(url, content = nil)
       catch(:done) do
-        base_url  = base_url_of(url)
-        extract_favicon_from_head_of(base_url, content)
-        extract_favicon_from_naive_guess(base_url)
-        raise Iconoclasm::MissingFavicon.new(base_url)
+        extract_favicon_from_head_of(url, content)
+        extract_favicon_from_naive_guess(base_url_of(url))
+        raise Iconoclasm::MissingFavicon.new(url)
       end
     end
     
     private
     
-    def extract_favicon_from_head_of(base_url, content = nil)
-      if document = document_from(base_url, content)
+    def extract_favicon_from_head_of(url, content = nil)
+      if document = document_from(url, content)
         favicon_links = find_favicon_links_in(document)
         throw(:done, {
-          :url          => href_of(favicon_links.first),
+          :url          => href_of(favicon_links.first, :base_url => base_url_of(url)),
           :content_type => type_of(favicon_links.first)
         }) unless favicon_links.empty?
       end
     end
 
-    def document_from(base_url, content = nil)
+    def document_from(url, content = nil)
       if content
         Nokogiri::XML(content)
       else
-        response = get(base_url)
-        Nokogiri::XML(response.body_str) if response.response_code == 200
+        response = get(url)
+        Nokogiri::XML(response.body) if response.code == 200
       end
     end
     
     def extract_favicon_from_naive_guess(base_url)
       naive_url = "#{base_url}/favicon.ico"
       response  = get(naive_url)
-      headers   = Iconoclasm::Headers.new(response.header_str)
-      if response.response_code == 200
+      headers   = Iconoclasm::Headers.new(response.headers)
+      if response.code == 200
         throw(:done, {
           :url            => naive_url,
-          :content_length => header.content_length,
+          :content_length => headers.content_length,
           :content_type   => headers.content_type,
           :data           => response.body_str
         })
@@ -53,7 +52,7 @@ module Iconoclasm
     end
     
     def find_favicon_links_in(document)
-      document.xpath('//link[favicon_link(.)]', Class.new {
+      document.css('link:favicon_link', Class.new {
         def favicon_link(node_set)
           node_set.find_all { |node| node['rel'] && node['rel'] =~ /^(?:shortcut\s)?icon$/i }
         end
@@ -65,9 +64,15 @@ module Iconoclasm
       "#{uri.scheme}://#{uri.host}"
     end
     
-    def href_of(node)
+    def href_of(node, options = {})
       href = normal_node_attributes(node)['href']
-      href.value if href
+      if href
+        relative?(href.value) ? "#{options[:base_url]}#{href.value}" : href.value
+      end
+    end
+    
+    def relative?(href)
+      href =~ /^[\.\/]/
     end
     
     def type_of(node)

data/lib/iconoclasm/favicon.rb

@@ -14,8 +14,8 @@ module Iconoclasm
       @data         = attributes[:data]
       @name         = attributes[:name]           || parse_name_from(@url)
       headers       = attributes[:headers]
-      @content_type = attributes[:content_type]   || headers ? headers.content_type : nil
-      @size         = attributes[:content_length] || headers ? headers.content_length : nil
+      @content_type = attributes[:content_type]   ? attributes[:content_type] : headers ? headers.content_type : nil
+      @size         = attributes[:content_length] ? attributes[:content_length] : headers ? headers.content_length : nil
       @save_path    = nil
     end
     
@@ -58,8 +58,8 @@ module Iconoclasm
     
     def fetch_data
       response = get(url)
-      if response.response_code == 200
-        response.body_str
+      if response.code == 200
+        response.body
       else
         raise Iconoclasm::HTTPError.new(url, response)
       end

data/spec/iconoclasm/downloader_spec.rb

@@ -10,39 +10,30 @@ describe Iconoclasm::Downloader do
   end
   
   describe "GETting a url" do
-    it "should GET the url using curl easy" do
-      Curl::Easy.expects(:http_get).with(@url)
+    it "should GET the url using Typheous" do
+      Typhoeus::Request.expects(:get).with(@url, instance_of(Hash))
       @thing.get(@url)
     end
     
     it "should set the user agent to the default user agent" do
-      @curl.stubs(:follow_location=)
-      headers = mock('headers')
-      Curl::Easy.stubs(:http_get).yields(@curl)
-      @curl.expects(:headers).returns(headers)
-      headers.expects(:[]=).with('User-Agent', Iconoclasm::Downloader.user_agent)
+      Typhoeus::Request.expects(:get).with(instance_of(String), has_entry(:user_agent => Iconoclasm::Downloader.user_agent))
       @thing.get(@url)
     end
     
     it "should follow redirects" do
-      @curl.stubs(:headers).returns({})
-      Curl::Easy.stubs(:http_get).yields(@curl)
-      @curl.expects(:follow_location=).with(true)
+      Typhoeus::Request.expects(:get).with(instance_of(String), has_entry(:follow_location => true))
       @thing.get(@url)
     end
   end
   
   describe "HEADing a url" do
-    it "should HEAD the url using curl easy" do
-      Curl::Easy.expects(:http_head).with(@url)
+    it "should HEAD the url using Typhoeus" do
+      Typhoeus::Request.expects(:head).with(@url, instance_of(Hash))
       @thing.head(@url)
     end
     
     it "should set the user agent to the default user agent" do
-      headers = mock('headers')
-      Curl::Easy.stubs(:http_head).yields(@curl)
-      @curl.expects(:headers).returns(headers)
-      headers.expects(:[]=).with('User-Agent', Iconoclasm::Downloader.user_agent)
+      Typhoeus::Request.expects(:head).with(instance_of(String), has_entry(:user_agent => Iconoclasm::Downloader.user_agent))
       @thing.head(@url)
     end
   end

data/spec/iconoclasm/extractor_spec.rb

@@ -20,7 +20,7 @@ describe Iconoclasm::Extractor do
     end
     
     it "should try to find the favicon path in the head of the content" do
-      @thing.expects(:extract_favicon_from_head_of).with(@base_url, nil).throws(:done)
+      @thing.expects(:extract_favicon_from_head_of).with(@url, nil).throws(:done)
       @thing.extract_favicon_from(@url)
     end
     
@@ -69,7 +69,7 @@ describe Iconoclasm::Extractor do
     
     describe "when content isn't already provided" do
       before do
-        @response = mock('http response', :response_code => 200, :body_str => "")
+        @response = mock('http response', :code => 200, :body => "")
       end
       
       it "should go get the content" do
@@ -100,7 +100,7 @@ describe Iconoclasm::Extractor do
           before do
             @href = 'http://www.website.com/images/favicon.ico'
             @type = 'image/vnd.microsoft.icon'
-            @thing.expects(:href_of).with(@link).returns(@href)
+            @thing.expects(:href_of).with(@link, instance_of(Hash)).returns(@href)
             @thing.expects(:type_of).with(@link).returns(@type)
             @hash = catch(:done) { @thing.__send__(:extract_favicon_from_head_of, @url, @content) } 
           end

data/spec/iconoclasm/favicon_spec.rb

@@ -149,7 +149,7 @@ describe Iconoclasm::Favicon do
     
     it "should request the icon image" do
       @favicon.expects(:get).returns(@response)
-      @response.stubs(:response_code => 200, :body_str => "IMAGE DATA!")
+      @response.stubs(:code => 200, :body => "IMAGE DATA!")
       @favicon.fetch_data      
     end
     
@@ -157,11 +157,11 @@ describe Iconoclasm::Favicon do
       before do
         @favicon.stubs(:get).returns(@response)
         @data = "THIS IS ALSO TOTALLY SOME IMAGE DATA HAR HAR HAR!"
-        @response.expects(:response_code).returns(200)
+        @response.expects(:code).returns(200)
       end
       
       it "should return the content of the request (the binary image data)" do
-        @response.expects(:body_str).returns(@data)
+        @response.expects(:body).returns(@data)
         @favicon.fetch_data.should == @data
       end
     end
@@ -169,7 +169,7 @@ describe Iconoclasm::Favicon do
     describe "when the HTTP request is not successful" do
       before do
         @favicon.stubs(:get).returns(@response)
-        @response.expects(:response_code).returns(400)
+        @response.expects(:code).returns(400)
       end
       
       it "should raise an HTTP error" do

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: iconoclasm
 version: !ruby/object:Gem::Version 
-  version: 1.0.0
+  version: 1.0.1
 platform: ruby
 authors: 
 - Sander Hartlage