rgabo-readability 0.1.3 → 0.1.4

data/VERSION

@@ -1 +1 @@
-0.1.3
+0.1.4

data/lib/readability/harmonizable.rb

@@ -35,8 +35,19 @@ module Readability
     end
     
     def harmony_page
+      html = self.to_html
+
+      # unless encoding is already utf[-8], convert using iconv
+      unless meta_encoding =~ /utf/i
+        begin
+          require 'iconv'
+          html = Iconv.new("UTF8", "LATIN1//TRANSLIT//IGNORE").iconv(self.to_html)
+        rescue 
+        end
+      end
+      
       # load document into a page
-      page = Harmony::Page.new(self.to_html)
+      page = Harmony::Page.new(html)
       
       # yield the page and reparse if a block is given
       if block_given?

data/readability.gems

@@ -1,30 +1,8 @@
 # readability.gems generated gem export file. Note that any env variable settings will be missing. Append these after using a ';' field separator
 
-# nokogiri
 nokogiri -v1.4.1
-
-# harmony (johnson & envjs)
-stackdeck -v0.2.0
-johnson -v2.0.0.pre3
-envjs -v0.3.1
 harmony -v0.5.5
-
-# tomdoc
-hoe -v2.6.0
-ParseTree -v3.0.5
-RubyInline -v3.7.0
-ruby_parser -v2.0.4
-sexp_processor -v3.0.4
-colored -v1.2
 tomdoc -v0.1.0
-
-# jeweler
-gemcutter -v0.5.0
-git -v1.2.5
-json_pure -v1.4.3
-rubyforge -v2.0.4
 jeweler -v1.4.0
-
-# rspec
 rspec -v1.3.0
 syntax -v1.0.0

data/rgabo-readability.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{rgabo-readability}
-  s.version = "0.1.3"
+  s.version = "0.1.4"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Gabor Ratky"]
-  s.date = %q{2010-05-17}
+  s.date = %q{2010-05-19}
   s.description = %q{Extends Nokogiri::HTML::Document to run Arc90's Readability script and produces easy to read HTML documents.}
   s.email = %q{rgabo@rgabostyle.com}
   s.extra_rdoc_files = [

data/spec/readability/readable_spec.rb

@@ -72,16 +72,28 @@ describe Readability::Readable do
 end
 
 describe "Readability.js" do
-  it "should not fail on any article" do
-    urls = YAML.load(File.open(File.join(File.dirname(__FILE__), 'urls.yaml')))
-    
-    urls.each do |url|
-      # load webpage
-      @doc = Nokogiri::HTML(open(url))
+  class << self
+    def should_not_fail_on url
+      self.class_eval <<-EOF
+      it "should not fail on #{url}" do
+        @doc = Nokogiri::HTML(open("#{url}"))
+
+        # run readability in place
+        @doc.to_readable!
       
+        @doc.to_html.should include('Readability version 1.5.0')
+      end
+      EOF
+    end
+  end
+  
+  YAML.load(File.open(File.join(File.dirname(__FILE__), 'urls.yaml'))).each do |url|
+    it "should not fail on #{url}" do
+      @doc = Nokogiri::HTML(open("#{url}"))
+
       # run readability in place
       @doc.to_readable!
-      
+    
       @doc.to_html.should include('Readability version 1.5.0')
     end
   end

data/spec/readability/urls.yaml

@@ -1,8 +1,9 @@
 # Array of URLs to test readability with
- - 'http://techcrunch.com/2010/05/16/groupon-invades-europe-with-acquisition-of-citydeal'
  - 'http://blogs.wsj.com/venturecapital/2010/05/14/despite-short-term-improvement-vc-10-year-index-goes-negative/?mod=rss_WSJBlog'
  - 'http://www.engadget.com/2010/04/03/entelligence-the-ipad-as-a-productivity-tool/'
-# - 'http://www.huffingtonpost.com/2010/05/15/delete-facebook-account-q_n_576956.html'
+ - 'http://www.huffingtonpost.com/2010/05/15/delete-facebook-account-q_n_576956.html'
  - 'http://mashable.com/2010/05/16/in-defense-of-facebook/'
  - 'http://feeds.venturebeat.com/~r/Venturebeat/~3/rYPVBROMiEI/'
- - 'http://eu.techcrunch.com/2010/04/25/new-eu-rules-could-kill-off-european-vc-and-screw-startups-lets-stop-them/' 
+ - 'http://eu.techcrunch.com/2010/04/25/new-eu-rules-could-kill-off-european-vc-and-screw-startups-lets-stop-them/' 
+ - 'http://www.j-learning.org/promote_it/page/surveys/'
+ 

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: rgabo-readability
 version: !ruby/object:Gem::Version 
-  hash: 29
+  hash: 19
   prerelease: false
   segments: 
   - 0
   - 1
-  - 3
-  version: 0.1.3
+  - 4
+  version: 0.1.4
 platform: ruby
 authors: 
 - Gabor Ratky
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-05-17 00:00:00 +02:00
+date: 2010-05-19 00:00:00 +02:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency