From 621995f35a5315271ec8e7b0447587113859e1f7 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?R=C3=A9gis=20Hanol?= <regis@hanol.fr>
Date: Thu, 7 Mar 2013 03:30:40 +0100
Subject: [PATCH] globalized wikipedia onebox

---
 lib/oneboxer/wikipedia_onebox.rb              | 11 ++++---
 .../oneboxer/wikipedia_onebox_spec.rb         | 30 +++++++++++++++----
 2 files changed, 29 insertions(+), 12 deletions(-)

diff --git a/lib/oneboxer/wikipedia_onebox.rb b/lib/oneboxer/wikipedia_onebox.rb
index 54457a5bb..51cc54b94 100644
--- a/lib/oneboxer/wikipedia_onebox.rb
+++ b/lib/oneboxer/wikipedia_onebox.rb
@@ -3,7 +3,7 @@ require_dependency 'oneboxer/handlebars_onebox'
 module Oneboxer
   class WikipediaOnebox < HandlebarsOnebox
 
-    matcher /^https?:\/\/.*wikipedia.(com|org)\/.*$/
+    matcher /^https?:\/\/.*wikipedia\.(com|org)\/.*$/
     favicon 'wikipedia.png'
 
     def template
@@ -11,11 +11,10 @@ module Oneboxer
     end
 
     def translate_url
-      m = @url.match(/wiki\/(?<identifier>[^#\/]+)/mi)
-
+      m = @url.match(/^https?:\/\/((?<subdomain>.+)\.)?wikipedia\.(com|org)\/wiki\/(?<identifier>[^#\/]+)/mi)
+      subdomain = m[:subdomain] || "en"
       article_id = CGI::unescape(m[:identifier])
-      return "http://en.m.wikipedia.org/w/index.php?title=#{URI::encode(article_id)}"
-      @url
+      "http://#{subdomain}.m.wikipedia.org/w/index.php?title=#{URI::encode(article_id)}"
     end
 
     def parse(data)
@@ -25,7 +24,7 @@ module Oneboxer
       result = {}
 
       title = html_doc.at('title').inner_html
-      result[:title] = title.gsub!(/ - Wikipedia, the free encyclopedia/, '') if title.present?
+      result[:title] = title.gsub!(/ - Wikipedia.*$/, '') if title.present?
 
       # get the first image > 150 pix high
       images = html_doc.search("img").select { |img| img['height'].to_i > 150 }
diff --git a/spec/components/oneboxer/wikipedia_onebox_spec.rb b/spec/components/oneboxer/wikipedia_onebox_spec.rb
index b27aa7b71..43d2e8055 100644
--- a/spec/components/oneboxer/wikipedia_onebox_spec.rb
+++ b/spec/components/oneboxer/wikipedia_onebox_spec.rb
@@ -5,14 +5,32 @@ require 'oneboxer'
 require 'oneboxer/wikipedia_onebox'
 
 describe Oneboxer::WikipediaOnebox do
-  before(:each) do
-    @o = Oneboxer::WikipediaOnebox.new("http://en.wikipedia.org/wiki/Ruby")
-    FakeWeb.register_uri(:get, @o.translate_url, :response => fixture_file('oneboxer/wikipedia.response'))
-    FakeWeb.register_uri(:get, 'http://en.m.wikipedia.org/wiki/Ruby', :response => fixture_file('oneboxer/wikipedia_redirected.response'))
-  end
 
   it "generates the expected onebox for Wikipedia" do
-    @o.onebox.should == expected_wikipedia_result
+    o = Oneboxer::WikipediaOnebox.new('http://en.wikipedia.org/wiki/Ruby')
+    FakeWeb.register_uri(:get, o.translate_url, :response => fixture_file('oneboxer/wikipedia.response'))
+    FakeWeb.register_uri(:get, 'http://en.m.wikipedia.org/wiki/Ruby', :response => fixture_file('oneboxer/wikipedia_redirected.response'))
+    o.onebox.should == expected_wikipedia_result
+  end
+
+  it "accepts .com extention" do
+    o = Oneboxer::WikipediaOnebox.new('http://en.wikipedia.com/wiki/Postgres')
+    o.translate_url.should == 'http://en.m.wikipedia.org/w/index.php?title=Postgres'
+  end
+
+  it "encodes identifier" do
+    o = Oneboxer::WikipediaOnebox.new('http://en.wikipedia.com/wiki/Café')
+    o.translate_url.should == 'http://en.m.wikipedia.org/w/index.php?title=Caf%C3%A9'
+  end
+
+  it "defaults to en locale" do
+    o = Oneboxer::WikipediaOnebox.new('http://wikipedia.org/wiki/Ruby_on_rails')
+    o.translate_url.should == 'http://en.m.wikipedia.org/w/index.php?title=Ruby_on_rails'
+  end
+
+  it "generates localized url" do
+    o = Oneboxer::WikipediaOnebox.new('http://fr.wikipedia.org/wiki/Ruby')
+    o.translate_url.should == 'http://fr.m.wikipedia.org/w/index.php?title=Ruby'
   end
 
 private