Resolve urls.

config/initializers/05-string.rb

@@ -1,3 +1,6 @@
+require "net/http"
+require "uri"
+
 class String
   def to_line
     self.gsub("\n", " ")
@@ -32,4 +35,59 @@ def url_ext
     uri = URI.parse(self)
     File.extname(uri.path)
   end
+
+  def normalize_url
+    uri = URI.parse(self)
+    port = uri.port if (uri.scheme == "http" and uri.port != 80) or (uri.scheme == "https" and uri.port != 443)
+    URI::HTTP.new(uri.scheme.downcase, uri.userinfo, uri.host.downcase, port, uri.registry, uri.path, uri.opaque, uri.query, uri.fragment).to_s
+  end
+
+  def short_host
+    uri = URI.parse(self)
+    if uri.host[0..3] == "www."
+      uri.host[4..-1]
+    else
+      uri.host
+    end
+  end
+
+  def resolve_url
+    url = self.normalize_url
+    dest = $redis.hget("urls", url)
+    return dest if dest
+
+    dest = url
+    catch :done do
+      5.times do
+        uri = URI.parse(dest)
+        Net::HTTP.start(uri.host, uri.port, use_ssl: uri.scheme == "https") do |http|
+          response = http.head(uri.request_uri)
+          case response
+          when Net::HTTPRedirection then
+            if %w[
+              ://www.youtube.com/das_captcha
+              ://www.nytimes.com/glogin
+              ://www.facebook.com/unsupportedbrowser
+              ://play.spotify.com/error/browser-not-supported.php
+              ://www.linkedin.com/uas/login
+              ://www.theaustralian.com.au/remote/check_cookie.html
+            ].any? { |s| response["location"].include?(s) }
+              throw :done
+            end
+            if response["location"][0] == "/"
+              uri = URI.parse(url)
+              dest = uri.scheme + "://" + uri.host + response["location"]
+            else
+              dest = response["location"]
+            end
+          else
+            throw :done
+          end
+        end
+      end
+    end
+
+    $redis.hset("urls", url, dest)
+    dest
+  end
 end

views/twitter_feed.erb

@@ -6,16 +6,35 @@
   <link href="<%= request.original_url.esc %>" rel="self" />
   <link href="https://twitter.com/<%= @username %>" rel="alternate" />
   <updated><%= Time.parse(@data[0]["created_at"]) if @data[0] %></updated>
-<%- @data.each do |tweet| -%>
+<%-
+@data.each do |tweet|
+  if tweet["retweeted_status"]
+    text = "RT #{tweet["retweeted_status"]["user"]["screen_name"]}: #{tweet["retweeted_status"]["text"]}"
+    entities = tweet["retweeted_status"]["entities"]["urls"]
+  else
+    text = tweet["text"]
+    entities = tweet["entities"]["urls"]
+  end
+  for entity in entities
+    text.gsub!(entity["url"], entity["expanded_url"])
+  end
+  body = text.dup
+  title = text.dup
+  URI.extract(-text, %w[http https]) do |url|
+    dest = url.resolve_url
+    title.gsub!(url, "[#{dest.short_host}]")
+    body.gsub!(url, "<a href='#{dest}' title='#{url}' rel='noreferrer'>#{dest}</a>")
+  end
+-%>
 
   <entry>
     <id>twitter:tweet:<%= tweet["id_str"] %><%= ":#{params[:cachebuster]}" if params[:cachebuster] %></id>
-    <title><%= tweet["text"].to_line.esc %></title>
+    <title><%= title.to_line.esc %></title>
     <link href="https://twitter.com/<%= @username %>/status/<%= tweet["id_str"] %>" />
     <updated><%= Time.parse(tweet["created_at"]) %></updated>
     <author><name><%= @username %></name></author>
     <content type="html">
-<%= tweet["text"].esc %>
+<%= body.esc %>
     </content>
   </entry>
 <%- end -%>