lib/postrank-uri.rb in postrank-uri-1.0.13 vs lib/postrank-uri.rb in postrank-uri-1.0.14

@@ -16,10 +16,26 @@
       end
 
       dom = dp.public_suffix
       dom = dp.domain.downcase + "." + dom unless dp.domain.empty?
     end
+
+    def normalized_query
+      @normalized_query ||= (begin
+        if self.query && self.query.strip != ''
+          (self.query.strip.split("&", -1).map do |pair|
+            Addressable::URI.normalize_component(
+              pair,
+              Addressable::URI::CharacterClasses::QUERY.sub("\\&", "")
+            )
+          end).join("&")
+        else
+          nil
+        end
+      end)
+    end
+
   end
 end
 
 module PostRank
   module URI
@@ -31,11 +47,11 @@
     C18N[:hosts]  = c18ndb[:hosts].inject({}) {|h,(k,v)| h[/#{Regexp.escape(k)}$/.freeze] = v; h}
 
     URIREGEX = {}
     URIREGEX[:protocol] = /https?:\/\//i
     URIREGEX[:valid_preceding_chars] = /(?:|\.|[^-\/"':!=A-Z0-9_@＠]|^|\:)/i
-    URIREGEX[:valid_domain] = /(?:[^[:punct:]\s][\.-](?=[^[:punct:]\s])|[^[:punct:]\s]){1,}\.[a-z]{2,}(?::[0-9]+)?/i
+    URIREGEX[:valid_domain] = /\b(?:[a-z0-9-]{1,63}\.){1,}[a-z]{2,63}(?::[0-9]+)?/i
     URIREGEX[:valid_general_url_path_chars] = /[a-z0-9!\*';:=\+\,\$\/%#\[\]\-_~]/i
 
     # Allow URL paths to contain balanced parens
     #  1. Used in Wikipedia URLs like /Primer_(film)
     #  2. Used in IIS sessions like /S(dfd346)/
@@ -201,11 +217,10 @@
           end
         end
       end
 
       uri.scheme = 'http' if uri.host && !uri.scheme
-
-      uri.normalize
+      uri.normalize!
     end
 
   end
 end