lib/postrank-uri.rb in postrank-uri-1.0.17 vs lib/postrank-uri.rb in postrank-uri-1.0.18

@@ -1,6 +1,5 @@
-# -*- encoding: utf-8 -*-
 
 require 'addressable/uri'
 require 'digest/md5'
 require 'nokogiri'
 require 'public_suffix'
@@ -32,15 +31,15 @@
 end
 
 module PostRank
   module URI
 
-    c18ndb = YAML.load_file(File.dirname(__FILE__) + '/postrank-uri/c18n.yml')
+    c14ndb = YAML.load_file(File.dirname(__FILE__) + '/postrank-uri/c14n.yml')
 
-    C18N = {}
-    C18N[:global] = c18ndb[:all].freeze
-    C18N[:hosts]  = c18ndb[:hosts].inject({}) {|h,(k,v)| h[/#{Regexp.escape(k)}$/.freeze] = v; h}
+    C14N = {}
+    C14N[:global] = c14ndb[:all].freeze
+    C14N[:hosts]  = c14ndb[:hosts].inject({}) {|h,(k,v)| h[/#{Regexp.escape(k)}$/.freeze] = v; h}
 
     URIREGEX = {}
     URIREGEX[:protocol] = /https?:\/\//i
     URIREGEX[:valid_preceding_chars] = /(?:|\.|[^-\/"':!=A-Z0-9_@＠]|^|\:)/i
     URIREGEX[:valid_domain] = /\b(?:[a-z0-9-]{1,63}\.){1,}[a-z]{2,63}(?::[0-9]+)?/i
@@ -128,17 +127,19 @@
         '%' + $1.unpack('H2' * $1.size).join('%').upcase
       end.gsub(' ','%20')
     end
 
     def unescape(uri)
-      uri.tr('+', ' ').gsub(URIREGEX[:unescape]) do
+      u = parse(uri)
+      u.query = u.query.tr('+', ' ') if u.query
+      u.to_s.gsub(URIREGEX[:unescape]) do
         [$1.delete('%')].pack('H*')
       end
     end
 
     def clean(uri, opts = {})
-      uri = normalize(c18n(unescape(uri), opts))
+      uri = normalize(c14n(unescape(uri), opts))
       opts[:raw] ? uri : uri.to_s
     end
 
     def hash(uri, opts = {})
       Digest::MD5.hexdigest(opts[:clean] == true ? clean(uri) : uri)
@@ -151,17 +152,17 @@
       u.query = nil if u.query && u.query.empty?
       u.fragment = nil
       u
     end
 
-    def c18n(uri, opts = {})
+    def c14n(uri, opts = {})
       u = parse(uri, opts)
       u = embedded(u)
 
       if q = u.query_values(Array)
-        q.delete_if { |k,v| C18N[:global].include?(k) }
-        q.delete_if { |k,v| C18N[:hosts].find {|r,p| u.host =~ r && p.include?(k) } }
+        q.delete_if { |k,v| C14N[:global].include?(k) }
+        q.delete_if { |k,v| C14N[:hosts].find {|r,p| u.host =~ r && p.include?(k) } }
       end
       u.query_values = q
 
       if u.host =~ /^(mobile\.)?twitter\.com$/ && u.fragment && u.fragment.match(/^!(.*)/)
         u.fragment = nil
@@ -222,10 +223,10 @@
 
       is_valid = false
       cleaned_uri = clean(uri, :raw => true)
 
       if host = cleaned_uri.host
-        is_valid = PublicSuffix.valid?(host)
+        is_valid = PublicSuffix.valid?(Addressable::IDNA.to_unicode(host))
       end
 
       is_valid
     end
   end