page.rb in mechanize-0.9.3

- old
+ new

@@ -19,24 +19,32 @@
     #
     class Page < WWW::Mechanize::File
       extend Forwardable
 
       attr_accessor :mech
-      attr_accessor :encoding
 
       def initialize(uri=nil, response=nil, body=nil, code=nil, mech=nil)
         @encoding = nil
-        response.each do |header,v|
+
+        method = response.respond_to?(:each_header) ? :each_header : :each
+        response.send(method) do |header,v|
           next unless v =~ /charset/i
-          @encoding = v.split('=').last.strip
+          encoding = v.split('=').last.strip
+          @encoding = encoding unless encoding == 'none'
         end
+
+        # Force the encoding to be 8BIT so we can perform regular expressions.
+        # We'll set it to the detected encoding later
+        body.force_encoding('ASCII-8BIT') if defined?(Encoding) && body
+
         @encoding ||= Util.detect_charset(body)
-        body = Util.to_native_charset(body, @encoding) rescue body
 
         super(uri, response, body, code)
         @mech           ||= mech
 
+        @encoding = nil if html_body =~ /<meta[^>]*charset[^>]*>/i
+
         raise Mechanize::ContentTypeError.new(response['content-type']) unless
            response['content-type'] =~ /^(text\/html)|(application\/xhtml\+xml)/i
         @parser = @links = @forms = @meta = @bases = @frames = @iframes = nil
       end
 
@@ -44,19 +52,31 @@
         @title ||= if parser && search('title').inner_text.length > 0
           search('title').inner_text
         end
       end
 
+      def encoding=(encoding)
+        @encoding = encoding
+
+        if @parser && @parser.encoding.downcase != encoding.downcase
+          # lazy reinitialize the parser with the new encoding
+          @parser = nil
+        end
+      end
+
+      def encoding
+        parser.respond_to?(:encoding) ? parser.encoding : nil
+      end
+
       def parser
         return @parser if @parser
 
         if body && response
-          html_body = body.length > 0 ? body : '<html></html>'
-          if WWW::Mechanize.html_parser == Nokogiri::HTML
-            @parser = Mechanize.html_parser.parse(html_body, nil, @encoding)
+          if mech.html_parser == Nokogiri::HTML
+            @parser = mech.html_parser.parse(html_body, nil, @encoding)
           else
-            @parser = Mechanize.html_parser.parse(html_body)
+            @parser = mech.html_parser.parse(html_body)
           end
         end
 
         @parser
       end
@@ -116,12 +136,13 @@
       def meta
         @meta ||= search('meta').map do |node|
           next unless node['http-equiv'] && node['content']
           (equiv, content) = node['http-equiv'], node['content']
           if equiv && equiv.downcase == 'refresh'
-            if content && content =~ /^\d+\s*;\s*url\s*=\s*'?([^\s']+)/i
-              node['href'] = $1
+            Meta.parse(content, uri) do |delay, href|
+              node['delay'] = delay
+              node['href'] = href
               Meta.new(node, @mech, self)
             end
           end
         end.compact
       end
@@ -137,9 +158,19 @@
       end
 
       def iframes
         @iframes ||= 
           search('iframe').map { |node| Frame.new(node, @mech, self) }
+      end
+
+      private
+
+      def html_body
+        if body
+          body.length > 0 ? body : '<html></html>'
+        else
+          ''
+        end
       end
     end
   end
 end