lib/spidr/page.rb in spidr-0.5.0 vs lib/spidr/page.rb in spidr-0.6.0

@@ -1,9 +1,5 @@
-require 'spidr/page/headers'
-require 'spidr/page/body'
-require 'spidr/page/links'
-
 module Spidr
   #
   # Represents a requested page from a website.
   #
   class Page
@@ -32,47 +28,94 @@
       @headers  = response.to_hash
       @doc      = nil
     end
 
     #
-    # The meta-redirect links of the page.
+    # The body of the response.
     #
-    # @return [Array<String>]
-    #   All meta-redirect links in the page.
+    # @return [String]
+    #   The body of the response.
     #
-    # @deprecated
-    #   Deprecated in 0.3.0 and will be removed in 0.4.0.
-    #   Use {#meta_redirects} instead.
+    def body
+      (response.body || '')
+    end
+
+    alias to_s body
+
     #
-    def meta_redirect
-      STDERR.puts 'DEPRECATION: Spidr::Page#meta_redirect will be removed in 0.3.0'
-      STDERR.puts 'DEPRECATION: Use Spidr::Page#meta_redirects instead'
+    # Returns a parsed document object for HTML, XML, RSS and Atom pages.
+    #
+    # @return [Nokogiri::HTML::Document, Nokogiri::XML::Document, nil]
+    #   The document that represents HTML or XML pages.
+    #   Returns `nil` if the page is neither HTML, XML, RSS, Atom or if
+    #   the page could not be parsed properly.
+    #
+    # @see http://nokogiri.rubyforge.org/nokogiri/Nokogiri/XML/Document.html
+    # @see http://nokogiri.rubyforge.org/nokogiri/Nokogiri/HTML/Document.html
+    #
+    def doc
+      unless body.empty?
+        doc_class = if html?
+                      Nokogiri::HTML::Document
+                    elsif rss? || atom? || xml? || xsl?
+                      Nokogiri::XML::Document
+                    end
 
-      meta_redirects
+        if doc_class
+          begin
+            @doc ||= doc_class.parse(body, @url.to_s, content_charset)
+          rescue
+          end
+        end
+      end
     end
 
     #
-    # Determines if the response code is `300`, `301`, `302`, `303`
-    # or `307`. Also checks for "soft" redirects added at the page 
-    # level by a meta refresh tag.
+    # Searches the document for XPath or CSS Path paths.
     #
-    # @return [Boolean]
-    #   Specifies whether the response code is a HTTP Redirect code.
+    # @param [Array<String>] paths
+    #   CSS or XPath expressions to search the document with.
     #
-    def is_redirect?
-      case code
-      when 300..303, 307
-        true
-      when 200
-        meta_redirect?
+    # @return [Array]
+    #   The matched nodes from the document.
+    #   Returns an empty Array if no nodes were matched, or if the page
+    #   is not an HTML or XML document.
+    #
+    # @example
+    #   page.search('//a[@href]')
+    #
+    # @see http://nokogiri.rubyforge.org/nokogiri/Nokogiri/XML/Node.html#M000239
+    #
+    def search(*paths)
+      if doc
+        doc.search(*paths)
       else
-        false
+        []
       end
     end
 
-    alias redirect? is_redirect?
+    #
+    # Searches for the first occurrence an XPath or CSS Path expression.
+    #
+    # @return [Nokogiri::HTML::Node, Nokogiri::XML::Node, nil]
+    #   The first matched node. Returns `nil` if no nodes could be matched,
+    #   or if the page is not a HTML or XML document.
+    #
+    # @example
+    #   page.at('//title')
+    #
+    # @see http://nokogiri.rubyforge.org/nokogiri/Nokogiri/XML/Node.html#M000251
+    #
+    def at(*arguments)
+      if doc
+        doc.at(*arguments)
+      end
+    end
 
+    alias / search
+    alias % at
+
     protected
 
     #
     # Provides transparent access to the values in {#headers}.
     #
@@ -88,11 +131,11 @@
     # @raise [NoMethodError]
     #   The missing method did not map to a header in {#headers}.
     #
     def method_missing(name,*arguments,&block)
       if (arguments.empty? && block.nil?)
-        header_name = name.to_s.sub('_','-')
+        header_name = name.to_s.tr('_','-')
 
         if @response.key?(header_name)
           return @response[header_name]
         end
       end
@@ -100,5 +143,10 @@
       return super(name,*arguments,&block)
     end
   
   end
 end
+
+require 'spidr/page/status_codes'
+require 'spidr/page/content_types'
+require 'spidr/page/cookies'
+require 'spidr/page/html'