Add a generic parser that will hopefully supplant some site-specific subclasses.

[dead/whatever-dl.git] / src / website.rb

diff --git a/src/website.rb b/src/website.rb
index 4e20466d30d95f0bf5a95370bcaa674a4adf5f73..e9e65ca1909f7add3c8b2c6ecc800a3856145528 100644 (file)
--- a/src/website.rb
+++ b/src/website.rb
@@ -30,12 +30,12 @@ require 'net/http'
 # We use this to loop through every "website" in an
 # attempt to determine to which site a URL belongs.
 class Website
- 
+
   protected;
-  
+
   @url = nil
 
-  
+
   def self.inherited(subclass)
     if superclass.respond_to? :inherited
       superclass.inherited(subclass)
@@ -55,7 +55,7 @@ class Website
   end
 
 
-  
+
   def get_page_data(url)
     # A naive implementation that just grabs the
     # data from a page.
@@ -65,11 +65,14 @@ class Website
       http.get(uri.request_uri, self.headers)
     end
 
+    # Set the referer in case it is needed for some later request.
+    self.headers['Referer'] = uri.request_uri
+
     return response.body
   end
 
-  
-  
+
+
   public;
 
   # Additional headers used when requesting data from the website.
@@ -86,19 +89,29 @@ class Website
   def self.create(url)
     # Factory method returning an instance of
     # the appropriate subclass.
-    
+
+    # While we're looping through the list of subclasses,
+    # we'll set this to the Generic class.
+    generic = nil
+
     # Check the URL against each website's class.
     # The class will know whether or not the URL
     # "belongs" to its website.
     @subclasses.each do |w|
       if w.owns_url?(url)
-        return w.new(url)
+        if w.to_s == 'Generic'
+          generic = w
+        else
+          # We don't want to return Generic here because some
+          # other subclasses further down the list might match
+          # the URL.
+          return w.new(url)
+        end
       end
     end
 
-    # If nothing matched, we don't return an instance
-    # of anything.
-    return nil
+    # If nothing matched, try the generic parser.
+    return generic.new(url)
   end
 
 
@@ -108,13 +121,13 @@ class Website
     raise NotImplementedError
   end
 
-  
+
   # Same here. Abstract.
   def get_video_url()
     raise NotImplementedError
   end
 
-  
+
   # The website class should be responsible for determining the
   # video's filename. By default, we can take the last component
   # of the video URL, but in some cases, subclasses will want
@@ -125,10 +138,10 @@ class Website
 
     # Unless it contains URL parameters. We don't want those.
     return file_and_params unless file_and_params.include?('?')
-    
+
     # There must be some parameters. Strip them off.
     param_start_idx = file_and_params.index('?')
     return file_and_params[0...(param_start_idx)]
   end
-  
+
 end