pass config to cleaners init

Author: grangier

goose/cleaners.py

@@ -26,8 +26,8 @@
 
 class DocumentCleaner(object):
 
-    def __init__(self):
-
+    def __init__(self, config):
+        self.config = config
         self.remove_nodes_re = (
         "^side$|combx|retweet|mediaarticlerelated|menucontainer|navbar"
         "|comment|PopularQuestions|contact|foot|footer|Footer|footnote"

goose/crawler.py

@@ -120,7 +120,7 @@ def get_output_formatter(self):
         return StandardOutputFormatter(self.config)
 
     def get_document_cleaner(self):
-        return StandardDocumentCleaner()
+        return StandardDocumentCleaner(self.config)
 
     def get_document(self, raw_html):
         doc = Parser.fromstring(raw_html)