Fix strings caching

add body as last fallback

beautiful_soup/beautiful_soup.py

@@ -15,17 +15,17 @@ import requests
 '''
 
 def get_url_content( url ):
-    file = 'page-content/' + uuid.uuid5( uuid.NAMESPACE_URL, url ).hex + '.json'
-    makedirs(dirname(file), exist_ok=True)
-    if exists( file ):
-        with open( file, 'r' ) as file_content:
+    file_path = 'page-content/' + uuid.uuid5( uuid.NAMESPACE_URL, url ).hex + '.json'
+    makedirs(dirname(file_path), exist_ok=True)
+    if exists( file_path ):
+        with open( file_path, 'r' ) as file:
             strings = json.load( file )
     else:
         try:
             strings = extract_strings( url )
         except Exception as exception:
             raise exception
-        with open( file, 'w' ) as file:
+        with open( file_path, 'w' ) as file:
             json.dump( strings, file )
 
     return strings
@@ -133,6 +133,11 @@ def get_main_content( soup ):
         print('Has article tag.')
         return content
 
+    content = soup.find( "body" )
+    if content is not None:
+        print('Has body tag.')
+        return content
+
     return None
 
 def get_tags_text( soup ):