@@ -64,22 +64,18 @@ def _not_in_sphinx():
6464class ReutersParser (html_parser .HTMLParser ):
6565 """Utility class to parse a SGML file and yield documents one at a time."""
6666
67- def __init__ (self , verbose = 0 , encoding = 'latin-1' ):
68- html_parser .HTMLParser .__init__ (self , verbose )
67+ def __init__ (self , encoding = 'latin-1' ):
68+ html_parser .HTMLParser .__init__ (self )
6969 self ._reset ()
7070 self .encoding = encoding
7171
72- if not six .PY2 :
73- # In Python 3 need to be defined explicitly
74- def handle_starttag (tag , attrs ):
75- method = 'start_' + tag
76- getattr (self , method , lambda x : None )(attrs )
77- self .handle_starttag = handle_starttag
78-
79- def handle_endtag (tag ):
80- method = 'end_' + tag
81- getattr (self , method , lambda : None )()
82- self .handle_endtag = handle_endtag
72+ def handle_starttag (self , tag , attrs ):
73+ method = 'start_' + tag
74+ getattr (self , method , lambda x : None )(attrs )
75+
76+ def handle_endtag (self , tag ):
77+ method = 'end_' + tag
78+ getattr (self , method , lambda : None )()
8379
8480 def _reset (self ):
8581 self .in_title = 0
0 commit comments