merge from 3.3

#17403: urllib.parse.robotparser normalizes the urls before adding to
ruleline. This helps in handling certain types invalid urls in a conservative
manner. Patch contributed by Mher Movsisyan.

Lib/urllib/robotparser.py

@@ -157,6 +157,7 @@ class RuleLine:
         if path == '' and not allowance:
             # an empty value means allow all
             allowance = True
+        path = urllib.parse.urlunparse(urllib.parse.urlparse(path))
         self.path = urllib.parse.quote(path)
         self.allowance = allowance