-my_read_opts :: SysConfigList
-my_read_opts = [ withValidate no,
- withParseHTML yes,
- withWarnings no ]
-
--- | My version of HandsomeSoup's parseHTML.
-my_read :: String -> IOStateArrow s b XmlTree
-my_read = readString my_read_opts
-
--- | Try to parse the given article using HXT. We try a few different
--- methods; if none of them work, we return 'Nothing'.
-get_xml_from_article :: String -> IO (Maybe (IOStateArrow s b XmlTree))
-get_xml_from_article s = do
- article <- real_article_path s
- is_file <- doesFileExist article
- case is_file of
- True -> do
- contents <- readFile article
- return $ Just $ my_read contents
- False -> do
- -- Download the URL and try to parse it.
- return Nothing
-