-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 :: Cfg -> IO (Maybe (IOStateArrow s b XmlTree))
-get_xml_from_article cfg = do
- my_article <- real_article_path (article cfg)
- is_file <- doesFileExist my_article
- case is_file of
- True -> do
- contents <- readFile my_article
- return $ Just $ my_read contents
- False -> do
- -- Download the URL and try to parse it.
- if use_account cfg then do
- -- use_account would be false if these fromJusts would fail.
- cj <- make_cookie_jar
- li_result <- log_in cj
- (fromJust $ username cfg)
- (fromJust $ password cfg)
- when (not li_result) $ do
- hPutStrLn stderr "Failed to log in."
-
- html <- get_page (Just cj) my_article
- print $ fromJust $ html
- return $
- case html of
- Nothing -> Nothing
- Just h -> Just $ my_read h
- else do
- html <- get_page Nothing my_article
- putStrLn "Not logged in."
- print $ fromJust $ html
- return $
- case html of
- Nothing -> Nothing
- Just h -> Just $ my_read h
-