+-- | 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
+
+-- | If we're given an empty path, return a handle to
+-- 'stdout'. Otherwise, open the given file and return a read/write
+-- handle to that.
+get_output_handle :: FilePath -> IO Handle
+get_output_handle path =
+ if (null path) then
+ return stdout
+ else
+ openBinaryFile path WriteMode
+
+
+
+-- | Convert the given article to either a URL or a filesystem
+-- path. If the given article exists on the filesystem, we assume
+-- it's a file. Otherwise, we check to see if it's a URL. Failing
+-- that, we try to construct a URL from what we're given and do our
+-- best.
+real_article_path :: String -> IO String
+real_article_path s = do
+ is_file <- doesFileExist s
+ return $ if is_file then s else add_trailing_slash check_cases
+ where
+ abs_current =
+ case make_absolute_url "current" of
+ Nothing -> s
+ Just ac -> ac
+ abs_article =
+ case make_absolute_url ("Articles/" ++ s) of
+ Nothing -> s
+ Just as -> as
+
+ check_cases :: String
+ check_cases
+ | is_lwn_url s = make_https s
+ | s `contains` "current" = abs_current
+ | s =~ "^[0-9]+$" = abs_article
+ | otherwise = s -- Give up
+