Add config file parsing.

[dead/lwn-epub.git] / src / Main.hs

diff --git a/src/Main.hs b/src/Main.hs
index 373d423c9f322ba885db966fecff5b7e190c4dd4..47c69fa1cba9ade063d2b63bea22fe800b33a4e9 100644 (file)
--- a/src/Main.hs
+++ b/src/Main.hs
@@ -2,24 +2,48 @@
 module Main
 where
 
+import Prelude hiding (readFile)
+import System.Directory (doesFileExist)
 import System.IO (
   Handle,
   IOMode (WriteMode),
   openBinaryFile,
   stdout
   )
+import System.IO.UTF8 (readFile)
+import Text.Regex.Posix ((=~))
 import Text.XML.HXT.Core
 
-import CommandLine (Args(..), apply_args)
+import CommandLine (show_help)
+import Configuration (Cfg(..), get_cfg)
 import LWN.Page
+import LWN.URI (is_lwn_url, make_absolute_url)
+import Misc (contains)
 
-my_read :: String -> IOStateArrow s b XmlTree
-my_read =
-  readDocument [ withValidate  no,
+
+my_read_opts :: SysConfigList
+my_read_opts = [ withValidate  no,
                  withParseHTML yes,
-                 withInputEncoding utf8,
                  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
+
 -- | 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.
@@ -31,24 +55,49 @@ get_output_handle path =
     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 = return . id
+real_article_path s = do
+  is_file <- doesFileExist s
+  return $ if is_file then s else check_cases
+  where
+    abs_current =
+      case make_absolute_url "current" of
+        Nothing -> s
+        Just ac -> ac
+    abs_s =
+      case make_absolute_url s of
+        Nothing -> s
+        Just as -> as
+
+    check_cases :: String
+    check_cases
+      | is_lwn_url s = s
+      | s `contains` "current" = abs_current
+      | s =~ "^[0-9]+$" = abs_s
+      | otherwise = s -- Give up
 
 main :: IO ()
 main = do
-  Args{..} <- apply_args
+  Cfg{..} <- get_cfg
   output_handle <- get_output_handle output
-  input_path <- real_article_path article
-  let html = my_read input_path
-  result <- parse html
+  maybe_html <- get_xml_from_article article
+
+  case maybe_html of
 
-  case result of
-    Just stuff  -> epublish stuff output_handle
-    Nothing     -> return ()
+    Just html -> do
+      result <- parse html
+      case result of
+        Just stuff  -> epublish stuff output_handle
+        Nothing     -> do
+         _ <- show_help
+         return ()
 
-  putStrLn "Done."
+    Nothing -> do
+      _ <- show_help
+      return ()