Add some tests for the article URL construction.

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

diff --git a/src/Main.hs b/src/Main.hs
index da760e45af7a58676b8b6cfdd2c0198de6c41d78..924b9a558dd59df682d9cbdb3713fceaad0b27f1 100644 (file)
--- a/src/Main.hs
+++ b/src/Main.hs
@@ -1,7 +1,142 @@
+{-# LANGUAGE ScopedTypeVariables, RecordWildCards #-}
 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 Test.HUnit (Assertion, assertEqual)
+import Test.Framework (Test, testGroup)
+import Test.Framework.Providers.HUnit (testCase)
+import Text.Regex.Posix ((=~))
+import Text.XML.HXT.Core
+
+import CommandLine (show_help)
+import Configuration (Cfg(..), get_cfg)
+import LWN.Page
+import LWN.URI (is_lwn_url, make_absolute_url, make_https)
+import Misc (contains)
+
+
+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
+
+-- | 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 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
 
 main :: IO ()
 main = do
-  putStrLn "Hello, world."
+  Cfg{..} <- get_cfg
+  output_handle <- get_output_handle output
+  maybe_html <- get_xml_from_article article
+
+  case maybe_html of
+
+    Just html -> do
+      result <- parse html
+      case result of
+        Just stuff  -> epublish stuff output_handle
+        Nothing     -> do
+         _ <- show_help
+         return ()
+
+    Nothing -> do
+      _ <- show_help
+      return ()
+
+
+
+test_current_article_path :: Assertion
+test_current_article_path = do
+  let expected = "https://lwn.net/current"
+  actual <- real_article_path "current"
+  assertEqual "Current article path constructed" expected actual
+
+test_numbered_article_path :: Assertion
+test_numbered_article_path = do
+  let expected = "https://lwn.net/Articles/69"
+  actual <- real_article_path "69" -- I'm twelve
+  assertEqual "Numbered article path constructed" expected actual
+
+
+test_full_article_path :: Assertion
+test_full_article_path = do
+  let expected = "https://lwn.net/Articles/502979/"
+  actual <- real_article_path "https://lwn.net/Articles/502979/"
+  assertEqual "Full article path left alone" expected actual
+
+test_non_https_article_path :: Assertion
+test_non_https_article_path = do
+  let expected = "https://lwn.net/Articles/502979/"
+  actual <- real_article_path "http://lwn.net/Articles/502979/"
+  assertEqual "Non-https URL made https" expected actual
+
+main_tests :: Test
+main_tests =
+  testGroup "Main Tests" [
+    testCase "Current article path constructed" test_current_article_path,
+    testCase "Numbered article path constructed" test_numbered_article_path,
+    testCase "Full article path left alone" test_full_article_path,
+    testCase "Non-https URL made https" test_non_https_article_path ]
+     
\ No newline at end of file