Source reorganization and cleanup.

[dead/lwn-epub.git] / src / LWN / Article.hs

diff --git a/src/LWN/Article.hs b/src/LWN/Article.hs
index 70c68a5190d724fc2663531c57a63d02efb7c227..9d4c85868be9f25299eabd79604249e56474f1ca 100644 (file)
--- a/src/LWN/Article.hs
+++ b/src/LWN/Article.hs
@@ -1,7 +1,27 @@
-module LWN.Article
+module LWN.Article (
+  Article(..),
+  Byline(..),
+  Title(..),
+  BodyHtml(..),
+  article_tests,
+  real_article_path
+  )
 where
 
-import XHTML
+import Data.List (isPrefixOf)
+import System.Directory (doesFileExist)
+import Test.HUnit (Assertion, assertEqual)
+import Test.Framework (Test, testGroup)
+import Test.Framework.Providers.HUnit (testCase)
+import Text.Regex.Posix ((=~))
+
+import LWN.URI (
+  add_trailing_slash,
+  is_lwn_url,
+  try_make_absolute_url,
+  make_https)
+
+import LWN.XHTML (XHTML, to_xhtml)
 
 newtype Title    = Title    { getTitle    :: String }
 newtype Byline   = Byline   { getByline   :: Maybe String }
@@ -36,3 +56,69 @@ instance XHTML Article where
     (to_xhtml t)  ++
     (to_xhtml bl) ++
     (to_xhtml b)
+
+
+
+-- | 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 path = do
+  is_file <- doesFileExist path
+  return $ if is_file then path else add_trailing_slash check_cases
+  where
+    abs_current = try_make_absolute_url ("/" ++ path)
+    abs_article = try_make_absolute_url ("Articles/" ++ path)
+
+    check_cases :: String
+    check_cases
+      | is_lwn_url path = make_https path
+      | isPrefixOf "current" path = abs_current
+      | path =~ "^[0-9]+$" = abs_article
+      | otherwise = path -- Give up
+
+
+
+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_current_bigpage_article_path :: Assertion
+test_current_bigpage_article_path = do
+  let expected = "https://lwn.net/current/bigpage"
+  actual <- real_article_path "current/bigpage"
+  assertEqual "Current bigpage 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
+
+article_tests :: Test
+article_tests =
+  testGroup "Article Tests" [
+    testCase "Current article path constructed" test_current_article_path,
+    testCase
+      "Current bigpage article path constructed"
+      test_current_bigpage_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 ]