Attempt to implement article downloading.

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

diff --git a/src/Main.hs b/src/Main.hs
index 924b9a558dd59df682d9cbdb3713fceaad0b27f1..4cc4216d0e02e3c81ccaff617acb97ff26033bc5 100644 (file)
--- a/src/Main.hs
+++ b/src/Main.hs
@@ -1,13 +1,17 @@
-{-# LANGUAGE ScopedTypeVariables, RecordWildCards #-}
+{-# LANGUAGE ScopedTypeVariables, RecordWildCards, DoAndIfThenElse #-}

 module Main
 where
 
 module Main
 where
 

+import Control.Monad (when)
+import Data.Maybe (fromJust)

 import Prelude hiding (readFile)
 import System.Directory (doesFileExist)
 import System.IO (
   Handle,
   IOMode (WriteMode),
 import Prelude hiding (readFile)
 import System.Directory (doesFileExist)
 import System.IO (
   Handle,
   IOMode (WriteMode),

+  hPutStrLn,

   openBinaryFile,
   openBinaryFile,

+  stderr,

   stdout
   )
 import System.IO.UTF8 (readFile)
   stdout
   )
 import System.IO.UTF8 (readFile)


@@ -15,11 +19,12 @@ import Test.HUnit (Assertion, assertEqual)
 import Test.Framework (Test, testGroup)
 import Test.Framework.Providers.HUnit (testCase)
 import Text.Regex.Posix ((=~))
 import Test.Framework (Test, testGroup)
 import Test.Framework.Providers.HUnit (testCase)
 import Text.Regex.Posix ((=~))

-import Text.XML.HXT.Core
+import Text.XML.HXT.Core hiding (when)

 
 import CommandLine (show_help)
 
 import CommandLine (show_help)

-import Configuration (Cfg(..), get_cfg)
-import LWN.Page
+import Configuration (Cfg(..), get_cfg, use_account)
+import LWN.HTTP (get_page, log_in, make_cookie_jar)
+import LWN.Page (epublish, parse)

 import LWN.URI (is_lwn_url, make_absolute_url, make_https)
 import Misc (contains)
 
 import LWN.URI (is_lwn_url, make_absolute_url, make_https)
 import Misc (contains)
 


@@ -35,17 +40,36 @@ 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'.
 
 -- | 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
+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
   case is_file of
     True -> do

-      contents <- readFile article
+      contents <- readFile my_article

       return $ Just $ my_read contents
     False -> do
       -- Download the URL and try to parse it.
       return $ Just $ my_read contents
     False -> do
       -- Download the URL and try to parse it.

-      return Nothing
+      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
+        return $
+          case html of
+            Nothing -> Nothing
+            Just h  -> Just $ my_read h
+      else do
+        html <- get_page Nothing my_article
+        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
 
 -- | If we're given an empty path, return a handle to
 --   'stdout'. Otherwise, open the given file and return a read/write


@@ -87,12 +111,12 @@ real_article_path s = do
 
 main :: IO ()
 main = do
 
 main :: IO ()
 main = do

-  Cfg{..} <- get_cfg
-  output_handle <- get_output_handle output
-  maybe_html <- get_xml_from_article article
+  cfg <- get_cfg
+  output_handle <- get_output_handle (output cfg)

 
 

-  case maybe_html of
+  maybe_html <- get_xml_from_article cfg

 
 

+  case maybe_html of

     Just html -> do
       result <- parse html
       case result of
     Just html -> do
       result <- parse html
       case result of


@@ -106,7 +130,6 @@ main = do
       return ()
 
 
       return ()
 
 

-

 test_current_article_path :: Assertion
 test_current_article_path = do
   let expected = "https://lwn.net/current"
 test_current_article_path :: Assertion
 test_current_article_path = do
   let expected = "https://lwn.net/current"


@@ -139,4 +162,3 @@ main_tests =
     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 ]
     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