{-# LANGUAGE NoMonomorphismRestriction #-} {-# LANGUAGE DoAndIfThenElse #-} module Main where import Control.Arrow ( (&&&), arr, returnA ) import Control.Concurrent ( threadDelay ) import Control.Exception ( SomeException, catch ) import Control.Monad ( when ) import Database.Groundhog.Generic ( runDbConn ) import Database.Groundhog.Sqlite ( withSqliteConn ) import Database.Groundhog.Postgresql ( withPostgresqlConn ) import Data.Monoid ( (<>) ) import Network.Services.TSN.Logging ( init_logging ) import System.Console.CmdArgs ( def ) import System.Directory ( removeFile ) import System.Exit ( exitWith, ExitCode (ExitFailure) ) import System.IO.Error ( catchIOError ) import Text.XML.HXT.Core ( ArrowXml, IOStateArrow, XmlTree, (>>>), (/>), getAttrl, getText, hasName, readDocument, runX ) import Backend ( Backend(..) ) import CommandLine ( get_args ) import Configuration ( Configuration(..), merge_optional ) import ConnectionString ( ConnectionString(..) ) import ExitCodes ( exit_no_xml_files ) import qualified OptionalConfiguration as OC ( OptionalConfiguration ( xml_files ), from_rc ) import Network.Services.TSN.Report ( report_info, report_error ) import TSN.DbImport ( DbImport(..), ImportResult(..) ) import qualified TSN.XML.Heartbeat as Heartbeat ( verify ) import qualified TSN.XML.Injuries as Injuries ( Listing ) import qualified TSN.XML.InjuriesDetail as InjuriesDetail ( PlayerListing ) import qualified TSN.XML.News as News ( Message ) import Xml ( parse_opts ) import_file :: Configuration -> FilePath -> IO Bool import_file cfg path = do results <- parse_and_import `catch` exception_handler case results of -- If results' is empty, one of the arrows return "nothing." [] -> do report_error $ "Unable to determine DTD for file " ++ path ++ "." return False (Err errmsg:_) -> do report_error errmsg return False (Info infomsg:_) -> do report_info infomsg return True (Succ count:_) -> do report_info $ "Successfully imported " ++ (show count) ++ " records from " ++ path ++ "." return True where exception_handler :: SomeException -> IO [ImportResult] exception_handler e = do report_error (show e) let errdesc = "Failed to import file " ++ path ++ "." -- Return a nonempty list so we don't claim incorrectly that -- we couldn't parse the DTD. return [Err errdesc] -- | An arrow that reads a document into an 'XmlTree'. readA :: IOStateArrow s a XmlTree readA = readDocument parse_opts path -- | An arrow which parses the doctype "SYSTEM" of an 'XmlTree'. -- We use these to determine the parser to use. doctypeA :: ArrowXml a => a XmlTree String doctypeA = getAttrl >>> hasName "doctype-SYSTEM" /> getText -- | Combine the arrows above as well as the function below -- (arrowized with 'arr') into an IO action that does everything -- (parses and then runs the import on what was parsed). -- -- The result of runX has type IO [IO ImportResult]. We thus use -- bind (>>=) and sequence to combine all of the IOs into one -- big one outside of the list. parse_and_import :: IO [ImportResult] parse_and_import = runX (readA >>> (doctypeA &&& returnA) >>> (arr import_with_dtd)) >>= sequence -- | Takes a 'Doctype', 'XmlTree' pair and uses the 'Doctype' to -- determine which function to call on the 'XmlTree'. import_with_dtd :: (String, XmlTree) -> IO ImportResult import_with_dtd (dtd,xml) | dtd == "Heartbeat.dtd" = Heartbeat.verify xml | otherwise = -- We need NoMonomorphismRestriction here. if backend cfg == Postgres then withPostgresqlConn cs $ runDbConn $ importer xml else withSqliteConn cs $ runDbConn $ importer xml where -- | Pull the real connection String out of the configuration. cs :: String cs = get_connection_string $ connection_string cfg importer | dtd == "injuriesxml.dtd" = dbimport (undefined :: Injuries.Listing) | dtd == "Injuries_Detail_XML.dtd" = dbimport (undefined :: InjuriesDetail.PlayerListing) | dtd == "newsxml.dtd" = dbimport (undefined :: News.Message) | otherwise = \_ -> do -- Dummy arg simplifies the other cases. let infomsg = "Unrecognized DTD in " ++ path ++ ": " ++ dtd ++ "." return $ Info infomsg main :: IO () main = do rc_cfg <- OC.from_rc cmd_cfg <- get_args -- Merge the config file options with the command-line ones, -- prefering the command-line ones. let opt_config = rc_cfg <> cmd_cfg -- Update a default config with any options that have been set in -- either the config file or on the command-line. We initialize -- logging before the missing parameter checks below so that we can -- log the errors. let cfg = (def :: Configuration) `merge_optional` opt_config init_logging (log_file cfg) (log_level cfg) (syslog cfg) -- Check the optional config for missing required options. when (null $ OC.xml_files opt_config) $ do report_error "No XML files given." exitWith (ExitFailure exit_no_xml_files) -- We don't do this in parallel (for now?) to keep the error -- messages nice and linear. results <- mapM (import_file cfg) (OC.xml_files opt_config) -- Zip the results with the files list to find out which ones can be -- deleted. let result_pairs = zip (OC.xml_files opt_config) results let victims = filter (\(_,result) -> result) result_pairs mapM_ ((kill True) . fst) victims where kill try_again path = do removeFile path `catchIOError` exception_handler report_info $ "Removed imported file " ++ path ++ "." where -- | A wrapper around threadDelay which takes seconds instead of -- microseconds as its argument. thread_sleep :: Int -> IO () thread_sleep seconds = do let microseconds = seconds * (10 ^ (6 :: Int)) threadDelay microseconds exception_handler :: IOError -> IO () exception_handler e = do report_error (show e) report_error $ "Failed to remove imported file " ++ path ++ "." if try_again then do report_info $ "Waiting 5 seconds to attempt removal again..." thread_sleep 5 kill False path else report_info $ "Giving up on " ++ path ++ "."