+{-# LANGUAGE NoMonomorphismRestriction #-}
+{-# LANGUAGE DoAndIfThenElse #-}
module Main
where
import Control.Arrow ( (&&&), arr, returnA )
-import Database.Groundhog.Core ( PersistEntity )
+import Control.Concurrent ( threadDelay )
+import Control.Exception ( SomeException, catch )
+import Control.Monad ( when )
+import Control.Monad.IO.Class ( MonadIO, liftIO )
+import Database.Groundhog.Generic ( runDbConn )
import Database.Groundhog.Sqlite (
- defaultMigrationLogger,
- insert,
- migrate,
- runDbConn,
- runMigration,
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,
- SysConfigList,
- XmlPickler,
XmlTree,
(>>>),
(/>),
getAttrl,
getText,
hasName,
- no,
readDocument,
- runX,
- unpickleDoc,
- withPreserveComment,
- withRemoveWS,
- withSubstDTDEntities,
- withValidate,
- xpickle,
- yes )
-import System.Environment ( getArgs )
-import qualified TSN.Injuries as Injuries (
- Listing,
- Message ( listings ) )
-import qualified TSN.InjuriesDetail as InjuriesDetail (
- Listing ( player_listings ),
- Message ( listings ),
- PlayerListing )
-
-
--- | A list of options passed to 'readDocument' when we parse an XML
--- document. We don't validate because the DTDs from TSN are
--- wrong. As a result, we don't want to keep useless DTDs
--- areound. Thus we disable 'withSubstDTDEntities' which, when
--- combined with "withValidate no", prevents HXT from trying to read
--- the DTD at all.
---
-parse_opts :: SysConfigList
-parse_opts =
- [ withPreserveComment no,
- withRemoveWS yes,
- withSubstDTDEntities no,
- withValidate no ]
-
-
--- | We put the 'XmlTree' argument last so that it's easy to eta
--- reduce all of the import_foo functions that call this.
---
-import_generic :: (XmlPickler a, PersistEntity b)
- => b -- ^ Dummy Listing instance needed for 'migrate'
- -> (a -> [b]) -- ^ listings getter
- -> XmlTree
- -> IO ()
-import_generic dummy g xml =
- withSqliteConn "foo.sqlite3" $ runDbConn $ do
- runMigration defaultMigrationLogger $ do
- migrate dummy
- let msg = unpickleDoc xpickle xml
- case msg of
- Nothing -> error "Should unpickle!"
- Just m -> mapM_ (\l -> insert l) (g m)
-
--- | Import TSN.Injuries from an 'XmlTree'.
-import_injuries :: XmlTree -> IO ()
-import_injuries =
- import_generic
- (undefined :: Injuries.Listing)
- Injuries.listings
-
--- | Import TSN.InjuriesDetail from an 'XmlTree'.
-import_injuries_detail :: XmlTree -> IO ()
-import_injuries_detail =
- import_generic
- (undefined :: InjuriesDetail.PlayerListing)
- ( (concatMap InjuriesDetail.player_listings) . InjuriesDetail.listings)
-
-import_file :: FilePath -> IO ()
-import_file path = do
- results <- runX $ readA >>> (doctypeA &&& returnA) >>> (arr import_with_dtd)
+ 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
+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
- [] -> error "ERROR: Unable to determine DOCTYPE."
- (r:_) -> r -- Need to do something with the result or it gets GCed?
- -- We do only expect one result fortunately.
+ -- If results' is empty, one of the arrows return "nothing."
+ [] -> do
+ report_error $ "Unable to determine DTD for file " ++ path ++ "."
+ return False
+ (r:_) ->
+ case r of
+ Nothing -> return False
+ Just cnt -> do
+ report_info $ "Successfully imported " ++
+ (show cnt) ++
+ " records from " ++ path ++ "."
+ return True
where
+ exception_handler :: SomeException -> IO [Maybe Int]
+ exception_handler e = do
+ report_error (show e)
+ report_error $ "Failed to import file " ++ path ++ "."
+ -- Return a nonempty list so we don't claim incorrectly that
+ -- we couldn't parse the DTD.
+ return [Nothing]
+
-- | An arrow that reads a document into an 'XmlTree'.
readA :: IOStateArrow s a XmlTree
readA = readDocument parse_opts path
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 (Maybe Int)]. We thus use
+ -- bind (>>=) and sequence to combine all of the IOs into one
+ -- big one outside of the list.
+ parse_and_import :: IO [Maybe Int]
+ 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 ()
- import_with_dtd (dtd,xml)
- | dtd == "injuriesxml.dtd" = import_injuries xml
- | dtd == "Injuries_Detail_XML.dtd" = import_injuries_detail xml
- | otherwise = error "ERROR: Unrecognized DTD."
+ import_with_dtd :: (String, XmlTree) -> IO (Maybe Int)
+ import_with_dtd (dtd,xml) =
+ -- 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 errmsg = "Unrecognized DTD in " ++ path ++ ": " ++ dtd ++ "."
+ liftIO $ report_info errmsg
+ return Nothing
+
main :: IO ()
main = do
- args <- getArgs
- import_file (args !! 0)
+ 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 ++ "."
projects / dead / htsn-import.git / blobdiff