module Main
where
+-- System imports.
import Control.Arrow ( (&&&), (>>^), arr, returnA )
import Control.Concurrent ( threadDelay )
import Control.Exception ( SomeException, catch )
getText,
hasName,
readDocument,
- runX )
+ runX,
+ unpickleDoc )
+-- Local imports.
import Backend ( Backend(..) )
import CommandLine ( get_args )
import Configuration ( Configuration(..), merge_optional )
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 qualified TSN.XML.Odds as Odds ( Message )
+import qualified TSN.XML.AutoRacingResults as AutoRacingResults (
+ dtd,
+ pickle_message )
+import qualified TSN.XML.AutoRacingSchedule as AutoRacingSchedule (
+ dtd,
+ pickle_message )
+import qualified TSN.XML.GameInfo as GameInfo ( dtds, parse_xml )
+import qualified TSN.XML.Heartbeat as Heartbeat ( dtd, verify )
+import qualified TSN.XML.Injuries as Injuries ( dtd, pickle_message )
+import qualified TSN.XML.InjuriesDetail as InjuriesDetail (
+ dtd,
+ pickle_message )
+import qualified TSN.XML.News as News ( dtd, pickle_message )
+import qualified TSN.XML.Odds as Odds ( dtd, pickle_message )
+import qualified TSN.XML.Scores as Scores ( dtd, pickle_message )
+import qualified TSN.XML.SportInfo as SportInfo ( dtds, parse_xml )
+import qualified TSN.XML.Weather as Weather ( dtd, pickle_message )
import Xml ( DtdName(..), parse_opts )
-- | This is where most of the work happens. This function is called
-- on every file that we would like to import. It determines which
--- importer to use based on the DTD, processes the file, and then
--- returns whether or not any records were imported. If the file was
--- processed, the number of records imported is returned (wrapped in
--- a Just). Otherwise, if the file was not processed, 'Nothing' is
+-- importer to use based on the DTD, attempts to process the file,
+-- and then returns whether or not it was successful. If the file
+-- was processed, 'True' is returned. Otherwise, 'False' is
-- returned.
--
--- Since we are already in arrow world with HXT, the
--- 'import_with_dtd' function is lifted to an 'Arrow' as well with
--- 'arr'. This prevents us from having to do a bunch of unwrapping
--- and rewrapping with the associated error checking.
+-- The implementation is straightforward with one exception: since
+-- we are already in arrow world with HXT, the @import_with_dtd@
+-- function is lifted to an 'Arrow' as well with 'arr'. This
+-- prevents us from having to do a bunch of unwrapping and
+-- rewrapping with the associated error checking.
--
import_file :: Configuration -- ^ A configuration object needed for the
-- 'backend' and 'connection_string'.
-> FilePath -- ^ The path of the XML file to import.
- -> IO (Maybe Int) -- ^ If we processed the file, Just the number
- -- of records imported. Otherwise, Nothing.
+ -> IO Bool -- ^ True if we processed the file, False otherwise.
import_file cfg path = do
results <- parse_and_import `catch` exception_handler
case results of
[] -> do
-- One of the arrows returned "nothing."
report_error $ "Unable to determine DTD for file " ++ path ++ "."
- return Nothing
+ return False
(ImportFailed errmsg:_) -> do
report_error errmsg
- return Nothing
+ return False
(ImportSkipped infomsg:_) -> do
-- We processed the message but didn't import anything. Return
-- "success" so that the XML file is deleted.
report_info infomsg
- return $ Just 0
- (ImportSucceeded count:_) -> do
- report_info $ "Successfully imported " ++ (show count) ++
- " records from " ++ path ++ "."
- return $ Just count
+ return True
+ (ImportSucceeded:_) -> do
+ report_info $ "Successfully imported " ++ path ++ "."
+ return True
(ImportUnsupported infomsg:_) -> do
-- For now we return "success" for these too, since we know we don't
-- support a bunch of DTDs and we want them to get deleted.
report_info infomsg
- return $ Just 0
+ return True
where
-- | This will catch *any* exception, even the ones thrown by
-- Haskell's 'error' (which should never occur under normal
import_with_dtd (DtdName dtd,xml)
-- We special-case the heartbeat so it doesn't have to run in
-- the database monad.
- | dtd == "Heartbeat.dtd" = Heartbeat.verify 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
+ then withPostgresqlConn cs $ runDbConn importer
+ else withSqliteConn cs $ runDbConn importer
where
-- | Pull the real connection String out of the configuration.
+ --
cs :: String
cs = get_connection_string $ connection_string cfg
+ -- | Convenience; we use this everywhere below in 'importer'.
+ --
+ migrate_and_import m = dbmigrate m >> dbimport m
+
importer
- | dtd == "injuriesxml.dtd" =
- dbimport (undefined :: Injuries.Listing)
+ | dtd == AutoRacingResults.dtd = do
+ let m = unpickleDoc AutoRacingResults.pickle_message xml
+ maybe (return $ ImportFailed errmsg) migrate_and_import m
+
+ | dtd == AutoRacingSchedule.dtd = do
+ let m = unpickleDoc AutoRacingSchedule.pickle_message xml
+ maybe (return $ ImportFailed errmsg) migrate_and_import m
+
+ -- GameInfo and SportInfo appear least in the guards
+ | dtd == Injuries.dtd = do
+ let m = unpickleDoc Injuries.pickle_message xml
+ maybe (return $ ImportFailed errmsg) migrate_and_import m
+
+ | dtd == InjuriesDetail.dtd = do
+ let m = unpickleDoc InjuriesDetail.pickle_message xml
+ maybe (return $ ImportFailed errmsg) migrate_and_import m
+
- | dtd == "Injuries_Detail_XML.dtd" =
- dbimport (undefined :: InjuriesDetail.PlayerListing)
+ | dtd == News.dtd = do
+ let m = unpickleDoc News.pickle_message xml
+ maybe (return $ ImportFailed errmsg) migrate_and_import m
- | dtd == "newsxml.dtd" =
- dbimport (undefined :: News.Message)
+ | dtd == Odds.dtd = do
+ let m = unpickleDoc Odds.pickle_message xml
+ maybe (return $ ImportFailed errmsg) migrate_and_import m
- | dtd == "Odds_XML.dtd" = undefined
+ | dtd == Scores.dtd = do
+ let m = unpickleDoc Scores.pickle_message xml
+ maybe (return $ ImportFailed errmsg) migrate_and_import m
- | otherwise = \_ -> do -- Dummy arg simplifies the other cases.
+ -- SportInfo and GameInfo appear least in the guards
+ | dtd == Weather.dtd = do
+ let m = unpickleDoc Weather.pickle_message xml
+ maybe (return $ ImportFailed errmsg) migrate_and_import m
+
+ | dtd `elem` GameInfo.dtds = do
+ let either_m = GameInfo.parse_xml dtd xml
+ case either_m of
+ -- This might give us a slightly better error
+ -- message than the default 'errmsg'.
+ Left err -> return $ ImportFailed err
+ Right m -> migrate_and_import m
+
+ | dtd `elem` SportInfo.dtds = do
+ let either_m = SportInfo.parse_xml dtd xml
+ case either_m of
+ -- This might give us a slightly better error
+ -- message than the default 'errmsg'.
+ Left err -> return $ ImportFailed err
+ Right m -> migrate_and_import m
+
+ | otherwise = do
let infomsg =
"Unrecognized DTD in " ++ path ++ ": " ++ dtd ++ "."
return $ ImportUnsupported infomsg
+ where
+ errmsg = "Could not unpickle " ++ dtd ++ "."
+
-- | Entry point of the program. It twiddles some knobs for
--- configuration options and then calls 'import_file' on each XML file
--- given on the command-line.
+-- configuration options and then calls 'import_file' on each XML
+-- file given on the command-line.
--
--- Any file successfully processed is then removed, and we're done.
+-- Any file successfully processed is then optionally removed, and
+-- we're done.
--
main :: IO ()
main = do
-- 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 = [ (p,c) | (p, Just c) <- result_pairs ]
- let imported_count = sum $ map snd victims
- report_info $ "Imported " ++ (show imported_count) ++ " records total."
- mapM_ ((kill True) . fst) victims
+ let victims = [ p | (p, True) <- result_pairs ]
+ let processed_count = length victims
+ report_info $ "Processed " ++ (show processed_count) ++ " document(s) total."
+ when (remove cfg) $ mapM_ (kill True) victims
where
-- | Wrap these two actions into one function so that we don't