-{-# LANGUAGE NoMonomorphismRestriction #-}
{-# LANGUAGE DoAndIfThenElse #-}
+{-# LANGUAGE NoMonomorphismRestriction #-}
module Main
where
-import Control.Arrow ( (&&&), arr, returnA )
+-- System imports.
+import Control.Arrow ( (&&&), (>>^), arr, returnA )
import Control.Concurrent ( threadDelay )
import Control.Exception ( SomeException, catch )
import Control.Monad ( when )
getText,
hasName,
readDocument,
- runX )
+ runX,
+ unpickleDoc )
+-- Local imports.
import Backend ( Backend(..) )
import CommandLine ( get_args )
import Configuration ( Configuration(..), merge_optional )
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 qualified TSN.XML.Injuries as Injuries ( pickle_message )
+import qualified TSN.XML.InjuriesDetail as InjuriesDetail ( pickle_message )
+import qualified TSN.XML.News as News ( pickle_message )
+import qualified TSN.XML.Odds as Odds ( 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, 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.
+--
+-- 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 Bool -- ^ True if we processed the file, False otherwise.
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
+ -- One of the arrows returned "nothing."
report_error $ "Unable to determine DTD for file " ++ path ++ "."
return False
- (Err errmsg:_) -> do
+ (ImportFailed errmsg:_) -> do
report_error errmsg
return False
- (Info infomsg:_) -> do
+ (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 True
- (Succ count:_) -> do
- report_info $ "Successfully imported " ++ (show count) ++
- " records from " ++ path ++ "."
+ (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 True
where
+ -- | This will catch *any* exception, even the ones thrown by
+ -- Haskell's 'error' (which should never occur under normal
+ -- circumstances).
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]
+ return [ImportFailed errdesc]
-- | An arrow that reads a document into an 'XmlTree'.
readA :: IOStateArrow s a XmlTree
-- | 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
+ dtdnameA :: ArrowXml a => a XmlTree DtdName
+ dtdnameA = getAttrl >>> hasName "doctype-SYSTEM" /> getText >>^ DtdName
-- | Combine the arrows above as well as the function below
-- (arrowized with 'arr') into an IO action that does everything
-- big one outside of the list.
parse_and_import :: IO [ImportResult]
parse_and_import =
- runX (readA >>> (doctypeA &&& returnA) >>> (arr import_with_dtd))
+ runX (readA >>> (dtdnameA &&& 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)
+ -- | Takes a ('DtdName', 'XmlTree') pair and uses the 'DtdName'
+ -- to determine which function to call on the 'XmlTree'.
+ import_with_dtd :: (DtdName, XmlTree) -> IO ImportResult
+ 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
| 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 == "injuriesxml.dtd" = do
+ let m = unpickleDoc Injuries.pickle_message xml
+ let errmsg = "Could not unpickle injuriesxml."
+ maybe (return $ ImportFailed errmsg) migrate_and_import m
+
+ | dtd == "Injuries_Detail_XML.dtd" = do
+ let m = unpickleDoc InjuriesDetail.pickle_message xml
+ let errmsg = "Could not unpickle Injuries_Detail_XML."
+ maybe (return $ ImportFailed errmsg) migrate_and_import m
+
- | dtd == "Injuries_Detail_XML.dtd" =
- dbimport (undefined :: InjuriesDetail.PlayerListing)
+ | dtd == "newsxml.dtd" = do
+ let m = unpickleDoc News.pickle_message xml
+ let errmsg = "Could not unpickle newsxml."
+ maybe (return $ ImportFailed errmsg) migrate_and_import m
- | dtd == "newsxml.dtd" =
- dbimport (undefined :: News.Message)
+ | dtd == "Odds_XML.dtd" = do
+ let m = unpickleDoc Odds.pickle_message xml
+ let errmsg = "Could not unpickle Odds_XML."
+ maybe (return $ ImportFailed errmsg) migrate_and_import m
- | otherwise = \_ -> do -- Dummy arg simplifies the other cases.
+ | otherwise = do
let infomsg =
"Unrecognized DTD in " ++ path ++ ": " ++ dtd ++ "."
- return $ Info infomsg
+ return $ ImportUnsupported infomsg
+-- | 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.
+--
+-- Any file successfully processed is then optionally removed, and
+-- we're done.
+--
main :: IO ()
main = do
rc_cfg <- OC.from_rc
-- 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)
+ init_logging (log_level cfg) (log_file cfg) (syslog cfg)
-- Check the optional config for missing required options.
when (null $ OC.xml_files opt_config) $ 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 = filter (\(_,result) -> result) result_pairs
- mapM_ ((kill True) . fst) victims
+ let victims = [ p | (p, True) <- result_pairs ]
+ let imported_count = length victims
+ report_info $ "Imported " ++ (show imported_count) ++ " document(s) total."
+ when (remove cfg) $ mapM_ (kill True) victims
where
- kill try_again path = do
- removeFile path `catchIOError` exception_handler
- report_info $ "Removed imported file " ++ path ++ "."
+ -- | Wrap these two actions into one function so that we don't
+ -- report that the file was removed if the exception handler is
+ -- run.
+ remove_and_report path = do
+ removeFile path
+ report_info $ "Removed processed file " ++ path ++ "."
+
+ -- | Try to remove @path@ and potentially try again.
+ kill try_again path =
+ (remove_and_report path) `catchIOError` exception_handler
where
-- | A wrapper around threadDelay which takes seconds instead of
-- microseconds as its argument.
let microseconds = seconds * (10 ^ (6 :: Int))
threadDelay microseconds
+ -- | If we can't remove the file, report that, and try once
+ -- more after waiting a few seconds.
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..."
+ report_info "Waiting 5 seconds to attempt removal again..."
thread_sleep 5
kill False path
else