+{-# 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 )
-import Control.Monad.IO.Class ( liftIO )
-import Database.Groundhog (
- defaultMigrationLogger,
- insert,
- migrate,
- runMigration )
-import Database.Groundhog.Core ( PersistEntity )
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,
- SysConfigList,
- XmlPickler,
XmlTree,
(>>>),
(/>),
getAttrl,
getText,
hasName,
- no,
readDocument,
runX,
- unpickleDoc,
- withPreserveComment,
- withRemoveWS,
- withSubstDTDEntities,
- withValidate,
- xpickle,
- yes )
+ unpickleDoc )
+-- Local imports.
import Backend ( Backend(..) )
import CommandLine ( get_args )
import Configuration ( Configuration(..), merge_optional )
import ConnectionString ( ConnectionString(..) )
import ExitCodes ( exit_no_xml_files )
-import Network.Services.TSN.Logging ( init_logging )
import qualified OptionalConfiguration as OC (
OptionalConfiguration ( xml_files ),
from_rc )
import Network.Services.TSN.Report (
report_info,
report_error )
-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.
+import TSN.DbImport ( DbImport(..), ImportResult(..) )
+import qualified TSN.XML.Heartbeat as Heartbeat ( verify )
+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.
--
-parse_opts :: SysConfigList
-parse_opts =
- [ withPreserveComment no,
- withRemoveWS yes,
- withSubstDTDEntities no,
- withValidate no ]
+-- 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.
--- | We put the 'Configuration' and 'XmlTree' arguments 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
- -> Configuration
- -> XmlTree
- -> IO (Maybe Int) -- ^ Return the number of records inserted.
-import_generic dummy g cfg xml
- | backend cfg == Postgres = withPostgresqlConn cs go
- | otherwise = withSqliteConn cs go
- where
- -- | Pull the real connection String out of the configuration.
- cs :: String
- cs = get_connection_string $ connection_string cfg
-
- -- Needs NoMonomorphismRestriction to be allowed to return
- -- different types in the two cases above.
- go = runDbConn $ do
- runMigration defaultMigrationLogger $ migrate dummy
- let root_element = unpickleDoc xpickle xml
- case root_element of
- Nothing -> do
- let msg = "Could not unpickle document in import_generic."
- liftIO $ report_error msg
- return Nothing
- Just elt -> do
- ids <- mapM insert (g elt)
- return $ Just (length ids)
-
-
-
--- | Import TSN.Injuries from an 'XmlTree'.
-import_injuries :: Configuration -> XmlTree -> IO (Maybe Int)
-import_injuries =
- import_generic
- (undefined :: Injuries.Listing)
- Injuries.listings
-
--- | Import TSN.InjuriesDetail from an 'XmlTree'.
-import_injuries_detail :: Configuration -> XmlTree -> IO (Maybe Int)
-import_injuries_detail =
- import_generic
- (undefined :: InjuriesDetail.PlayerListing)
- ( (concatMap InjuriesDetail.player_listings) . InjuriesDetail.listings)
-
-import_file :: Configuration -> FilePath -> IO ()
+ -> IO Bool -- ^ True if we processed the file, False otherwise.
import_file cfg path = do
- results <- catchIOError
- parse_and_import
- (\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 ] )
-
+ results <- parse_and_import `catch` exception_handler
case results of
- -- If results' is empty, one of the arrows return "nothing."
- [] -> report_error $ "Unable to determine DTD for file " ++ path ++ "."
- (r:_) ->
- case r of
- Nothing -> return ()
- Just cnt -> report_info $ "Successfully imported " ++
- (show cnt) ++
- " records from " ++ path ++ "."
+ [] -> do
+ -- One of the arrows returned "nothing."
+ report_error $ "Unable to determine DTD for file " ++ path ++ "."
+ return False
+ (ImportFailed errmsg:_) -> do
+ report_error errmsg
+ 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 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 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 [ImportFailed 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
+ 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
-- (parses and then runs the import on what was parsed).
--
- -- The result of runX has type IO [IO (Maybe Int)]. We thus use
+ -- 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 [Maybe Int]
+ 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 (Maybe Int)
- import_with_dtd (dtd,xml)
- | dtd == "injuriesxml.dtd" = import_injuries cfg xml
- | dtd == "Injuries_Detail_XML.dtd" = import_injuries_detail cfg xml
- | otherwise = do
- report_info $ "Unrecognized DTD in " ++ path ++ ": " ++ dtd ++ "."
- return Nothing
-
+ -- | 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
+ 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" = 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 == "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 == "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
+ let infomsg =
+ "Unrecognized DTD in " ++ path ++ ": " ++ dtd ++ "."
+ 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
-- We don't do this in parallel (for now?) to keep the error
-- messages nice and linear.
- mapM_ (import_file cfg) (OC.xml_files opt_config)
+ 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 = [ 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
+ -- | 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.
+ thread_sleep :: Int -> IO ()
+ thread_sleep seconds = do
+ 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..."
+ thread_sleep 5
+ kill False path
+ else
+ report_info $ "Giving up on " ++ path ++ "."