X-Git-Url: http://gitweb.michael.orlitzky.com/?a=blobdiff_plain;f=src%2FMain.hs;h=a2277057940edabe70df7517f250ed1f7d99649e;hb=HEAD;hp=6c8223db5628b397405eaeeaec454e571e8ceac8;hpb=ef96e8bf0cadf5d602022f8c91914d3cabeb35a0;p=dead%2Fhtsn-import.git diff --git a/src/Main.hs b/src/Main.hs index 6c8223d..a227705 100644 --- a/src/Main.hs +++ b/src/Main.hs @@ -22,6 +22,7 @@ import System.IO.Error ( catchIOError ) import Text.XML.HXT.Core ( ArrowXml, IOStateArrow, + SysConfigList, XmlTree, (>>>), (/>), @@ -45,12 +46,45 @@ 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 ( 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 ) +import TSN.Parse ( format_parse_error ) +import qualified TSN.XML.AutoRacingDriverList as AutoRacingDriverList ( + dtd, + pickle_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.EarlyLine as EarlyLine ( + 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.MLBEarlyLine as MLBEarlyLine ( + dtd, + pickle_message ) +import qualified TSN.XML.JFile as JFile ( dtd, pickle_message ) +import qualified TSN.XML.News as News ( + dtd, + has_only_single_sms, + pickle_message ) +import qualified TSN.XML.Odds as Odds ( dtd, pickle_message ) +import qualified TSN.XML.ScheduleChanges as ScheduleChanges ( + 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, + is_type1, + pickle_message, + teams_are_normal ) +import Xml ( DtdName(..), parse_opts, parse_opts_novalidate ) -- | This is where most of the work happens. This function is called @@ -76,11 +110,15 @@ 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 False + -- One of the arrows returned "nothing." Now that we're + -- validating against the DTDs, this will almost always be + -- caused by a document whose DTD is not present (i.e. is + -- unsupported). So we return "success" to allow the XML file to + -- be deleted. + report_info $ "No DTD for file " ++ path ++ "." + return True (ImportFailed errmsg:_) -> do - report_error errmsg + report_error $ errmsg ++ " (" ++ path ++ ")" return False (ImportSkipped infomsg:_) -> do -- We processed the message but didn't import anything. Return @@ -107,9 +145,11 @@ import_file cfg path = do -- 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 that reads a document into an 'XmlTree'. We take a + -- SysConfigList so our caller can decide whether or not to + -- e.g. validate the document against its DTD. + readA :: SysConfigList -> IOStateArrow s a XmlTree + readA scl = readDocument scl path -- | An arrow which parses the doctype "SYSTEM" of an 'XmlTree'. -- We use these to determine the parser to use. @@ -123,11 +163,27 @@ import_file cfg path = do -- 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. + -- + -- Before we actually run the import, we check it against a list + -- of problem DTDs. These can produce weird errors, and we have + -- checks for them. But with DTD validation enabled, we can't + -- even look inside the document to see what's wrong -- parsing + -- will fail! So for those special document types, we proceed + -- using 'parse_opts_novalidate' instead of the default + -- 'parse_opts'. + -- parse_and_import :: IO [ImportResult] - parse_and_import = - runX (readA >>> (dtdnameA &&& returnA) >>> (arr import_with_dtd)) - >>= - sequence + parse_and_import = do + -- Get the DTD name without validating against it. + ((DtdName dtd) : _) <- runX $ (readA parse_opts_novalidate) >>> dtdnameA + + let problem_dtds = [ News.dtd, Weather.dtd ] + let opts = if dtd `elem` problem_dtds + then parse_opts_novalidate + else parse_opts + + runX ((readA opts) >>> (dtdnameA &&& returnA) >>> (arr import_with_dtd)) + >>= sequence -- | Takes a ('DtdName', 'XmlTree') pair and uses the 'DtdName' -- to determine which function to call on the 'XmlTree'. @@ -135,7 +191,7 @@ import_file cfg path = do 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 @@ -143,40 +199,108 @@ import_file cfg path = do 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 + -- | The error message we return if unpickling fails. + -- + errmsg = "Could not unpickle " ++ dtd ++ "." + + -- | Try to migrate and import using the given pickler @f@; + -- if it works, return the result. Otherwise, return an + -- 'ImportFailed' along with our error message. + -- + go f = maybe + (return $ ImportFailed errmsg) + migrate_and_import + (unpickleDoc f xml) + 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 == AutoRacingDriverList.dtd = + go AutoRacingDriverList.pickle_message + + | dtd == AutoRacingResults.dtd = + go AutoRacingResults.pickle_message - | 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 == AutoRacingSchedule.dtd = + go AutoRacingSchedule.pickle_message + | dtd == EarlyLine.dtd = + go EarlyLine.pickle_message - | 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 + -- GameInfo and SportInfo appear last in the guards + | dtd == Injuries.dtd = go Injuries.pickle_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 + | dtd == InjuriesDetail.dtd = go InjuriesDetail.pickle_message + + | dtd == JFile.dtd = go JFile.pickle_message + + | dtd == MLBEarlyLine.dtd = go MLBEarlyLine.pickle_message + + | dtd == News.dtd = + -- Some of the newsxml docs are busted in predictable ways. + -- We want them to "succeed" so that they're deleted. + -- We already know we can't parse them. + if News.has_only_single_sms xml + then go News.pickle_message + else do + let msg = "Unsupported newsxml.dtd with multiple SMS " ++ + "(" ++ path ++ ")" + return $ ImportUnsupported msg + | dtd == Odds.dtd = go Odds.pickle_message + + | dtd == ScheduleChanges.dtd = go ScheduleChanges.pickle_message + + | dtd == Scores.dtd = go Scores.pickle_message + + -- SportInfo and GameInfo appear last in the guards + | dtd == Weather.dtd = + -- Some of the weatherxml docs are busted in predictable ways. + -- We want them to "succeed" so that they're deleted. + -- We already know we can't parse them. + if Weather.is_type1 xml + then if Weather.teams_are_normal xml + then go Weather.pickle_message + else do + let msg = "Teams in reverse order in weatherxml.dtd" ++ + " (" ++ path ++ ")" + return $ ImportUnsupported msg + else do + let msg = "Unsupported weatherxml.dtd type (" ++ path ++ ")" + return $ ImportUnsupported msg + + | 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 (format_parse_error 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 (format_parse_error err) + Right m -> migrate_and_import m | otherwise = do let infomsg = "Unrecognized DTD in " ++ path ++ ": " ++ dtd ++ "." + -- This should be an impossible case while DTD + -- validation is enabled. If we can parse the file at + -- all, then we have a DTD for it sitting around. And we + -- only have DTDs for supported types. 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. @@ -213,8 +337,8 @@ main = do -- 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." + let processed_count = length victims + report_info $ "Processed " ++ (show processed_count) ++ " document(s) total." when (remove cfg) $ mapM_ (kill True) victims where