X-Git-Url: http://gitweb.michael.orlitzky.com/?a=blobdiff_plain;f=src%2FMain.hs;h=a2277057940edabe70df7517f250ed1f7d99649e;hb=HEAD;hp=fc6d14b8a116aef185d0c7d362b2204c912ff135;hpb=3768238ca528db394e4777ac1e19b2091979e1c8;p=dead%2Fhtsn-import.git diff --git a/src/Main.hs b/src/Main.hs index fc6d14b..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, (>>>), (/>), @@ -45,28 +46,45 @@ import Network.Services.TSN.Report ( report_info, report_error ) import TSN.DbImport ( DbImport(..), ImportResult(..) ) +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, 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, pickle_message ) -import Xml ( DtdName(..), parse_opts ) +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 @@ -92,9 +110,13 @@ 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 ++ " (" ++ path ++ ")" return False @@ -123,9 +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. @@ -139,11 +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'. @@ -181,12 +221,18 @@ import_file cfg path = do (unpickleDoc f xml) importer + | dtd == AutoRacingDriverList.dtd = + go AutoRacingDriverList.pickle_message + | dtd == AutoRacingResults.dtd = go AutoRacingResults.pickle_message | dtd == AutoRacingSchedule.dtd = go AutoRacingSchedule.pickle_message + | dtd == EarlyLine.dtd = + go EarlyLine.pickle_message + -- GameInfo and SportInfo appear last in the guards | dtd == Injuries.dtd = go Injuries.pickle_message @@ -194,8 +240,18 @@ import_file cfg path = do | dtd == JFile.dtd = go JFile.pickle_message - | dtd == News.dtd = go News.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 @@ -203,14 +259,27 @@ import_file cfg path = do | dtd == Scores.dtd = go Scores.pickle_message -- SportInfo and GameInfo appear last in the guards - | dtd == Weather.dtd = go Weather.pickle_message + | 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 err + Left err -> return $ ImportFailed (format_parse_error err) Right m -> migrate_and_import m | dtd `elem` SportInfo.dtds = do @@ -218,12 +287,16 @@ import_file cfg path = do case either_m of -- This might give us a slightly better error -- message than the default 'errmsg'. - Left err -> return $ ImportFailed err + 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