src/Main.hs

   1 {-# LANGUAGE DoAndIfThenElse #-}
   2 {-# LANGUAGE NoMonomorphismRestriction #-}
   3 module Main
   4 where
   5
   6 -- System imports.
   7 import Control.Arrow ( (&&&), (>>^), arr, returnA )
   8 import Control.Concurrent ( threadDelay )
   9 import Control.Exception ( SomeException, catch )
  10 import Control.Monad ( when )
  11 import Database.Groundhog.Generic ( runDbConn )
  12 import Database.Groundhog.Sqlite (
  13   withSqliteConn )
  14 import Database.Groundhog.Postgresql (
  15   withPostgresqlConn )
  16 import Data.Monoid ( (<>) )
  17 import Network.Services.TSN.Logging ( init_logging )
  18 import System.Console.CmdArgs ( def )
  19 import System.Directory ( removeFile )
  20 import System.Exit ( exitWith, ExitCode (ExitFailure) )
  21 import System.IO.Error ( catchIOError )
  22 import Text.XML.HXT.Core (
  23   ArrowXml,
  24   IOStateArrow,
  25   XmlTree,
  26   (>>>),
  27   (/>),
  28   getAttrl,
  29   getText,
  30   hasName,
  31   readDocument,
  32   runX,
  33   unpickleDoc )
  34
  35 -- Local imports.
  36 import Backend ( Backend(..) )
  37 import CommandLine ( get_args )
  38 import Configuration ( Configuration(..), merge_optional )
  39 import ConnectionString ( ConnectionString(..) )
  40 import ExitCodes ( exit_no_xml_files )
  41 import qualified OptionalConfiguration as OC (
  42   OptionalConfiguration ( xml_files ),
  43   from_rc )
  44 import Network.Services.TSN.Report (
  45   report_info,
  46   report_error )
  47 import TSN.DbImport ( DbImport(..), ImportResult(..) )
  48 import qualified TSN.XML.AutoRacingResults as AutoRacingResults (
  49   dtd,
  50   pickle_message )
  51 import qualified TSN.XML.AutoRacingSchedule as AutoRacingSchedule (
  52   dtd,
  53   pickle_message )
  54 import qualified TSN.XML.GameInfo as GameInfo ( dtds, parse_xml )
  55 import qualified TSN.XML.Heartbeat as Heartbeat ( dtd, verify )
  56 import qualified TSN.XML.Injuries as Injuries ( dtd, pickle_message )
  57 import qualified TSN.XML.InjuriesDetail as InjuriesDetail (
  58   dtd,
  59   pickle_message )
  60 import qualified TSN.XML.JFile as JFile ( dtd, pickle_message )
  61 import qualified TSN.XML.News as News ( dtd, pickle_message )
  62 import qualified TSN.XML.Odds as Odds ( dtd, pickle_message )
  63 import qualified TSN.XML.ScheduleChanges as ScheduleChanges (
  64   dtd,
  65   pickle_message )
  66 import qualified TSN.XML.Scores as Scores ( dtd, pickle_message )
  67 import qualified TSN.XML.SportInfo as SportInfo ( dtds, parse_xml )
  68 import qualified TSN.XML.Weather as Weather (
  69   dtd,
  70   is_type1,
  71   pickle_message,
  72   teams_are_normal )
  73 import Xml ( DtdName(..), parse_opts )
  74
  75
  76 -- | This is where most of the work happens. This function is called
  77 --   on every file that we would like to import. It determines which
  78 --   importer to use based on the DTD, attempts to process the file,
  79 --   and then returns whether or not it was successful. If the file
  80 --   was processed, 'True' is returned. Otherwise, 'False' is
  81 --   returned.
  82 --
  83 --   The implementation is straightforward with one exception: since
  84 --   we are already in arrow world with HXT, the @import_with_dtd@
  85 --   function is lifted to an 'Arrow' as well with 'arr'. This
  86 --   prevents us from having to do a bunch of unwrapping and
  87 --   rewrapping with the associated error checking.
  88 --
  89 import_file :: Configuration -- ^ A configuration object needed for the
  90                              --   'backend' and 'connection_string'.
  91
  92             -> FilePath -- ^ The path of the XML file to import.
  93
  94             -> IO Bool -- ^ True if we processed the file, False otherwise.
  95 import_file cfg path = do
  96   results <- parse_and_import `catch` exception_handler
  97   case results of
  98     []    -> do
  99       -- One of the arrows returned "nothing."
 100       report_error $ "Unable to determine DTD for file " ++ path ++ "."
 101       return False
 102     (ImportFailed errmsg:_) -> do
 103       report_error $ errmsg ++ " (" ++ path ++ ")"
 104       return False
 105     (ImportSkipped infomsg:_) -> do
 106       -- We processed the message but didn't import anything. Return
 107       -- "success" so that the XML file is deleted.
 108       report_info infomsg
 109       return True
 110     (ImportSucceeded:_) -> do
 111       report_info $ "Successfully imported " ++ path ++ "."
 112       return True
 113     (ImportUnsupported infomsg:_) -> do
 114       -- For now we return "success" for these too, since we know we don't
 115       -- support a bunch of DTDs and we want them to get deleted.
 116       report_info infomsg
 117       return True
 118   where
 119     -- | This will catch *any* exception, even the ones thrown by
 120     --   Haskell's 'error' (which should never occur under normal
 121     --   circumstances).
 122     exception_handler :: SomeException -> IO [ImportResult]
 123     exception_handler e = do
 124       report_error (show e)
 125       let errdesc = "Failed to import file " ++ path ++ "."
 126       -- Return a nonempty list so we don't claim incorrectly that
 127       -- we couldn't parse the DTD.
 128       return [ImportFailed errdesc]
 129
 130     -- | An arrow that reads a document into an 'XmlTree'.
 131     readA :: IOStateArrow s a XmlTree
 132     readA = readDocument parse_opts path
 133
 134     -- | An arrow which parses the doctype "SYSTEM" of an 'XmlTree'.
 135     --   We use these to determine the parser to use.
 136     dtdnameA :: ArrowXml a => a XmlTree DtdName
 137     dtdnameA = getAttrl >>> hasName "doctype-SYSTEM" /> getText >>^ DtdName
 138
 139     -- | Combine the arrows above as well as the function below
 140     --   (arrowized with 'arr') into an IO action that does everything
 141     --   (parses and then runs the import on what was parsed).
 142     --
 143     --   The result of runX has type IO [IO ImportResult]. We thus use
 144     --   bind (>>=) and sequence to combine all of the IOs into one
 145     --   big one outside of the list.
 146     parse_and_import :: IO [ImportResult]
 147     parse_and_import =
 148       runX (readA >>> (dtdnameA &&& returnA) >>> (arr import_with_dtd))
 149       >>=
 150       sequence
 151
 152     -- | Takes a ('DtdName', 'XmlTree') pair and uses the 'DtdName'
 153     --   to determine which function to call on the 'XmlTree'.
 154     import_with_dtd :: (DtdName, XmlTree) -> IO ImportResult
 155     import_with_dtd (DtdName dtd,xml)
 156         -- We special-case the heartbeat so it doesn't have to run in
 157         -- the database monad.
 158       | dtd == Heartbeat.dtd = Heartbeat.verify xml
 159       | otherwise =
 160         -- We need NoMonomorphismRestriction here.
 161         if backend cfg == Postgres
 162         then withPostgresqlConn cs $ runDbConn importer
 163         else withSqliteConn cs $ runDbConn importer
 164         where
 165           -- | Pull the real connection String out  of the configuration.
 166           --
 167           cs :: String
 168           cs = get_connection_string $ connection_string cfg
 169
 170           -- | Convenience; we use this everywhere below in 'importer'.
 171           --
 172           migrate_and_import m = dbmigrate m >> dbimport m
 173
 174           -- | The error message we return if unpickling fails.
 175           --
 176           errmsg = "Could not unpickle " ++ dtd ++ "."
 177
 178           -- | Try to migrate and import using the given pickler @f@;
 179           --   if it works, return the result. Otherwise, return an
 180           --   'ImportFailed' along with our error message.
 181           --
 182           go f = maybe
 183                    (return $ ImportFailed errmsg)
 184                    migrate_and_import
 185                    (unpickleDoc f xml)
 186
 187           importer
 188             | dtd == AutoRacingResults.dtd =
 189                 go AutoRacingResults.pickle_message
 190
 191             | dtd == AutoRacingSchedule.dtd =
 192                 go AutoRacingSchedule.pickle_message
 193
 194             -- GameInfo and SportInfo appear last in the guards
 195             | dtd == Injuries.dtd = go Injuries.pickle_message
 196
 197             | dtd == InjuriesDetail.dtd = go InjuriesDetail.pickle_message
 198
 199             | dtd == JFile.dtd = go JFile.pickle_message
 200
 201             | dtd == News.dtd = go News.pickle_message
 202
 203             | dtd == Odds.dtd = go Odds.pickle_message
 204
 205             | dtd == ScheduleChanges.dtd = go ScheduleChanges.pickle_message
 206
 207             | dtd == Scores.dtd = go Scores.pickle_message
 208
 209             -- SportInfo and GameInfo appear last in the guards
 210             | dtd == Weather.dtd =
 211                 -- Some of the weatherxml docs are busted in predictable ways.
 212                 -- We want them to "succeed" so that they're deleted.
 213                 -- We already know we can't parse them.
 214                 if Weather.is_type1 xml
 215                 then if Weather.teams_are_normal xml
 216                      then go Weather.pickle_message
 217                      else do
 218                        let msg = "Teams in reverse order in weatherxml.dtd" ++
 219                                  " (" ++ path ++ ")"
 220                        return $ ImportUnsupported msg
 221                 else do
 222                   let msg = "Unsupported weatherxml.dtd type (" ++ path ++ ")"
 223                   return $ ImportUnsupported msg
 224
 225             | dtd `elem` GameInfo.dtds = do
 226                 let either_m = GameInfo.parse_xml dtd xml
 227                 case either_m of
 228                   -- This might give us a slightly better error
 229                   -- message than the default 'errmsg'.
 230                   Left err -> return $ ImportFailed err
 231                   Right m     -> migrate_and_import m
 232
 233             | dtd `elem` SportInfo.dtds = do
 234                 let either_m = SportInfo.parse_xml dtd xml
 235                 case either_m of
 236                   -- This might give us a slightly better error
 237                   -- message than the default 'errmsg'.
 238                   Left err -> return $ ImportFailed err
 239                   Right m     -> migrate_and_import m
 240
 241             | otherwise = do
 242               let infomsg =
 243                     "Unrecognized DTD in " ++ path ++ ": " ++ dtd ++ "."
 244               return $ ImportUnsupported infomsg
 245
 246
 247
 248 -- | Entry point of the program. It twiddles some knobs for
 249 --   configuration options and then calls 'import_file' on each XML
 250 --   file given on the command-line.
 251 --
 252 --   Any file successfully processed is then optionally removed, and
 253 --   we're done.
 254 --
 255 main :: IO ()
 256 main = do
 257   rc_cfg <- OC.from_rc
 258   cmd_cfg <- get_args
 259
 260   -- Merge the config file options with the command-line ones,
 261   -- prefering the command-line ones.
 262   let opt_config = rc_cfg <> cmd_cfg
 263
 264   -- Update a default config with any options that have been set in
 265   -- either the config file or on the command-line.  We initialize
 266   -- logging before the missing parameter checks below so that we can
 267   -- log the errors.
 268   let cfg = (def :: Configuration) `merge_optional` opt_config
 269   init_logging (log_level cfg) (log_file cfg) (syslog cfg)
 270
 271   -- Check the optional config for missing required options.
 272   when (null $ OC.xml_files opt_config) $ do
 273     report_error "No XML files given."
 274     exitWith (ExitFailure exit_no_xml_files)
 275
 276   -- We don't do this in parallel (for now?) to keep the error
 277   -- messages nice and linear.
 278   results <- mapM (import_file cfg) (OC.xml_files opt_config)
 279
 280   -- Zip the results with the files list to find out which ones can be
 281   -- deleted.
 282   let result_pairs = zip (OC.xml_files opt_config) results
 283   let victims = [ p | (p, True) <- result_pairs ]
 284   let processed_count = length victims
 285   report_info $ "Processed " ++ (show processed_count) ++ " document(s) total."
 286   when (remove cfg) $ mapM_ (kill True) victims
 287
 288   where
 289     -- | Wrap these two actions into one function so that we don't
 290     --   report that the file was removed if the exception handler is
 291     --   run.
 292     remove_and_report path = do
 293       removeFile path
 294       report_info $ "Removed processed file " ++ path ++ "."
 295
 296     -- | Try to remove @path@ and potentially try again.
 297     kill try_again path =
 298       (remove_and_report path) `catchIOError` exception_handler
 299       where
 300         -- | A wrapper around threadDelay which takes seconds instead of
 301         --   microseconds as its argument.
 302         thread_sleep :: Int -> IO ()
 303         thread_sleep seconds = do
 304           let microseconds = seconds * (10 ^ (6 :: Int))
 305           threadDelay microseconds
 306
 307         -- | If we can't remove the file, report that, and try once
 308         --   more after waiting a few seconds.
 309         exception_handler :: IOError -> IO ()
 310         exception_handler e = do
 311           report_error (show e)
 312           report_error $ "Failed to remove imported file " ++ path ++ "."
 313           if try_again then do
 314             report_info "Waiting 5 seconds to attempt removal again..."
 315             thread_sleep 5
 316             kill False path
 317           else
 318             report_info $ "Giving up on " ++ path ++ "."