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.AutoRacingSchedule as AutoRacingSchedule (
  49   dtd,
  50   pickle_message )
  51 import qualified TSN.XML.Heartbeat as Heartbeat ( dtd, verify )
  52 import qualified TSN.XML.Injuries as Injuries ( dtd, pickle_message )
  53 import qualified TSN.XML.InjuriesDetail as InjuriesDetail (
  54   dtd,
  55   pickle_message )
  56 import qualified TSN.XML.News as News ( dtd, pickle_message )
  57 import qualified TSN.XML.Odds as Odds ( dtd, pickle_message )
  58 import qualified TSN.XML.Scores as Scores ( dtd, pickle_message )
  59 import qualified TSN.XML.Weather as Weather ( dtd, pickle_message )
  60 import Xml ( DtdName(..), parse_opts )
  61
  62
  63 -- | This is where most of the work happens. This function is called
  64 --   on every file that we would like to import. It determines which
  65 --   importer to use based on the DTD, attempts to process the file,
  66 --   and then returns whether or not it was successful. If the file
  67 --   was processed, 'True' is returned. Otherwise, 'False' is
  68 --   returned.
  69 --
  70 --   The implementation is straightforward with one exception: since
  71 --   we are already in arrow world with HXT, the @import_with_dtd@
  72 --   function is lifted to an 'Arrow' as well with 'arr'. This
  73 --   prevents us from having to do a bunch of unwrapping and
  74 --   rewrapping with the associated error checking.
  75 --
  76 import_file :: Configuration -- ^ A configuration object needed for the
  77                              --   'backend' and 'connection_string'.
  78
  79             -> FilePath -- ^ The path of the XML file to import.
  80
  81             -> IO Bool -- ^ True if we processed the file, False otherwise.
  82 import_file cfg path = do
  83   results <- parse_and_import `catch` exception_handler
  84   case results of
  85     []    -> do
  86       -- One of the arrows returned "nothing."
  87       report_error $ "Unable to determine DTD for file " ++ path ++ "."
  88       return False
  89     (ImportFailed errmsg:_) -> do
  90       report_error errmsg
  91       return False
  92     (ImportSkipped infomsg:_) -> do
  93       -- We processed the message but didn't import anything. Return
  94       -- "success" so that the XML file is deleted.
  95       report_info infomsg
  96       return True
  97     (ImportSucceeded:_) -> do
  98       report_info $ "Successfully imported " ++ path ++ "."
  99       return True
 100     (ImportUnsupported infomsg:_) -> do
 101       -- For now we return "success" for these too, since we know we don't
 102       -- support a bunch of DTDs and we want them to get deleted.
 103       report_info infomsg
 104       return True
 105   where
 106     -- | This will catch *any* exception, even the ones thrown by
 107     --   Haskell's 'error' (which should never occur under normal
 108     --   circumstances).
 109     exception_handler :: SomeException -> IO [ImportResult]
 110     exception_handler e = do
 111       report_error (show e)
 112       let errdesc = "Failed to import file " ++ path ++ "."
 113       -- Return a nonempty list so we don't claim incorrectly that
 114       -- we couldn't parse the DTD.
 115       return [ImportFailed errdesc]
 116
 117     -- | An arrow that reads a document into an 'XmlTree'.
 118     readA :: IOStateArrow s a XmlTree
 119     readA = readDocument parse_opts path
 120
 121     -- | An arrow which parses the doctype "SYSTEM" of an 'XmlTree'.
 122     --   We use these to determine the parser to use.
 123     dtdnameA :: ArrowXml a => a XmlTree DtdName
 124     dtdnameA = getAttrl >>> hasName "doctype-SYSTEM" /> getText >>^ DtdName
 125
 126     -- | Combine the arrows above as well as the function below
 127     --   (arrowized with 'arr') into an IO action that does everything
 128     --   (parses and then runs the import on what was parsed).
 129     --
 130     --   The result of runX has type IO [IO ImportResult]. We thus use
 131     --   bind (>>=) and sequence to combine all of the IOs into one
 132     --   big one outside of the list.
 133     parse_and_import :: IO [ImportResult]
 134     parse_and_import =
 135       runX (readA >>> (dtdnameA &&& returnA) >>> (arr import_with_dtd))
 136       >>=
 137       sequence
 138
 139     -- | Takes a ('DtdName', 'XmlTree') pair and uses the 'DtdName'
 140     --   to determine which function to call on the 'XmlTree'.
 141     import_with_dtd :: (DtdName, XmlTree) -> IO ImportResult
 142     import_with_dtd (DtdName dtd,xml)
 143         -- We special-case the heartbeat so it doesn't have to run in
 144         -- the database monad.
 145       | dtd == Heartbeat.dtd = Heartbeat.verify xml
 146       | otherwise =
 147         -- We need NoMonomorphismRestriction here.
 148         if backend cfg == Postgres
 149         then withPostgresqlConn cs $ runDbConn importer
 150         else withSqliteConn cs $ runDbConn importer
 151         where
 152           -- | Pull the real connection String out  of the configuration.
 153           --
 154           cs :: String
 155           cs = get_connection_string $ connection_string cfg
 156
 157           -- | Convenience; we use this everywhere below in 'importer'.
 158           --
 159           migrate_and_import m = dbmigrate m >> dbimport m
 160
 161           importer
 162             | dtd == AutoRacingSchedule.dtd = do
 163                let m = unpickleDoc AutoRacingSchedule.pickle_message xml
 164                maybe (return $ ImportFailed errmsg) migrate_and_import m
 165
 166             | dtd == Injuries.dtd = do
 167                let m = unpickleDoc Injuries.pickle_message xml
 168                maybe (return $ ImportFailed errmsg) migrate_and_import m
 169
 170             | dtd == InjuriesDetail.dtd = do
 171                 let m = unpickleDoc InjuriesDetail.pickle_message xml
 172                 maybe (return $ ImportFailed errmsg) migrate_and_import m
 173
 174
 175             | dtd == News.dtd = do
 176                 let m = unpickleDoc News.pickle_message xml
 177                 maybe (return $ ImportFailed errmsg) migrate_and_import m
 178
 179             | dtd == Odds.dtd = do
 180                 let m = unpickleDoc Odds.pickle_message xml
 181                 maybe (return $ ImportFailed errmsg) migrate_and_import m
 182
 183             | dtd == Scores.dtd = do
 184                 let m = unpickleDoc Scores.pickle_message xml
 185                 maybe (return $ ImportFailed errmsg) migrate_and_import m
 186
 187             | dtd == Weather.dtd = do
 188                 let m = unpickleDoc Weather.pickle_message xml
 189                 maybe (return $ ImportFailed errmsg) migrate_and_import m
 190
 191             | otherwise = do
 192               let infomsg =
 193                     "Unrecognized DTD in " ++ path ++ ": " ++ dtd ++ "."
 194               return $ ImportUnsupported infomsg
 195
 196             where
 197               errmsg = "Could not unpickle " ++ dtd ++ "."
 198
 199
 200 -- | Entry point of the program. It twiddles some knobs for
 201 --   configuration options and then calls 'import_file' on each XML
 202 --   file given on the command-line.
 203 --
 204 --   Any file successfully processed is then optionally removed, and
 205 --   we're done.
 206 --
 207 main :: IO ()
 208 main = do
 209   rc_cfg <- OC.from_rc
 210   cmd_cfg <- get_args
 211
 212   -- Merge the config file options with the command-line ones,
 213   -- prefering the command-line ones.
 214   let opt_config = rc_cfg <> cmd_cfg
 215
 216   -- Update a default config with any options that have been set in
 217   -- either the config file or on the command-line.  We initialize
 218   -- logging before the missing parameter checks below so that we can
 219   -- log the errors.
 220   let cfg = (def :: Configuration) `merge_optional` opt_config
 221   init_logging (log_level cfg) (log_file cfg) (syslog cfg)
 222
 223   -- Check the optional config for missing required options.
 224   when (null $ OC.xml_files opt_config) $ do
 225     report_error "No XML files given."
 226     exitWith (ExitFailure exit_no_xml_files)
 227
 228   -- We don't do this in parallel (for now?) to keep the error
 229   -- messages nice and linear.
 230   results <- mapM (import_file cfg) (OC.xml_files opt_config)
 231
 232   -- Zip the results with the files list to find out which ones can be
 233   -- deleted.
 234   let result_pairs = zip (OC.xml_files opt_config) results
 235   let victims = [ p | (p, True) <- result_pairs ]
 236   let processed_count = length victims
 237   report_info $ "Processed " ++ (show processed_count) ++ " document(s) total."
 238   when (remove cfg) $ mapM_ (kill True) victims
 239
 240   where
 241     -- | Wrap these two actions into one function so that we don't
 242     --   report that the file was removed if the exception handler is
 243     --   run.
 244     remove_and_report path = do
 245       removeFile path
 246       report_info $ "Removed processed file " ++ path ++ "."
 247
 248     -- | Try to remove @path@ and potentially try again.
 249     kill try_again path =
 250       (remove_and_report path) `catchIOError` exception_handler
 251       where
 252         -- | A wrapper around threadDelay which takes seconds instead of
 253         --   microseconds as its argument.
 254         thread_sleep :: Int -> IO ()
 255         thread_sleep seconds = do
 256           let microseconds = seconds * (10 ^ (6 :: Int))
 257           threadDelay microseconds
 258
 259         -- | If we can't remove the file, report that, and try once
 260         --   more after waiting a few seconds.
 261         exception_handler :: IOError -> IO ()
 262         exception_handler e = do
 263           report_error (show e)
 264           report_error $ "Failed to remove imported file " ++ path ++ "."
 265           if try_again then do
 266             report_info "Waiting 5 seconds to attempt removal again..."
 267             thread_sleep 5
 268             kill False path
 269           else
 270             report_info $ "Giving up on " ++ path ++ "."