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