1 {-# LANGUAGE DoAndIfThenElse #-}
2 {-# LANGUAGE NoMonomorphismRestriction #-}
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 (
14 import Database.Groundhog.Postgresql (
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 (
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 ),
44 import Network.Services.TSN.Report (
47 import TSN.DbImport ( DbImport(..), ImportResult(..) )
48 import qualified TSN.XML.AutoRacingResults as AutoRacingResults (
51 import qualified TSN.XML.AutoRacingSchedule as AutoRacingSchedule (
54 import qualified TSN.XML.EarlyLine as EarlyLine (
57 import qualified TSN.XML.GameInfo as GameInfo ( dtds, parse_xml )
58 import qualified TSN.XML.Heartbeat as Heartbeat ( dtd, verify )
59 import qualified TSN.XML.Injuries as Injuries ( dtd, pickle_message )
60 import qualified TSN.XML.InjuriesDetail as InjuriesDetail (
63 import qualified TSN.XML.MLBEarlyLine as MLBEarlyLine (
66 import qualified TSN.XML.JFile as JFile ( dtd, pickle_message )
67 import qualified TSN.XML.News as News (
71 import qualified TSN.XML.Odds as Odds ( dtd, pickle_message )
72 import qualified TSN.XML.ScheduleChanges as ScheduleChanges (
75 import qualified TSN.XML.Scores as Scores ( dtd, pickle_message )
76 import qualified TSN.XML.SportInfo as SportInfo ( dtds, parse_xml )
77 import qualified TSN.XML.Weather as Weather (
82 import Xml ( DtdName(..), parse_opts )
85 -- | This is where most of the work happens. This function is called
86 -- on every file that we would like to import. It determines which
87 -- importer to use based on the DTD, attempts to process the file,
88 -- and then returns whether or not it was successful. If the file
89 -- was processed, 'True' is returned. Otherwise, 'False' is
92 -- The implementation is straightforward with one exception: since
93 -- we are already in arrow world with HXT, the @import_with_dtd@
94 -- function is lifted to an 'Arrow' as well with 'arr'. This
95 -- prevents us from having to do a bunch of unwrapping and
96 -- rewrapping with the associated error checking.
98 import_file :: Configuration -- ^ A configuration object needed for the
99 -- 'backend' and 'connection_string'.
101 -> FilePath -- ^ The path of the XML file to import.
103 -> IO Bool -- ^ True if we processed the file, False otherwise.
104 import_file cfg path = do
105 results <- parse_and_import `catch` exception_handler
108 -- One of the arrows returned "nothing."
109 report_error $ "Unable to determine DTD for file " ++ path ++ "."
111 (ImportFailed errmsg:_) -> do
112 report_error $ errmsg ++ " (" ++ path ++ ")"
114 (ImportSkipped infomsg:_) -> do
115 -- We processed the message but didn't import anything. Return
116 -- "success" so that the XML file is deleted.
119 (ImportSucceeded:_) -> do
120 report_info $ "Successfully imported " ++ path ++ "."
122 (ImportUnsupported infomsg:_) -> do
123 -- For now we return "success" for these too, since we know we don't
124 -- support a bunch of DTDs and we want them to get deleted.
128 -- | This will catch *any* exception, even the ones thrown by
129 -- Haskell's 'error' (which should never occur under normal
131 exception_handler :: SomeException -> IO [ImportResult]
132 exception_handler e = do
133 report_error (show e)
134 let errdesc = "Failed to import file " ++ path ++ "."
135 -- Return a nonempty list so we don't claim incorrectly that
136 -- we couldn't parse the DTD.
137 return [ImportFailed errdesc]
139 -- | An arrow that reads a document into an 'XmlTree'.
140 readA :: IOStateArrow s a XmlTree
141 readA = readDocument parse_opts path
143 -- | An arrow which parses the doctype "SYSTEM" of an 'XmlTree'.
144 -- We use these to determine the parser to use.
145 dtdnameA :: ArrowXml a => a XmlTree DtdName
146 dtdnameA = getAttrl >>> hasName "doctype-SYSTEM" /> getText >>^ DtdName
148 -- | Combine the arrows above as well as the function below
149 -- (arrowized with 'arr') into an IO action that does everything
150 -- (parses and then runs the import on what was parsed).
152 -- The result of runX has type IO [IO ImportResult]. We thus use
153 -- bind (>>=) and sequence to combine all of the IOs into one
154 -- big one outside of the list.
155 parse_and_import :: IO [ImportResult]
157 runX (readA >>> (dtdnameA &&& returnA) >>> (arr import_with_dtd))
161 -- | Takes a ('DtdName', 'XmlTree') pair and uses the 'DtdName'
162 -- to determine which function to call on the 'XmlTree'.
163 import_with_dtd :: (DtdName, XmlTree) -> IO ImportResult
164 import_with_dtd (DtdName dtd,xml)
165 -- We special-case the heartbeat so it doesn't have to run in
166 -- the database monad.
167 | dtd == Heartbeat.dtd = Heartbeat.verify xml
169 -- We need NoMonomorphismRestriction here.
170 if backend cfg == Postgres
171 then withPostgresqlConn cs $ runDbConn importer
172 else withSqliteConn cs $ runDbConn importer
174 -- | Pull the real connection String out of the configuration.
177 cs = get_connection_string $ connection_string cfg
179 -- | Convenience; we use this everywhere below in 'importer'.
181 migrate_and_import m = dbmigrate m >> dbimport m
183 -- | The error message we return if unpickling fails.
185 errmsg = "Could not unpickle " ++ dtd ++ "."
187 -- | Try to migrate and import using the given pickler @f@;
188 -- if it works, return the result. Otherwise, return an
189 -- 'ImportFailed' along with our error message.
192 (return $ ImportFailed errmsg)
197 | dtd == AutoRacingResults.dtd =
198 go AutoRacingResults.pickle_message
200 | dtd == AutoRacingSchedule.dtd =
201 go AutoRacingSchedule.pickle_message
203 | dtd == EarlyLine.dtd =
204 go EarlyLine.pickle_message
206 -- GameInfo and SportInfo appear last in the guards
207 | dtd == Injuries.dtd = go Injuries.pickle_message
209 | dtd == InjuriesDetail.dtd = go InjuriesDetail.pickle_message
211 | dtd == JFile.dtd = go JFile.pickle_message
213 | dtd == MLBEarlyLine.dtd =
214 go MLBEarlyLine.pickle_message
217 -- Some of the newsxml docs are busted in predictable ways.
218 -- We want them to "succeed" so that they're deleted.
219 -- We already know we can't parse them.
220 if News.has_only_single_sms xml
221 then go News.pickle_message
223 let msg = "Unsupported newsxml.dtd with multiple SMS " ++
225 return $ ImportUnsupported msg
226 | dtd == Odds.dtd = go Odds.pickle_message
228 | dtd == ScheduleChanges.dtd = go ScheduleChanges.pickle_message
230 | dtd == Scores.dtd = go Scores.pickle_message
232 -- SportInfo and GameInfo appear last in the guards
233 | dtd == Weather.dtd =
234 -- Some of the weatherxml docs are busted in predictable ways.
235 -- We want them to "succeed" so that they're deleted.
236 -- We already know we can't parse them.
237 if Weather.is_type1 xml
238 then if Weather.teams_are_normal xml
239 then go Weather.pickle_message
241 let msg = "Teams in reverse order in weatherxml.dtd" ++
243 return $ ImportUnsupported msg
245 let msg = "Unsupported weatherxml.dtd type (" ++ path ++ ")"
246 return $ ImportUnsupported msg
248 | dtd `elem` GameInfo.dtds = do
249 let either_m = GameInfo.parse_xml dtd xml
251 -- This might give us a slightly better error
252 -- message than the default 'errmsg'.
253 Left err -> return $ ImportFailed err
254 Right m -> migrate_and_import m
256 | dtd `elem` SportInfo.dtds = do
257 let either_m = SportInfo.parse_xml dtd xml
259 -- This might give us a slightly better error
260 -- message than the default 'errmsg'.
261 Left err -> return $ ImportFailed err
262 Right m -> migrate_and_import m
266 "Unrecognized DTD in " ++ path ++ ": " ++ dtd ++ "."
267 return $ ImportUnsupported infomsg
271 -- | Entry point of the program. It twiddles some knobs for
272 -- configuration options and then calls 'import_file' on each XML
273 -- file given on the command-line.
275 -- Any file successfully processed is then optionally removed, and
283 -- Merge the config file options with the command-line ones,
284 -- prefering the command-line ones.
285 let opt_config = rc_cfg <> cmd_cfg
287 -- Update a default config with any options that have been set in
288 -- either the config file or on the command-line. We initialize
289 -- logging before the missing parameter checks below so that we can
291 let cfg = (def :: Configuration) `merge_optional` opt_config
292 init_logging (log_level cfg) (log_file cfg) (syslog cfg)
294 -- Check the optional config for missing required options.
295 when (null $ OC.xml_files opt_config) $ do
296 report_error "No XML files given."
297 exitWith (ExitFailure exit_no_xml_files)
299 -- We don't do this in parallel (for now?) to keep the error
300 -- messages nice and linear.
301 results <- mapM (import_file cfg) (OC.xml_files opt_config)
303 -- Zip the results with the files list to find out which ones can be
305 let result_pairs = zip (OC.xml_files opt_config) results
306 let victims = [ p | (p, True) <- result_pairs ]
307 let processed_count = length victims
308 report_info $ "Processed " ++ (show processed_count) ++ " document(s) total."
309 when (remove cfg) $ mapM_ (kill True) victims
312 -- | Wrap these two actions into one function so that we don't
313 -- report that the file was removed if the exception handler is
315 remove_and_report path = do
317 report_info $ "Removed processed file " ++ path ++ "."
319 -- | Try to remove @path@ and potentially try again.
320 kill try_again path =
321 (remove_and_report path) `catchIOError` exception_handler
323 -- | A wrapper around threadDelay which takes seconds instead of
324 -- microseconds as its argument.
325 thread_sleep :: Int -> IO ()
326 thread_sleep seconds = do
327 let microseconds = seconds * (10 ^ (6 :: Int))
328 threadDelay microseconds
330 -- | If we can't remove the file, report that, and try once
331 -- more after waiting a few seconds.
332 exception_handler :: IOError -> IO ()
333 exception_handler e = do
334 report_error (show e)
335 report_error $ "Failed to remove imported file " ++ path ++ "."
337 report_info "Waiting 5 seconds to attempt removal again..."
341 report_info $ "Giving up on " ++ path ++ "."