src/TSN/XML/News.hs

   1 {-# LANGUAGE BangPatterns #-}
   2 {-# LANGUAGE DeriveDataTypeable #-}
   3 {-# LANGUAGE FlexibleInstances #-}
   4 {-# LANGUAGE GADTs #-}
   5 {-# LANGUAGE QuasiQuotes #-}
   6 {-# LANGUAGE RecordWildCards #-}
   7 {-# LANGUAGE ScopedTypeVariables #-}
   8 {-# LANGUAGE StandaloneDeriving #-}
   9 {-# LANGUAGE TemplateHaskell #-}
  10 {-# LANGUAGE TypeFamilies #-}
  11
  12 -- | Parse TSN XML for the DTD "newsxml.dtd". Each document contains a
  13 --   root element \<message\> that contains an entire news item.
  14 --
  15 module TSN.XML.News (
  16   Message,
  17   news_tests )
  18 where
  19
  20 import Data.Data ( Data, constrFields, dataTypeConstrs, dataTypeOf )
  21 import Data.List.Utils ( join, split )
  22 import Data.Tuple.Curry ( uncurryN )
  23 import Data.Typeable ( Typeable )
  24 import Database.Groundhog (
  25   defaultMigrationLogger,
  26   insert,
  27   migrate,
  28   runMigration )
  29 import Database.Groundhog.Core ( DefaultKey )
  30 import Database.Groundhog.TH (
  31   groundhog,
  32   mkPersist )
  33 import System.Console.CmdArgs.Default ( Default(..) )
  34 import Test.Tasty ( TestTree, testGroup )
  35 import Test.Tasty.HUnit ( (@?=), testCase )
  36 import Text.XML.HXT.Core (
  37   PU,
  38   XmlPickler(..),
  39   unpickleDoc,
  40   xp12Tuple,
  41   xpAttr,
  42   xpElem,
  43   xpInt,
  44   xpList,
  45   xpOption,
  46   xpPair,
  47   xpText,
  48   xpTriple,
  49   xpWrap )
  50
  51 import TSN.Codegen (
  52   tsn_codegen_config,
  53   tsn_db_field_namer ) -- Used in a test
  54 import TSN.DbImport ( DbImport(..), ImportResult(..) )
  55 import Xml ( ToFromXml(..), pickle_unpickle, unpickleable )
  56
  57
  58
  59 -- | The database type for teams as they show up in the news.  We need
  60 --   this separate from its XML representation because of the
  61 --   DefaultKey pointing to a message. We don't know how to create one
  62 --   of those unless we've just inserted a message into the database,
  63 --   so it screws up pickling.
  64 data NewsTeam =
  65   NewsTeam {
  66     nt_news_id :: DefaultKey Message, -- ^ foreign key.
  67     db_team_name :: String }
  68 deriving instance Eq NewsTeam  -- Standalone instances necessary for
  69 deriving instance Show NewsTeam  -- Groundhog types with DefaultKeys
  70
  71 -- | The XML type for teams as they show up in the news. See
  72 --   'NewsTeam' for why there are two types.
  73 data NewsTeamXml =
  74   NewsTeamXml {
  75     xml_team_name :: String }
  76   deriving (Eq, Show)
  77
  78 -- | Specify how to convert between the two representations NewsTeam
  79 --   (database) and NewsTeamXml (XML).
  80 instance ToFromXml NewsTeam where
  81   type Xml NewsTeam = NewsTeamXml
  82   type Container NewsTeam = Message
  83   -- Use a record wildcard here so GHC doesn't complain that we never
  84   -- used our named fields.
  85   to_xml (NewsTeam {..}) = NewsTeamXml db_team_name
  86   -- We can't create a DefaultKey Message...
  87   from_xml = error "Called from_xml on a NewsTeam."
  88   -- unless we're handed one.
  89   from_xml_fk key = (NewsTeam key) . xml_team_name
  90
  91
  92 -- | The database type for locations as they show up in the news.  We
  93 --   need this separate from its XML representation because of the
  94 --   DefaultKey pointing to a message. We don't know how to create one
  95 --   of those unless we've just inserted a message into the database,
  96 --   so it screws up pickling.
  97 data NewsLocation =
  98   NewsLocation {
  99     loc_news_id :: DefaultKey Message, -- ^ foreign key.
 100     db_city ::String,
 101     db_state :: String,
 102     db_country :: String }
 103 deriving instance Eq NewsLocation -- Standalone instances necessary for
 104 deriving instance Show NewsLocation -- Groundhog types with DefaultKeys
 105
 106 -- | The XML type for locations as they show up in the news. See
 107 --   'NewsLocation' for why there are two types.
 108 data NewsLocationXml =
 109   NewsLocationXml {
 110     xml_city :: String,
 111     xml_state :: String,
 112     xml_country :: String }
 113   deriving (Eq, Show)
 114
 115
 116 -- | Specify how to convert between the two representations
 117 --   NewsLocation (database) and NewsLocationXml (XML).
 118 instance ToFromXml NewsLocation where
 119   type Xml NewsLocation = NewsLocationXml
 120   type Container NewsLocation = Message
 121   -- Use a record wildcard here so GHC doesn't complain that we never
 122   -- used our named fields.
 123   to_xml (NewsLocation {..}) = NewsLocationXml db_city db_state db_country
 124   -- We can't create a DefaultKey Message...
 125   from_xml = error "Called from_xml on a NewsLocation."
 126   -- unless we're given one.
 127   from_xml_fk key (NewsLocationXml x y z) = NewsLocation key x y z
 128
 129
 130 -- | The msg_id child of <message> contains an event_id attribute; we
 131 --   embed it into the 'Message' type. We (pointlessly) use the "db_"
 132 --   prefix here so that the two names collide on "id" when Groundhog
 133 --   is creating its fields using our field namer.
 134 data MsgId =
 135   MsgId {
 136     db_msg_id       :: Int,
 137     db_event_id     :: Maybe Int }
 138   deriving (Data, Eq, Show, Typeable)
 139
 140
 141 data MessageXml =
 142   MessageXml {
 143     xml_xml_file_id :: Int,
 144     xml_heading :: String,
 145     xml_mid :: MsgId,
 146     xml_category :: String,
 147     xml_sport :: String,
 148     xml_url :: String,
 149     xml_teams :: [NewsTeamXml],
 150     xml_locations :: [NewsLocationXml],
 151     xml_sms :: String,
 152     xml_text :: String,
 153     xml_continue :: String,
 154     xml_time_stamp :: String }
 155   deriving (Eq, Show)
 156
 157 data Message =
 158   Message {
 159     db_mid :: MsgId,
 160     db_sport :: String,
 161     db_url :: String,
 162     db_sms :: String,
 163     db_text :: String,
 164     db_continue :: String }
 165   deriving (Data, Eq, Show, Typeable)
 166
 167 instance ToFromXml Message where
 168   type Xml Message = MessageXml
 169   type Container Message = ()
 170
 171   -- Use a record wildcard here so GHC doesn't complain that we never
 172   -- used our named fields.
 173   to_xml (Message {..}) =
 174     MessageXml
 175       def
 176       def
 177       db_mid
 178       def
 179       db_sport
 180       db_url
 181       def
 182       def
 183       db_sms
 184       db_text
 185       db_continue
 186       def
 187
 188   -- We don't need the key argument (from_xml_fk) since the XML type
 189   -- contains more information in this case.
 190   from_xml (MessageXml _ _ c _ e f _ _ i j k _) =
 191     Message c e f i j k
 192
 193
 194 mkPersist tsn_codegen_config [groundhog|
 195 - entity: NewsTeam
 196   dbName: news_teams
 197
 198 - entity: NewsLocation
 199   dbName: news_locations
 200
 201 - entity: Message
 202   dbName: news
 203   constructors:
 204     - name: Message
 205       fields:
 206         - name: db_mid
 207           embeddedType:
 208             - {name: msg_id, dbName: msg_id}
 209             - {name: event_id, dbName: event_id}
 210 - embedded: MsgId
 211   fields:
 212     - name: db_msg_id
 213       dbName: msg_id
 214     - name: db_event_id
 215       dbName: event_id
 216 |]
 217
 218 pickle_news_team :: PU NewsTeamXml
 219 pickle_news_team =
 220   xpElem "team" $
 221     xpWrap (from_string, to_string) xpText
 222   where
 223     to_string :: NewsTeamXml -> String
 224     to_string = xml_team_name
 225
 226     from_string :: String -> NewsTeamXml
 227     from_string = NewsTeamXml
 228
 229 instance XmlPickler NewsTeamXml where
 230   xpickle = pickle_news_team
 231
 232 pickle_msg_id :: PU MsgId
 233 pickle_msg_id =
 234   xpElem "msg_id" $
 235     xpWrap (from_tuple, to_tuple) $
 236     xpPair xpInt (xpAttr "EventId" (xpOption xpInt))
 237   where
 238     from_tuple = uncurryN MsgId
 239     to_tuple m = (db_msg_id m, db_event_id m)
 240
 241 instance XmlPickler MsgId where
 242   xpickle = pickle_msg_id
 243
 244 pickle_location :: PU NewsLocationXml
 245 pickle_location =
 246   xpElem "location" $
 247     xpWrap (from_tuple, to_tuple) $
 248     xpTriple (xpElem "city" xpText)
 249              (xpElem "state" xpText)
 250              (xpElem "country" xpText)
 251   where
 252     from_tuple =
 253       uncurryN NewsLocationXml
 254     to_tuple l = (xml_city l, xml_state l, xml_country l)
 255
 256 instance XmlPickler NewsLocationXml where
 257   xpickle = pickle_location
 258
 259
 260 pickle_message :: PU MessageXml
 261 pickle_message =
 262   xpElem "message" $
 263     xpWrap (from_tuple, to_tuple) $
 264     xp12Tuple (xpElem "XML_File_ID" xpInt)
 265               (xpElem "heading" xpText)
 266               pickle_msg_id
 267               (xpElem "category" xpText)
 268               (xpElem "sport" xpText)
 269               (xpElem "url" xpText)
 270               (xpList $ pickle_news_team)
 271               (xpList $ pickle_location)
 272               (xpElem "SMS" xpText)
 273               (xpElem "text" xpText)
 274               pickle_continue
 275               (xpElem "time_stamp" xpText)
 276   where
 277     from_tuple = uncurryN MessageXml
 278     to_tuple m = (xml_xml_file_id m,
 279                   xml_heading m,
 280                   xml_mid m,
 281                   xml_category m,
 282                   xml_sport m,
 283                   xml_url m,
 284                   xml_teams m,
 285                   xml_locations m,
 286                   xml_sms m,
 287                   xml_text m,
 288                   xml_continue m,
 289                   xml_time_stamp m)
 290
 291     pickle_continue :: PU String
 292     pickle_continue =
 293       xpWrap (to_string, from_string) $
 294         xpElem "continue" $
 295           (xpList $ xpElem "P" xpText)
 296       where
 297         from_string :: String -> [String]
 298         from_string = split "\n"
 299
 300         to_string :: [String] -> String
 301         to_string = join "\n"
 302
 303 instance XmlPickler MessageXml where
 304   xpickle = pickle_message
 305
 306
 307
 308 instance DbImport Message where
 309   dbimport _ xml = do
 310     runMigration defaultMigrationLogger $ do
 311       migrate (undefined :: Message)
 312       migrate (undefined :: NewsTeam)
 313       migrate (undefined :: NewsLocation)
 314     let root_element = unpickleDoc xpickle xml :: Maybe MessageXml
 315     case root_element of
 316       Nothing -> do
 317         let errmsg = "Could not unpickle News message in dbimport."
 318         return $ ImportFailed errmsg
 319       Just message  -> do
 320         news_id <- insert (from_xml message :: Message)
 321         let nts :: [NewsTeam] = map (from_xml_fk news_id)
 322                                     (xml_teams message)
 323         let nlocs :: [NewsLocation] = map (from_xml_fk news_id)
 324                                           (xml_locations message)
 325         nt_ids <- mapM insert nts
 326         loc_ids <- mapM insert nlocs
 327
 328         return $ ImportSucceeded (1 + (length nt_ids) + (length loc_ids))
 329
 330
 331 -- * Tasty Tests
 332 news_tests :: TestTree
 333 news_tests =
 334   testGroup
 335     "News tests"
 336     [ test_news_fields_have_correct_names,
 337       test_pickle_of_unpickle_is_identity,
 338       test_unpickle_succeeds ]
 339
 340
 341 test_news_fields_have_correct_names :: TestTree
 342 test_news_fields_have_correct_names =
 343   testCase "news fields get correct database names" $ do
 344     mapM_ check (zip actual expected)
 345   where
 346     -- This is cool, it uses the (derived) Data instance of
 347     -- News.Message to get its constructor names.
 348     field_names :: [String]
 349     field_names =
 350       constrFields . head $ dataTypeConstrs $ dataTypeOf (undefined :: Message)
 351
 352     expected :: [String]
 353     expected =
 354       map (\x -> tsn_db_field_namer "herp" "derp" 8675309 x 90210) field_names
 355
 356     actual :: [String]
 357     actual = ["mid", "sport", "url", "sms", "text", "continue"]
 358
 359     check (x,y) = (x @?= y)
 360
 361
 362 -- | Warning, succeess of this test does not mean that unpickling
 363 --   succeeded.
 364 test_pickle_of_unpickle_is_identity :: TestTree
 365 test_pickle_of_unpickle_is_identity =
 366   testCase "pickle composed with unpickle is the identity" $ do
 367     let path = "test/xml/newsxml.xml"
 368     (expected :: [MessageXml], actual) <- pickle_unpickle "message" path
 369     actual @?= expected
 370
 371
 372 test_unpickle_succeeds :: TestTree
 373 test_unpickle_succeeds =
 374   testCase "unpickling succeeds" $ do
 375   let path = "test/xml/newsxml.xml"
 376   actual <- unpickleable path pickle_message
 377   let expected = True
 378   actual @?= expected