X-Git-Url: http://gitweb.michael.orlitzky.com/?a=blobdiff_plain;f=src%2FTSN%2FXML%2FNews.hs;h=3f9fef5375b58ec440eed4c041985871ed0d6300;hb=83bad08d7f28143cdaae42156d951b421fa15a8a;hp=a388415ff7109264e418d3338a68faea77654ee1;hpb=8fd79929d4e139608ecde40fc70703e0efd56f30;p=dead%2Fhtsn-import.git diff --git a/src/TSN/XML/News.hs b/src/TSN/XML/News.hs index a388415..3f9fef5 100644 --- a/src/TSN/XML/News.hs +++ b/src/TSN/XML/News.hs @@ -3,26 +3,40 @@ {-# LANGUAGE GADTs #-} {-# LANGUAGE QuasiQuotes #-} {-# LANGUAGE RecordWildCards #-} -{-# LANGUAGE StandaloneDeriving #-} {-# LANGUAGE TemplateHaskell #-} {-# LANGUAGE TypeFamilies #-} --- | Parse TSN XML for the DTD "newsxml.dtd". Each document contains a --- root element \ that contains an entire news item. +-- | Parse TSN XML for the DTD \"newsxml.dtd\". Each document contains +-- a root element \ that contains an entire news item. -- module TSN.XML.News ( + dtd, + pickle_message, + -- * Tests news_tests, - pickle_message ) + -- * WARNING: these are private but exported to silence warnings + News_LocationConstructor(..), + News_NewsTeamConstructor(..), + NewsConstructor(..), + NewsTeamConstructor(..) ) where +-- System imports. import Data.Data ( Data, constrFields, dataTypeConstrs, dataTypeOf ) +import Data.Time.Clock ( UTCTime ) import Data.List.Utils ( join, split ) import Data.Tuple.Curry ( uncurryN ) import Data.Typeable ( Typeable ) import Database.Groundhog ( + countAll, + deleteAll, insert_, - migrate ) + migrate, + runMigration, + silentMigrationLogger ) import Database.Groundhog.Core ( DefaultKey ) +import Database.Groundhog.Generic ( runDbConn ) +import Database.Groundhog.Sqlite ( withSqliteConn ) import Database.Groundhog.TH ( defaultCodegenConfig, groundhog, @@ -39,66 +53,42 @@ import Text.XML.HXT.Core ( xpOption, xpPair, xpText, - xpTriple, xpWrap ) +-- Local imports. import TSN.Codegen ( tsn_codegen_config, tsn_db_field_namer ) -- Used in a test +import TSN.Database ( insert_or_select ) import TSN.DbImport ( DbImport(..), ImportResult(..), run_dbmigrate ) +import TSN.Picklers ( xp_time_stamp ) +import TSN.Location ( Location(..), pickle_location ) import TSN.XmlImport ( XmlImport(..) ) -import Xml ( FromXml(..), pickle_unpickle, unpickleable ) +import Xml ( + FromXml(..), + ToDb(..), + pickle_unpickle, + unpickleable, + unsafe_unpickle ) - --- | The database type for teams as they show up in the news. -data NewsTeam = - NewsTeam { team_name :: String } - deriving (Eq, Show) - -instance FromXml NewsTeam where - type Db NewsTeam = NewsTeam - from_xml = id - -instance XmlImport NewsTeam - - --- | Mapping between News records and NewsTeam records in the --- database. We name the fields (even though they're never used) for --- Groundhog's benefit. -data News_NewsTeam = - News_NewsTeam { - nnt_news_id :: DefaultKey News, - nnt_news_team_id :: DefaultKey NewsTeam } - --- | The database type for locations as they show up in the news. -data NewsLocation = - NewsLocation { - city :: Maybe String, - state :: Maybe String, - country :: String } - deriving (Eq, Show) - -instance FromXml NewsLocation where - type Db NewsLocation = NewsLocation - from_xml = id - -instance XmlImport NewsLocation +-- | The DTD to which this module corresponds. Used to invoke dbimport. +-- +dtd :: String +dtd = "newsxml.dtd" --- | Mapping between News records and NewsLocation records in the --- database. We name the fields (even though they're never used) for --- Groundhog's benefit. -data News_NewsLocation = - News_NewsLocation { - nnl_news_id :: DefaultKey News, - nnl_news_location_id :: DefaultKey NewsLocation } +-- +-- DB/XML Data types +-- +-- * News/Message --- | The msg_id child of contains an event_id attribute; we --- embed it into the 'News' type. We (pointlessly) use the "db_" --- prefix here so that the two names collide on "id" when Groundhog --- is creating its fields using our field namer. +-- | The msg_id child of \ contains an event_id attribute; we +-- embed it into the 'News' type. We (pointlessly) use the \"db_\" +-- prefix here so that the two names don't collide on \"id\" when +-- Groundhog is creating its fields using our field namer. +-- data MsgId = MsgId { db_msg_id :: Int, @@ -106,6 +96,8 @@ data MsgId = deriving (Data, Eq, Show, Typeable) +-- | The XML representation of a news item (\). +-- data Message = Message { xml_xml_file_id :: Int, @@ -115,70 +107,143 @@ data Message = xml_sport :: String, xml_url :: Maybe String, xml_teams :: [NewsTeam], - xml_locations :: [NewsLocation], + xml_locations :: [Location], xml_sms :: String, xml_editor :: Maybe String, xml_text :: Maybe String, -- Text and continue seem to show up in pairs, xml_continue :: Maybe String, -- either both present or both missing. - xml_time_stamp :: String } + xml_time_stamp :: UTCTime } deriving (Eq, Show) + +-- | The database representation of a news item. We drop several +-- uninteresting fields from 'Message', and omit the list fields which +-- will be represented as join tables. +-- data News = News { + db_xml_file_id :: Int, db_mid :: MsgId, db_sport :: String, db_url :: Maybe String, db_sms :: String, db_editor :: Maybe String, db_text :: Maybe String, - db_continue :: Maybe String } + db_continue :: Maybe String, + db_time_stamp :: UTCTime } deriving (Data, Eq, Show, Typeable) -instance FromXml Message where - type Db Message = News - -- We don't need the key argument (from_xml_fk) since the XML type - -- contains more information in this case. - from_xml (Message _ _ c _ e f _ _ i j k l _) = - News c e f i j k l +instance ToDb Message where + -- | The database representation of 'Message' is 'News'. + type Db Message = News + +-- | Convert the XML representation 'Message' to the database +-- representation 'News'. +-- +instance FromXml Message where + -- | We use a record wildcard so GHC doesn't complain that we never + -- used the field names. + -- + from_xml Message{..} = News { db_xml_file_id = xml_xml_file_id, + db_mid = xml_mid, + db_sport = xml_sport, + db_url = xml_url, + db_sms = xml_sms, + db_editor = xml_editor, + db_text = xml_text, + db_continue = xml_continue, + db_time_stamp = xml_time_stamp } + +-- | This lets us insert the XML representation 'Message' directly. +-- instance XmlImport Message + +-- * NewsTeam + +-- | The database/XML type for teams as they show up in the news. We +-- can't reuse the representation from "TSN.Team" because they +-- require a team id. We wouldn't want to make the team ID optional +-- and then insert a team with no id, only to find the same team +-- later with an id and be unable to update the record. (We could +-- add the update logic, but it would be more trouble than it's +-- worth.) +-- +data NewsTeam = + NewsTeam { team_name :: String } + deriving (Eq, Show) + + + +-- * News_NewsTeam + +-- | Mapping between News records and NewsTeam records in the +-- database. We don't name the fields because we don't use the names +-- explicitly; that means we have to give them nice database names +-- via groundhog. +-- +data News_NewsTeam = News_NewsTeam + (DefaultKey News) + (DefaultKey NewsTeam) + + +-- * News_Location + +-- | Mapping between 'News' records and 'Location' records in the +-- database. We don't name the fields because we don't use the names +-- explicitly; that means we have to give them nice database names +-- via groundhog. +-- +data News_Location = News_Location + (DefaultKey News) + (DefaultKey Location) + + + +-- +-- Database code +-- + +-- | Define 'dbmigrate' and 'dbimport' for 'Message's. The import is +-- slightly non-generic because of our 'News_NewsTeam' and +-- 'News_Location' join tables. +-- instance DbImport Message where dbmigrate _ = run_dbmigrate $ do - migrate (undefined :: NewsTeam) - migrate (undefined :: NewsLocation) + migrate (undefined :: Location) migrate (undefined :: News) + migrate (undefined :: NewsTeam) migrate (undefined :: News_NewsTeam) - migrate (undefined :: News_NewsLocation) + migrate (undefined :: News_Location) dbimport message = do -- Insert the message and acquire its primary key (unique ID) news_id <- insert_xml message - -- And insert each one into its own table. We use insert_xml_or_select - -- because we know that most teams will already exist, and we - -- want to get back the id for the existing team when - -- there's a collision. - nt_ids <- mapM insert_xml_or_select (xml_teams message) + -- Now insert the teams. We use insert_or_select because we know + -- that most teams will already exist, and we want to get back the + -- id for the existing team when there's a collision. + nt_ids <- mapM insert_or_select (xml_teams message) -- Now that the teams have been inserted, create -- news__news_team records mapping beween the two. let news_news_teams = map (News_NewsTeam news_id) nt_ids mapM_ insert_ news_news_teams - -- Do all of that over again for the NewsLocations. - loc_ids <- mapM insert_xml_or_select (xml_locations message) - let news_news_locations = map (News_NewsLocation news_id) loc_ids + -- Do all of that over again for the Locations. + loc_ids <- mapM insert_or_select (xml_locations message) + let news_news_locations = map (News_Location news_id) loc_ids mapM_ insert_ news_news_locations return ImportSucceeded -- These types don't have special XML representations or field name --- collisions so we use the defaultCodegenConfig and give their fields --- nice simple names. +-- collisions so we use the defaultCodegenConfig and give their +-- fields nice simple names. mkPersist defaultCodegenConfig [groundhog| - entity: NewsTeam dbName: news_teams @@ -189,27 +254,26 @@ mkPersist defaultCodegenConfig [groundhog| type: constraint fields: [team_name] -- entity: NewsLocation - dbName: news_locations - constructors: - - name: NewsLocation - uniques: - - name: unique_news_location - type: constraint - fields: [city, state, country] - |] + +-- These types have fields with e.g. db_ and xml_ prefixes, so we +-- use our own codegen to peel those off before naming the columns. mkPersist tsn_codegen_config [groundhog| - entity: News - dbName: news constructors: - name: News + uniques: + - name: unique_news + type: constraint + # Prevent multiple imports of the same message. + fields: [db_xml_file_id] fields: - name: db_mid embeddedType: - {name: msg_id, dbName: msg_id} - {name: event_id, dbName: event_id} + - embedded: MsgId fields: - name: db_msg_id @@ -217,14 +281,42 @@ mkPersist tsn_codegen_config [groundhog| - name: db_event_id dbName: event_id - - entity: News_NewsTeam dbName: news__news_teams - -- entity: News_NewsLocation - dbName: news__news_locations + constructors: + - name: News_NewsTeam + fields: + - name: news_NewsTeam0 # Default created by mkNormalFieldName + dbName: news_id + reference: + onDelete: cascade + - name: news_NewsTeam1 # Default created by mkNormalFieldName + dbName: news_teams_id + reference: + onDelete: cascade + +- entity: News_Location + dbName: news__locations + constructors: + - name: News_Location + fields: + - name: news_Location0 # Default created by mkNormalFieldName + dbName: news_id + reference: + onDelete: cascade + - name: news_Location1 # Default created by mkNormalFieldName + dbName: locations_id + reference: + onDelete: cascade |] + +-- +-- XML Picklers +-- + +-- | Convert a 'NewsTeam' to/from XML. +-- pickle_news_team :: PU NewsTeam pickle_news_team = xpElem "team" $ @@ -237,6 +329,8 @@ pickle_news_team = from_string = NewsTeam +-- | Convert a 'MsgId' to/from XML. +-- pickle_msg_id :: PU MsgId pickle_msg_id = xpElem "msg_id" $ @@ -247,20 +341,9 @@ pickle_msg_id = to_tuple m = (db_msg_id m, db_event_id m) -pickle_location :: PU NewsLocation -pickle_location = - xpElem "location" $ - xpWrap (from_tuple, to_tuple) $ - xpTriple (xpOption (xpElem "city" xpText)) - (xpOption (xpElem "state" xpText)) - (xpElem "country" xpText) - where - from_tuple = - uncurryN NewsLocation - to_tuple l = (city l, state l, country l) - - +-- | Convert a 'Message' to/from XML. +-- pickle_message :: PU Message pickle_message = xpElem "message" $ @@ -277,23 +360,26 @@ pickle_message = (xpOption (xpElem "Editor" xpText)) (xpOption (xpElem "text" xpText)) pickle_continue - (xpElem "time_stamp" xpText) + (xpElem "time_stamp" xp_time_stamp) where from_tuple = uncurryN Message - to_tuple m = (xml_xml_file_id m, - xml_heading m, - xml_mid m, - xml_category m, - xml_sport m, - xml_url m, - xml_teams m, - xml_locations m, + to_tuple m = (xml_xml_file_id m, -- Verbose, + xml_heading m, -- but + xml_mid m, -- eliminates + xml_category m, -- GHC + xml_sport m, -- warnings + xml_url m, -- . + xml_teams m, -- . + xml_locations m, -- . xml_sms m, xml_editor m, xml_text m, xml_continue m, xml_time_stamp m) + -- | We combine all of the \ elements into one 'String' + -- while unpickling and do the reverse while pickling. + -- pickle_continue :: PU (Maybe String) pickle_continue = xpOption $ @@ -308,17 +394,24 @@ pickle_message = to_string = join "\n" +-- +-- Tasty Tests +-- --- * Tasty Tests +-- | A list of all tests for this module. +-- news_tests :: TestTree news_tests = testGroup "News tests" [ test_news_fields_have_correct_names, + test_on_delete_cascade, test_pickle_of_unpickle_is_identity, test_unpickle_succeeds ] +-- | Make sure our codegen is producing the correct database names. +-- test_news_fields_have_correct_names :: TestTree test_news_fields_have_correct_names = testCase "news fields get correct database names" $ @@ -335,13 +428,22 @@ test_news_fields_have_correct_names = map (\x -> tsn_db_field_namer "herp" "derp" 8675309 x 90210) field_names actual :: [String] - actual = ["mid", "sport", "url", "sms", "editor", "text", "continue"] + actual = ["xml_file_id", + "mid", + "sport", + "url", + "sms", + "editor", + "text", + "continue"] check (x,y) = (x @?= y) --- | Warning, succeess of this test does not mean that unpickling --- succeeded. +-- | If we unpickle something and then pickle it, we should wind up +-- with the same thing we started with. WARNING: success of this +-- test does not mean that unpickling succeeded. +-- test_pickle_of_unpickle_is_identity :: TestTree test_pickle_of_unpickle_is_identity = testGroup "pickle-unpickle tests" [ check "pickle composed with unpickle is the identity" @@ -355,6 +457,8 @@ test_pickle_of_unpickle_is_identity = testGroup "pickle-unpickle tests" actual @?= expected +-- | Make sure we can actually unpickle these things. +-- test_unpickle_succeeds :: TestTree test_unpickle_succeeds = testGroup "unpickle tests" [ check "unpickling succeeds" @@ -367,3 +471,38 @@ test_unpickle_succeeds = testGroup "unpickle tests" actual <- unpickleable path pickle_message let expected = True actual @?= expected + + +-- | Make sure everything gets deleted when we delete the top-level +-- record. +-- +test_on_delete_cascade :: TestTree +test_on_delete_cascade = testGroup "cascading delete tests" + [ check "deleting news deletes its children" + "test/xml/newsxml.xml" + 4 -- 2 news_teams and 2 news_locations that should remain. + ] + where + check desc path expected = testCase desc $ do + news <- unsafe_unpickle path pickle_message + let a = undefined :: Location + let b = undefined :: News + let c = undefined :: NewsTeam + let d = undefined :: News_NewsTeam + let e = undefined :: News_Location + actual <- withSqliteConn ":memory:" $ runDbConn $ do + runMigration silentMigrationLogger $ do + migrate a + migrate b + migrate c + migrate d + migrate e + _ <- dbimport news + deleteAll b + count_a <- countAll a + count_b <- countAll b + count_c <- countAll c + count_d <- countAll d + count_e <- countAll e + return $ count_a + count_b + count_c + count_d + count_e + actual @?= expected