{-# 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 \<message\> that contains an entire news item.
+-- | Parse TSN XML for the DTD \"newsxml.dtd\". Each document contains
+-- a root element \<message\> that contains an entire news item.
--
module TSN.XML.News (
+ dtd,
pickle_message,
-- * Tests
news_tests,
-- * WARNING: these are private but exported to silence warnings
- News_NewsLocationConstructor(..),
+ News_LocationConstructor(..),
News_NewsTeamConstructor(..),
NewsConstructor(..),
- NewsLocationConstructor(..),
NewsTeamConstructor(..) )
where
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,
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(..) )
import TSN.XmlImport ( XmlImport(..) )
-import Xml ( FromXml(..), ToDb(..), pickle_unpickle, unpickleable )
+import Xml (
+ FromXml(..),
+ ToDb(..),
+ pickle_unpickle,
+ unpickleable,
+ unsafe_unpickle )
-
--- | The database type for teams as they show up in the news.
+-- | The DTD to which this module corresponds. Used to invoke dbimport.
--
-data NewsTeam =
- NewsTeam { team_name :: String }
- deriving (Eq, Show)
-
+dtd :: String
+dtd = "newsxml.dtd"
-instance ToDb NewsTeam where
- -- | The database representaion of a 'NewsTeam' is itself.
- type Db NewsTeam = NewsTeam
--- | This is needed to define the XmlImport instance for NewsTeam.
--
-instance FromXml NewsTeam where
- -- | How to we get a 'NewsTeam' from itself?
- from_xml = id
-
--- | Allow us to call 'insert_xml' on the XML representation of
--- NewsTeams.
+-- DB/XML Data types
--
-instance XmlImport NewsTeam
+-- * News/Message
--- | 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)
-
-
--- | 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 ToDb NewsLocation where
- -- | The database representation of a 'NewsLocation' is itself.
- type Db NewsLocation = NewsLocation
-
--- | This is needed to define the XmlImport instance for NewsLocation.
---
-instance FromXml NewsLocation where
- -- | How to we get a 'NewsLocation' from itself?
- from_xml = id
-
--- | Allow us to call 'insert_xml' on the XML representation of
--- NewsLocations.
---
-instance XmlImport NewsLocation
-
-
--- | Mapping between News records and NewsLocation 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_NewsLocation = News_NewsLocation
- (DefaultKey News)
- (DefaultKey NewsLocation)
-
-
--- | The msg_id child of <message> 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
+-- | The msg_id child of \<message\> 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,
deriving (Data, Eq, Show, Typeable)
--- | The XML representation of a news item (message).
+-- | The XML representation of a news item (\<message\>).
--
data Message =
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,
--
data News =
News {
+ db_xml_file_id :: Int,
db_mid :: MsgId,
db_sport :: String,
db_url :: Maybe String,
instance ToDb Message where
+ -- | The database representation of 'Message' is 'News'.
type Db Message = News
-- | Convert the XML representation 'Message' to the database
-- | We use a record wildcard so GHC doesn't complain that we never
-- used the field names.
--
- from_xml Message{..} = News { db_mid = xml_mid,
+ 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_continue = xml_continue,
db_time_stamp = xml_time_stamp }
--- | This lets us call 'insert_xml' on a 'Message'.
+-- | 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_NewsLocation' join tables.
+-- '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
type: constraint
fields: [team_name]
-- entity: NewsLocation
- dbName: news_locations
- constructors:
- - name: NewsLocation
- uniques:
- - name: unique_news_location
- type: constraint
- fields: [city, state, country]
-
|]
-- 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
- name: db_event_id
dbName: event_id
-
- entity: News_NewsTeam
dbName: news__news_teams
constructors:
reference:
onDelete: cascade
-- entity: News_NewsLocation
- dbName: news__news_locations
+- entity: News_Location
+ dbName: news__locations
constructors:
- - name: News_NewsLocation
+ - name: News_Location
fields:
- - name: news_NewsLocation0 # Default created by mkNormalFieldName
+ - name: news_Location0 # Default created by mkNormalFieldName
dbName: news_id
reference:
onDelete: cascade
- - name: news_NewsLocation1 # Default created by mkNormalFieldName
- dbName: news_locations_id
+ - 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
to_tuple m = (db_msg_id m, db_event_id m)
--- | Convert a 'NewsLocation' to/from XML.
+-- | Convert a 'Location' to/from XML.
--
-pickle_location :: PU NewsLocation
+pickle_location :: PU Location
pickle_location =
xpElem "location" $
xpWrap (from_tuple, to_tuple) $
(xpElem "country" xpText)
where
from_tuple =
- uncurryN NewsLocation
+ uncurryN Location
to_tuple l = (city l, state l, country l)
testGroup
"News tests"
[ test_news_fields_have_correct_names,
+ test_on_delete_cascade,
test_pickle_of_unpickle_is_identity,
test_unpickle_succeeds ]
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)
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
projects / dead / htsn-import.git / blobdiff