graphql-engine/server/src-lib/Hasura/RQL/DDL/Schema/Table.hs
Vamshi Surabhi e8e4f30dd6 server: support remote relationships on SQL Server and BigQuery (#1497)
Remote relationships are now supported on SQL Server and BigQuery. The major change though is the re-architecture of remote join execution logic. Prior to this PR, each backend is responsible for processing the remote relationships that are part of their AST.

This is not ideal as there is nothing specific about a remote join's execution that ties it to a backend. The only backend specific part is whether or not the specification of the remote relationship is valid (i.e, we'll need to validate whether the scalars are compatible).

The approach now changes to this:

1. Before delegating the AST to the backend, we traverse the AST, collect all the remote joins while modifying the AST to add necessary join fields where needed.

1. Once the remote joins are collected from the AST, the database call is made to fetch the response. The necessary data for the remote join(s) is collected from the database's response and one or more remote schema calls are constructed as necessary.

1. The remote schema calls are then executed and the data from the database and from the remote schemas is joined to produce the final response.

### Known issues

1. Ideally the traversal of the IR to collect remote joins should return an AST which does not include remote join fields. This operation can be type safe but isn't taken up as part of the PR.

1. There is a lot of code duplication between `Transport/HTTP.hs` and `Transport/Websocket.hs` which needs to be fixed ASAP. This too hasn't been taken up by this PR.

1. The type which represents the execution plan is only modified to handle our current remote joins and as such it will have to be changed to accommodate general remote joins.

1. Use of lenses would have reduced the boilerplate code to collect remote joins from the base AST.

1. The current remote join logic assumes that the join columns of a remote relationship appear with their names in the database response. This however is incorrect as they could be aliased. This can be taken up by anyone, I've left a comment in the code.

### Notes to the reviewers

I think it is best reviewed commit by commit.

1. The first one is very straight forward.

1. The second one refactors the remote join execution logic but other than moving things around, it doesn't change the user facing functionality.  This moves Postgres specific parts to `Backends/Postgres` module from `Execute`. Some IR related code to `Hasura.RQL.IR` module.  Simplifies various type class function signatures as a backend doesn't have to handle remote joins anymore

1. The third one fixes partial case matches that for some weird reason weren't shown as warnings before this refactor

1. The fourth one generalizes the validation logic of remote relationships and implements `scalarTypeGraphQLName` function on SQL Server and BigQuery which is used by the validation logic. This enables remote relationships on BigQuery and SQL Server.

https://github.com/hasura/graphql-engine-mono/pull/1497

GitOrigin-RevId: 77dd8eed326602b16e9a8496f52f46d22b795598
2021-06-11 03:27:39 +00:00

546 lines
21 KiB
Haskell
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

{-# LANGUAGE Arrows #-}
-- | Description: Create/delete SQL tables to/from Hasura metadata.
module Hasura.RQL.DDL.Schema.Table
( TrackTable(..)
, runTrackTableQ
, TrackTableV2(..)
, runTrackTableV2Q
, UntrackTable(..)
, runUntrackTableQ
, dropTableInMetadata
, SetTableIsEnum(..)
, runSetExistingTableIsEnumQ
, SetTableCustomFields(..)
, runSetTableCustomFieldsQV2
, SetTableCustomization(..)
, runSetTableCustomization
, buildTableCache
, checkConflictingNode
) where
import Hasura.Prelude
import qualified Data.Aeson.Ordered as JO
import qualified Data.HashMap.Strict.Extended as Map
import qualified Data.HashMap.Strict.InsOrd as OMap
import qualified Data.HashSet as S
import qualified Language.GraphQL.Draft.Syntax as G
import Control.Arrow.Extended
import Control.Lens.Extended hiding ((.=))
import Control.Monad.Trans.Control (MonadBaseControl)
import Data.Aeson
import Data.Aeson.TH
import Data.Text.Extended
import qualified Hasura.Incremental as Inc
import qualified Hasura.SQL.AnyBackend as AB
import Hasura.Backends.Postgres.SQL.Types (QualifiedTable)
import Hasura.Base.Error
import Hasura.EncJSON
import Hasura.GraphQL.Context
import Hasura.GraphQL.Schema.Common (textToName)
import Hasura.RQL.DDL.Deps
import Hasura.RQL.DDL.Schema.Cache.Common
import Hasura.RQL.DDL.Schema.Common
import Hasura.RQL.DDL.Schema.Enum (resolveEnumReferences)
import Hasura.RQL.IR
import Hasura.RQL.Types hiding (fmFunction)
import Hasura.Server.Utils
data TrackTable b
= TrackTable
{ tSource :: !SourceName
, tName :: !(TableName b)
, tIsEnum :: !Bool
}
deriving instance (Backend b) => Show (TrackTable b)
deriving instance (Backend b) => Eq (TrackTable b)
instance (Backend b) => FromJSON (TrackTable b) where
parseJSON v = withOptions <|> withoutOptions
where
withOptions = flip (withObject "TrackTable") v $ \o -> TrackTable
<$> o .:? "source" .!= defaultSource
<*> o .: "table"
<*> o .:? "is_enum" .!= False
withoutOptions = TrackTable defaultSource <$> parseJSON v <*> pure False
instance (Backend b) => ToJSON (TrackTable b) where
toJSON (TrackTable source name isEnum)
| isEnum = object [ "source" .= source, "table" .= name, "is_enum" .= isEnum ]
| otherwise = toJSON name
data SetTableIsEnum
= SetTableIsEnum
{ stieSource :: !SourceName
, stieTable :: !QualifiedTable
, stieIsEnum :: !Bool
} deriving (Show, Eq)
$(deriveToJSON hasuraJSON ''SetTableIsEnum)
instance FromJSON SetTableIsEnum where
parseJSON = withObject "Object" $ \o ->
SetTableIsEnum
<$> o .:? "source" .!= defaultSource
<*> o .: "table"
<*> o .: "is_enum"
data UntrackTable b =
UntrackTable
{ utSource :: !SourceName
, utTable :: !(TableName b)
, utCascade :: !Bool
} deriving (Generic)
deriving instance (Backend b) => Show (UntrackTable b)
deriving instance (Backend b) => Eq (UntrackTable b)
instance (Backend b) => ToJSON (UntrackTable b) where
toJSON = genericToJSON hasuraJSON{omitNothingFields=True}
instance (Backend b) => FromJSON (UntrackTable b) where
parseJSON = withObject "Object" $ \o ->
UntrackTable
<$> o .:? "source" .!= defaultSource
<*> o .: "table"
<*> o .:? "cascade" .!= False
isTableTracked :: forall b. (Backend b) => SourceInfo b -> TableName b -> Bool
isTableTracked sourceInfo tableName =
isJust $ Map.lookup tableName $ _siTables sourceInfo
-- | Track table/view, Phase 1:
-- Validate table tracking operation. Fails if table is already being tracked,
-- or if a function with the same name is being tracked.
trackExistingTableOrViewP1
:: forall b m
. (QErrM m, CacheRWM m, Backend b, MetadataM m)
=> SourceName
-> TableName b
-> m ()
trackExistingTableOrViewP1 source tableName = do
sourceInfo <- askSourceInfo source
when (isTableTracked @b sourceInfo tableName) $
throw400 AlreadyTracked $ "view/table already tracked : " <>> tableName
let functionName = tableToFunction @b tableName
when (isJust $ Map.lookup functionName $ _siFunctions @b sourceInfo) $
throw400 NotSupported $ "function with name " <> tableName <<> " already exists"
-- | Check whether a given name would conflict with the current schema by doing
-- an internal introspection
checkConflictingNode
:: forall m. MonadError QErr m
=> SchemaCache
-> Text
-> m ()
checkConflictingNode sc tnGQL = do
let GQLContext queryParser _ = scUnauthenticatedGQLContext sc
-- {
-- __schema {
-- queryType {
-- fields {
-- name
-- }
-- }
-- }
-- }
introspectionQuery =
[ G.SelectionField $ G.Field Nothing $$(G.litName "__schema") mempty []
[ G.SelectionField $ G.Field Nothing $$(G.litName "queryType") mempty []
[ G.SelectionField $ G.Field Nothing $$(G.litName "fields") mempty []
[ G.SelectionField $ G.Field Nothing $$(G.litName "name") mempty []
[]
]
]
]
]
case queryParser introspectionQuery of
Left _ -> pure ()
Right (results, _reusability) -> do
case OMap.lookup $$(G.litName "__schema") results of
Just (RFRaw (JO.Object schema)) -> do
let names = do
JO.Object queryType <- JO.lookup "queryType" schema
JO.Array fields <- JO.lookup "fields" queryType
for fields \case
JO.Object field -> do
JO.String name <- JO.lookup "name" field
pure name
_ -> Nothing
case names of
Nothing -> pure ()
Just ns ->
if tnGQL `elem` ns
then throw400 RemoteSchemaConflicts $
"node " <> tnGQL <>
" already exists in current graphql schema"
else pure ()
_ -> pure ()
trackExistingTableOrViewP2
:: forall b m
. (MonadError QErr m, CacheRWM m, MetadataM m, BackendMetadata b)
=> SourceName -> TableName b -> Bool -> TableConfig b -> m EncJSON
trackExistingTableOrViewP2 source tableName isEnum config = do
sc <- askSchemaCache
{-
The next line does more than what it says on the tin. Removing the following
call to 'checkConflictingNode' causes memory usage to spike when newly
tracking a large amount (~100) of tables. The memory usage can be triggered
by first creating a large amount of tables through SQL, without tracking the
tables, and then clicking "track all" in the console. Curiously, this high
memory usage happens even when no substantial GraphQL schema is generated.
-}
checkConflictingNode sc $ snakeCaseTableName @b tableName
let metadata = mkTableMeta tableName isEnum config
buildSchemaCacheFor
(MOSourceObjId source
$ AB.mkAnyBackend
$ SMOTable @b tableName)
$ MetadataModifier
$ metaSources.ix source.toSourceMetadata.smTables %~ OMap.insert tableName metadata
pure successMsg
runTrackTableQ
:: forall b m
. (MonadError QErr m, CacheRWM m, MetadataM m, BackendMetadata b)
=> TrackTable b
-> m EncJSON
runTrackTableQ (TrackTable source qt isEnum) = do
trackExistingTableOrViewP1 @b source qt
trackExistingTableOrViewP2 @b source qt isEnum emptyTableConfig
data TrackTableV2 b
= TrackTableV2
{ ttv2Table :: !(TrackTable b)
, ttv2Configuration :: !(TableConfig b)
} deriving (Show, Eq, Generic)
instance (Backend b) => ToJSON (TrackTableV2 b) where
toJSON = genericToJSON hasuraJSON
instance (Backend b) => FromJSON (TrackTableV2 b) where
parseJSON = withObject "Object" $ \o -> do
table <- parseJSON $ Object o
configuration <- o .:? "configuration" .!= emptyTableConfig
pure $ TrackTableV2 table configuration
runTrackTableV2Q
:: forall b m
. (MonadError QErr m, CacheRWM m, MetadataM m, BackendMetadata b)
=> TrackTableV2 b
-> m EncJSON
runTrackTableV2Q (TrackTableV2 (TrackTable source qt isEnum) config) = do
trackExistingTableOrViewP1 @b source qt
trackExistingTableOrViewP2 @b source qt isEnum config
runSetExistingTableIsEnumQ :: (MonadError QErr m, CacheRWM m, MetadataM m) => SetTableIsEnum -> m EncJSON
runSetExistingTableIsEnumQ (SetTableIsEnum source tableName isEnum) = do
void $ askTabInfo @('Postgres 'Vanilla) source tableName -- assert that table is tracked
buildSchemaCacheFor
(MOSourceObjId source $ AB.mkAnyBackend $ SMOTable @('Postgres 'Vanilla) tableName)
$ MetadataModifier
$ tableMetadataSetter @('Postgres 'Vanilla) source tableName.tmIsEnum .~ isEnum
return successMsg
data SetTableCustomization
= SetTableCustomization
{ _stcSource :: !SourceName
, _stcTable :: !QualifiedTable
, _stcConfiguration :: !(TableConfig ('Postgres 'Vanilla))
} deriving (Show, Eq)
$(deriveToJSON hasuraJSON ''SetTableCustomization)
instance FromJSON SetTableCustomization where
parseJSON = withObject "Object" $ \o ->
SetTableCustomization
<$> o .:? "source" .!= defaultSource
<*> o .: "table"
<*> o .: "configuration"
data SetTableCustomFields
= SetTableCustomFields
{ _stcfSource :: !SourceName
, _stcfTable :: !QualifiedTable
, _stcfCustomRootFields :: !TableCustomRootFields
, _stcfCustomColumnNames :: !(CustomColumnNames ('Postgres 'Vanilla))
} deriving (Show, Eq)
$(deriveToJSON hasuraJSON ''SetTableCustomFields)
instance FromJSON SetTableCustomFields where
parseJSON = withObject "SetTableCustomFields" $ \o ->
SetTableCustomFields
<$> o .:? "source" .!= defaultSource
<*> o .: "table"
<*> o .:? "custom_root_fields" .!= emptyCustomRootFields
<*> o .:? "custom_column_names" .!= Map.empty
runSetTableCustomFieldsQV2
:: (QErrM m, CacheRWM m, MetadataM m) => SetTableCustomFields -> m EncJSON
runSetTableCustomFieldsQV2 (SetTableCustomFields source tableName rootFields columnNames) = do
void $ askTabInfo @('Postgres 'Vanilla) source tableName -- assert that table is tracked
let tableConfig = TableConfig @('Postgres 'Vanilla) rootFields columnNames Nothing
buildSchemaCacheFor
(MOSourceObjId source $ AB.mkAnyBackend $ SMOTable @('Postgres 'Vanilla) tableName)
$ MetadataModifier
$ tableMetadataSetter source tableName.tmConfiguration .~ tableConfig
return successMsg
runSetTableCustomization
:: (QErrM m, CacheRWM m, MetadataM m) => SetTableCustomization -> m EncJSON
runSetTableCustomization (SetTableCustomization source table config) = do
void $ askTabInfo @('Postgres 'Vanilla) source table
buildSchemaCacheFor
(MOSourceObjId source $ AB.mkAnyBackend $ SMOTable @('Postgres 'Vanilla) table)
$ MetadataModifier
$ tableMetadataSetter source table.tmConfiguration .~ config
return successMsg
unTrackExistingTableOrViewP1
:: forall b m
. (CacheRM m, QErrM m, Backend b)
=> UntrackTable b
-> m ()
unTrackExistingTableOrViewP1 (UntrackTable source vn _) = do
schemaCache <- askSchemaCache
tableInfo <- unsafeTableInfo @b source vn (scSources schemaCache)
`onNothing` throw400 AlreadyUntracked ("view/table already untracked : " <>> vn)
when (isSystemDefined $ _tciSystemDefined $ _tiCoreInfo tableInfo) $
throw400 NotSupported $ vn <<> " is system defined, cannot untrack"
unTrackExistingTableOrViewP2
:: forall b m
. (CacheRWM m, QErrM m, MetadataM m, BackendMetadata b)
=> UntrackTable b -> m EncJSON
unTrackExistingTableOrViewP2 (UntrackTable source qtn cascade) = withNewInconsistentObjsCheck do
sc <- askSchemaCache
-- Get relational, query template and function dependants
let allDeps =
getDependentObjs
sc
(SOSourceObj source $ AB.mkAnyBackend $ SOITable @b qtn)
indirectDeps = mapMaybe getIndirectDep allDeps
-- Report bach with an error if cascade is not set
when (indirectDeps /= [] && not cascade)
$ reportDepsExt
(map (SOSourceObj source . AB.mkAnyBackend) indirectDeps)
[]
-- Purge all the dependents from state
metadataModifier <- execWriterT do
mapM_ (purgeDependentObject source >=> tell) indirectDeps
tell $ dropTableInMetadata @b source qtn
-- delete the table and its direct dependencies
buildSchemaCache metadataModifier
pure successMsg
where
getIndirectDep :: SchemaObjId -> Maybe (SourceObjId b)
getIndirectDep = \case
SOSourceObj s exists ->
-- If the dependency is to any other source, it automatically is an
-- indirect dependency, hence the cast is safe here. However, we don't
-- have these cross source dependencies yet
AB.unpackAnyBackend exists >>= \case
v@(SOITableObj dtn _) ->
if not (s == source && qtn == dtn) then Just v else Nothing
v -> Just v
_ -> Nothing
dropTableInMetadata
:: forall b
. (BackendMetadata b)
=> SourceName
-> TableName b
-> MetadataModifier
dropTableInMetadata source table =
MetadataModifier $ metaSources.ix source.(toSourceMetadata @b).smTables %~ OMap.delete table
runUntrackTableQ
:: forall b m
. (CacheRWM m, QErrM m, MetadataM m, BackendMetadata b)
=> UntrackTable b
-> m EncJSON
runUntrackTableQ q = do
unTrackExistingTableOrViewP1 @b q
unTrackExistingTableOrViewP2 @b q
-- | Builds an initial @'TableCache' 'ColumnInfo'@ from catalog information. Does not fill in
-- '_tiRolePermInfoMap' or '_tiEventTriggerInfoMap' at all, and '_tiFieldInfoMap' only contains
-- columns, not relationships; those pieces of information are filled in by later stages.
buildTableCache
:: forall arr m b
. ( ArrowChoice arr, Inc.ArrowDistribute arr, ArrowWriter (Seq CollectedInfo) arr
, Inc.ArrowCache m arr, MonadIO m, MonadBaseControl IO m, BackendMetadata b
)
=> ( SourceName
, SourceConfig b
, DBTablesMetadata b
, [TableBuildInput b]
, Inc.Dependency Inc.InvalidationKey
) `arr` Map.HashMap (TableName b) (TableCoreInfoG b (ColumnInfo b) (ColumnInfo b))
buildTableCache = Inc.cache proc (source, sourceConfig, dbTablesMeta, tableBuildInputs, reloadMetadataInvalidationKey) -> do
rawTableInfos <-
(| Inc.keyed (| withTable (\tables -> do
table <- noDuplicateTables -< tables
let maybeInfo = Map.lookup (_tbiName table) dbTablesMeta
buildRawTableInfo -< (table, maybeInfo, sourceConfig, reloadMetadataInvalidationKey)
)
|)
|) (withSourceInKey source $ Map.groupOnNE _tbiName tableBuildInputs)
let rawTableCache = removeSourceInKey $ Map.catMaybes rawTableInfos
enumTables = flip Map.mapMaybe rawTableCache \rawTableInfo ->
(,) <$> _tciPrimaryKey rawTableInfo <*> _tciEnumValues rawTableInfo
tableInfos <-
(| Inc.keyed (| withTable (\table -> processTableInfo -< (enumTables, table)) |)
|) (withSourceInKey source rawTableCache)
returnA -< removeSourceInKey (Map.catMaybes tableInfos)
where
withSourceInKey :: (Eq k, Hashable k) => SourceName -> HashMap k v -> HashMap (SourceName, k) v
withSourceInKey source = mapKeys (source,)
removeSourceInKey :: (Eq k, Hashable k) => HashMap (SourceName, k) v -> HashMap k v
removeSourceInKey = mapKeys snd
withTable :: ErrorA QErr arr (e, s) a -> arr (e, ((SourceName, TableName b), s)) (Maybe a)
withTable f = withRecordInconsistency f <<<
second (first $ arr \(source, name) ->
MetadataObject
(MOSourceObjId source
$ AB.mkAnyBackend
$ SMOTable @b name)
(toJSON name))
noDuplicateTables = proc tables -> case tables of
table :| [] -> returnA -< table
_ -> throwA -< err400 AlreadyExists "duplication definition for table"
-- Step 1: Build the raw table cache from metadata information.
buildRawTableInfo
:: ErrorA QErr arr
( TableBuildInput b
, Maybe (DBTableMetadata b)
, SourceConfig b
, Inc.Dependency Inc.InvalidationKey
) (TableCoreInfoG b (RawColumnInfo b) (Column b))
buildRawTableInfo = Inc.cache proc (tableBuildInput, maybeInfo, sourceConfig, reloadMetadataInvalidationKey) -> do
let TableBuildInput name isEnum config = tableBuildInput
metadataTable <-
(| onNothingA (throwA -<
err400 NotExists $ "no such table/view exists in source: " <>> name)
|) maybeInfo
let columns :: [RawColumnInfo b] = _ptmiColumns metadataTable
columnMap = mapFromL (FieldName . toTxt . prciName) columns
primaryKey = _ptmiPrimaryKey metadataTable
rawPrimaryKey <- liftEitherA -< traverse (resolvePrimaryKeyColumns columnMap) primaryKey
enumValues <- if isEnum
then do
-- We want to make sure we reload enum values whenever someone explicitly calls
-- `reload_metadata`.
Inc.dependOn -< reloadMetadataInvalidationKey
eitherEnums <- bindA -< fetchAndValidateEnumValues sourceConfig name rawPrimaryKey columns
liftEitherA -< Just <$> eitherEnums
else returnA -< Nothing
returnA -< TableCoreInfo
{ _tciName = name
, _tciSystemDefined = SystemDefined False
, _tciFieldInfoMap = columnMap
, _tciPrimaryKey = primaryKey
, _tciUniqueConstraints = _ptmiUniqueConstraints metadataTable
, _tciForeignKeys = S.map unForeignKeyMetadata $ _ptmiForeignKeys metadataTable
, _tciViewInfo = _ptmiViewInfo metadataTable
, _tciEnumValues = enumValues
, _tciCustomConfig = config
, _tciDescription = _ptmiDescription metadataTable
, _tciExtraTableMetadata = _ptmiExtraTableMetadata metadataTable
}
-- Step 2: Process the raw table cache to replace Postgres column types with logical column
-- types.
processTableInfo
:: ErrorA QErr arr
( Map.HashMap (TableName b) (PrimaryKey b (Column b), EnumValues)
, TableCoreInfoG b (RawColumnInfo b) (Column b)
) (TableCoreInfoG b (ColumnInfo b) (ColumnInfo b))
processTableInfo = proc (enumTables, rawInfo) -> liftEitherA -< do
let columns = _tciFieldInfoMap rawInfo
enumReferences = resolveEnumReferences enumTables (_tciForeignKeys rawInfo)
columnInfoMap <-
alignCustomColumnNames columns (_tcCustomColumnNames $ _tciCustomConfig rawInfo)
>>= traverse (processColumnInfo enumReferences (_tciName rawInfo))
assertNoDuplicateFieldNames (Map.elems columnInfoMap)
primaryKey <- traverse (resolvePrimaryKeyColumns columnInfoMap) (_tciPrimaryKey rawInfo)
pure rawInfo
{ _tciFieldInfoMap = columnInfoMap
, _tciPrimaryKey = primaryKey
}
resolvePrimaryKeyColumns
:: forall n a . (QErrM n) => HashMap FieldName a -> PrimaryKey b (Column b) -> n (PrimaryKey b a)
resolvePrimaryKeyColumns columnMap = traverseOf (pkColumns.traverse) \columnName ->
Map.lookup (FieldName (toTxt columnName)) columnMap
`onNothing` throw500 "column in primary key not in table!"
alignCustomColumnNames
:: (QErrM n)
=> FieldInfoMap (RawColumnInfo b)
-> CustomColumnNames b
-> n (FieldInfoMap (RawColumnInfo b, G.Name))
alignCustomColumnNames columns customNames = do
let customNamesByFieldName = mapKeys (fromCol @b) customNames
flip Map.traverseWithKey (align columns customNamesByFieldName) \columnName -> \case
This column -> (column,) <$> textToName (getFieldNameTxt columnName)
These column customName -> pure (column, customName)
That customName -> throw400 NotExists $ "the custom field name " <> customName
<<> " was given for the column " <> columnName <<> ", but no such column exists"
-- | “Processes” a '(RawColumnInfo ('Postgres 'Vanilla))' into a 'PGColumnInfo' by resolving its type using a map of
-- known enum tables.
processColumnInfo
:: (QErrM n)
=> Map.HashMap (Column b) (NonEmpty (EnumReference b))
-> TableName b -- ^ the table this column belongs to
-> (RawColumnInfo b, G.Name)
-> n (ColumnInfo b)
processColumnInfo tableEnumReferences tableName (rawInfo, name) = do
resolvedType <- resolveColumnType
pure ColumnInfo
{ pgiColumn = pgCol
, pgiName = name
, pgiPosition = prciPosition rawInfo
, pgiType = resolvedType
, pgiIsNullable = prciIsNullable rawInfo
, pgiDescription = prciDescription rawInfo
}
where
pgCol = prciName rawInfo
resolveColumnType =
case Map.lookup pgCol tableEnumReferences of
-- no references? not an enum
Nothing -> pure $ ColumnScalar (prciType rawInfo)
-- one reference? is an enum
Just (enumReference:|[]) -> pure $ ColumnEnumReference enumReference
-- multiple referenced enums? the schema is strange, so lets reject it
Just enumReferences -> throw400 ConstraintViolation
$ "column " <> prciName rawInfo <<> " in table " <> tableName
<<> " references multiple enum tables ("
<> commaSeparated (map (dquote . erTable) $ toList enumReferences) <> ")"
assertNoDuplicateFieldNames columns =
flip Map.traverseWithKey (Map.groupOn pgiName columns) \name columnsWithName ->
case columnsWithName of
one:two:more -> throw400 AlreadyExists $ "the definitions of columns "
<> englishList "and" (dquote . pgiColumn <$> (one:|two:more))
<> " are in conflict: they are mapped to the same field name, " <>> name
_ -> pure ()