graphql-engine/server/src-lib/Hasura/RQL/DDL/Relationship.hs

336 lines
12 KiB
Haskell
Raw Normal View History

module Hasura.RQL.DDL.Relationship
Clean metadata arguments ## Description Thanks to #1664, the Metadata API types no longer require a `ToJSON` instance. This PR follows up with a cleanup of the types of the arguments to the metadata API: - whenever possible, it moves those argument types to where they're used (RQL.DDL.*) - it removes all unrequired instances (mostly `ToJSON`) This PR does not attempt to do it for _all_ such argument types. For some of the metadata operations, the type used to describe the argument to the API and used to represent the value in the metadata are one and the same (like for `CreateEndpoint`). Sometimes, the two types are intertwined in complex ways (`RemoteRelationship` and `RemoteRelationshipDef`). In the spirit of only doing uncontroversial cleaning work, this PR only moves types that are not used outside of RQL.DDL. Furthermore, this is a small step towards separating the different types all jumbled together in RQL.Types. ## Notes This PR also improves several `FromJSON` instances to make use of `withObject`, and to use a human readable string instead of a type name in error messages whenever possible. For instance: - before: `expected Object for Object, but encountered X` after: `expected Object for add computed field, but encountered X` - before: `Expecting an object for update query` after: `expected Object for update query, but encountered X` This PR also renames `CreateFunctionPermission` to `FunctionPermissionArgument`, to remove the quite surprising `type DropFunctionPermission = CreateFunctionPermission`. This PR also deletes some dead code, mostly in RQL.DML. This PR also moves a PG-specific source resolving function from DDL.Schema.Source to the only place where it is used: App.hs. https://github.com/hasura/graphql-engine-mono/pull/1844 GitOrigin-RevId: a594521194bb7fe6a111b02a9e099896f9fed59c
2021-07-27 13:41:42 +03:00
( CreateArrRel(..)
, CreateObjRel(..)
, runCreateRelationship
, objRelP2Setup
, arrRelP2Setup
Clean metadata arguments ## Description Thanks to #1664, the Metadata API types no longer require a `ToJSON` instance. This PR follows up with a cleanup of the types of the arguments to the metadata API: - whenever possible, it moves those argument types to where they're used (RQL.DDL.*) - it removes all unrequired instances (mostly `ToJSON`) This PR does not attempt to do it for _all_ such argument types. For some of the metadata operations, the type used to describe the argument to the API and used to represent the value in the metadata are one and the same (like for `CreateEndpoint`). Sometimes, the two types are intertwined in complex ways (`RemoteRelationship` and `RemoteRelationshipDef`). In the spirit of only doing uncontroversial cleaning work, this PR only moves types that are not used outside of RQL.DDL. Furthermore, this is a small step towards separating the different types all jumbled together in RQL.Types. ## Notes This PR also improves several `FromJSON` instances to make use of `withObject`, and to use a human readable string instead of a type name in error messages whenever possible. For instance: - before: `expected Object for Object, but encountered X` after: `expected Object for add computed field, but encountered X` - before: `Expecting an object for update query` after: `expected Object for update query, but encountered X` This PR also renames `CreateFunctionPermission` to `FunctionPermissionArgument`, to remove the quite surprising `type DropFunctionPermission = CreateFunctionPermission`. This PR also deletes some dead code, mostly in RQL.DML. This PR also moves a PG-specific source resolving function from DDL.Schema.Source to the only place where it is used: App.hs. https://github.com/hasura/graphql-engine-mono/pull/1844 GitOrigin-RevId: a594521194bb7fe6a111b02a9e099896f9fed59c
2021-07-27 13:41:42 +03:00
, DropRel
, runDropRel
, dropRelationshipInMetadata
Clean metadata arguments ## Description Thanks to #1664, the Metadata API types no longer require a `ToJSON` instance. This PR follows up with a cleanup of the types of the arguments to the metadata API: - whenever possible, it moves those argument types to where they're used (RQL.DDL.*) - it removes all unrequired instances (mostly `ToJSON`) This PR does not attempt to do it for _all_ such argument types. For some of the metadata operations, the type used to describe the argument to the API and used to represent the value in the metadata are one and the same (like for `CreateEndpoint`). Sometimes, the two types are intertwined in complex ways (`RemoteRelationship` and `RemoteRelationshipDef`). In the spirit of only doing uncontroversial cleaning work, this PR only moves types that are not used outside of RQL.DDL. Furthermore, this is a small step towards separating the different types all jumbled together in RQL.Types. ## Notes This PR also improves several `FromJSON` instances to make use of `withObject`, and to use a human readable string instead of a type name in error messages whenever possible. For instance: - before: `expected Object for Object, but encountered X` after: `expected Object for add computed field, but encountered X` - before: `Expecting an object for update query` after: `expected Object for update query, but encountered X` This PR also renames `CreateFunctionPermission` to `FunctionPermissionArgument`, to remove the quite surprising `type DropFunctionPermission = CreateFunctionPermission`. This PR also deletes some dead code, mostly in RQL.DML. This PR also moves a PG-specific source resolving function from DDL.Schema.Source to the only place where it is used: App.hs. https://github.com/hasura/graphql-engine-mono/pull/1844 GitOrigin-RevId: a594521194bb7fe6a111b02a9e099896f9fed59c
2021-07-27 13:41:42 +03:00
, SetRelComment
, runSetRelComment
)
where
import Hasura.Prelude
import qualified Data.HashMap.Strict as HM
import qualified Data.HashMap.Strict.InsOrd as OMap
import qualified Data.HashSet as HS
2018-06-27 16:11:32 +03:00
import Control.Lens ((.~))
2018-06-27 16:11:32 +03:00
import Data.Aeson.Types
import Data.Text.Extended
import Data.Tuple (swap)
import qualified Hasura.SQL.AnyBackend as AB
import Hasura.Base.Error
import Hasura.EncJSON
import Hasura.RQL.DDL.Deps
import Hasura.RQL.DDL.Permission
import Hasura.RQL.Types
2018-06-27 16:11:32 +03:00
Clean metadata arguments ## Description Thanks to #1664, the Metadata API types no longer require a `ToJSON` instance. This PR follows up with a cleanup of the types of the arguments to the metadata API: - whenever possible, it moves those argument types to where they're used (RQL.DDL.*) - it removes all unrequired instances (mostly `ToJSON`) This PR does not attempt to do it for _all_ such argument types. For some of the metadata operations, the type used to describe the argument to the API and used to represent the value in the metadata are one and the same (like for `CreateEndpoint`). Sometimes, the two types are intertwined in complex ways (`RemoteRelationship` and `RemoteRelationshipDef`). In the spirit of only doing uncontroversial cleaning work, this PR only moves types that are not used outside of RQL.DDL. Furthermore, this is a small step towards separating the different types all jumbled together in RQL.Types. ## Notes This PR also improves several `FromJSON` instances to make use of `withObject`, and to use a human readable string instead of a type name in error messages whenever possible. For instance: - before: `expected Object for Object, but encountered X` after: `expected Object for add computed field, but encountered X` - before: `Expecting an object for update query` after: `expected Object for update query, but encountered X` This PR also renames `CreateFunctionPermission` to `FunctionPermissionArgument`, to remove the quite surprising `type DropFunctionPermission = CreateFunctionPermission`. This PR also deletes some dead code, mostly in RQL.DML. This PR also moves a PG-specific source resolving function from DDL.Schema.Source to the only place where it is used: App.hs. https://github.com/hasura/graphql-engine-mono/pull/1844 GitOrigin-RevId: a594521194bb7fe6a111b02a9e099896f9fed59c
2021-07-27 13:41:42 +03:00
--------------------------------------------------------------------------------
-- Create local relationship
newtype CreateArrRel b = CreateArrRel { unCreateArrRel :: WithTable b (ArrRelDef b) }
deriving newtype (FromJSON)
newtype CreateObjRel b = CreateObjRel { unCreateObjRel :: WithTable b (ObjRelDef b) }
deriving newtype (FromJSON)
runCreateRelationship
:: forall m b a
. (MonadError QErr m, CacheRWM m, ToJSON a, MetadataM m, Backend b, BackendMetadata b)
Clean metadata arguments ## Description Thanks to #1664, the Metadata API types no longer require a `ToJSON` instance. This PR follows up with a cleanup of the types of the arguments to the metadata API: - whenever possible, it moves those argument types to where they're used (RQL.DDL.*) - it removes all unrequired instances (mostly `ToJSON`) This PR does not attempt to do it for _all_ such argument types. For some of the metadata operations, the type used to describe the argument to the API and used to represent the value in the metadata are one and the same (like for `CreateEndpoint`). Sometimes, the two types are intertwined in complex ways (`RemoteRelationship` and `RemoteRelationshipDef`). In the spirit of only doing uncontroversial cleaning work, this PR only moves types that are not used outside of RQL.DDL. Furthermore, this is a small step towards separating the different types all jumbled together in RQL.Types. ## Notes This PR also improves several `FromJSON` instances to make use of `withObject`, and to use a human readable string instead of a type name in error messages whenever possible. For instance: - before: `expected Object for Object, but encountered X` after: `expected Object for add computed field, but encountered X` - before: `Expecting an object for update query` after: `expected Object for update query, but encountered X` This PR also renames `CreateFunctionPermission` to `FunctionPermissionArgument`, to remove the quite surprising `type DropFunctionPermission = CreateFunctionPermission`. This PR also deletes some dead code, mostly in RQL.DML. This PR also moves a PG-specific source resolving function from DDL.Schema.Source to the only place where it is used: App.hs. https://github.com/hasura/graphql-engine-mono/pull/1844 GitOrigin-RevId: a594521194bb7fe6a111b02a9e099896f9fed59c
2021-07-27 13:41:42 +03:00
=> RelType
-> WithTable b (RelDef a)
-> m EncJSON
runCreateRelationship relType (WithTable source tableName relDef) = do
let relName = _rdName relDef
-- Check if any field with relationship name already exists in the table
tableFields <- _tciFieldInfoMap <$> askTableCoreInfo @b source tableName
onJust (HM.lookup (fromRel relName) tableFields) $ const $
throw400 AlreadyExists $
"field with name " <> relName <<> " already exists in table " <>> tableName
tableCache <- askSchemaCache
>>= flip onNothing (throw400 NotFound "Could not find source.")
. unsafeTableCache source
. scSources
let comment = _rdComment relDef
metadataObj = MOSourceObjId source
$ AB.mkAnyBackend
$ SMOTableObj @b tableName
$ MTORel relName relType
addRelationshipToMetadata <- case relType of
ObjRel -> do
value <- decodeValue $ toJSON relDef
validateRelationship @b
tableCache
tableName
(Left value)
pure $ tmObjectRelationships %~ OMap.insert relName (RelDef relName (_rdUsing value) comment)
ArrRel -> do
value <- decodeValue $ toJSON relDef
validateRelationship @b
tableCache
tableName
(Right value)
pure $ tmArrayRelationships %~ OMap.insert relName (RelDef relName (_rdUsing value) comment)
buildSchemaCacheFor metadataObj
$ MetadataModifier
$ tableMetadataSetter @b source tableName %~ addRelationshipToMetadata
pure successMsg
objRelP2Setup
:: forall b m
. (QErrM m, Backend b)
=> SourceName
-> TableName b
-> HashMap (TableName b) (HashSet (ForeignKey b))
-> RelDef (ObjRelUsing b)
-> m (RelInfo b, [SchemaDependency])
server: fix the nullability of object relationships (fix hasura/graphql-engine#7201) When adding object relationships, we set the nullability of the generated GraphQL field based on whether the database backend enforces that the referenced data always exists. For manual relationships (corresponding to `manual_configuration`), the database backend is unaware of any relationship between data, and hence such fields are always set to be nullable. For relationships generated from foreign key constraints (corresponding to `foreign_key_constraint_on`), we distinguish between two cases: 1. The "forward" object relationship from a referencing table (i.e. which has the foreign key constraint) to a referenced table. This should be set to be non-nullable when all referencing columns are non-nullable. But in fact, it used to set it to be non-nullable if *any* referencing column is non-nullable, which is only correct in Postgres when `MATCH FULL` is set (a flag we don't consider). This fixes that by changing a boolean conjunction to a disjunction. 2. The "reverse" object relationship from a referenced table to a referencing table which has the foreign key constraint. This should always be set to be nullable. But in fact, it used to always be set to non-nullable, as was reported in hasura/graphql-engine#7201. This fixes that. Moreover, we have moved the computation of the nullability from `Hasura.RQL.DDL.Relationship` to `Hasura.GraphQL.Schema.Select`: this nullability used to be passed through the `riIsNullable` field of `RelInfo`, but for array relationships this information is not actually used, and moreover the remaining fields of `RelInfo` are already enough to deduce the nullability. This also adds regression tests for both (1) and (2) above. https://github.com/hasura/graphql-engine-mono/pull/2159 GitOrigin-RevId: 617f12765614f49746d18d3368f41dfae2f3e6ca
2021-08-26 18:26:43 +03:00
objRelP2Setup source qt foreignKeys (RelDef rn ru _) = case ru of
RUManual rm -> do
let refqt = rmTable rm
(lCols, rCols) = unzip $ HM.toList $ rmColumns rm
io = fromMaybe BeforeParent $ rmInsertOrder rm
mkDependency tableName reason col = SchemaDependency
(SOSourceObj source
$ AB.mkAnyBackend
$ SOITableObj @b tableName
$ TOCol @b col)
reason
dependencies = map (mkDependency qt DRLeftColumn) lCols
<> map (mkDependency refqt DRRightColumn) rCols
server: fix the nullability of object relationships (fix hasura/graphql-engine#7201) When adding object relationships, we set the nullability of the generated GraphQL field based on whether the database backend enforces that the referenced data always exists. For manual relationships (corresponding to `manual_configuration`), the database backend is unaware of any relationship between data, and hence such fields are always set to be nullable. For relationships generated from foreign key constraints (corresponding to `foreign_key_constraint_on`), we distinguish between two cases: 1. The "forward" object relationship from a referencing table (i.e. which has the foreign key constraint) to a referenced table. This should be set to be non-nullable when all referencing columns are non-nullable. But in fact, it used to set it to be non-nullable if *any* referencing column is non-nullable, which is only correct in Postgres when `MATCH FULL` is set (a flag we don't consider). This fixes that by changing a boolean conjunction to a disjunction. 2. The "reverse" object relationship from a referenced table to a referencing table which has the foreign key constraint. This should always be set to be nullable. But in fact, it used to always be set to non-nullable, as was reported in hasura/graphql-engine#7201. This fixes that. Moreover, we have moved the computation of the nullability from `Hasura.RQL.DDL.Relationship` to `Hasura.GraphQL.Schema.Select`: this nullability used to be passed through the `riIsNullable` field of `RelInfo`, but for array relationships this information is not actually used, and moreover the remaining fields of `RelInfo` are already enough to deduce the nullability. This also adds regression tests for both (1) and (2) above. https://github.com/hasura/graphql-engine-mono/pull/2159 GitOrigin-RevId: 617f12765614f49746d18d3368f41dfae2f3e6ca
2021-08-26 18:26:43 +03:00
pure (RelInfo rn ObjRel (rmColumns rm) refqt True io, dependencies)
RUFKeyOn (SameTable columns) -> do
foreignTableForeignKeys <- findTable @b qt foreignKeys
ForeignKey constraint foreignTable colMap <- getRequiredFkey columns (HS.toList foreignTableForeignKeys)
let dependencies =
[ SchemaDependency
(SOSourceObj source
$ AB.mkAnyBackend
$ SOITableObj @b qt
$ TOForeignKey @b (_cName constraint))
DRFkey
-- this needs to be added explicitly to handle the remote table being untracked. In this case,
-- neither the using_col nor the constraint name will help.
, SchemaDependency
(SOSourceObj source
$ AB.mkAnyBackend
$ SOITable @b foreignTable)
DRRemoteTable
] <> fmap (drUsingColumnDep @b source qt) (toList columns)
server: fix the nullability of object relationships (fix hasura/graphql-engine#7201) When adding object relationships, we set the nullability of the generated GraphQL field based on whether the database backend enforces that the referenced data always exists. For manual relationships (corresponding to `manual_configuration`), the database backend is unaware of any relationship between data, and hence such fields are always set to be nullable. For relationships generated from foreign key constraints (corresponding to `foreign_key_constraint_on`), we distinguish between two cases: 1. The "forward" object relationship from a referencing table (i.e. which has the foreign key constraint) to a referenced table. This should be set to be non-nullable when all referencing columns are non-nullable. But in fact, it used to set it to be non-nullable if *any* referencing column is non-nullable, which is only correct in Postgres when `MATCH FULL` is set (a flag we don't consider). This fixes that by changing a boolean conjunction to a disjunction. 2. The "reverse" object relationship from a referenced table to a referencing table which has the foreign key constraint. This should always be set to be nullable. But in fact, it used to always be set to non-nullable, as was reported in hasura/graphql-engine#7201. This fixes that. Moreover, we have moved the computation of the nullability from `Hasura.RQL.DDL.Relationship` to `Hasura.GraphQL.Schema.Select`: this nullability used to be passed through the `riIsNullable` field of `RelInfo`, but for array relationships this information is not actually used, and moreover the remaining fields of `RelInfo` are already enough to deduce the nullability. This also adds regression tests for both (1) and (2) above. https://github.com/hasura/graphql-engine-mono/pull/2159 GitOrigin-RevId: 617f12765614f49746d18d3368f41dfae2f3e6ca
2021-08-26 18:26:43 +03:00
pure (RelInfo rn ObjRel colMap foreignTable False BeforeParent, dependencies)
RUFKeyOn (RemoteTable remoteTable remoteCols) ->
mkFkeyRel ObjRel AfterParent source rn qt remoteTable remoteCols foreignKeys
arrRelP2Setup
:: forall b m
. (QErrM m, Backend b)
=> HashMap (TableName b) (HashSet (ForeignKey b))
-> SourceName
-> TableName b
-> ArrRelDef b
-> m (RelInfo b, [SchemaDependency])
arrRelP2Setup foreignKeys source qt (RelDef rn ru _) = case ru of
RUManual rm -> do
let refqt = rmTable rm
(lCols, rCols) = unzip $ HM.toList $ rmColumns rm
deps = map (\c -> SchemaDependency
(SOSourceObj source
$ AB.mkAnyBackend
$ SOITableObj @b qt
$ TOCol @b c) DRLeftColumn)
lCols
<> map (\c -> SchemaDependency
(SOSourceObj source
$ AB.mkAnyBackend
$ SOITableObj @b refqt
$ TOCol @b c)
DRRightColumn)
rCols
server: fix the nullability of object relationships (fix hasura/graphql-engine#7201) When adding object relationships, we set the nullability of the generated GraphQL field based on whether the database backend enforces that the referenced data always exists. For manual relationships (corresponding to `manual_configuration`), the database backend is unaware of any relationship between data, and hence such fields are always set to be nullable. For relationships generated from foreign key constraints (corresponding to `foreign_key_constraint_on`), we distinguish between two cases: 1. The "forward" object relationship from a referencing table (i.e. which has the foreign key constraint) to a referenced table. This should be set to be non-nullable when all referencing columns are non-nullable. But in fact, it used to set it to be non-nullable if *any* referencing column is non-nullable, which is only correct in Postgres when `MATCH FULL` is set (a flag we don't consider). This fixes that by changing a boolean conjunction to a disjunction. 2. The "reverse" object relationship from a referenced table to a referencing table which has the foreign key constraint. This should always be set to be nullable. But in fact, it used to always be set to non-nullable, as was reported in hasura/graphql-engine#7201. This fixes that. Moreover, we have moved the computation of the nullability from `Hasura.RQL.DDL.Relationship` to `Hasura.GraphQL.Schema.Select`: this nullability used to be passed through the `riIsNullable` field of `RelInfo`, but for array relationships this information is not actually used, and moreover the remaining fields of `RelInfo` are already enough to deduce the nullability. This also adds regression tests for both (1) and (2) above. https://github.com/hasura/graphql-engine-mono/pull/2159 GitOrigin-RevId: 617f12765614f49746d18d3368f41dfae2f3e6ca
2021-08-26 18:26:43 +03:00
pure (RelInfo rn ArrRel (rmColumns rm) refqt True AfterParent, deps)
RUFKeyOn (ArrRelUsingFKeyOn refqt refCols) ->
mkFkeyRel ArrRel AfterParent source rn qt refqt refCols foreignKeys
mkFkeyRel
:: forall b m
. QErrM m
=> Backend b
=> RelType
-> InsertOrder
-> SourceName
-> RelName
-> TableName b
-> TableName b
-> NonEmpty (Column b)
-> HashMap (TableName b) (HashSet (ForeignKey b))
-> m (RelInfo b, [SchemaDependency])
mkFkeyRel relType io source rn sourceTable remoteTable remoteColumns foreignKeys = do
foreignTableForeignKeys <- findTable @b remoteTable foreignKeys
let keysThatReferenceUs = filter ((== sourceTable) . _fkForeignTable) (HS.toList foreignTableForeignKeys)
ForeignKey constraint _foreignTable colMap <- getRequiredFkey remoteColumns keysThatReferenceUs
let dependencies =
[ SchemaDependency
(SOSourceObj source
$ AB.mkAnyBackend
$ SOITableObj @b remoteTable
$ TOForeignKey @b (_cName constraint))
DRRemoteFkey
, SchemaDependency
(SOSourceObj source
$ AB.mkAnyBackend
$ SOITable @b remoteTable)
DRRemoteTable
] <> fmap (drUsingColumnDep @b source remoteTable) (toList remoteColumns)
server: fix the nullability of object relationships (fix hasura/graphql-engine#7201) When adding object relationships, we set the nullability of the generated GraphQL field based on whether the database backend enforces that the referenced data always exists. For manual relationships (corresponding to `manual_configuration`), the database backend is unaware of any relationship between data, and hence such fields are always set to be nullable. For relationships generated from foreign key constraints (corresponding to `foreign_key_constraint_on`), we distinguish between two cases: 1. The "forward" object relationship from a referencing table (i.e. which has the foreign key constraint) to a referenced table. This should be set to be non-nullable when all referencing columns are non-nullable. But in fact, it used to set it to be non-nullable if *any* referencing column is non-nullable, which is only correct in Postgres when `MATCH FULL` is set (a flag we don't consider). This fixes that by changing a boolean conjunction to a disjunction. 2. The "reverse" object relationship from a referenced table to a referencing table which has the foreign key constraint. This should always be set to be nullable. But in fact, it used to always be set to non-nullable, as was reported in hasura/graphql-engine#7201. This fixes that. Moreover, we have moved the computation of the nullability from `Hasura.RQL.DDL.Relationship` to `Hasura.GraphQL.Schema.Select`: this nullability used to be passed through the `riIsNullable` field of `RelInfo`, but for array relationships this information is not actually used, and moreover the remaining fields of `RelInfo` are already enough to deduce the nullability. This also adds regression tests for both (1) and (2) above. https://github.com/hasura/graphql-engine-mono/pull/2159 GitOrigin-RevId: 617f12765614f49746d18d3368f41dfae2f3e6ca
2021-08-26 18:26:43 +03:00
pure (RelInfo rn relType (reverseHM colMap) remoteTable False io, dependencies)
where
reverseHM :: Eq y => Hashable y => HashMap x y -> HashMap y x
reverseHM = HM.fromList . fmap swap . HM.toList
server: fix the nullability of object relationships (fix hasura/graphql-engine#7201) When adding object relationships, we set the nullability of the generated GraphQL field based on whether the database backend enforces that the referenced data always exists. For manual relationships (corresponding to `manual_configuration`), the database backend is unaware of any relationship between data, and hence such fields are always set to be nullable. For relationships generated from foreign key constraints (corresponding to `foreign_key_constraint_on`), we distinguish between two cases: 1. The "forward" object relationship from a referencing table (i.e. which has the foreign key constraint) to a referenced table. This should be set to be non-nullable when all referencing columns are non-nullable. But in fact, it used to set it to be non-nullable if *any* referencing column is non-nullable, which is only correct in Postgres when `MATCH FULL` is set (a flag we don't consider). This fixes that by changing a boolean conjunction to a disjunction. 2. The "reverse" object relationship from a referenced table to a referencing table which has the foreign key constraint. This should always be set to be nullable. But in fact, it used to always be set to non-nullable, as was reported in hasura/graphql-engine#7201. This fixes that. Moreover, we have moved the computation of the nullability from `Hasura.RQL.DDL.Relationship` to `Hasura.GraphQL.Schema.Select`: this nullability used to be passed through the `riIsNullable` field of `RelInfo`, but for array relationships this information is not actually used, and moreover the remaining fields of `RelInfo` are already enough to deduce the nullability. This also adds regression tests for both (1) and (2) above. https://github.com/hasura/graphql-engine-mono/pull/2159 GitOrigin-RevId: 617f12765614f49746d18d3368f41dfae2f3e6ca
2021-08-26 18:26:43 +03:00
-- | Try to find a foreign key constraint, identifying a constraint by its set of columns
Clean metadata arguments ## Description Thanks to #1664, the Metadata API types no longer require a `ToJSON` instance. This PR follows up with a cleanup of the types of the arguments to the metadata API: - whenever possible, it moves those argument types to where they're used (RQL.DDL.*) - it removes all unrequired instances (mostly `ToJSON`) This PR does not attempt to do it for _all_ such argument types. For some of the metadata operations, the type used to describe the argument to the API and used to represent the value in the metadata are one and the same (like for `CreateEndpoint`). Sometimes, the two types are intertwined in complex ways (`RemoteRelationship` and `RemoteRelationshipDef`). In the spirit of only doing uncontroversial cleaning work, this PR only moves types that are not used outside of RQL.DDL. Furthermore, this is a small step towards separating the different types all jumbled together in RQL.Types. ## Notes This PR also improves several `FromJSON` instances to make use of `withObject`, and to use a human readable string instead of a type name in error messages whenever possible. For instance: - before: `expected Object for Object, but encountered X` after: `expected Object for add computed field, but encountered X` - before: `Expecting an object for update query` after: `expected Object for update query, but encountered X` This PR also renames `CreateFunctionPermission` to `FunctionPermissionArgument`, to remove the quite surprising `type DropFunctionPermission = CreateFunctionPermission`. This PR also deletes some dead code, mostly in RQL.DML. This PR also moves a PG-specific source resolving function from DDL.Schema.Source to the only place where it is used: App.hs. https://github.com/hasura/graphql-engine-mono/pull/1844 GitOrigin-RevId: a594521194bb7fe6a111b02a9e099896f9fed59c
2021-07-27 13:41:42 +03:00
getRequiredFkey
:: (QErrM m, Backend b)
=> NonEmpty (Column b)
-> [ForeignKey b]
-> m (ForeignKey b)
getRequiredFkey cols fkeys =
case filteredFkeys of
[k] -> return k
server: fix the nullability of object relationships (fix hasura/graphql-engine#7201) When adding object relationships, we set the nullability of the generated GraphQL field based on whether the database backend enforces that the referenced data always exists. For manual relationships (corresponding to `manual_configuration`), the database backend is unaware of any relationship between data, and hence such fields are always set to be nullable. For relationships generated from foreign key constraints (corresponding to `foreign_key_constraint_on`), we distinguish between two cases: 1. The "forward" object relationship from a referencing table (i.e. which has the foreign key constraint) to a referenced table. This should be set to be non-nullable when all referencing columns are non-nullable. But in fact, it used to set it to be non-nullable if *any* referencing column is non-nullable, which is only correct in Postgres when `MATCH FULL` is set (a flag we don't consider). This fixes that by changing a boolean conjunction to a disjunction. 2. The "reverse" object relationship from a referenced table to a referencing table which has the foreign key constraint. This should always be set to be nullable. But in fact, it used to always be set to non-nullable, as was reported in hasura/graphql-engine#7201. This fixes that. Moreover, we have moved the computation of the nullability from `Hasura.RQL.DDL.Relationship` to `Hasura.GraphQL.Schema.Select`: this nullability used to be passed through the `riIsNullable` field of `RelInfo`, but for array relationships this information is not actually used, and moreover the remaining fields of `RelInfo` are already enough to deduce the nullability. This also adds regression tests for both (1) and (2) above. https://github.com/hasura/graphql-engine-mono/pull/2159 GitOrigin-RevId: 617f12765614f49746d18d3368f41dfae2f3e6ca
2021-08-26 18:26:43 +03:00
[] -> throw400 ConstraintError "no foreign constraint exists on the given column(s)"
_ -> throw400 ConstraintError "more than one foreign key constraint exists on the given column(s)"
Clean metadata arguments ## Description Thanks to #1664, the Metadata API types no longer require a `ToJSON` instance. This PR follows up with a cleanup of the types of the arguments to the metadata API: - whenever possible, it moves those argument types to where they're used (RQL.DDL.*) - it removes all unrequired instances (mostly `ToJSON`) This PR does not attempt to do it for _all_ such argument types. For some of the metadata operations, the type used to describe the argument to the API and used to represent the value in the metadata are one and the same (like for `CreateEndpoint`). Sometimes, the two types are intertwined in complex ways (`RemoteRelationship` and `RemoteRelationshipDef`). In the spirit of only doing uncontroversial cleaning work, this PR only moves types that are not used outside of RQL.DDL. Furthermore, this is a small step towards separating the different types all jumbled together in RQL.Types. ## Notes This PR also improves several `FromJSON` instances to make use of `withObject`, and to use a human readable string instead of a type name in error messages whenever possible. For instance: - before: `expected Object for Object, but encountered X` after: `expected Object for add computed field, but encountered X` - before: `Expecting an object for update query` after: `expected Object for update query, but encountered X` This PR also renames `CreateFunctionPermission` to `FunctionPermissionArgument`, to remove the quite surprising `type DropFunctionPermission = CreateFunctionPermission`. This PR also deletes some dead code, mostly in RQL.DML. This PR also moves a PG-specific source resolving function from DDL.Schema.Source to the only place where it is used: App.hs. https://github.com/hasura/graphql-engine-mono/pull/1844 GitOrigin-RevId: a594521194bb7fe6a111b02a9e099896f9fed59c
2021-07-27 13:41:42 +03:00
where
filteredFkeys = filter ((== HS.fromList (toList cols)) . HM.keysSet . _fkColumnMapping) fkeys
drUsingColumnDep
:: forall b
. Backend b
=> SourceName
-> TableName b
-> Column b
-> SchemaDependency
drUsingColumnDep source qt col =
SchemaDependency
(SOSourceObj source
$ AB.mkAnyBackend
$ SOITableObj @b qt
$ TOCol @b col)
DRUsingColumn
Clean metadata arguments ## Description Thanks to #1664, the Metadata API types no longer require a `ToJSON` instance. This PR follows up with a cleanup of the types of the arguments to the metadata API: - whenever possible, it moves those argument types to where they're used (RQL.DDL.*) - it removes all unrequired instances (mostly `ToJSON`) This PR does not attempt to do it for _all_ such argument types. For some of the metadata operations, the type used to describe the argument to the API and used to represent the value in the metadata are one and the same (like for `CreateEndpoint`). Sometimes, the two types are intertwined in complex ways (`RemoteRelationship` and `RemoteRelationshipDef`). In the spirit of only doing uncontroversial cleaning work, this PR only moves types that are not used outside of RQL.DDL. Furthermore, this is a small step towards separating the different types all jumbled together in RQL.Types. ## Notes This PR also improves several `FromJSON` instances to make use of `withObject`, and to use a human readable string instead of a type name in error messages whenever possible. For instance: - before: `expected Object for Object, but encountered X` after: `expected Object for add computed field, but encountered X` - before: `Expecting an object for update query` after: `expected Object for update query, but encountered X` This PR also renames `CreateFunctionPermission` to `FunctionPermissionArgument`, to remove the quite surprising `type DropFunctionPermission = CreateFunctionPermission`. This PR also deletes some dead code, mostly in RQL.DML. This PR also moves a PG-specific source resolving function from DDL.Schema.Source to the only place where it is used: App.hs. https://github.com/hasura/graphql-engine-mono/pull/1844 GitOrigin-RevId: a594521194bb7fe6a111b02a9e099896f9fed59c
2021-07-27 13:41:42 +03:00
--------------------------------------------------------------------------------
-- Drop local relationship
data DropRel b
= DropRel
{ _drSource :: !SourceName
, _drTable :: !(TableName b)
, _drRelationship :: !RelName
, _drCascade :: !Bool
}
instance (Backend b) => FromJSON (DropRel b) where
parseJSON = withObject "DropRel" $ \o ->
Clean metadata arguments ## Description Thanks to #1664, the Metadata API types no longer require a `ToJSON` instance. This PR follows up with a cleanup of the types of the arguments to the metadata API: - whenever possible, it moves those argument types to where they're used (RQL.DDL.*) - it removes all unrequired instances (mostly `ToJSON`) This PR does not attempt to do it for _all_ such argument types. For some of the metadata operations, the type used to describe the argument to the API and used to represent the value in the metadata are one and the same (like for `CreateEndpoint`). Sometimes, the two types are intertwined in complex ways (`RemoteRelationship` and `RemoteRelationshipDef`). In the spirit of only doing uncontroversial cleaning work, this PR only moves types that are not used outside of RQL.DDL. Furthermore, this is a small step towards separating the different types all jumbled together in RQL.Types. ## Notes This PR also improves several `FromJSON` instances to make use of `withObject`, and to use a human readable string instead of a type name in error messages whenever possible. For instance: - before: `expected Object for Object, but encountered X` after: `expected Object for add computed field, but encountered X` - before: `Expecting an object for update query` after: `expected Object for update query, but encountered X` This PR also renames `CreateFunctionPermission` to `FunctionPermissionArgument`, to remove the quite surprising `type DropFunctionPermission = CreateFunctionPermission`. This PR also deletes some dead code, mostly in RQL.DML. This PR also moves a PG-specific source resolving function from DDL.Schema.Source to the only place where it is used: App.hs. https://github.com/hasura/graphql-engine-mono/pull/1844 GitOrigin-RevId: a594521194bb7fe6a111b02a9e099896f9fed59c
2021-07-27 13:41:42 +03:00
DropRel
<$> o .:? "source" .!= defaultSource
<*> o .: "table"
<*> o .: "relationship"
<*> o .:? "cascade" .!= False
runDropRel
:: forall b m
. (MonadError QErr m, CacheRWM m, MetadataM m, BackendMetadata b)
=> DropRel b -> m EncJSON
runDropRel (DropRel source qt rn cascade) = do
depObjs <- collectDependencies
withNewInconsistentObjsCheck do
metadataModifiers <- traverse purgeRelDep depObjs
buildSchemaCache $ MetadataModifier $
tableMetadataSetter @b source qt %~
dropRelationshipInMetadata rn . foldr (.) id metadataModifiers
pure successMsg
where
collectDependencies = do
tabInfo <- askTableCoreInfo @b source qt
void $ askRelType (_tciFieldInfoMap tabInfo) rn ""
sc <- askSchemaCache
let depObjs = getDependentObjs
sc
(SOSourceObj source
$ AB.mkAnyBackend
$ SOITableObj @b qt
$ TORel rn)
when (depObjs /= [] && not cascade) $ reportDeps depObjs
pure depObjs
dropRelationshipInMetadata
:: RelName -> TableMetadata b -> TableMetadata b
dropRelationshipInMetadata relName =
-- Since the name of a relationship is unique in a table, the relationship
-- with given name may present in either array or object relationships but
-- not in both.
(tmObjectRelationships %~ OMap.delete relName)
. (tmArrayRelationships %~ OMap.delete relName)
purgeRelDep
:: forall b m
. QErrM m
=> Backend b
=> SchemaObjId -> m (TableMetadata b -> TableMetadata b)
purgeRelDep (SOSourceObj _ exists)
| Just (SOITableObj _ (TOPerm rn pt)) <- AB.unpackAnyBackend @b exists =
pure $ dropPermissionInMetadata rn pt
purgeRelDep d = throw500 $ "unexpected dependency of relationship : "
<> reportSchemaObj d
Clean metadata arguments ## Description Thanks to #1664, the Metadata API types no longer require a `ToJSON` instance. This PR follows up with a cleanup of the types of the arguments to the metadata API: - whenever possible, it moves those argument types to where they're used (RQL.DDL.*) - it removes all unrequired instances (mostly `ToJSON`) This PR does not attempt to do it for _all_ such argument types. For some of the metadata operations, the type used to describe the argument to the API and used to represent the value in the metadata are one and the same (like for `CreateEndpoint`). Sometimes, the two types are intertwined in complex ways (`RemoteRelationship` and `RemoteRelationshipDef`). In the spirit of only doing uncontroversial cleaning work, this PR only moves types that are not used outside of RQL.DDL. Furthermore, this is a small step towards separating the different types all jumbled together in RQL.Types. ## Notes This PR also improves several `FromJSON` instances to make use of `withObject`, and to use a human readable string instead of a type name in error messages whenever possible. For instance: - before: `expected Object for Object, but encountered X` after: `expected Object for add computed field, but encountered X` - before: `Expecting an object for update query` after: `expected Object for update query, but encountered X` This PR also renames `CreateFunctionPermission` to `FunctionPermissionArgument`, to remove the quite surprising `type DropFunctionPermission = CreateFunctionPermission`. This PR also deletes some dead code, mostly in RQL.DML. This PR also moves a PG-specific source resolving function from DDL.Schema.Source to the only place where it is used: App.hs. https://github.com/hasura/graphql-engine-mono/pull/1844 GitOrigin-RevId: a594521194bb7fe6a111b02a9e099896f9fed59c
2021-07-27 13:41:42 +03:00
--------------------------------------------------------------------------------
-- Set local relationship comment
data SetRelComment b
= SetRelComment
{ arSource :: !SourceName
, arTable :: !(TableName b)
, arRelationship :: !RelName
, arComment :: !(Maybe Text)
} deriving (Generic)
deriving instance (Backend b) => Show (SetRelComment b)
deriving instance (Backend b) => Eq (SetRelComment b)
instance (Backend b) => FromJSON (SetRelComment b) where
parseJSON = withObject "SetRelComment" $ \o ->
Clean metadata arguments ## Description Thanks to #1664, the Metadata API types no longer require a `ToJSON` instance. This PR follows up with a cleanup of the types of the arguments to the metadata API: - whenever possible, it moves those argument types to where they're used (RQL.DDL.*) - it removes all unrequired instances (mostly `ToJSON`) This PR does not attempt to do it for _all_ such argument types. For some of the metadata operations, the type used to describe the argument to the API and used to represent the value in the metadata are one and the same (like for `CreateEndpoint`). Sometimes, the two types are intertwined in complex ways (`RemoteRelationship` and `RemoteRelationshipDef`). In the spirit of only doing uncontroversial cleaning work, this PR only moves types that are not used outside of RQL.DDL. Furthermore, this is a small step towards separating the different types all jumbled together in RQL.Types. ## Notes This PR also improves several `FromJSON` instances to make use of `withObject`, and to use a human readable string instead of a type name in error messages whenever possible. For instance: - before: `expected Object for Object, but encountered X` after: `expected Object for add computed field, but encountered X` - before: `Expecting an object for update query` after: `expected Object for update query, but encountered X` This PR also renames `CreateFunctionPermission` to `FunctionPermissionArgument`, to remove the quite surprising `type DropFunctionPermission = CreateFunctionPermission`. This PR also deletes some dead code, mostly in RQL.DML. This PR also moves a PG-specific source resolving function from DDL.Schema.Source to the only place where it is used: App.hs. https://github.com/hasura/graphql-engine-mono/pull/1844 GitOrigin-RevId: a594521194bb7fe6a111b02a9e099896f9fed59c
2021-07-27 13:41:42 +03:00
SetRelComment
<$> o .:? "source" .!= defaultSource
<*> o .: "table"
<*> o .: "relationship"
<*> o .:? "comment"
runSetRelComment
:: forall m b
. (CacheRWM m, MonadError QErr m, MetadataM m, BackendMetadata b)
Clean metadata arguments ## Description Thanks to #1664, the Metadata API types no longer require a `ToJSON` instance. This PR follows up with a cleanup of the types of the arguments to the metadata API: - whenever possible, it moves those argument types to where they're used (RQL.DDL.*) - it removes all unrequired instances (mostly `ToJSON`) This PR does not attempt to do it for _all_ such argument types. For some of the metadata operations, the type used to describe the argument to the API and used to represent the value in the metadata are one and the same (like for `CreateEndpoint`). Sometimes, the two types are intertwined in complex ways (`RemoteRelationship` and `RemoteRelationshipDef`). In the spirit of only doing uncontroversial cleaning work, this PR only moves types that are not used outside of RQL.DDL. Furthermore, this is a small step towards separating the different types all jumbled together in RQL.Types. ## Notes This PR also improves several `FromJSON` instances to make use of `withObject`, and to use a human readable string instead of a type name in error messages whenever possible. For instance: - before: `expected Object for Object, but encountered X` after: `expected Object for add computed field, but encountered X` - before: `Expecting an object for update query` after: `expected Object for update query, but encountered X` This PR also renames `CreateFunctionPermission` to `FunctionPermissionArgument`, to remove the quite surprising `type DropFunctionPermission = CreateFunctionPermission`. This PR also deletes some dead code, mostly in RQL.DML. This PR also moves a PG-specific source resolving function from DDL.Schema.Source to the only place where it is used: App.hs. https://github.com/hasura/graphql-engine-mono/pull/1844 GitOrigin-RevId: a594521194bb7fe6a111b02a9e099896f9fed59c
2021-07-27 13:41:42 +03:00
=> SetRelComment b
-> m EncJSON
runSetRelComment defn = do
tabInfo <- askTableCoreInfo @b source qt
relType <- riType <$> askRelType (_tciFieldInfoMap tabInfo) rn ""
let metadataObj = MOSourceObjId source
$ AB.mkAnyBackend
$ SMOTableObj @b qt
$ MTORel rn relType
buildSchemaCacheFor metadataObj
$ MetadataModifier
$ tableMetadataSetter @b source qt %~ case relType of
ObjRel -> tmObjectRelationships.ix rn.rdComment .~ comment
ArrRel -> tmArrayRelationships.ix rn.rdComment .~ comment
pure successMsg
where
SetRelComment source qt rn comment = defn