graphql-engine/server/src-lib/Hasura/GraphQL/Schema/Update.hs

341 lines
13 KiB
Haskell
Raw Normal View History

{-# LANGUAGE ApplicativeDo #-}
{-# LANGUAGE TemplateHaskell #-}
-- | This module provides common building blocks for composing Schema Parsers
-- used in the schema of Update Mutations.
module Hasura.GraphQL.Schema.Update
( UpdateOperator (..),
updateOperator,
buildUpdateOperators,
presetColumns,
setOp,
incOp,
updateTable,
updateTableByPk,
)
where
import Data.Has (Has)
import Data.HashMap.Strict qualified as M
import Data.HashMap.Strict.Extended qualified as M
import Data.List.NonEmpty qualified as NE
import Data.Text.Extended (commaSeparated, dquote, (<>>))
import Hasura.Base.Error (QErr)
import Hasura.GraphQL.Parser qualified as P
import Hasura.GraphQL.Schema.Backend (BackendSchema, MonadBuildSchema, columnParser)
import Hasura.GraphQL.Schema.BoolExp (boolExp)
import Hasura.GraphQL.Schema.Common (mapField, partialSQLExpToUnpreparedValue)
import Hasura.GraphQL.Schema.Mutation (mutationSelectionSet, primaryKeysArguments)
import Hasura.GraphQL.Schema.Select (tableSelectionSet)
Role-invariant schema constructors We build the GraphQL schema by combining building blocks such as `tableSelectionSet` and `columnParser`. These building blocks individually build `{InputFields,Field,}Parser` objects. Those object specify the valid GraphQL schema. Since the GraphQL schema is role-dependent, at some point we need to know what fragment of the GraphQL schema a specific role is allowed to access, and this is stored in `{Sel,Upd,Ins,Del}PermInfo` objects. We have passed around these permission objects as function arguments to the schema building blocks since we first started dealing with permissions during the PDV refactor - see hasura/graphql-engine@5168b99e463199b1934d8645bd6cd37eddb64ae1 in hasura/graphql-engine#4111. This means that, for instance, `tableSelectionSet` has as its type: ```haskell tableSelectionSet :: forall b r m n. MonadBuildSchema b r m n => SourceName -> TableInfo b -> SelPermInfo b -> m (Parser 'Output n (AnnotatedFields b)) ``` There are three reasons to change this. 1. We often pass a `Maybe (xPermInfo b)` instead of a proper `xPermInfo b`, and it's not clear what the intended semantics of this is. Some potential improvements on the data types involved are discussed in issue hasura/graphql-engine-mono#3125. 2. In most cases we also already pass a `TableInfo b`, and together with the `MonadRole` that is usually also in scope, this means that we could look up the required permissions regardless: so passing the permissions explicitly undermines the "single source of truth" principle. Breaking this principle also makes the code more difficult to read. 3. We are working towards role-based parsers (see hasura/graphql-engine-mono#2711), where the `{InputFields,Field,}Parser` objects are constructed in a role-invariant way, so that we have a single object that can be used for all roles. In particular, this means that the schema building blocks _need_ to be constructed in a role-invariant way. While this PR doesn't accomplish that, it does reduce the amount of role-specific arguments being passed, thus fixing hasura/graphql-engine-mono#3068. Concretely, this PR simply drops the `xPermInfo b` argument from almost all schema building blocks. Instead these objects are looked up from the `TableInfo b` as-needed. The resulting code is considerably simpler and shorter. One way to interpret this change is as follows. Before this PR, we figured out permissions at the top-level in `Hasura.GraphQL.Schema`, passing down the obtained `xPermInfo` objects as required. After this PR, we have a bottom-up approach where the schema building blocks themselves decide whether they want to be included for a particular role. So this moves some permission logic out of `Hasura.GraphQL.Schema`, which is very complex. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3608 GitOrigin-RevId: 51a744f34ec7d57bc8077667ae7f9cb9c4f6c962
2022-02-17 11:16:20 +03:00
import Hasura.GraphQL.Schema.Table (getTableGQLName, tableColumns, tablePermissions, tableUpdateColumns)
import Hasura.Prelude
import Hasura.RQL.IR.BoolExp (AnnBoolExp, annBoolExpTrue)
import Hasura.RQL.IR.Returning (MutationOutputG (..))
import Hasura.RQL.IR.Root (RemoteRelationshipField)
import Hasura.RQL.IR.Update (AnnotatedUpdateG (..))
import Hasura.RQL.Types.Backend (Backend (..))
import Hasura.RQL.Types.Column (ColumnInfo (..), isNumCol)
import Hasura.RQL.Types.Common (SourceName)
Role-invariant schema constructors We build the GraphQL schema by combining building blocks such as `tableSelectionSet` and `columnParser`. These building blocks individually build `{InputFields,Field,}Parser` objects. Those object specify the valid GraphQL schema. Since the GraphQL schema is role-dependent, at some point we need to know what fragment of the GraphQL schema a specific role is allowed to access, and this is stored in `{Sel,Upd,Ins,Del}PermInfo` objects. We have passed around these permission objects as function arguments to the schema building blocks since we first started dealing with permissions during the PDV refactor - see hasura/graphql-engine@5168b99e463199b1934d8645bd6cd37eddb64ae1 in hasura/graphql-engine#4111. This means that, for instance, `tableSelectionSet` has as its type: ```haskell tableSelectionSet :: forall b r m n. MonadBuildSchema b r m n => SourceName -> TableInfo b -> SelPermInfo b -> m (Parser 'Output n (AnnotatedFields b)) ``` There are three reasons to change this. 1. We often pass a `Maybe (xPermInfo b)` instead of a proper `xPermInfo b`, and it's not clear what the intended semantics of this is. Some potential improvements on the data types involved are discussed in issue hasura/graphql-engine-mono#3125. 2. In most cases we also already pass a `TableInfo b`, and together with the `MonadRole` that is usually also in scope, this means that we could look up the required permissions regardless: so passing the permissions explicitly undermines the "single source of truth" principle. Breaking this principle also makes the code more difficult to read. 3. We are working towards role-based parsers (see hasura/graphql-engine-mono#2711), where the `{InputFields,Field,}Parser` objects are constructed in a role-invariant way, so that we have a single object that can be used for all roles. In particular, this means that the schema building blocks _need_ to be constructed in a role-invariant way. While this PR doesn't accomplish that, it does reduce the amount of role-specific arguments being passed, thus fixing hasura/graphql-engine-mono#3068. Concretely, this PR simply drops the `xPermInfo b` argument from almost all schema building blocks. Instead these objects are looked up from the `TableInfo b` as-needed. The resulting code is considerably simpler and shorter. One way to interpret this change is as follows. Before this PR, we figured out permissions at the top-level in `Hasura.GraphQL.Schema`, passing down the obtained `xPermInfo` objects as required. After this PR, we have a bottom-up approach where the schema building blocks themselves decide whether they want to be included for a particular role. So this moves some permission logic out of `Hasura.GraphQL.Schema`, which is very complex. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3608 GitOrigin-RevId: 51a744f34ec7d57bc8077667ae7f9cb9c4f6c962
2022-02-17 11:16:20 +03:00
import Hasura.RQL.Types.Table
import Language.GraphQL.Draft.Syntax (Description (..), Name (..), Nullability (..), litName)
-- | @UpdateOperator b m n op@ represents one single update operator for a
-- backend @b@.
--
-- The type variable @op@ is the backend-specific data type that represents
-- update operators, typically in the form of a sum-type with an
-- @UnpreparedValue b@ in each constructor.
--
-- The @UpdateOperator b m n@ is a @Functor@. There exist building blocks of
-- common update operators (such as 'setOp', etc.) which have @op ~
-- UnpreparedValue b@. The Functor instance lets you wrap the generic update
-- operators in backend-specific tags.
data UpdateOperator b m n op = UpdateOperator
{ updateOperatorApplicableColumn :: ColumnInfo b -> Bool,
updateOperatorParser ::
Name ->
TableName b ->
NonEmpty (ColumnInfo b) ->
m (P.InputFieldsParser n (HashMap (Column b) op))
}
deriving (Functor)
-- | The top-level component for building update operators parsers.
--
-- * It implements the @preset@ functionality from Update Permissions (see
-- <https://hasura.io/docs/latest/graphql/core/auth/authorization/permission-rules.html#column-presets
-- Permissions user docs>). Use the 'presetColumns' function to extract those from the update permissions.
-- * It validates that that the update fields parsed are sound when taken as a
-- whole, i.e. that some changes are actually specified (either in the
-- mutation query text or in update preset columns) and that each column is
-- only used in one operator.
buildUpdateOperators ::
forall b r m n op.
MonadBuildSchema b r m n =>
-- | Columns with @preset@ expressions
(HashMap (Column b) op) ->
-- | Update operators to include in the Schema
[UpdateOperator b m n op] ->
TableInfo b ->
m (P.InputFieldsParser n (HashMap (Column b) op))
Role-invariant schema constructors We build the GraphQL schema by combining building blocks such as `tableSelectionSet` and `columnParser`. These building blocks individually build `{InputFields,Field,}Parser` objects. Those object specify the valid GraphQL schema. Since the GraphQL schema is role-dependent, at some point we need to know what fragment of the GraphQL schema a specific role is allowed to access, and this is stored in `{Sel,Upd,Ins,Del}PermInfo` objects. We have passed around these permission objects as function arguments to the schema building blocks since we first started dealing with permissions during the PDV refactor - see hasura/graphql-engine@5168b99e463199b1934d8645bd6cd37eddb64ae1 in hasura/graphql-engine#4111. This means that, for instance, `tableSelectionSet` has as its type: ```haskell tableSelectionSet :: forall b r m n. MonadBuildSchema b r m n => SourceName -> TableInfo b -> SelPermInfo b -> m (Parser 'Output n (AnnotatedFields b)) ``` There are three reasons to change this. 1. We often pass a `Maybe (xPermInfo b)` instead of a proper `xPermInfo b`, and it's not clear what the intended semantics of this is. Some potential improvements on the data types involved are discussed in issue hasura/graphql-engine-mono#3125. 2. In most cases we also already pass a `TableInfo b`, and together with the `MonadRole` that is usually also in scope, this means that we could look up the required permissions regardless: so passing the permissions explicitly undermines the "single source of truth" principle. Breaking this principle also makes the code more difficult to read. 3. We are working towards role-based parsers (see hasura/graphql-engine-mono#2711), where the `{InputFields,Field,}Parser` objects are constructed in a role-invariant way, so that we have a single object that can be used for all roles. In particular, this means that the schema building blocks _need_ to be constructed in a role-invariant way. While this PR doesn't accomplish that, it does reduce the amount of role-specific arguments being passed, thus fixing hasura/graphql-engine-mono#3068. Concretely, this PR simply drops the `xPermInfo b` argument from almost all schema building blocks. Instead these objects are looked up from the `TableInfo b` as-needed. The resulting code is considerably simpler and shorter. One way to interpret this change is as follows. Before this PR, we figured out permissions at the top-level in `Hasura.GraphQL.Schema`, passing down the obtained `xPermInfo` objects as required. After this PR, we have a bottom-up approach where the schema building blocks themselves decide whether they want to be included for a particular role. So this moves some permission logic out of `Hasura.GraphQL.Schema`, which is very complex. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3608 GitOrigin-RevId: 51a744f34ec7d57bc8077667ae7f9cb9c4f6c962
2022-02-17 11:16:20 +03:00
buildUpdateOperators presetCols ops tableInfo = do
parsers :: P.InputFieldsParser n [HashMap (Column b) op] <-
Role-invariant schema constructors We build the GraphQL schema by combining building blocks such as `tableSelectionSet` and `columnParser`. These building blocks individually build `{InputFields,Field,}Parser` objects. Those object specify the valid GraphQL schema. Since the GraphQL schema is role-dependent, at some point we need to know what fragment of the GraphQL schema a specific role is allowed to access, and this is stored in `{Sel,Upd,Ins,Del}PermInfo` objects. We have passed around these permission objects as function arguments to the schema building blocks since we first started dealing with permissions during the PDV refactor - see hasura/graphql-engine@5168b99e463199b1934d8645bd6cd37eddb64ae1 in hasura/graphql-engine#4111. This means that, for instance, `tableSelectionSet` has as its type: ```haskell tableSelectionSet :: forall b r m n. MonadBuildSchema b r m n => SourceName -> TableInfo b -> SelPermInfo b -> m (Parser 'Output n (AnnotatedFields b)) ``` There are three reasons to change this. 1. We often pass a `Maybe (xPermInfo b)` instead of a proper `xPermInfo b`, and it's not clear what the intended semantics of this is. Some potential improvements on the data types involved are discussed in issue hasura/graphql-engine-mono#3125. 2. In most cases we also already pass a `TableInfo b`, and together with the `MonadRole` that is usually also in scope, this means that we could look up the required permissions regardless: so passing the permissions explicitly undermines the "single source of truth" principle. Breaking this principle also makes the code more difficult to read. 3. We are working towards role-based parsers (see hasura/graphql-engine-mono#2711), where the `{InputFields,Field,}Parser` objects are constructed in a role-invariant way, so that we have a single object that can be used for all roles. In particular, this means that the schema building blocks _need_ to be constructed in a role-invariant way. While this PR doesn't accomplish that, it does reduce the amount of role-specific arguments being passed, thus fixing hasura/graphql-engine-mono#3068. Concretely, this PR simply drops the `xPermInfo b` argument from almost all schema building blocks. Instead these objects are looked up from the `TableInfo b` as-needed. The resulting code is considerably simpler and shorter. One way to interpret this change is as follows. Before this PR, we figured out permissions at the top-level in `Hasura.GraphQL.Schema`, passing down the obtained `xPermInfo` objects as required. After this PR, we have a bottom-up approach where the schema building blocks themselves decide whether they want to be included for a particular role. So this moves some permission logic out of `Hasura.GraphQL.Schema`, which is very complex. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3608 GitOrigin-RevId: 51a744f34ec7d57bc8077667ae7f9cb9c4f6c962
2022-02-17 11:16:20 +03:00
sequenceA . catMaybes <$> traverse (runUpdateOperator tableInfo) ops
pure $
parsers
`P.bindFields` ( \opExps -> do
let withPreset = presetCols : opExps
mergeDisjoint @b withPreset
)
-- | The columns that have 'preset' definitions applied to them. (see
-- <https://hasura.io/docs/latest/graphql/core/auth/authorization/permission-rules.html#column-presets
-- Permissions user docs>)
presetColumns :: UpdPermInfo b -> HashMap (Column b) (P.UnpreparedValue b)
presetColumns = fmap partialSQLExpToUnpreparedValue . upiSet
-- | Produce an InputFieldsParser from an UpdateOperator, but only if the operator
-- applies to the table (i.e., it admits a non-empty column set).
runUpdateOperator ::
forall b r m n op.
MonadBuildSchema b r m n =>
TableInfo b ->
UpdateOperator b m n op ->
m
( Maybe
( P.InputFieldsParser
n
(HashMap (Column b) op)
)
)
Role-invariant schema constructors We build the GraphQL schema by combining building blocks such as `tableSelectionSet` and `columnParser`. These building blocks individually build `{InputFields,Field,}Parser` objects. Those object specify the valid GraphQL schema. Since the GraphQL schema is role-dependent, at some point we need to know what fragment of the GraphQL schema a specific role is allowed to access, and this is stored in `{Sel,Upd,Ins,Del}PermInfo` objects. We have passed around these permission objects as function arguments to the schema building blocks since we first started dealing with permissions during the PDV refactor - see hasura/graphql-engine@5168b99e463199b1934d8645bd6cd37eddb64ae1 in hasura/graphql-engine#4111. This means that, for instance, `tableSelectionSet` has as its type: ```haskell tableSelectionSet :: forall b r m n. MonadBuildSchema b r m n => SourceName -> TableInfo b -> SelPermInfo b -> m (Parser 'Output n (AnnotatedFields b)) ``` There are three reasons to change this. 1. We often pass a `Maybe (xPermInfo b)` instead of a proper `xPermInfo b`, and it's not clear what the intended semantics of this is. Some potential improvements on the data types involved are discussed in issue hasura/graphql-engine-mono#3125. 2. In most cases we also already pass a `TableInfo b`, and together with the `MonadRole` that is usually also in scope, this means that we could look up the required permissions regardless: so passing the permissions explicitly undermines the "single source of truth" principle. Breaking this principle also makes the code more difficult to read. 3. We are working towards role-based parsers (see hasura/graphql-engine-mono#2711), where the `{InputFields,Field,}Parser` objects are constructed in a role-invariant way, so that we have a single object that can be used for all roles. In particular, this means that the schema building blocks _need_ to be constructed in a role-invariant way. While this PR doesn't accomplish that, it does reduce the amount of role-specific arguments being passed, thus fixing hasura/graphql-engine-mono#3068. Concretely, this PR simply drops the `xPermInfo b` argument from almost all schema building blocks. Instead these objects are looked up from the `TableInfo b` as-needed. The resulting code is considerably simpler and shorter. One way to interpret this change is as follows. Before this PR, we figured out permissions at the top-level in `Hasura.GraphQL.Schema`, passing down the obtained `xPermInfo` objects as required. After this PR, we have a bottom-up approach where the schema building blocks themselves decide whether they want to be included for a particular role. So this moves some permission logic out of `Hasura.GraphQL.Schema`, which is very complex. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3608 GitOrigin-RevId: 51a744f34ec7d57bc8077667ae7f9cb9c4f6c962
2022-02-17 11:16:20 +03:00
runUpdateOperator tableInfo UpdateOperator {..} = do
let tableName = tableInfoName tableInfo
tableGQLName <- getTableGQLName tableInfo
Role-invariant schema constructors We build the GraphQL schema by combining building blocks such as `tableSelectionSet` and `columnParser`. These building blocks individually build `{InputFields,Field,}Parser` objects. Those object specify the valid GraphQL schema. Since the GraphQL schema is role-dependent, at some point we need to know what fragment of the GraphQL schema a specific role is allowed to access, and this is stored in `{Sel,Upd,Ins,Del}PermInfo` objects. We have passed around these permission objects as function arguments to the schema building blocks since we first started dealing with permissions during the PDV refactor - see hasura/graphql-engine@5168b99e463199b1934d8645bd6cd37eddb64ae1 in hasura/graphql-engine#4111. This means that, for instance, `tableSelectionSet` has as its type: ```haskell tableSelectionSet :: forall b r m n. MonadBuildSchema b r m n => SourceName -> TableInfo b -> SelPermInfo b -> m (Parser 'Output n (AnnotatedFields b)) ``` There are three reasons to change this. 1. We often pass a `Maybe (xPermInfo b)` instead of a proper `xPermInfo b`, and it's not clear what the intended semantics of this is. Some potential improvements on the data types involved are discussed in issue hasura/graphql-engine-mono#3125. 2. In most cases we also already pass a `TableInfo b`, and together with the `MonadRole` that is usually also in scope, this means that we could look up the required permissions regardless: so passing the permissions explicitly undermines the "single source of truth" principle. Breaking this principle also makes the code more difficult to read. 3. We are working towards role-based parsers (see hasura/graphql-engine-mono#2711), where the `{InputFields,Field,}Parser` objects are constructed in a role-invariant way, so that we have a single object that can be used for all roles. In particular, this means that the schema building blocks _need_ to be constructed in a role-invariant way. While this PR doesn't accomplish that, it does reduce the amount of role-specific arguments being passed, thus fixing hasura/graphql-engine-mono#3068. Concretely, this PR simply drops the `xPermInfo b` argument from almost all schema building blocks. Instead these objects are looked up from the `TableInfo b` as-needed. The resulting code is considerably simpler and shorter. One way to interpret this change is as follows. Before this PR, we figured out permissions at the top-level in `Hasura.GraphQL.Schema`, passing down the obtained `xPermInfo` objects as required. After this PR, we have a bottom-up approach where the schema building blocks themselves decide whether they want to be included for a particular role. So this moves some permission logic out of `Hasura.GraphQL.Schema`, which is very complex. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3608 GitOrigin-RevId: 51a744f34ec7d57bc8077667ae7f9cb9c4f6c962
2022-02-17 11:16:20 +03:00
columns <- tableUpdateColumns tableInfo
let applicableCols :: Maybe (NonEmpty (ColumnInfo b)) =
nonEmpty . filter updateOperatorApplicableColumn $ columns
(sequenceA :: Maybe (m a) -> m (Maybe a))
(applicableCols <&> updateOperatorParser tableGQLName tableName)
-- | Ensure that /some/ updates have been specified in a mutation.
ensureNonEmpty ::
forall b m t.
(P.MonadParse m, Backend b) =>
[Text] ->
[HashMap (Column b) t] ->
m ()
ensureNonEmpty allowedOperators parsedResults =
when (null $ M.unions parsedResults) $
P.parseError $
"At least any one of "
<> commaSeparated allowedOperators
<> " is expected"
-- | Merge the results of parsed update operators. Throws an error if the same
-- column has been specified in multiple operators.
mergeDisjoint ::
forall b m t.
(Backend b, P.MonadParse m) =>
[HashMap (Column b) t] ->
m (HashMap (Column b) t)
mergeDisjoint parsedResults = do
let unioned = M.unionsAll parsedResults
duplicates = M.keys $ M.filter (not . null . NE.tail) unioned
unless (null duplicates) $
P.parseError
( "Column found in multiple operators: "
<> commaSeparated (map dquote duplicates)
<> "."
)
return $ M.map NE.head unioned
-- | Construct a parser for a single update operator.
--
-- @updateOperator _ "op" fp MkOp ["col1","col2"]@ gives a parser that accepts
-- objects in the shape of:
--
-- > op: {
-- > col1: "x",
-- > col2: "y"
-- > }
--
-- And (morally) parses into values:
--
-- > M.fromList [("col1", MkOp (fp "x")), ("col2", MkOp (fp "y"))]
updateOperator ::
forall n r m b a.
(P.MonadParse n, MonadReader r m, Has P.MkTypename r, Backend b) =>
Name ->
Name ->
(ColumnInfo b -> m (P.Parser 'P.Both n a)) ->
NonEmpty (ColumnInfo b) ->
Description ->
Description ->
m (P.InputFieldsParser n (HashMap (Column b) a))
updateOperator tableGQLName opName mkParser columns opDesc objDesc = do
fieldParsers :: NonEmpty (P.InputFieldsParser n (Maybe (Column b, a))) <-
for columns \columnInfo -> do
let fieldName = ciName columnInfo
fieldDesc = ciDescription columnInfo
fieldParser <- mkParser columnInfo
pure $
P.fieldOptional fieldName fieldDesc fieldParser
`mapField` \value -> (ciColumn columnInfo, value)
objName <- P.mkTypename $ tableGQLName <> opName <> $$(litName "_input")
pure $
fmap (M.fromList . (fold :: Maybe [(Column b, a)] -> [(Column b, a)])) $
P.fieldOptional opName (Just opDesc) $
P.object objName (Just objDesc) $
(catMaybes . toList) <$> sequenceA fieldParsers
{-# ANN updateOperator ("HLint: ignore Use tuple-section" :: String) #-}
setOp ::
forall b n r m.
( BackendSchema b,
MonadReader r m,
Has P.MkTypename r,
MonadError QErr m,
P.MonadSchema n m
) =>
UpdateOperator b m n (P.UnpreparedValue b)
setOp = UpdateOperator {..}
where
updateOperatorApplicableColumn = const True
updateOperatorParser tableGQLName tableName columns = do
let typedParser columnInfo =
fmap P.mkParameter
<$> columnParser
(ciType columnInfo)
(Nullability $ ciIsNullable columnInfo)
updateOperator
tableGQLName
$$(litName "_set")
typedParser
columns
"sets the columns of the filtered rows to the given values"
(Description $ "input type for updating data in table " <>> tableName)
incOp ::
forall b m n r.
( Backend b,
MonadReader r m,
MonadError QErr m,
P.MonadSchema n m,
BackendSchema b,
Has P.MkTypename r
) =>
UpdateOperator b m n (P.UnpreparedValue b)
incOp = UpdateOperator {..}
where
updateOperatorApplicableColumn = isNumCol
updateOperatorParser tableGQLName tableName columns = do
let typedParser columnInfo =
fmap P.mkParameter
<$> columnParser
(ciType columnInfo)
(Nullability $ ciIsNullable columnInfo)
updateOperator
tableGQLName
$$(litName "_inc")
typedParser
columns
"increments the numeric columns with given value of the filtered values"
(Description $ "input type for incrementing numeric columns in table " <>> tableName)
-- | Construct a root field, normally called update_tablename, that can be used
-- to update rows in a DB table specified by filters. Only returns a parser if
-- there are columns the user is allowed to update; otherwise returns Nothing.
updateTable ::
forall b r m n.
MonadBuildSchema b r m n =>
-- | backend-specific data needed to perform an update mutation
P.InputFieldsParser n (BackendUpdate b (P.UnpreparedValue b)) ->
-- | table source
SourceName ->
-- | table info
TableInfo b ->
-- | field display name
Name ->
-- | field description, if any
Maybe Description ->
Role-invariant schema constructors We build the GraphQL schema by combining building blocks such as `tableSelectionSet` and `columnParser`. These building blocks individually build `{InputFields,Field,}Parser` objects. Those object specify the valid GraphQL schema. Since the GraphQL schema is role-dependent, at some point we need to know what fragment of the GraphQL schema a specific role is allowed to access, and this is stored in `{Sel,Upd,Ins,Del}PermInfo` objects. We have passed around these permission objects as function arguments to the schema building blocks since we first started dealing with permissions during the PDV refactor - see hasura/graphql-engine@5168b99e463199b1934d8645bd6cd37eddb64ae1 in hasura/graphql-engine#4111. This means that, for instance, `tableSelectionSet` has as its type: ```haskell tableSelectionSet :: forall b r m n. MonadBuildSchema b r m n => SourceName -> TableInfo b -> SelPermInfo b -> m (Parser 'Output n (AnnotatedFields b)) ``` There are three reasons to change this. 1. We often pass a `Maybe (xPermInfo b)` instead of a proper `xPermInfo b`, and it's not clear what the intended semantics of this is. Some potential improvements on the data types involved are discussed in issue hasura/graphql-engine-mono#3125. 2. In most cases we also already pass a `TableInfo b`, and together with the `MonadRole` that is usually also in scope, this means that we could look up the required permissions regardless: so passing the permissions explicitly undermines the "single source of truth" principle. Breaking this principle also makes the code more difficult to read. 3. We are working towards role-based parsers (see hasura/graphql-engine-mono#2711), where the `{InputFields,Field,}Parser` objects are constructed in a role-invariant way, so that we have a single object that can be used for all roles. In particular, this means that the schema building blocks _need_ to be constructed in a role-invariant way. While this PR doesn't accomplish that, it does reduce the amount of role-specific arguments being passed, thus fixing hasura/graphql-engine-mono#3068. Concretely, this PR simply drops the `xPermInfo b` argument from almost all schema building blocks. Instead these objects are looked up from the `TableInfo b` as-needed. The resulting code is considerably simpler and shorter. One way to interpret this change is as follows. Before this PR, we figured out permissions at the top-level in `Hasura.GraphQL.Schema`, passing down the obtained `xPermInfo` objects as required. After this PR, we have a bottom-up approach where the schema building blocks themselves decide whether they want to be included for a particular role. So this moves some permission logic out of `Hasura.GraphQL.Schema`, which is very complex. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3608 GitOrigin-RevId: 51a744f34ec7d57bc8077667ae7f9cb9c4f6c962
2022-02-17 11:16:20 +03:00
m (Maybe (P.FieldParser n (AnnotatedUpdateG b (RemoteRelationshipField P.UnpreparedValue) (P.UnpreparedValue b))))
updateTable backendUpdate sourceName tableInfo fieldName description = runMaybeT do
let tableName = tableInfoName tableInfo
columns = tableColumns tableInfo
whereName = $$(litName "where")
whereDesc = "filter the rows which have to be updated"
Role-invariant schema constructors We build the GraphQL schema by combining building blocks such as `tableSelectionSet` and `columnParser`. These building blocks individually build `{InputFields,Field,}Parser` objects. Those object specify the valid GraphQL schema. Since the GraphQL schema is role-dependent, at some point we need to know what fragment of the GraphQL schema a specific role is allowed to access, and this is stored in `{Sel,Upd,Ins,Del}PermInfo` objects. We have passed around these permission objects as function arguments to the schema building blocks since we first started dealing with permissions during the PDV refactor - see hasura/graphql-engine@5168b99e463199b1934d8645bd6cd37eddb64ae1 in hasura/graphql-engine#4111. This means that, for instance, `tableSelectionSet` has as its type: ```haskell tableSelectionSet :: forall b r m n. MonadBuildSchema b r m n => SourceName -> TableInfo b -> SelPermInfo b -> m (Parser 'Output n (AnnotatedFields b)) ``` There are three reasons to change this. 1. We often pass a `Maybe (xPermInfo b)` instead of a proper `xPermInfo b`, and it's not clear what the intended semantics of this is. Some potential improvements on the data types involved are discussed in issue hasura/graphql-engine-mono#3125. 2. In most cases we also already pass a `TableInfo b`, and together with the `MonadRole` that is usually also in scope, this means that we could look up the required permissions regardless: so passing the permissions explicitly undermines the "single source of truth" principle. Breaking this principle also makes the code more difficult to read. 3. We are working towards role-based parsers (see hasura/graphql-engine-mono#2711), where the `{InputFields,Field,}Parser` objects are constructed in a role-invariant way, so that we have a single object that can be used for all roles. In particular, this means that the schema building blocks _need_ to be constructed in a role-invariant way. While this PR doesn't accomplish that, it does reduce the amount of role-specific arguments being passed, thus fixing hasura/graphql-engine-mono#3068. Concretely, this PR simply drops the `xPermInfo b` argument from almost all schema building blocks. Instead these objects are looked up from the `TableInfo b` as-needed. The resulting code is considerably simpler and shorter. One way to interpret this change is as follows. Before this PR, we figured out permissions at the top-level in `Hasura.GraphQL.Schema`, passing down the obtained `xPermInfo` objects as required. After this PR, we have a bottom-up approach where the schema building blocks themselves decide whether they want to be included for a particular role. So this moves some permission logic out of `Hasura.GraphQL.Schema`, which is very complex. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3608 GitOrigin-RevId: 51a744f34ec7d57bc8077667ae7f9cb9c4f6c962
2022-02-17 11:16:20 +03:00
viewInfo = _tciViewInfo $ _tiCoreInfo tableInfo
guard $ isMutable viIsUpdatable viewInfo
updatePerms <- MaybeT $ _permUpd <$> tablePermissions tableInfo
Role-invariant schema constructors We build the GraphQL schema by combining building blocks such as `tableSelectionSet` and `columnParser`. These building blocks individually build `{InputFields,Field,}Parser` objects. Those object specify the valid GraphQL schema. Since the GraphQL schema is role-dependent, at some point we need to know what fragment of the GraphQL schema a specific role is allowed to access, and this is stored in `{Sel,Upd,Ins,Del}PermInfo` objects. We have passed around these permission objects as function arguments to the schema building blocks since we first started dealing with permissions during the PDV refactor - see hasura/graphql-engine@5168b99e463199b1934d8645bd6cd37eddb64ae1 in hasura/graphql-engine#4111. This means that, for instance, `tableSelectionSet` has as its type: ```haskell tableSelectionSet :: forall b r m n. MonadBuildSchema b r m n => SourceName -> TableInfo b -> SelPermInfo b -> m (Parser 'Output n (AnnotatedFields b)) ``` There are three reasons to change this. 1. We often pass a `Maybe (xPermInfo b)` instead of a proper `xPermInfo b`, and it's not clear what the intended semantics of this is. Some potential improvements on the data types involved are discussed in issue hasura/graphql-engine-mono#3125. 2. In most cases we also already pass a `TableInfo b`, and together with the `MonadRole` that is usually also in scope, this means that we could look up the required permissions regardless: so passing the permissions explicitly undermines the "single source of truth" principle. Breaking this principle also makes the code more difficult to read. 3. We are working towards role-based parsers (see hasura/graphql-engine-mono#2711), where the `{InputFields,Field,}Parser` objects are constructed in a role-invariant way, so that we have a single object that can be used for all roles. In particular, this means that the schema building blocks _need_ to be constructed in a role-invariant way. While this PR doesn't accomplish that, it does reduce the amount of role-specific arguments being passed, thus fixing hasura/graphql-engine-mono#3068. Concretely, this PR simply drops the `xPermInfo b` argument from almost all schema building blocks. Instead these objects are looked up from the `TableInfo b` as-needed. The resulting code is considerably simpler and shorter. One way to interpret this change is as follows. Before this PR, we figured out permissions at the top-level in `Hasura.GraphQL.Schema`, passing down the obtained `xPermInfo` objects as required. After this PR, we have a bottom-up approach where the schema building blocks themselves decide whether they want to be included for a particular role. So this moves some permission logic out of `Hasura.GraphQL.Schema`, which is very complex. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3608 GitOrigin-RevId: 51a744f34ec7d57bc8077667ae7f9cb9c4f6c962
2022-02-17 11:16:20 +03:00
whereArg <- lift $ P.field whereName (Just whereDesc) <$> boolExp sourceName tableInfo
selection <- lift $ mutationSelectionSet sourceName tableInfo
let argsParser = liftA2 (,) backendUpdate whereArg
pure $
P.subselection fieldName description argsParser selection
<&> mkUpdateObject tableName columns updatePerms . fmap MOutMultirowFields
-- | Construct a root field, normally called 'update_tablename_by_pk', that can be used
-- to update a single in a DB table, specified by primary key. Only returns a
-- parser if there are columns the user is allowed to update and if the user has
-- select permissions on all primary keys; otherwise returns Nothing.
updateTableByPk ::
forall b r m n.
MonadBuildSchema b r m n =>
-- | backend-specific data needed to perform an update mutation
P.InputFieldsParser n (BackendUpdate b (P.UnpreparedValue b)) ->
-- | table source
SourceName ->
-- | table info
TableInfo b ->
-- | field display name
Name ->
-- | field description, if any
Maybe Description ->
m (Maybe (P.FieldParser n (AnnotatedUpdateG b (RemoteRelationshipField P.UnpreparedValue) (P.UnpreparedValue b))))
Role-invariant schema constructors We build the GraphQL schema by combining building blocks such as `tableSelectionSet` and `columnParser`. These building blocks individually build `{InputFields,Field,}Parser` objects. Those object specify the valid GraphQL schema. Since the GraphQL schema is role-dependent, at some point we need to know what fragment of the GraphQL schema a specific role is allowed to access, and this is stored in `{Sel,Upd,Ins,Del}PermInfo` objects. We have passed around these permission objects as function arguments to the schema building blocks since we first started dealing with permissions during the PDV refactor - see hasura/graphql-engine@5168b99e463199b1934d8645bd6cd37eddb64ae1 in hasura/graphql-engine#4111. This means that, for instance, `tableSelectionSet` has as its type: ```haskell tableSelectionSet :: forall b r m n. MonadBuildSchema b r m n => SourceName -> TableInfo b -> SelPermInfo b -> m (Parser 'Output n (AnnotatedFields b)) ``` There are three reasons to change this. 1. We often pass a `Maybe (xPermInfo b)` instead of a proper `xPermInfo b`, and it's not clear what the intended semantics of this is. Some potential improvements on the data types involved are discussed in issue hasura/graphql-engine-mono#3125. 2. In most cases we also already pass a `TableInfo b`, and together with the `MonadRole` that is usually also in scope, this means that we could look up the required permissions regardless: so passing the permissions explicitly undermines the "single source of truth" principle. Breaking this principle also makes the code more difficult to read. 3. We are working towards role-based parsers (see hasura/graphql-engine-mono#2711), where the `{InputFields,Field,}Parser` objects are constructed in a role-invariant way, so that we have a single object that can be used for all roles. In particular, this means that the schema building blocks _need_ to be constructed in a role-invariant way. While this PR doesn't accomplish that, it does reduce the amount of role-specific arguments being passed, thus fixing hasura/graphql-engine-mono#3068. Concretely, this PR simply drops the `xPermInfo b` argument from almost all schema building blocks. Instead these objects are looked up from the `TableInfo b` as-needed. The resulting code is considerably simpler and shorter. One way to interpret this change is as follows. Before this PR, we figured out permissions at the top-level in `Hasura.GraphQL.Schema`, passing down the obtained `xPermInfo` objects as required. After this PR, we have a bottom-up approach where the schema building blocks themselves decide whether they want to be included for a particular role. So this moves some permission logic out of `Hasura.GraphQL.Schema`, which is very complex. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3608 GitOrigin-RevId: 51a744f34ec7d57bc8077667ae7f9cb9c4f6c962
2022-02-17 11:16:20 +03:00
updateTableByPk backendUpdate sourceName tableInfo fieldName description = runMaybeT $ do
let columns = tableColumns tableInfo
tableName = tableInfoName tableInfo
Role-invariant schema constructors We build the GraphQL schema by combining building blocks such as `tableSelectionSet` and `columnParser`. These building blocks individually build `{InputFields,Field,}Parser` objects. Those object specify the valid GraphQL schema. Since the GraphQL schema is role-dependent, at some point we need to know what fragment of the GraphQL schema a specific role is allowed to access, and this is stored in `{Sel,Upd,Ins,Del}PermInfo` objects. We have passed around these permission objects as function arguments to the schema building blocks since we first started dealing with permissions during the PDV refactor - see hasura/graphql-engine@5168b99e463199b1934d8645bd6cd37eddb64ae1 in hasura/graphql-engine#4111. This means that, for instance, `tableSelectionSet` has as its type: ```haskell tableSelectionSet :: forall b r m n. MonadBuildSchema b r m n => SourceName -> TableInfo b -> SelPermInfo b -> m (Parser 'Output n (AnnotatedFields b)) ``` There are three reasons to change this. 1. We often pass a `Maybe (xPermInfo b)` instead of a proper `xPermInfo b`, and it's not clear what the intended semantics of this is. Some potential improvements on the data types involved are discussed in issue hasura/graphql-engine-mono#3125. 2. In most cases we also already pass a `TableInfo b`, and together with the `MonadRole` that is usually also in scope, this means that we could look up the required permissions regardless: so passing the permissions explicitly undermines the "single source of truth" principle. Breaking this principle also makes the code more difficult to read. 3. We are working towards role-based parsers (see hasura/graphql-engine-mono#2711), where the `{InputFields,Field,}Parser` objects are constructed in a role-invariant way, so that we have a single object that can be used for all roles. In particular, this means that the schema building blocks _need_ to be constructed in a role-invariant way. While this PR doesn't accomplish that, it does reduce the amount of role-specific arguments being passed, thus fixing hasura/graphql-engine-mono#3068. Concretely, this PR simply drops the `xPermInfo b` argument from almost all schema building blocks. Instead these objects are looked up from the `TableInfo b` as-needed. The resulting code is considerably simpler and shorter. One way to interpret this change is as follows. Before this PR, we figured out permissions at the top-level in `Hasura.GraphQL.Schema`, passing down the obtained `xPermInfo` objects as required. After this PR, we have a bottom-up approach where the schema building blocks themselves decide whether they want to be included for a particular role. So this moves some permission logic out of `Hasura.GraphQL.Schema`, which is very complex. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3608 GitOrigin-RevId: 51a744f34ec7d57bc8077667ae7f9cb9c4f6c962
2022-02-17 11:16:20 +03:00
viewInfo = _tciViewInfo $ _tiCoreInfo tableInfo
guard $ isMutable viIsUpdatable viewInfo
updatePerms <- MaybeT $ _permUpd <$> tablePermissions tableInfo
Role-invariant schema constructors We build the GraphQL schema by combining building blocks such as `tableSelectionSet` and `columnParser`. These building blocks individually build `{InputFields,Field,}Parser` objects. Those object specify the valid GraphQL schema. Since the GraphQL schema is role-dependent, at some point we need to know what fragment of the GraphQL schema a specific role is allowed to access, and this is stored in `{Sel,Upd,Ins,Del}PermInfo` objects. We have passed around these permission objects as function arguments to the schema building blocks since we first started dealing with permissions during the PDV refactor - see hasura/graphql-engine@5168b99e463199b1934d8645bd6cd37eddb64ae1 in hasura/graphql-engine#4111. This means that, for instance, `tableSelectionSet` has as its type: ```haskell tableSelectionSet :: forall b r m n. MonadBuildSchema b r m n => SourceName -> TableInfo b -> SelPermInfo b -> m (Parser 'Output n (AnnotatedFields b)) ``` There are three reasons to change this. 1. We often pass a `Maybe (xPermInfo b)` instead of a proper `xPermInfo b`, and it's not clear what the intended semantics of this is. Some potential improvements on the data types involved are discussed in issue hasura/graphql-engine-mono#3125. 2. In most cases we also already pass a `TableInfo b`, and together with the `MonadRole` that is usually also in scope, this means that we could look up the required permissions regardless: so passing the permissions explicitly undermines the "single source of truth" principle. Breaking this principle also makes the code more difficult to read. 3. We are working towards role-based parsers (see hasura/graphql-engine-mono#2711), where the `{InputFields,Field,}Parser` objects are constructed in a role-invariant way, so that we have a single object that can be used for all roles. In particular, this means that the schema building blocks _need_ to be constructed in a role-invariant way. While this PR doesn't accomplish that, it does reduce the amount of role-specific arguments being passed, thus fixing hasura/graphql-engine-mono#3068. Concretely, this PR simply drops the `xPermInfo b` argument from almost all schema building blocks. Instead these objects are looked up from the `TableInfo b` as-needed. The resulting code is considerably simpler and shorter. One way to interpret this change is as follows. Before this PR, we figured out permissions at the top-level in `Hasura.GraphQL.Schema`, passing down the obtained `xPermInfo` objects as required. After this PR, we have a bottom-up approach where the schema building blocks themselves decide whether they want to be included for a particular role. So this moves some permission logic out of `Hasura.GraphQL.Schema`, which is very complex. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3608 GitOrigin-RevId: 51a744f34ec7d57bc8077667ae7f9cb9c4f6c962
2022-02-17 11:16:20 +03:00
pkArgs <- MaybeT $ primaryKeysArguments tableInfo
selection <- MaybeT $ tableSelectionSet sourceName tableInfo
lift $ do
Role-invariant schema constructors We build the GraphQL schema by combining building blocks such as `tableSelectionSet` and `columnParser`. These building blocks individually build `{InputFields,Field,}Parser` objects. Those object specify the valid GraphQL schema. Since the GraphQL schema is role-dependent, at some point we need to know what fragment of the GraphQL schema a specific role is allowed to access, and this is stored in `{Sel,Upd,Ins,Del}PermInfo` objects. We have passed around these permission objects as function arguments to the schema building blocks since we first started dealing with permissions during the PDV refactor - see hasura/graphql-engine@5168b99e463199b1934d8645bd6cd37eddb64ae1 in hasura/graphql-engine#4111. This means that, for instance, `tableSelectionSet` has as its type: ```haskell tableSelectionSet :: forall b r m n. MonadBuildSchema b r m n => SourceName -> TableInfo b -> SelPermInfo b -> m (Parser 'Output n (AnnotatedFields b)) ``` There are three reasons to change this. 1. We often pass a `Maybe (xPermInfo b)` instead of a proper `xPermInfo b`, and it's not clear what the intended semantics of this is. Some potential improvements on the data types involved are discussed in issue hasura/graphql-engine-mono#3125. 2. In most cases we also already pass a `TableInfo b`, and together with the `MonadRole` that is usually also in scope, this means that we could look up the required permissions regardless: so passing the permissions explicitly undermines the "single source of truth" principle. Breaking this principle also makes the code more difficult to read. 3. We are working towards role-based parsers (see hasura/graphql-engine-mono#2711), where the `{InputFields,Field,}Parser` objects are constructed in a role-invariant way, so that we have a single object that can be used for all roles. In particular, this means that the schema building blocks _need_ to be constructed in a role-invariant way. While this PR doesn't accomplish that, it does reduce the amount of role-specific arguments being passed, thus fixing hasura/graphql-engine-mono#3068. Concretely, this PR simply drops the `xPermInfo b` argument from almost all schema building blocks. Instead these objects are looked up from the `TableInfo b` as-needed. The resulting code is considerably simpler and shorter. One way to interpret this change is as follows. Before this PR, we figured out permissions at the top-level in `Hasura.GraphQL.Schema`, passing down the obtained `xPermInfo` objects as required. After this PR, we have a bottom-up approach where the schema building blocks themselves decide whether they want to be included for a particular role. So this moves some permission logic out of `Hasura.GraphQL.Schema`, which is very complex. PR-URL: https://github.com/hasura/graphql-engine-mono/pull/3608 GitOrigin-RevId: 51a744f34ec7d57bc8077667ae7f9cb9c4f6c962
2022-02-17 11:16:20 +03:00
tableGQLName <- getTableGQLName tableInfo
pkObjectName <- P.mkTypename $ tableGQLName <> $$(litName "_pk_columns_input")
let pkFieldName = $$(litName "pk_columns")
pkObjectDesc = Description $ "primary key columns input for table: " <> unName tableGQLName
pkParser = P.object pkObjectName (Just pkObjectDesc) pkArgs
argsParser = (,) <$> backendUpdate <*> P.field pkFieldName Nothing pkParser
pure $
P.subselection fieldName description argsParser selection
<&> mkUpdateObject tableName columns updatePerms . fmap MOutSinglerowObject
mkUpdateObject ::
Backend b =>
TableName b ->
[ColumnInfo b] ->
UpdPermInfo b ->
( ( BackendUpdate b (P.UnpreparedValue b),
AnnBoolExp b (P.UnpreparedValue b)
),
MutationOutputG b (RemoteRelationshipField P.UnpreparedValue) (P.UnpreparedValue b)
) ->
AnnotatedUpdateG b (RemoteRelationshipField P.UnpreparedValue) (P.UnpreparedValue b)
mkUpdateObject _auTable _auAllCols updatePerms ((_auBackend, whereExp), _auOutput) =
AnnotatedUpdateG {..}
where
permissionFilter = fmap partialSQLExpToUnpreparedValue <$> upiFilter updatePerms
_auWhere = (permissionFilter, whereExp)
_auCheck = maybe annBoolExpTrue ((fmap . fmap) partialSQLExpToUnpreparedValue) $ upiCheck updatePerms