graphql-engine/server/src-lib/Hasura/GraphQL/Schema/SubscriptionStream.hs
Auke Booij 8ccf7724ce server: Metadata origin for definitions (type parameter version v2)
The code that builds the GraphQL schema, and `buildGQLContext` in particular, is partial: not every value of `(ServerConfigCtx, GraphQLQueryType, SourceCache, HashMap RemoteSchemaName (RemoteSchemaCtx, MetadataObject), ActionCache, AnnotatedCustomTypes)` results in a valid GraphQL schema. When it fails, we want to be able to return better error messages than we currently do.

The key thing that is missing is a way to trace back GraphQL type information to their origin from the Hasura metadata. Currently, we have a number of correctness checks of our GraphQL schema. But these correctness checks only have access to pure GraphQL type information, and hence can only report errors in terms of that. Possibly the worst is the "conflicting definitions" error, which, in practice, can only be debugged by Hasura engineers. This is terrible DX for customers.

This PR allows us to print better error messages, by adding a field to the `Definition` type that traces the GraphQL type to its origin in the metadata. So the idea is simple: just add `MetadataObjId`, or `Maybe` that, or some other sum type of that, to `Definition`.

However, we want to avoid having to import a `Hasura.RQL` module from `Hasura.GraphQL.Parser`. So we instead define this additional field of `Definition` through a new type parameter, which is threaded through in `Hasura.GraphQL.Parser`. We then define type synonyms in `Hasura.GraphQL.Schema.Parser` that fill in this type parameter, so that it is not visible for the majority of the codebase.

The idea of associating metadata information to `Definition`s really comes to fruition when combined with hasura/graphql-engine-mono#4517. Their combination would allow us to use the API of fatal errors (just like the current `MonadError QErr`) to report _inconsistencies_ in the metadata. Such inconsistencies are then _automatically_ ignored. So no ad-hoc decisions need to be made on how to cut out inconsistent metadata from the GraphQL schema. This will allow us to report much better errors, as well as improve the likelihood of a successful HGE startup.

PR-URL: https://github.com/hasura/graphql-engine-mono/pull/4770
Co-authored-by: Samir Talwar <47582+SamirTalwar@users.noreply.github.com>
GitOrigin-RevId: 728402b0cae83ae8e83463a826ceeb609001acae
2022-06-28 15:53:44 +00:00

268 lines
10 KiB
Haskell

{-# LANGUAGE ApplicativeDo #-}
{-# LANGUAGE TemplateHaskellQuotes #-}
-- | Generate the GraphQL schema types related to streaming subscriptions.
module Hasura.GraphQL.Schema.SubscriptionStream
( selectStreamTable,
)
where
import Data.Has
import Data.List.NonEmpty qualified as NE
import Data.Text.Extended ((<>>))
import Hasura.Base.Error (QErr)
import Hasura.GraphQL.Parser.Class
import Hasura.GraphQL.Schema.Backend
import Hasura.GraphQL.Schema.Common
import Hasura.GraphQL.Schema.Parser
( InputFieldsParser,
Kind (..),
Parser,
)
import Hasura.GraphQL.Schema.Parser qualified as P
import Hasura.GraphQL.Schema.Select (tablePermissionsInfo, tableSelectionList, tableWhereArg)
import Hasura.GraphQL.Schema.Table (getTableGQLName, tableSelectColumns, tableSelectPermissions)
import Hasura.Name qualified as Name
import Hasura.Prelude
import Hasura.RQL.IR.Select qualified as IR
import Hasura.RQL.IR.Value qualified as IR
import Hasura.RQL.Types.Column
import Hasura.RQL.Types.SchemaCache
import Hasura.RQL.Types.Source
import Hasura.RQL.Types.SourceCustomization (NamingCase, applyFieldNameCaseCust, applyTypeNameCaseCust)
import Hasura.RQL.Types.Subscription
import Hasura.RQL.Types.Table
import Language.GraphQL.Draft.Syntax qualified as G
-- | Argument to limit the maximum number of results returned in a single batch.
cursorBatchSizeArg ::
forall n.
MonadParse n =>
NamingCase ->
InputFieldsParser n Int
cursorBatchSizeArg tCase =
fromIntegral
<$> P.field batchSizeName batchSizeDesc P.nonNegativeInt
where
batchSizeName = applyFieldNameCaseCust tCase Name._batch_size
batchSizeDesc = Just $ G.Description "maximum number of rows returned in a single batch"
-- | Cursor ordering enum fields
--
-- > enum cursor_ordering {
-- > ASC
-- > DESC
-- > }
cursorOrderingArgParser ::
forall n m r.
(MonadSchema n m, Has P.MkTypename r, Has NamingCase r, MonadReader r m) =>
m (Parser 'Both n CursorOrdering)
cursorOrderingArgParser = do
tCase <- asks getter
enumName <- P.mkTypename $ applyTypeNameCaseCust tCase Name._cursor_ordering
let description =
Just $
G.Description $
"ordering argument of a cursor"
pure $
P.enum enumName description $
NE.fromList -- It's fine to use fromList here because we know the list is never empty.
[ ( define enumNameVal,
snd enumNameVal
)
| enumNameVal <- [(Name._ASC, COAscending), (Name._DESC, CODescending)]
]
where
define (name, val) =
let orderingTypeDesc = bool "descending" "ascending" $ val == COAscending
in P.Definition name (Just $ G.Description $ orderingTypeDesc <> " ordering of the cursor") Nothing P.EnumValueInfo
-- | Argument to specify the ordering of the cursor.
-- > ordering: cursor_ordering
cursorOrderingArg ::
forall n m r.
(MonadSchema n m, Has P.MkTypename r, Has NamingCase r, MonadReader r m) =>
m (InputFieldsParser n (Maybe CursorOrdering))
cursorOrderingArg = do
cursorOrderingParser' <- cursorOrderingArgParser
pure do
P.fieldOptional Name._ordering (Just $ G.Description "cursor ordering") cursorOrderingParser'
-- | Input fields parser to parse the value of a table's column
-- > column_name: column_type
streamColumnParserArg ::
forall b n m r.
(BackendSchema b, MonadSchema n m, Has P.MkTypename r, MonadReader r m, MonadError QErr m, Has NamingCase r) =>
ColumnInfo b ->
m (InputFieldsParser n (Maybe (ColumnInfo b, ColumnValue b)))
streamColumnParserArg colInfo = do
fieldParser <- typedParser colInfo
let fieldName = ciName colInfo
fieldDesc = ciDescription colInfo
pure do
P.fieldOptional fieldName fieldDesc fieldParser <&> fmap (colInfo,)
where
typedParser columnInfo = do
fmap IR.openValueOrigin <$> columnParser (ciType columnInfo) (G.Nullability $ ciIsNullable columnInfo)
-- | Input object parser whose keys are the column names and the values are the
-- initial values of those columns from where the streaming should start.
-- > input table_stream_cursor_value_input {
-- > col1: col1_type
-- > col2: col2_type
-- ...
-- > }
streamColumnValueParser ::
forall b n m r.
(BackendSchema b, MonadSchema n m, Has P.MkTypename r, MonadReader r m, MonadError QErr m, Has NamingCase r) =>
SourceInfo b ->
G.Name ->
[ColumnInfo b] ->
m (Parser 'Input n [(ColumnInfo b, ColumnValue b)])
streamColumnValueParser sourceInfo tableGQLName colInfos =
memoizeOn 'streamColumnValueParser (_siName sourceInfo, tableGQLName) $ do
tCase <- asks getter
columnVals <- sequenceA <$> traverse streamColumnParserArg colInfos
objName <- P.mkTypename $ tableGQLName <> applyTypeNameCaseCust tCase Name.__stream_cursor_value_input
pure do
let description = G.Description $ "Initial value of the column from where the streaming should start"
P.object objName (Just description) columnVals <&> catMaybes
-- | Argument to accept the initial value from where the streaming should start.
-- > initial_value: table_stream_cursor_value_input!
streamColumnValueParserArg ::
forall b n m r.
( BackendSchema b,
MonadSchema n m,
Has P.MkTypename r,
MonadReader r m,
MonadError QErr m,
Has NamingCase r
) =>
SourceInfo b ->
G.Name ->
[ColumnInfo b] ->
m (InputFieldsParser n [(ColumnInfo b, ColumnValue b)])
streamColumnValueParserArg sourceInfo tableGQLName colInfos = do
tCase <- asks getter
columnValueParser <- streamColumnValueParser sourceInfo tableGQLName colInfos
pure do
P.field (applyFieldNameCaseCust tCase Name._initial_value) (Just $ G.Description "Stream column input with initial value") columnValueParser
-- | Argument to accept the cursor data. At the time of writing this, only a single
-- column cursor is supported and if multiple column cursors are provided,
-- then a parse error is thrown.
-- >
tableStreamColumnArg ::
forall n m r b.
(BackendSchema b, MonadSchema n m, Has P.MkTypename r, MonadReader r m, MonadError QErr m, Has NamingCase r) =>
SourceInfo b ->
G.Name ->
[ColumnInfo b] ->
m (InputFieldsParser n [IR.StreamCursorItem b])
tableStreamColumnArg sourceInfo tableGQLName colInfos = do
cursorOrderingParser <- cursorOrderingArg
streamColumnParser <- streamColumnValueParserArg sourceInfo tableGQLName colInfos
pure $ do
orderingArg <- cursorOrderingParser
columnArg <- streamColumnParser
pure $ (uncurry (IR.StreamCursorItem (fromMaybe COAscending orderingArg))) <$> columnArg
-- | Input object that contains the initial value of a column
-- along with how it needs to be ordered.
-- > input table_stream_cursor_input {
-- > initial_value: table_stream_cursor_value_input!
-- > ordering: cursor_ordering
-- > }
tableStreamCursorExp ::
forall m n r b.
MonadBuildSchema b r m n =>
SourceInfo b ->
TableInfo b ->
m (Parser 'Input n [(IR.StreamCursorItem b)])
tableStreamCursorExp sourceInfo tableInfo =
memoizeOn 'tableStreamCursorExp (_siName sourceInfo, tableInfoName tableInfo) $ do
tCase <- asks getter
tableGQLName <- getTableGQLName tableInfo
columnInfos <- tableSelectColumns sourceInfo tableInfo
objName <- P.mkTypename $ tableGQLName <> applyTypeNameCaseCust tCase Name.__stream_cursor_input
let description =
G.Description $ "Streaming cursor of the table " <>> tableGQLName
columnParsers <- tableStreamColumnArg sourceInfo tableGQLName columnInfos
pure $ P.object objName (Just description) columnParsers
-- | Argument to accept the cursor input object.
-- > cursor: [table_stream_cursor_input]!
tableStreamCursorArg ::
forall b r m n.
MonadBuildSchema b r m n =>
SourceInfo b ->
TableInfo b ->
m (InputFieldsParser n [IR.StreamCursorItem b])
tableStreamCursorArg sourceInfo tableInfo = do
cursorParser <- tableStreamCursorExp sourceInfo tableInfo
pure $ do
cursorArgs <-
P.field cursorName cursorDesc $ P.list $ P.nullable cursorParser
pure $ concat $ catMaybes cursorArgs
where
cursorName = Name._cursor
cursorDesc = Just $ G.Description "cursor to stream the results returned by the query"
-- | Arguments to the streaming subscription field.
-- > table_stream (cursor: [table_stream_cursor_input]!, batch_size: Int!, where: table_bool_exp)
tableStreamArguments ::
forall b r m n.
MonadBuildSchema b r m n =>
SourceInfo b ->
TableInfo b ->
m (InputFieldsParser n (SelectStreamArgs b))
tableStreamArguments sourceInfo tableInfo = do
tCase <- asks getter
whereParser <- tableWhereArg sourceInfo tableInfo
cursorParser <- tableStreamCursorArg sourceInfo tableInfo
pure $ do
whereArg <- whereParser
cursorArg <-
cursorParser `P.bindFields` \case
[] -> parseError "one streaming column field is expected"
[c] -> pure c
_ -> parseError "multiple column cursors are not supported yet"
batchSizeArg <- cursorBatchSizeArg tCase
pure $
IR.SelectStreamArgsG whereArg batchSizeArg cursorArg
-- | Field parser for a streaming subscription for a table.
selectStreamTable ::
forall b r m n.
MonadBuildSchema b r m n =>
SourceInfo b ->
-- | table info
TableInfo b ->
-- | field display name
G.Name ->
-- | field description, if any
Maybe G.Description ->
m (Maybe (P.FieldParser n (StreamSelectExp b)))
selectStreamTable sourceInfo tableInfo fieldName description = runMaybeT $ do
selectPermissions <- MaybeT $ tableSelectPermissions tableInfo
xStreamSubscription <- hoistMaybe $ streamSubscriptionExtension @b
stringifyNum <- retrieve soStringifyNum
tableStreamArgsParser <- lift $ tableStreamArguments sourceInfo tableInfo
selectionSetParser <- MaybeT $ tableSelectionList sourceInfo tableInfo
lift $
memoizeOn 'selectStreamTable (_siName sourceInfo, tableName, fieldName) $ do
pure $
P.subselection fieldName description tableStreamArgsParser selectionSetParser
<&> \(args, fields) ->
IR.AnnSelectStreamG
{ IR._assnXStreamingSubscription = xStreamSubscription,
IR._assnFields = fields,
IR._assnFrom = IR.FromTable tableName,
IR._assnPerm = tablePermissionsInfo selectPermissions,
IR._assnArgs = args,
IR._assnStrfyNum = stringifyNum
}
where
tableName = tableInfoName tableInfo