import asyncio
import logging
import time
import warnings
from concurrent.futures import Future
from queue import Queue
from threading import Event, Thread
from typing import (
    Any,
    AsyncGenerator,
    Callable,
    Dict,
    Generator,
    List,
    Literal,
    Optional,
    Tuple,
    TypeVar,
    Union,
    cast,
    overload,
)

import backoff
from anyio import fail_after
from graphql import (
    ExecutionResult,
    GraphQLSchema,
    IntrospectionQuery,
    build_ast_schema,
    parse,
    validate,
)

from .graphql_request import GraphQLRequest, support_deprecated_request
from .transport.async_transport import AsyncTransport
from .transport.exceptions import TransportConnectionFailed, TransportQueryError
from .transport.local_schema import LocalSchemaTransport
from .transport.transport import Transport
from .utilities import build_client_schema, get_introspection_query_ast
from .utilities import parse_result as parse_result_fn
from .utils import str_first_element

log = logging.getLogger(__name__)


class Client:
    """The Client class is the main entrypoint to execute GraphQL requests
    on a GQL transport.

    It can take sync or async transports as argument and can either execute
    and subscribe to requests itself with the
    :func:`execute <gql.client.Client.execute>` and
    :func:`subscribe <gql.client.Client.subscribe>` methods
    OR can be used to get a sync or async session depending on the
    transport type.

    To connect to an :ref:`async transport <async_transports>` and get an
    :class:`async session <gql.client.AsyncClientSession>`,
    use :code:`async with client as session:`

    To connect to a :ref:`sync transport <sync_transports>` and get a
    :class:`sync session <gql.client.SyncClientSession>`,
    use :code:`with client as session:`
    """

    def __init__(
        self,
        *,
        schema: Optional[Union[str, GraphQLSchema]] = None,
        introspection: Optional[IntrospectionQuery] = None,
        transport: Optional[Union[Transport, AsyncTransport]] = None,
        fetch_schema_from_transport: bool = False,
        introspection_args: Optional[Dict] = None,
        execute_timeout: Optional[Union[int, float]] = 10,
        serialize_variables: bool = False,
        parse_results: bool = False,
        batch_interval: float = 0,
        batch_max: int = 10,
    ):
        """Initialize the client with the given parameters.

        :param schema: an optional GraphQL Schema for local validation
                See :ref:`schema_validation`
        :param transport: The provided :ref:`transport <Transports>`.
        :param fetch_schema_from_transport: Boolean to indicate that if we want to fetch
                the schema from the transport using an introspection query.
        :param introspection_args: arguments passed to the
                :meth:`gql.utilities.get_introspection_query_ast` method.
        :param execute_timeout: The maximum time in seconds for the execution of a
                request before a TimeoutError is raised. Only used for async transports.
                Passing None results in waiting forever for a response.
        :param serialize_variables: whether the variable values should be
            serialized. Used for custom scalars and/or enums. Default: False.
        :param parse_results: Whether gql will try to parse the serialized output
                sent by the backend. Can be used to deserialize custom scalars or enums.
        :param batch_interval: Time to wait in seconds for batching requests together.
                Batching is disabled (by default) if 0.
        :param batch_max: Maximum number of requests in a single batch.
        """

        if introspection:
            assert (
                not schema
            ), "Cannot provide introspection and schema at the same time."
            schema = build_client_schema(introspection)

        if isinstance(schema, str):
            type_def_ast = parse(schema)
            schema = build_ast_schema(type_def_ast)

        if transport and fetch_schema_from_transport:
            assert (
                not schema
            ), "Cannot fetch the schema from transport if is already provided."

            assert not type(transport).__name__ == "AppSyncWebsocketsTransport", (
                "fetch_schema_from_transport=True is not allowed "
                "for AppSyncWebsocketsTransport "
                "because only subscriptions are allowed on the realtime endpoint."
            )

        if schema and not transport:
            transport = LocalSchemaTransport(schema)

        # GraphQL schema
        self.schema: Optional[GraphQLSchema] = schema

        # Answer of the introspection query
        self.introspection: Optional[IntrospectionQuery] = introspection

        # GraphQL transport chosen
        assert (
            transport is not None
        ), "You need to provide either a transport or a schema to the Client."
        self.transport: Union[Transport, AsyncTransport] = transport

        # Flag to indicate that we need to fetch the schema from the transport
        # On async transports, we fetch the schema before executing the first query
        self.fetch_schema_from_transport: bool = fetch_schema_from_transport
        self.introspection_args = (
            {} if introspection_args is None else introspection_args
        )

        # Enforced timeout of the execute function (only for async transports)
        self.execute_timeout = execute_timeout

        self.serialize_variables = serialize_variables
        self.parse_results = parse_results
        self.batch_interval = batch_interval
        self.batch_max = batch_max

    @property
    def batching_enabled(self) -> bool:
        return self.batch_interval != 0

    def validate(self, request: GraphQLRequest) -> None:
        """:meta private:"""
        assert (
            self.schema
        ), "Cannot validate the document locally, you need to pass a schema."

        validation_errors = validate(self.schema, request.document)
        if validation_errors:
            raise validation_errors[0]

    def _build_schema_from_introspection(
        self, execution_result: ExecutionResult
    ) -> None:
        if execution_result.errors:
            raise TransportQueryError(
                (
                    "Error while fetching schema: "
                    f"{str_first_element(execution_result.errors)}\n"
                    "If you don't need the schema, you can try with: "
                    '"fetch_schema_from_transport=False"'
                ),
                errors=execution_result.errors,
                data=execution_result.data,
                extensions=execution_result.extensions,
            )

        self.introspection = cast(IntrospectionQuery, execution_result.data)
        self.schema = build_client_schema(self.introspection)

    @staticmethod
    def _get_event_loop() -> asyncio.AbstractEventLoop:
        """Get the current asyncio event loop.

        Or create a new event loop if there isn't one (in a new Thread).
        """
        try:
            with warnings.catch_warnings():
                warnings.filterwarnings(
                    "ignore", message="There is no current event loop"
                )
                loop = asyncio.get_event_loop()
        except RuntimeError:
            loop = asyncio.new_event_loop()
            asyncio.set_event_loop(loop)

        return loop

    @overload
    def execute_sync(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[False] = ...,
        **kwargs: Any,
    ) -> Dict[str, Any]: ...  # pragma: no cover

    @overload
    def execute_sync(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[True],
        **kwargs: Any,
    ) -> ExecutionResult: ...  # pragma: no cover

    @overload
    def execute_sync(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: bool,
        **kwargs: Any,
    ) -> Union[Dict[str, Any], ExecutionResult]: ...  # pragma: no cover

    def execute_sync(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool = False,
        **kwargs: Any,
    ) -> Union[Dict[str, Any], ExecutionResult]:
        """:meta private:"""
        with self as session:
            return session.execute(
                request,
                serialize_variables=serialize_variables,
                parse_result=parse_result,
                get_execution_result=get_execution_result,
                **kwargs,
            )

    @overload
    def execute_batch_sync(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: Literal[False] = ...,
        **kwargs: Any,
    ) -> List[Dict[str, Any]]: ...  # pragma: no cover

    @overload
    def execute_batch_sync(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: Literal[True],
        **kwargs: Any,
    ) -> List[ExecutionResult]: ...  # pragma: no cover

    @overload
    def execute_batch_sync(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool,
        **kwargs: Any,
    ) -> Union[List[Dict[str, Any]], List[ExecutionResult]]: ...  # pragma: no cover

    def execute_batch_sync(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool = False,
        **kwargs: Any,
    ) -> Union[List[Dict[str, Any]], List[ExecutionResult]]:
        """:meta private:"""
        with self as session:
            return session.execute_batch(
                requests,
                serialize_variables=serialize_variables,
                parse_result=parse_result,
                get_execution_result=get_execution_result,
                **kwargs,
            )

    @overload
    async def execute_async(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[False] = ...,
        **kwargs: Any,
    ) -> Dict[str, Any]: ...  # pragma: no cover

    @overload
    async def execute_async(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[True],
        **kwargs: Any,
    ) -> ExecutionResult: ...  # pragma: no cover

    @overload
    async def execute_async(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: bool,
        **kwargs: Any,
    ) -> Union[Dict[str, Any], ExecutionResult]: ...  # pragma: no cover

    async def execute_async(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool = False,
        **kwargs: Any,
    ) -> Union[Dict[str, Any], ExecutionResult]:
        """:meta private:"""
        async with self as session:
            return await session.execute(
                request,
                serialize_variables=serialize_variables,
                parse_result=parse_result,
                get_execution_result=get_execution_result,
                **kwargs,
            )

    @overload
    async def execute_batch_async(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: Literal[False] = ...,
        **kwargs: Any,
    ) -> List[Dict[str, Any]]: ...  # pragma: no cover

    @overload
    async def execute_batch_async(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: Literal[True],
        **kwargs: Any,
    ) -> List[ExecutionResult]: ...  # pragma: no cover

    @overload
    async def execute_batch_async(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool,
        **kwargs: Any,
    ) -> Union[List[Dict[str, Any]], List[ExecutionResult]]: ...  # pragma: no cover

    async def execute_batch_async(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool = False,
        **kwargs: Any,
    ) -> Union[List[Dict[str, Any]], List[ExecutionResult]]:
        """:meta private:"""
        async with self as session:
            return await session.execute_batch(
                requests,
                serialize_variables=serialize_variables,
                parse_result=parse_result,
                get_execution_result=get_execution_result,
                **kwargs,
            )

    @overload
    def execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[False] = ...,
        **kwargs: Any,
    ) -> Dict[str, Any]: ...  # pragma: no cover

    @overload
    def execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[True],
        **kwargs: Any,
    ) -> ExecutionResult: ...  # pragma: no cover

    @overload
    def execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: bool,
        **kwargs: Any,
    ) -> Union[Dict[str, Any], ExecutionResult]: ...  # pragma: no cover

    def execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool = False,
        **kwargs: Any,
    ) -> Union[Dict[str, Any], ExecutionResult]:
        """Execute the provided request against the remote server using
        the transport provided during init.

        This function **WILL BLOCK** until the result is received from the server.

        Either the transport is sync and we execute the query synchronously directly
        OR the transport is async and we execute the query in the asyncio loop
        (blocking here until answer).

        This method will:

         - connect using the transport to get a session
         - execute the GraphQL request on the transport session
         - close the session and close the connection to the server

         If you have multiple requests to send, it is better to get your own session
         and execute the requests in your session.

         The extra arguments passed in the method will be passed to the transport
         execute method.
        """

        if isinstance(self.transport, AsyncTransport):
            loop = self._get_event_loop()

            assert not loop.is_running(), (
                "Cannot run client.execute(query) if an asyncio loop is running."
                " Use 'await client.execute_async(query)' instead."
            )

            data = loop.run_until_complete(
                self.execute_async(
                    request,
                    serialize_variables=serialize_variables,
                    parse_result=parse_result,
                    get_execution_result=get_execution_result,
                    **kwargs,
                )
            )

            return data

        else:  # Sync transports
            return self.execute_sync(
                request,
                serialize_variables=serialize_variables,
                parse_result=parse_result,
                get_execution_result=get_execution_result,
                **kwargs,
            )

    @overload
    def execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: Literal[False] = ...,
        **kwargs: Any,
    ) -> List[Dict[str, Any]]: ...  # pragma: no cover

    @overload
    def execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: Literal[True],
        **kwargs: Any,
    ) -> List[ExecutionResult]: ...  # pragma: no cover

    @overload
    def execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool,
        **kwargs: Any,
    ) -> Union[List[Dict[str, Any]], List[ExecutionResult]]: ...  # pragma: no cover

    def execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool = False,
        **kwargs: Any,
    ) -> Union[List[Dict[str, Any]], List[ExecutionResult]]:
        """Execute multiple GraphQL requests in a batch against the remote server using
        the transport provided during init.

        This function **WILL BLOCK** until the result is received from the server.

        Either the transport is sync and we execute the query synchronously directly
        OR the transport is async and we execute the query in the asyncio loop
        (blocking here until answer).

        This method will:

         - connect using the transport to get a session
         - execute the GraphQL requests on the transport session
         - close the session and close the connection to the server

         If you want to perform multiple executions, it is better to use
         the context manager to keep a session active.

         The extra arguments passed in the method will be passed to the transport
         execute method.
        """

        if isinstance(self.transport, AsyncTransport):
            loop = self._get_event_loop()

            assert not loop.is_running(), (
                "Cannot run client.execute_batch(query) if an asyncio loop is running."
                " Use 'await client.execute_batch(query)' instead."
            )

            data = loop.run_until_complete(
                self.execute_batch_async(
                    requests,
                    serialize_variables=serialize_variables,
                    parse_result=parse_result,
                    get_execution_result=get_execution_result,
                    **kwargs,
                )
            )

            return data

        else:  # Sync transports
            return self.execute_batch_sync(
                requests,
                serialize_variables=serialize_variables,
                parse_result=parse_result,
                get_execution_result=get_execution_result,
                **kwargs,
            )

    @overload
    def subscribe_async(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[False] = ...,
        **kwargs: Any,
    ) -> AsyncGenerator[Dict[str, Any], None]: ...  # pragma: no cover

    @overload
    def subscribe_async(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[True],
        **kwargs: Any,
    ) -> AsyncGenerator[ExecutionResult, None]: ...  # pragma: no cover

    @overload
    def subscribe_async(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: bool,
        **kwargs: Any,
    ) -> Union[
        AsyncGenerator[Dict[str, Any], None], AsyncGenerator[ExecutionResult, None]
    ]: ...  # pragma: no cover

    async def subscribe_async(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool = False,
        **kwargs: Any,
    ) -> Union[
        AsyncGenerator[Dict[str, Any], None], AsyncGenerator[ExecutionResult, None]
    ]:
        """:meta private:"""
        async with self as session:
            generator = session.subscribe(
                request,
                serialize_variables=serialize_variables,
                parse_result=parse_result,
                get_execution_result=get_execution_result,
                **kwargs,
            )

            async for result in generator:
                yield result

    @overload
    def subscribe(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[False] = ...,
        **kwargs: Any,
    ) -> Generator[Dict[str, Any], None, None]: ...  # pragma: no cover

    @overload
    def subscribe(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[True],
        **kwargs: Any,
    ) -> Generator[ExecutionResult, None, None]: ...  # pragma: no cover

    @overload
    def subscribe(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: bool,
        **kwargs: Any,
    ) -> Union[
        Generator[Dict[str, Any], None, None], Generator[ExecutionResult, None, None]
    ]: ...  # pragma: no cover

    def subscribe(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool = False,
        **kwargs: Any,
    ) -> Union[
        Generator[Dict[str, Any], None, None], Generator[ExecutionResult, None, None]
    ]:
        """Execute a GraphQL subscription with a python generator.

        We need an async transport for this functionality.
        """

        loop = self._get_event_loop()

        assert not loop.is_running(), (
            "Cannot run client.subscribe(query) if an asyncio loop is running."
            " Use 'await client.subscribe_async(query)' instead."
        )

        async_generator: Union[
            AsyncGenerator[Dict[str, Any], None], AsyncGenerator[ExecutionResult, None]
        ] = self.subscribe_async(
            request,
            serialize_variables=serialize_variables,
            parse_result=parse_result,
            get_execution_result=get_execution_result,
            **kwargs,
        )

        try:
            while True:
                # Note: we need to create a task here in order to be able to close
                # the async generator properly on python 3.8
                # See https://bugs.python.org/issue38559
                generator_task = asyncio.ensure_future(
                    async_generator.__anext__(), loop=loop
                )
                result: Union[
                    Dict[str, Any], ExecutionResult
                ] = loop.run_until_complete(
                    generator_task
                )  # type: ignore
                yield result

        except StopAsyncIteration:
            pass

        except (KeyboardInterrupt, Exception, GeneratorExit):
            # Graceful shutdown
            asyncio.ensure_future(async_generator.aclose(), loop=loop)

            generator_task.cancel()

            loop.run_until_complete(loop.shutdown_asyncgens())

            # Then reraise the exception
            raise

    async def connect_async(self, reconnecting=False, **kwargs):
        r"""Connect asynchronously with the underlying async transport to
        produce a session.

        That session will be a permanent auto-reconnecting session
        if :code:`reconnecting=True`.

        If you call this method, you should call the
        :meth:`close_async <gql.client.Client.close_async>` method
        for cleanup.

        :param reconnecting: if True, create a permanent reconnecting session
        :param \**kwargs: additional arguments for the
            :meth:`ReconnectingAsyncClientSession init method
            <gql.client.ReconnectingAsyncClientSession.__init__>`.
        """

        assert isinstance(
            self.transport, AsyncTransport
        ), "Only a transport of type AsyncTransport can be used asynchronously"

        self.session: Union[AsyncClientSession, SyncClientSession]

        if reconnecting:
            self.session = ReconnectingAsyncClientSession(client=self, **kwargs)
        else:
            self.session = AsyncClientSession(client=self)

        await self.session.connect()

        # Get schema from transport if needed
        try:
            if self.fetch_schema_from_transport and not self.schema:
                await self.session.fetch_schema()
        except Exception:
            # we don't know what type of exception is thrown here because it
            # depends on the underlying transport; we just make sure that the
            # transport is closed and re-raise the exception
            await self.session.close()
            raise

        return self.session

    async def close_async(self):
        """Close the async transport and stop the optional reconnecting task."""

        await self.session.close()

    async def __aenter__(self):
        return await self.connect_async()

    async def __aexit__(self, exc_type, exc, tb):
        await self.close_async()

    def connect_sync(self):
        r"""Connect synchronously with the underlying sync transport to
        produce a session.

        If you call this method, you should call the
        :meth:`close_sync <gql.client.Client.close_sync>` method
        for cleanup.
        """

        assert not isinstance(self.transport, AsyncTransport), (
            "Only a sync transport can be used."
            " Use 'async with Client(...) as session:' instead"
        )

        if not hasattr(self, "session"):
            self.session = SyncClientSession(client=self)

        assert isinstance(self.session, SyncClientSession)

        self.session.connect()

        # Get schema from transport if needed
        try:
            if self.fetch_schema_from_transport and not self.schema:
                self.session.fetch_schema()
        except Exception:
            # we don't know what type of exception is thrown here because it
            # depends on the underlying transport; we just make sure that the
            # transport is closed and re-raise the exception
            self.session.close()
            raise

        return self.session

    def close_sync(self):
        """Close the sync session and the sync transport.

        If batching is enabled, this will block until the remaining queries in the
        batching queue have been processed.
        """
        assert isinstance(self.session, SyncClientSession)

        self.session.close()

    def __enter__(self):
        return self.connect_sync()

    def __exit__(self, *args):
        self.close_sync()


class SyncClientSession:
    """An instance of this class is created when using :code:`with` on the client.

    It contains the sync method execute to send queries
    on a sync transport using the same session.
    """

    def __init__(self, client: Client):
        """:param client: the :class:`client <gql.client.Client>` used"""
        self.client = client

    def _execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        **kwargs: Any,
    ) -> ExecutionResult:
        """Execute the provided request synchronously using
        the sync transport, returning an ExecutionResult object.

        :param request: GraphQL request as a
                        :class:`GraphQLRequest <gql.GraphQLRequest>` object.
        :param serialize_variables: whether the variable values should be
            serialized. Used for custom scalars and/or enums.
            By default use the serialize_variables argument of the client.
        :param parse_result: Whether gql will deserialize the result.
            By default use the parse_results argument of the client.

        The extra arguments are passed to the transport execute method."""

        # Still supporting for now old method of providing
        # variable_values and operation_name
        request = support_deprecated_request(request, kwargs)

        # Validate document
        if self.client.schema:
            self.client.validate(request)

            # Parse variable values for custom scalars if requested
            if request.variable_values is not None:
                if serialize_variables or (
                    serialize_variables is None and self.client.serialize_variables
                ):
                    request = request.serialize_variable_values(self.client.schema)

        if self.client.batching_enabled:
            future_result = self._execute_future(request)
            result = future_result.result()

        else:
            result = self.transport.execute(
                request,
                **kwargs,
            )

        # Unserialize the result if requested
        if self.client.schema:
            if parse_result or (parse_result is None and self.client.parse_results):
                result.data = parse_result_fn(
                    self.client.schema,
                    request.document,
                    result.data,
                    operation_name=request.operation_name,
                )

        return result

    @overload
    def execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[False] = ...,
        **kwargs: Any,
    ) -> Dict[str, Any]: ...  # pragma: no cover

    @overload
    def execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[True],
        **kwargs: Any,
    ) -> ExecutionResult: ...  # pragma: no cover

    @overload
    def execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: bool,
        **kwargs: Any,
    ) -> Union[Dict[str, Any], ExecutionResult]: ...  # pragma: no cover

    def execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool = False,
        **kwargs: Any,
    ) -> Union[Dict[str, Any], ExecutionResult]:
        """Execute the provided request synchronously using
        the sync transport.

        Raises a TransportQueryError if an error has been returned in
            the ExecutionResult.

        :param request: GraphQL query as :class:`GraphQLRequest <gql.GraphQLRequest>`.
        :param serialize_variables: whether the variable values should be
            serialized. Used for custom scalars and/or enums.
            By default use the serialize_variables argument of the client.
        :param parse_result: Whether gql will deserialize the result.
            By default use the parse_results argument of the client.
        :param get_execution_result: return the full ExecutionResult instance instead of
            only the "data" field. Necessary if you want to get the "extensions" field.

        The extra arguments are passed to the transport execute method."""

        # Validate and execute on the transport
        result = self._execute(
            request,
            serialize_variables=serialize_variables,
            parse_result=parse_result,
            **kwargs,
        )

        # Raise an error if an error is returned in the ExecutionResult object
        if result.errors:
            raise TransportQueryError(
                str_first_element(result.errors),
                errors=result.errors,
                data=result.data,
                extensions=result.extensions,
            )

        assert (
            result.data is not None
        ), "Transport returned an ExecutionResult without data or errors"

        if get_execution_result:
            return result

        return result.data

    def _execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        validate_document: Optional[bool] = True,
        **kwargs: Any,
    ) -> List[ExecutionResult]:
        """Execute multiple GraphQL requests in a batch, using
        the sync transport, returning a list of ExecutionResult objects.

        :param requests: List of requests that will be executed.
        :param serialize_variables: whether the variable values should be
            serialized. Used for custom scalars and/or enums.
            By default use the serialize_variables argument of the client.
        :param parse_result: Whether gql will deserialize the result.
            By default use the parse_results argument of the client.
        :param validate_document: Whether we still need to validate the document.

        The extra arguments are passed to the transport execute method."""

        # Validate document
        if self.client.schema:

            if validate_document:
                for req in requests:
                    self.client.validate(req)

            # Parse variable values for custom scalars if requested
            if serialize_variables or (
                serialize_variables is None and self.client.serialize_variables
            ):
                requests = [
                    (
                        req.serialize_variable_values(self.client.schema)
                        if req.variable_values is not None
                        else req
                    )
                    for req in requests
                ]

        results = self.transport.execute_batch(requests, **kwargs)

        # Unserialize the result if requested
        if self.client.schema:
            if parse_result or (parse_result is None and self.client.parse_results):
                for result in results:
                    result.data = parse_result_fn(
                        self.client.schema,
                        req.document,
                        result.data,
                        operation_name=req.operation_name,
                    )

        return results

    @overload
    def execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: Literal[False] = ...,
        **kwargs: Any,
    ) -> List[Dict[str, Any]]: ...  # pragma: no cover

    @overload
    def execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: Literal[True],
        **kwargs: Any,
    ) -> List[ExecutionResult]: ...  # pragma: no cover

    @overload
    def execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool,
        **kwargs: Any,
    ) -> Union[List[Dict[str, Any]], List[ExecutionResult]]: ...  # pragma: no cover

    def execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool = False,
        **kwargs: Any,
    ) -> Union[List[Dict[str, Any]], List[ExecutionResult]]:
        """Execute multiple GraphQL requests in a batch, using
        the sync transport. This method sends the requests to the server all at once.

        Raises a TransportQueryError if an error has been returned in any
          ExecutionResult.

        :param requests: List of requests that will be executed.
        :param serialize_variables: whether the variable values should be
            serialized. Used for custom scalars and/or enums.
            By default use the serialize_variables argument of the client.
        :param parse_result: Whether gql will deserialize the result.
            By default use the parse_results argument of the client.
        :param get_execution_result: return the full ExecutionResult instance instead of
            only the "data" field. Necessary if you want to get the "extensions" field.

        The extra arguments are passed to the transport execute method."""

        # Validate and execute on the transport
        results = self._execute_batch(
            requests,
            serialize_variables=serialize_variables,
            parse_result=parse_result,
            **kwargs,
        )

        for result in results:
            # Raise an error if an error is returned in the ExecutionResult object
            if result.errors:
                raise TransportQueryError(
                    str_first_element(result.errors),
                    errors=result.errors,
                    data=result.data,
                    extensions=result.extensions,
                )

            assert (
                result.data is not None
            ), "Transport returned an ExecutionResult without data or errors"

        if get_execution_result:
            return results

        return cast(List[Dict[str, Any]], [result.data for result in results])

    def _batch_loop(self) -> None:
        """main loop of the thread used to wait for requests
        to execute them in a batch"""

        stop_loop = False

        while not stop_loop:

            # First wait for a first request in from the batch queue
            requests_and_futures: List[Tuple[GraphQLRequest, Future]] = []
            request_and_future: Tuple[GraphQLRequest, Future] = self.batch_queue.get()
            if request_and_future is None:
                break
            requests_and_futures.append(request_and_future)

            # Then wait the requested batch interval except if we already
            # have the maximum number of requests in the queue
            if self.batch_queue.qsize() < self.client.batch_max - 1:
                time.sleep(self.client.batch_interval)

            # Then get the requests which had been made during that wait interval
            for _ in range(self.client.batch_max - 1):
                if self.batch_queue.empty():
                    break
                request_and_future = self.batch_queue.get()
                if request_and_future is None:
                    stop_loop = True
                    break
                requests_and_futures.append(request_and_future)

            requests = [request for request, _ in requests_and_futures]
            futures = [future for _, future in requests_and_futures]

            # Manually execute the requests in a batch
            try:
                results: List[ExecutionResult] = self._execute_batch(
                    requests,
                    serialize_variables=False,  # already done
                    parse_result=False,
                    validate_document=False,
                )
            except Exception as exc:
                for future in futures:
                    future.set_exception(exc)
                continue

            # Fill in the future results
            for result, future in zip(results, futures):
                future.set_result(result)

        # Indicate that the Thread has stopped
        self._batch_thread_stopped_event.set()

    def _execute_future(
        self,
        request: GraphQLRequest,
    ) -> Future:
        """If batching is enabled, this method will put a request in the batching queue
        instead of executing it directly so that the requests could be put in a batch.
        """

        assert hasattr(self, "batch_queue"), "Batching is not enabled"
        assert not self._batch_thread_stop_requested, "Batching thread has been stopped"

        future: Future = Future()
        self.batch_queue.put((request, future))

        return future

    def connect(self):
        """Connect the transport and initialize the batch threading loop if batching
        is enabled."""

        if self.client.batching_enabled:
            self.batch_queue: Queue = Queue()
            self._batch_thread_stop_requested = False
            self._batch_thread_stopped_event = Event()
            self._batch_thread = Thread(target=self._batch_loop, daemon=True)
            self._batch_thread.start()

        self.transport.connect()

    def close(self):
        """Close the transport and cleanup the batching thread if batching is enabled.

        Will wait until all the remaining requests in the batch processing queue
        have been executed.
        """
        if hasattr(self, "_batch_thread_stopped_event"):
            # Send a None in the queue to indicate that the batching Thread must stop
            # after having processed the remaining requests in the queue
            self._batch_thread_stop_requested = True
            self.batch_queue.put(None)

            # Wait for the Thread to stop
            self._batch_thread_stopped_event.wait()

        self.transport.close()

    def fetch_schema(self) -> None:
        """Fetch the GraphQL schema explicitly using introspection.

        Don't use this function and instead set the fetch_schema_from_transport
        attribute to True"""
        introspection_query = get_introspection_query_ast(
            **self.client.introspection_args
        )
        execution_result = self.transport.execute(GraphQLRequest(introspection_query))

        self.client._build_schema_from_introspection(execution_result)

    @property
    def transport(self):
        return self.client.transport


class AsyncClientSession:
    """An instance of this class is created when using :code:`async with` on a
    :class:`client <gql.client.Client>`.

    It contains the async methods (execute, subscribe) to send queries
    on an async transport using the same session.
    """

    def __init__(self, client: Client):
        """:param client: the :class:`client <gql.client.Client>` used"""
        self.client = client

    async def _subscribe(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        **kwargs: Any,
    ) -> AsyncGenerator[ExecutionResult, None]:
        """Coroutine to subscribe asynchronously to the provided request
        asynchronously using the async transport,
        returning an async generator producing ExecutionResult objects.

        * Validate the query with the schema if provided.
        * Serialize the variable_values if requested.

        :param request: GraphQL request as a
                        :class:`GraphQLRequest <gql.GraphQLRequest>` object.
        :param serialize_variables: whether the variable values should be
            serialized. Used for custom scalars and/or enums.
            By default use the serialize_variables argument of the client.
        :param parse_result: Whether gql will deserialize the result.
            By default use the parse_results argument of the client.

        The extra arguments are passed to the transport subscribe method."""

        # Still supporting for now old method of providing
        # variable_values and operation_name
        request = support_deprecated_request(request, kwargs)

        # Validate document
        if self.client.schema:
            self.client.validate(request)

            # Parse variable values for custom scalars if requested
            if request.variable_values is not None:
                if serialize_variables or (
                    serialize_variables is None and self.client.serialize_variables
                ):
                    request = request.serialize_variable_values(self.client.schema)

        # Subscribe to the transport
        inner_generator: AsyncGenerator[ExecutionResult, None] = (
            self.transport.subscribe(
                request,
                **kwargs,
            )
        )

        # Keep a reference to the inner generator
        # This is only used for the tests to simulate a KeyboardInterrupt event
        self._generator = inner_generator

        try:
            async for result in inner_generator:
                if self.client.schema:
                    if parse_result or (
                        parse_result is None and self.client.parse_results
                    ):
                        result.data = parse_result_fn(
                            self.client.schema,
                            request.document,
                            result.data,
                            operation_name=request.operation_name,
                        )

                yield result

        finally:
            await inner_generator.aclose()

    @overload
    def subscribe(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[False] = ...,
        **kwargs: Any,
    ) -> AsyncGenerator[Dict[str, Any], None]: ...  # pragma: no cover

    @overload
    def subscribe(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[True],
        **kwargs: Any,
    ) -> AsyncGenerator[ExecutionResult, None]: ...  # pragma: no cover

    @overload
    def subscribe(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: bool,
        **kwargs: Any,
    ) -> Union[
        AsyncGenerator[Dict[str, Any], None], AsyncGenerator[ExecutionResult, None]
    ]: ...  # pragma: no cover

    async def subscribe(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool = False,
        **kwargs: Any,
    ) -> Union[
        AsyncGenerator[Dict[str, Any], None], AsyncGenerator[ExecutionResult, None]
    ]:
        """Coroutine to subscribe asynchronously to the provided request
        asynchronously using the async transport.

        Raises a TransportQueryError if an error has been returned in
            the ExecutionResult.

        :param request: GraphQL query as :class:`GraphQLRequest <gql.GraphQLRequest>`.
        :param serialize_variables: whether the variable values should be
            serialized. Used for custom scalars and/or enums.
            By default use the serialize_variables argument of the client.
        :param parse_result: Whether gql will deserialize the result.
            By default use the parse_results argument of the client.
        :param get_execution_result: yield the full ExecutionResult instance instead of
            only the "data" field. Necessary if you want to get the "extensions" field.

        The extra arguments are passed to the transport subscribe method."""

        inner_generator: AsyncGenerator[ExecutionResult, None] = self._subscribe(
            request,
            serialize_variables=serialize_variables,
            parse_result=parse_result,
            **kwargs,
        )

        try:
            # Validate and subscribe on the transport
            async for result in inner_generator:
                # Raise an error if an error is returned in the ExecutionResult object
                if result.errors:
                    raise TransportQueryError(
                        str_first_element(result.errors),
                        errors=result.errors,
                        data=result.data,
                        extensions=result.extensions,
                    )

                elif result.data is not None:
                    if get_execution_result:
                        yield result
                    else:
                        yield result.data
        finally:
            await inner_generator.aclose()

    async def _execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        **kwargs: Any,
    ) -> ExecutionResult:
        """Coroutine to execute the provided request asynchronously using
        the async transport, returning an ExecutionResult object.

        * Validate the query with the schema if provided.
        * Serialize the variable_values if requested.

        :param request: GraphQL request as a
                        :class:`GraphQLRequest <gql.GraphQLRequest>` object.
        :param serialize_variables: whether the variable values should be
            serialized. Used for custom scalars and/or enums.
            By default use the serialize_variables argument of the client.
        :param parse_result: Whether gql will deserialize the result.
            By default use the parse_results argument of the client.

        The extra arguments are passed to the transport execute method."""

        # Still supporting for now old method of providing
        # variable_values and operation_name
        request = support_deprecated_request(request, kwargs)

        # Validate document
        if self.client.schema:
            self.client.validate(request)

            # Parse variable values for custom scalars if requested
            if request.variable_values is not None:
                if serialize_variables or (
                    serialize_variables is None and self.client.serialize_variables
                ):
                    request = request.serialize_variable_values(self.client.schema)

        # Check if batching is enabled
        if self.client.batching_enabled:
            future_result = await self._execute_future(request)
            result = await future_result
        else:
            # Execute the query with the transport with a timeout
            with fail_after(self.client.execute_timeout):
                result = await self.transport.execute(
                    request,
                    **kwargs,
                )

        # Unserialize the result if requested
        if self.client.schema:
            if parse_result or (parse_result is None and self.client.parse_results):
                result.data = parse_result_fn(
                    self.client.schema,
                    request.document,
                    result.data,
                    operation_name=request.operation_name,
                )

        return result

    @overload
    async def execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[False] = ...,
        **kwargs: Any,
    ) -> Dict[str, Any]: ...  # pragma: no cover

    @overload
    async def execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: Literal[True],
        **kwargs: Any,
    ) -> ExecutionResult: ...  # pragma: no cover

    @overload
    async def execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = ...,
        parse_result: Optional[bool] = ...,
        get_execution_result: bool,
        **kwargs: Any,
    ) -> Union[Dict[str, Any], ExecutionResult]: ...  # pragma: no cover

    async def execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool = False,
        **kwargs: Any,
    ) -> Union[Dict[str, Any], ExecutionResult]:
        """Coroutine to execute the provided request asynchronously using
        the async transport.

        Raises a TransportQueryError if an error has been returned in
            the ExecutionResult.

        :param request: GraphQL query as :class:`GraphQLRequest <gql.GraphQLRequest>`.
        :param serialize_variables: whether the variable values should be
            serialized. Used for custom scalars and/or enums.
            By default use the serialize_variables argument of the client.
        :param parse_result: Whether gql will deserialize the result.
            By default use the parse_results argument of the client.
        :param get_execution_result: return the full ExecutionResult instance instead of
            only the "data" field. Necessary if you want to get the "extensions" field.

        The extra arguments are passed to the transport execute method."""

        # Validate and execute on the transport
        result = await self._execute(
            request,
            serialize_variables=serialize_variables,
            parse_result=parse_result,
            **kwargs,
        )

        # Raise an error if an error is returned in the ExecutionResult object
        if result.errors:
            raise TransportQueryError(
                str_first_element(result.errors),
                errors=result.errors,
                data=result.data,
                extensions=result.extensions,
            )

        assert (
            result.data is not None
        ), "Transport returned an ExecutionResult without data or errors"

        if get_execution_result:
            return result

        return result.data

    async def _execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        validate_document: Optional[bool] = True,
        **kwargs: Any,
    ) -> List[ExecutionResult]:
        """Execute multiple GraphQL requests in a batch, using
        the async transport, returning a list of ExecutionResult objects.

        :param requests: List of requests that will be executed.
        :param serialize_variables: whether the variable values should be
            serialized. Used for custom scalars and/or enums.
            By default use the serialize_variables argument of the client.
        :param parse_result: Whether gql will deserialize the result.
            By default use the parse_results argument of the client.
        :param validate_document: Whether we still need to validate the document.

        The extra arguments are passed to the transport execute_batch method."""

        # Validate document
        if self.client.schema:

            if validate_document:
                for req in requests:
                    self.client.validate(req)

            # Parse variable values for custom scalars if requested
            if serialize_variables or (
                serialize_variables is None and self.client.serialize_variables
            ):
                requests = [
                    (
                        req.serialize_variable_values(self.client.schema)
                        if req.variable_values is not None
                        else req
                    )
                    for req in requests
                ]

        results = await self.transport.execute_batch(requests, **kwargs)

        # Unserialize the result if requested
        if self.client.schema:
            if parse_result or (parse_result is None and self.client.parse_results):
                for result in results:
                    result.data = parse_result_fn(
                        self.client.schema,
                        req.document,
                        result.data,
                        operation_name=req.operation_name,
                    )

        return results

    @overload
    async def execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: Literal[False] = ...,
        **kwargs: Any,
    ) -> List[Dict[str, Any]]: ...  # pragma: no cover

    @overload
    async def execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: Literal[True],
        **kwargs: Any,
    ) -> List[ExecutionResult]: ...  # pragma: no cover

    @overload
    async def execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool,
        **kwargs: Any,
    ) -> Union[List[Dict[str, Any]], List[ExecutionResult]]: ...  # pragma: no cover

    async def execute_batch(
        self,
        requests: List[GraphQLRequest],
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        get_execution_result: bool = False,
        **kwargs: Any,
    ) -> Union[List[Dict[str, Any]], List[ExecutionResult]]:
        """Execute multiple GraphQL requests in a batch, using
        the async transport. This method sends the requests to the server all at once.

        Raises a TransportQueryError if an error has been returned in any
          ExecutionResult.

        :param requests: List of requests that will be executed.
        :param serialize_variables: whether the variable values should be
            serialized. Used for custom scalars and/or enums.
            By default use the serialize_variables argument of the client.
        :param parse_result: Whether gql will deserialize the result.
            By default use the parse_results argument of the client.
        :param get_execution_result: return the full ExecutionResult instance instead of
            only the "data" field. Necessary if you want to get the "extensions" field.

        The extra arguments are passed to the transport execute method."""

        # Validate and execute on the transport
        results = await self._execute_batch(
            requests,
            serialize_variables=serialize_variables,
            parse_result=parse_result,
            **kwargs,
        )

        for result in results:
            # Raise an error if an error is returned in the ExecutionResult object
            if result.errors:
                raise TransportQueryError(
                    str_first_element(result.errors),
                    errors=result.errors,
                    data=result.data,
                    extensions=result.extensions,
                )

            assert (
                result.data is not None
            ), "Transport returned an ExecutionResult without data or errors"

        if get_execution_result:
            return results

        return cast(List[Dict[str, Any]], [result.data for result in results])

    async def _batch_loop(self) -> None:
        """Main loop of the task used to wait for requests
        to execute them in a batch"""

        stop_loop = False

        while not stop_loop:
            # First wait for a first request in from the batch queue
            requests_and_futures: List[Tuple[GraphQLRequest, asyncio.Future]] = []

            # Wait for the first request
            request_and_future: Optional[Tuple[GraphQLRequest, asyncio.Future]] = (
                await self.batch_queue.get()
            )

            if request_and_future is None:
                # None is our sentinel value to stop the loop
                break

            requests_and_futures.append(request_and_future)

            # Then wait the requested batch interval except if we already
            # have the maximum number of requests in the queue
            if self.batch_queue.qsize() < self.client.batch_max - 1:
                # Wait for the batch interval
                await asyncio.sleep(self.client.batch_interval)

            # Then get the requests which had been made during that wait interval
            for _ in range(self.client.batch_max - 1):
                try:
                    # Use get_nowait since we don't want to wait here
                    request_and_future = self.batch_queue.get_nowait()

                    if request_and_future is None:
                        # Sentinel value - stop after processing current batch
                        stop_loop = True
                        break

                    requests_and_futures.append(request_and_future)

                except asyncio.QueueEmpty:
                    # No more requests in queue, that's fine
                    break

            # Extract requests and futures
            requests = [request for request, _ in requests_and_futures]
            futures = [future for _, future in requests_and_futures]

            # Execute the batch
            try:
                results: List[ExecutionResult] = await self._execute_batch(
                    requests,
                    serialize_variables=False,  # already done
                    parse_result=False,  # will be done later
                    validate_document=False,  # already validated
                )

                # Set the result for each future
                for result, future in zip(results, futures):
                    if not future.cancelled():
                        future.set_result(result)

            except Exception as exc:
                # If batch execution fails, propagate the error to all futures
                for future in futures:
                    if not future.cancelled():
                        future.set_exception(exc)

        # Signal that the task has stopped
        self._batch_task_stopped_event.set()

    async def _execute_future(
        self,
        request: GraphQLRequest,
    ) -> asyncio.Future:
        """If batching is enabled, this method will put a request in the batching queue
        instead of executing it directly so that the requests could be put in a batch.
        """

        assert hasattr(self, "batch_queue"), "Batching is not enabled"
        assert not self._batch_task_stop_requested, "Batching task has been stopped"

        future: asyncio.Future = asyncio.Future()
        await self.batch_queue.put((request, future))

        return future

    async def _batch_init(self):
        """Initialize the batch task loop if batching is enabled."""
        if self.client.batching_enabled:
            self.batch_queue: asyncio.Queue = asyncio.Queue()
            self._batch_task_stop_requested = False
            self._batch_task_stopped_event = asyncio.Event()
            self._batch_task = asyncio.create_task(self._batch_loop())

    async def _batch_cleanup(self):
        """Cleanup the batching task if batching is enabled."""
        if hasattr(self, "_batch_task_stopped_event"):
            # Send a None in the queue to indicate that the batching task must stop
            # after having processed the remaining requests in the queue
            self._batch_task_stop_requested = True
            await self.batch_queue.put(None)

            # Wait for the task to process remaining requests and stop
            await self._batch_task_stopped_event.wait()

    async def connect(self):
        """Connect the transport and initialize the batch task loop if batching
        is enabled."""

        await self._batch_init()

        try:
            await self.transport.connect()
        except Exception as e:
            await self.transport.close()
            raise e

    async def close(self):
        """Close the transport and cleanup the batching task if batching is enabled.

        Will wait until all the remaining requests in the batch processing queue
        have been executed.
        """
        await self._batch_cleanup()

        await self.transport.close()

    async def fetch_schema(self) -> None:
        """Fetch the GraphQL schema explicitly using introspection.

        Don't use this function and instead set the fetch_schema_from_transport
        attribute to True"""
        introspection_query = get_introspection_query_ast(
            **self.client.introspection_args
        )
        execution_result = await self.transport.execute(
            GraphQLRequest(introspection_query)
        )

        self.client._build_schema_from_introspection(execution_result)

    @property
    def transport(self):
        return self.client.transport


_CallableT = TypeVar("_CallableT", bound=Callable[..., Any])
_Decorator = Callable[[_CallableT], _CallableT]


class ReconnectingAsyncClientSession(AsyncClientSession):
    """An instance of this class is created when using the
    :meth:`connect_async <gql.client.Client.connect_async>` method of the
    :class:`Client <gql.client.Client>` class with :code:`reconnecting=True`.

    It is used to provide a single session which will reconnect automatically if
    the connection fails.
    """

    def __init__(
        self,
        client: Client,
        *,
        retry_connect: Union[bool, _Decorator] = True,
        retry_execute: Union[bool, _Decorator] = True,
    ):
        """
        :param client: the :class:`client <gql.client.Client>` used.
        :param retry_connect: Either a Boolean to activate/deactivate the retries
            for the connection to the transport OR a backoff decorator to
            provide specific retries parameters for the connections.
        :param retry_execute: Either a Boolean to activate/deactivate the retries
            for the execute method OR a backoff decorator to
            provide specific retries parameters for this method.
        """
        self.client = client
        self._connect_task = None

        self._reconnect_request_event = asyncio.Event()
        self._connected_event = asyncio.Event()

        if retry_connect is True:
            # By default, retry again and again, with maximum 60 seconds
            # between retries
            self.retry_connect = backoff.on_exception(
                backoff.expo,
                Exception,
                max_value=60,
            )
        elif retry_connect is False:
            self.retry_connect = lambda e: e
        else:
            assert callable(retry_connect)
            self.retry_connect = retry_connect

        if retry_execute is True:
            # By default, retry 5 times, except if we receive a TransportQueryError
            self.retry_execute = backoff.on_exception(
                backoff.expo,
                Exception,
                max_tries=5,
                giveup=lambda e: isinstance(e, TransportQueryError),
            )
        elif retry_execute is False:
            self.retry_execute = lambda e: e
        else:
            assert callable(retry_execute)
            self.retry_execute = retry_execute

        # Creating the _execute_with_retries and _connect_with_retries  methods
        # using the provided backoff decorators
        self._execute_with_retries = self.retry_execute(self._execute_once)
        self._connect_with_retries = self.retry_connect(self.transport.connect)

    async def _connection_loop(self):
        """Coroutine used for the connection task.

        - try to connect to the transport with retries
        - send a connected event when the connection has been made
        - then wait for a reconnect request to try to connect again
        """

        while True:
            # Connect to the transport with the retry decorator
            # By default it should keep retrying until it connect
            await self._connect_with_retries()

            # Once connected, set the connected event
            self._connected_event.set()
            self._connected_event.clear()

            # Then wait for the reconnect event
            self._reconnect_request_event.clear()
            await self._reconnect_request_event.wait()
            await self.transport.close()

    async def start_connecting_task(self):
        """Start the task responsible to restart the connection
        of the transport when requested by an event.
        """
        if self._connect_task:
            log.warning("connect task already started!")
        else:
            self._connect_task = asyncio.ensure_future(self._connection_loop())

            await self._connected_event.wait()

    async def stop_connecting_task(self):
        """Stop the connecting task."""
        if self._connect_task is not None:
            self._connect_task.cancel()
            self._connect_task = None

    async def connect(self):
        """Start the connect task and initialize the batch task loop if batching
        is enabled."""

        await self._batch_init()

        await self.start_connecting_task()

    async def close(self):
        """Stop the connect task and cleanup the batching task
        if batching is enabled."""
        await self._batch_cleanup()

        await self.stop_connecting_task()

        await self.transport.close()

    async def _execute_once(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        **kwargs: Any,
    ) -> ExecutionResult:
        """Same Coroutine as parent method _execute but requesting a
        reconnection if we receive a TransportConnectionFailed exception.
        """

        try:
            answer = await super()._execute(
                request,
                serialize_variables=serialize_variables,
                parse_result=parse_result,
                **kwargs,
            )
        except TransportConnectionFailed:
            self._reconnect_request_event.set()
            raise

        return answer

    async def _execute(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        **kwargs: Any,
    ) -> ExecutionResult:
        """Same Coroutine as parent, but with optional retries
        and requesting a reconnection if we receive a
        TransportConnectionFailed exception.
        """

        return await self._execute_with_retries(
            request,
            serialize_variables=serialize_variables,
            parse_result=parse_result,
            **kwargs,
        )

    async def _subscribe(
        self,
        request: GraphQLRequest,
        *,
        serialize_variables: Optional[bool] = None,
        parse_result: Optional[bool] = None,
        **kwargs: Any,
    ) -> AsyncGenerator[ExecutionResult, None]:
        """Same Async generator as parent method _subscribe but requesting a
        reconnection if we receive a TransportConnectionFailed exception.
        """

        inner_generator: AsyncGenerator[ExecutionResult, None] = super()._subscribe(
            request,
            serialize_variables=serialize_variables,
            parse_result=parse_result,
            **kwargs,
        )

        try:
            async for result in inner_generator:
                yield result

        except TransportConnectionFailed:
            self._reconnect_request_event.set()
            raise

        finally:
            await inner_generator.aclose()