src/websockets/asyncio/connection.py

from __future__ import annotations

import asyncio
import collections
import contextlib
import logging
import random
import struct
import sys
import traceback
import uuid
from collections.abc import AsyncIterable, AsyncIterator, Awaitable, Iterable, Mapping
from types import TracebackType
from typing import Any, Literal, cast, overload

from ..exceptions import (
    ConcurrencyError,
    ConnectionClosed,
    ConnectionClosedOK,
    ProtocolError,
)
from ..frames import DATA_OPCODES, BytesLike, CloseCode, Frame, Opcode
from ..http11 import Request, Response
from ..protocol import CLOSED, OPEN, Event, Protocol, State
from ..typing import Data, LoggerLike, Subprotocol
from .compatibility import (
    TimeoutError,
    aiter,
    anext,
    asyncio_timeout,
    asyncio_timeout_at,
)
from .messages import Assembler


__all__ = ["Connection"]


class Connection(asyncio.Protocol):
    """
    :mod:`asyncio` implementation of a WebSocket connection.

    :class:`Connection` provides APIs shared between WebSocket servers and
    clients.

    You shouldn't use it directly. Instead, use
    :class:`~websockets.asyncio.client.ClientConnection` or
    :class:`~websockets.asyncio.server.ServerConnection`.

    """

    def __init__(
        self,
        protocol: Protocol,
        *,
        ping_interval: float | None = 20,
        ping_timeout: float | None = 20,
        close_timeout: float | None = 10,
        max_queue: int | None | tuple[int | None, int | None] = 16,
        write_limit: int | tuple[int, int | None] = 2**15,
    ) -> None:
        self.protocol = protocol
        self.ping_interval = ping_interval
        self.ping_timeout = ping_timeout
        self.close_timeout = close_timeout
        self.max_queue: tuple[int | None, int | None]
        if isinstance(max_queue, int) or max_queue is None:
            self.max_queue = (max_queue, None)
        else:
            self.max_queue = max_queue
        if isinstance(write_limit, int):
            write_limit = (write_limit, None)
        self.write_limit = write_limit

        # Inject reference to this instance in the protocol's logger.
        self.protocol.logger = logging.LoggerAdapter(
            self.protocol.logger,
            {"websocket": self},
        )

        # Copy attributes from the protocol for convenience.
        self.id: uuid.UUID = self.protocol.id
        """Unique identifier of the connection. Useful in logs."""
        self.logger: LoggerLike = self.protocol.logger
        """Logger for this connection."""
        self.debug = self.protocol.debug

        # HTTP handshake request and response.
        self.request: Request | None = None
        """Opening handshake request."""
        self.response: Response | None = None
        """Opening handshake response."""

        # Event loop running this connection.
        self.loop = asyncio.get_running_loop()

        # Assembler turning frames into messages and serializing reads.
        self.recv_messages: Assembler  # initialized in connection_made

        # Deadline for the closing handshake.
        self.close_deadline: float | None = None

        # Protect sending fragmented messages.
        self.fragmented_send_waiter: asyncio.Future[None] | None = None

        # Mapping of ping IDs to pong waiters, in chronological order.
        self.pong_waiters: dict[bytes, tuple[asyncio.Future[float], float]] = {}

        self.latency: float = 0
        """
        Latency of the connection, in seconds.

        Latency is defined as the round-trip time of the connection. It is
        measured by sending a Ping frame and waiting for a matching Pong frame.
        Before the first measurement, :attr:`latency` is ``0``.

        By default, websockets enables a :ref:`keepalive <keepalive>` mechanism
        that sends Ping frames automatically at regular intervals. You can also
        send Ping frames and measure latency with :meth:`ping`.
        """

        # Task that sends keepalive pings. None when ping_interval is None.
        self.keepalive_task: asyncio.Task[None] | None = None

        # Exception raised while reading from the connection, to be chained to
        # ConnectionClosed in order to show why the TCP connection dropped.
        self.recv_exc: BaseException | None = None

        # Completed when the TCP connection is closed and the WebSocket
        # connection state becomes CLOSED.
        self.connection_lost_waiter: asyncio.Future[None] = self.loop.create_future()

        # Adapted from asyncio.FlowControlMixin
        self.paused: bool = False
        self.drain_waiters: collections.deque[asyncio.Future[None]] = (
            collections.deque()
        )

    # Public attributes

    @property
    def local_address(self) -> Any:
        """
        Local address of the connection.

        For IPv4 connections, this is a ``(host, port)`` tuple.

        The format of the address depends on the address family.
        See :meth:`~socket.socket.getsockname`.

        """
        return self.transport.get_extra_info("sockname")

    @property
    def remote_address(self) -> Any:
        """
        Remote address of the connection.

        For IPv4 connections, this is a ``(host, port)`` tuple.

        The format of the address depends on the address family.
        See :meth:`~socket.socket.getpeername`.

        """
        return self.transport.get_extra_info("peername")

    @property
    def state(self) -> State:
        """
        State of the WebSocket connection, defined in :rfc:`6455`.

        This attribute is provided for completeness. Typical applications
        shouldn't check its value. Instead, they should call :meth:`~recv` or
        :meth:`send` and handle :exc:`~websockets.exceptions.ConnectionClosed`
        exceptions.

        """
        return self.protocol.state

    @property
    def subprotocol(self) -> Subprotocol | None:
        """
        Subprotocol negotiated during the opening handshake.

        :obj:`None` if no subprotocol was negotiated.

        """
        return self.protocol.subprotocol

    @property
    def close_code(self) -> int | None:
        """
        State of the WebSocket connection, defined in :rfc:`6455`.

        This attribute is provided for completeness. Typical applications
        shouldn't check its value. Instead, they should inspect attributes
        of :exc:`~websockets.exceptions.ConnectionClosed` exceptions.

        """
        return self.protocol.close_code

    @property
    def close_reason(self) -> str | None:
        """
        State of the WebSocket connection, defined in :rfc:`6455`.

        This attribute is provided for completeness. Typical applications
        shouldn't check its value. Instead, they should inspect attributes
        of :exc:`~websockets.exceptions.ConnectionClosed` exceptions.

        """
        return self.protocol.close_reason

    # Public methods

    async def __aenter__(self) -> Connection:
        return self

    async def __aexit__(
        self,
        exc_type: type[BaseException] | None,
        exc_value: BaseException | None,
        traceback: TracebackType | None,
    ) -> None:
        if exc_type is None:
            await self.close()
        else:
            await self.close(CloseCode.INTERNAL_ERROR)

    async def __aiter__(self) -> AsyncIterator[Data]:
        """
        Iterate on incoming messages.

        The iterator calls :meth:`recv` and yields messages asynchronously in an
        infinite loop.

        It exits when the connection is closed normally. It raises a
        :exc:`~websockets.exceptions.ConnectionClosedError` exception after a
        protocol error or a network failure.

        """
        try:
            while True:
                yield await self.recv()
        except ConnectionClosedOK:
            return

    @overload
    async def recv(self, decode: Literal[True]) -> str: ...

    @overload
    async def recv(self, decode: Literal[False]) -> bytes: ...

    @overload
    async def recv(self, decode: bool | None = None) -> Data: ...

    async def recv(self, decode: bool | None = None) -> Data:
        """
        Receive the next message.

        When the connection is closed, :meth:`recv` raises
        :exc:`~websockets.exceptions.ConnectionClosed`. Specifically, it raises
        :exc:`~websockets.exceptions.ConnectionClosedOK` after a normal closure
        and :exc:`~websockets.exceptions.ConnectionClosedError` after a protocol
        error or a network failure. This is how you detect the end of the
        message stream.

        Canceling :meth:`recv` is safe. There's no risk of losing data. The next
        invocation of :meth:`recv` will return the next message.

        This makes it possible to enforce a timeout by wrapping :meth:`recv` in
        :func:`~asyncio.timeout` or :func:`~asyncio.wait_for`.

        When the message is fragmented, :meth:`recv` waits until all fragments
        are received, reassembles them, and returns the whole message.

        Args:
            decode: Set this flag to override the default behavior of returning
                :class:`str` or :class:`bytes`. See below for details.

        Returns:
            A string (:class:`str`) for a Text_ frame or a bytestring
            (:class:`bytes`) for a Binary_ frame.

            .. _Text: https://datatracker.ietf.org/doc/html/rfc6455#section-5.6
            .. _Binary: https://datatracker.ietf.org/doc/html/rfc6455#section-5.6

            You may override this behavior with the ``decode`` argument:

            * Set ``decode=False`` to disable UTF-8 decoding of Text_ frames and
              return a bytestring (:class:`bytes`). This improves performance
              when decoding isn't needed, for example if the message contains
              JSON and you're using a JSON library that expects a bytestring.
            * Set ``decode=True`` to force UTF-8 decoding of Binary_ frames
              and return a string (:class:`str`). This may be useful for
              servers that send binary frames instead of text frames.

        Raises:
            ConnectionClosed: When the connection is closed.
            ConcurrencyError: If two coroutines call :meth:`recv` or
                :meth:`recv_streaming` concurrently.

        """
        try:
            return await self.recv_messages.get(decode)
        except EOFError:
            pass
            # fallthrough
        except ConcurrencyError:
            raise ConcurrencyError(
                "cannot call recv while another coroutine "
                "is already running recv or recv_streaming"
            ) from None
        except UnicodeDecodeError as exc:
            async with self.send_context():
                self.protocol.fail(
                    CloseCode.INVALID_DATA,
                    f"{exc.reason} at position {exc.start}",
                )
            # fallthrough

        # Wait for the protocol state to be CLOSED before accessing close_exc.
        await asyncio.shield(self.connection_lost_waiter)
        raise self.protocol.close_exc from self.recv_exc

    @overload
    def recv_streaming(self, decode: Literal[True]) -> AsyncIterator[str]: ...

    @overload
    def recv_streaming(self, decode: Literal[False]) -> AsyncIterator[bytes]: ...

    @overload
    def recv_streaming(self, decode: bool | None = None) -> AsyncIterator[Data]: ...

    async def recv_streaming(self, decode: bool | None = None) -> AsyncIterator[Data]:
        """
        Receive the next message frame by frame.

        This method is designed for receiving fragmented messages. It returns an
        asynchronous iterator that yields each fragment as it is received. This
        iterator must be fully consumed. Else, future calls to :meth:`recv` or
        :meth:`recv_streaming` will raise
        :exc:`~websockets.exceptions.ConcurrencyError`, making the connection
        unusable.

        :meth:`recv_streaming` raises the same exceptions as :meth:`recv`.

        Canceling :meth:`recv_streaming` before receiving the first frame is
        safe. Canceling it after receiving one or more frames leaves the
        iterator in a partially consumed state, making the connection unusable.
        Instead, you should close the connection with :meth:`close`.

        Args:
            decode: Set this flag to override the default behavior of returning
                :class:`str` or :class:`bytes`. See below for details.

        Returns:
            An iterator of strings (:class:`str`) for a Text_ frame or
            bytestrings (:class:`bytes`) for a Binary_ frame.

            .. _Text: https://datatracker.ietf.org/doc/html/rfc6455#section-5.6
            .. _Binary: https://datatracker.ietf.org/doc/html/rfc6455#section-5.6

            You may override this behavior with the ``decode`` argument:

            * Set ``decode=False`` to disable UTF-8 decoding of Text_ frames
              and return bytestrings (:class:`bytes`). This may be useful to
              optimize performance when decoding isn't needed.
            * Set ``decode=True`` to force UTF-8 decoding of Binary_ frames
              and return strings (:class:`str`). This is useful for servers
              that send binary frames instead of text frames.

        Raises:
            ConnectionClosed: When the connection is closed.
            ConcurrencyError: If two coroutines call :meth:`recv` or
                :meth:`recv_streaming` concurrently.

        """
        try:
            async for frame in self.recv_messages.get_iter(decode):
                yield frame
            return
        except EOFError:
            pass
            # fallthrough
        except ConcurrencyError:
            raise ConcurrencyError(
                "cannot call recv_streaming while another coroutine "
                "is already running recv or recv_streaming"
            ) from None
        except UnicodeDecodeError as exc:
            async with self.send_context():
                self.protocol.fail(
                    CloseCode.INVALID_DATA,
                    f"{exc.reason} at position {exc.start}",
                )
            # fallthrough

        # Wait for the protocol state to be CLOSED before accessing close_exc.
        await asyncio.shield(self.connection_lost_waiter)
        raise self.protocol.close_exc from self.recv_exc

    async def send(
        self,
        message: Data | Iterable[Data] | AsyncIterable[Data],
        text: bool | None = None,
    ) -> None:
        """
        Send a message.

        A string (:class:`str`) is sent as a Text_ frame. A bytestring or
        bytes-like object (:class:`bytes`, :class:`bytearray`, or
        :class:`memoryview`) is sent as a Binary_ frame.

        .. _Text: https://datatracker.ietf.org/doc/html/rfc6455#section-5.6
        .. _Binary: https://datatracker.ietf.org/doc/html/rfc6455#section-5.6

        You may override this behavior with the ``text`` argument:

        * Set ``text=True`` to send a bytestring or bytes-like object
          (:class:`bytes`, :class:`bytearray`, or :class:`memoryview`) as a
          Text_ frame. This improves performance when the message is already
          UTF-8 encoded, for example if the message contains JSON and you're
          using a JSON library that produces a bytestring.
        * Set ``text=False`` to send a string (:class:`str`) in a Binary_
          frame. This may be useful for servers that expect binary frames
          instead of text frames.

        :meth:`send` also accepts an iterable or an asynchronous iterable of
        strings, bytestrings, or bytes-like objects to enable fragmentation_.
        Each item is treated as a message fragment and sent in its own frame.
        All items must be of the same type, or else :meth:`send` will raise a
        :exc:`TypeError` and the connection will be closed.

        .. _fragmentation: https://datatracker.ietf.org/doc/html/rfc6455#section-5.4

        :meth:`send` rejects dict-like objects because this is often an error.
        (If you really want to send the keys of a dict-like object as fragments,
        call its :meth:`~dict.keys` method and pass the result to :meth:`send`.)

        Canceling :meth:`send` is discouraged. Instead, you should close the
        connection with :meth:`close`. Indeed, there are only two situations
        where :meth:`send` may yield control to the event loop and then get
        canceled; in both cases, :meth:`close` has the same effect and is
        more clear:

        1. The write buffer is full. If you don't want to wait until enough
           data is sent, your only alternative is to close the connection.
           :meth:`close` will likely time out then abort the TCP connection.
        2. ``message`` is an asynchronous iterator that yields control.
           Stopping in the middle of a fragmented message will cause a
           protocol error and the connection will be closed.

        When the connection is closed, :meth:`send` raises
        :exc:`~websockets.exceptions.ConnectionClosed`. Specifically, it
        raises :exc:`~websockets.exceptions.ConnectionClosedOK` after a normal
        connection closure and
        :exc:`~websockets.exceptions.ConnectionClosedError` after a protocol
        error or a network failure.

        Args:
            message: Message to send.

        Raises:
            ConnectionClosed: When the connection is closed.
            TypeError: If ``message`` doesn't have a supported type.

        """
        # While sending a fragmented message, prevent sending other messages
        # until all fragments are sent.
        while self.fragmented_send_waiter is not None:
            await asyncio.shield(self.fragmented_send_waiter)

        # Unfragmented message -- this case must be handled first because
        # strings and bytes-like objects are iterable.

        if isinstance(message, str):
            async with self.send_context():
                if text is False:
                    self.protocol.send_binary(message.encode())
                else:
                    self.protocol.send_text(message.encode())

        elif isinstance(message, BytesLike):
            async with self.send_context():
                if text is True:
                    self.protocol.send_text(message)
                else:
                    self.protocol.send_binary(message)

        # Catch a common mistake -- passing a dict to send().

        elif isinstance(message, Mapping):
            raise TypeError("data is a dict-like object")

        # Fragmented message -- regular iterator.

        elif isinstance(message, Iterable):
            chunks = iter(message)
            try:
                chunk = next(chunks)
            except StopIteration:
                return

            assert self.fragmented_send_waiter is None
            self.fragmented_send_waiter = self.loop.create_future()
            try:
                # First fragment.
                if isinstance(chunk, str):
                    async with self.send_context():
                        if text is False:
                            self.protocol.send_binary(chunk.encode(), fin=False)
                        else:
                            self.protocol.send_text(chunk.encode(), fin=False)
                    encode = True
                elif isinstance(chunk, BytesLike):
                    async with self.send_context():
                        if text is True:
                            self.protocol.send_text(chunk, fin=False)
                        else:
                            self.protocol.send_binary(chunk, fin=False)
                    encode = False
                else:
                    raise TypeError("iterable must contain bytes or str")

                # Other fragments
                for chunk in chunks:
                    if isinstance(chunk, str) and encode:
                        async with self.send_context():
                            self.protocol.send_continuation(chunk.encode(), fin=False)
                    elif isinstance(chunk, BytesLike) and not encode:
                        async with self.send_context():
                            self.protocol.send_continuation(chunk, fin=False)
                    else:
                        raise TypeError("iterable must contain uniform types")

                # Final fragment.
                async with self.send_context():
                    self.protocol.send_continuation(b"", fin=True)

            except Exception:
                # We're half-way through a fragmented message and we can't
                # complete it. This makes the connection unusable.
                async with self.send_context():
                    self.protocol.fail(
                        CloseCode.INTERNAL_ERROR,
                        "error in fragmented message",
                    )
                raise

            finally:
                self.fragmented_send_waiter.set_result(None)
                self.fragmented_send_waiter = None

        # Fragmented message -- async iterator.

        elif isinstance(message, AsyncIterable):
            achunks = aiter(message)
            try:
                chunk = await anext(achunks)
            except StopAsyncIteration:
                return

            assert self.fragmented_send_waiter is None
            self.fragmented_send_waiter = self.loop.create_future()
            try:
                # First fragment.
                if isinstance(chunk, str):
                    if text is False:
                        async with self.send_context():
                            self.protocol.send_binary(chunk.encode(), fin=False)
                    else:
                        async with self.send_context():
                            self.protocol.send_text(chunk.encode(), fin=False)
                    encode = True
                elif isinstance(chunk, BytesLike):
                    if text is True:
                        async with self.send_context():
                            self.protocol.send_text(chunk, fin=False)
                    else:
                        async with self.send_context():
                            self.protocol.send_binary(chunk, fin=False)
                    encode = False
                else:
                    raise TypeError("async iterable must contain bytes or str")

                # Other fragments
                async for chunk in achunks:
                    if isinstance(chunk, str) and encode:
                        async with self.send_context():
                            self.protocol.send_continuation(chunk.encode(), fin=False)
                    elif isinstance(chunk, BytesLike) and not encode:
                        async with self.send_context():
                            self.protocol.send_continuation(chunk, fin=False)
                    else:
                        raise TypeError("async iterable must contain uniform types")

                # Final fragment.
                async with self.send_context():
                    self.protocol.send_continuation(b"", fin=True)

            except Exception:
                # We're half-way through a fragmented message and we can't
                # complete it. This makes the connection unusable.
                async with self.send_context():
                    self.protocol.fail(
                        CloseCode.INTERNAL_ERROR,
                        "error in fragmented message",
                    )
                raise

            finally:
                self.fragmented_send_waiter.set_result(None)
                self.fragmented_send_waiter = None

        else:
            raise TypeError("data must be str, bytes, iterable, or async iterable")

    async def close(self, code: int = 1000, reason: str = "") -> None:
        """
        Perform the closing handshake.

        :meth:`close` waits for the other end to complete the handshake and
        for the TCP connection to terminate.

        :meth:`close` is idempotent: it doesn't do anything once the
        connection is closed.

        Args:
            code: WebSocket close code.
            reason: WebSocket close reason.

        """
        try:
            # The context manager takes care of waiting for the TCP connection
            # to terminate after calling a method that sends a close frame.
            async with self.send_context():
                if self.fragmented_send_waiter is not None:
                    self.protocol.fail(
                        CloseCode.INTERNAL_ERROR,
                        "close during fragmented message",
                    )
                else:
                    self.protocol.send_close(code, reason)
        except ConnectionClosed:
            # Ignore ConnectionClosed exceptions raised from send_context().
            # They mean that the connection is closed, which was the goal.
            pass

    async def wait_closed(self) -> None:
        """
        Wait until the connection is closed.

        :meth:`wait_closed` waits for the closing handshake to complete and for
        the TCP connection to terminate.

        """
        await asyncio.shield(self.connection_lost_waiter)

    async def ping(self, data: Data | None = None) -> Awaitable[float]:
        """
        Send a Ping_.

        .. _Ping: https://datatracker.ietf.org/doc/html/rfc6455#section-5.5.2

        A ping may serve as a keepalive or as a check that the remote endpoint
        received all messages up to this point

        Args:
            data: Payload of the ping. A :class:`str` will be encoded to UTF-8.
                If ``data`` is :obj:`None`, the payload is four random bytes.

        Returns:
            A future that will be completed when the corresponding pong is
            received. You can ignore it if you don't intend to wait. The result
            of the future is the latency of the connection in seconds.

            ::

                pong_waiter = await ws.ping()
                # only if you want to wait for the corresponding pong
                latency = await pong_waiter

        Raises:
            ConnectionClosed: When the connection is closed.
            ConcurrencyError: If another ping was sent with the same data and
                the corresponding pong wasn't received yet.

        """
        if isinstance(data, BytesLike):
            data = bytes(data)
        elif isinstance(data, str):
            data = data.encode()
        elif data is not None:
            raise TypeError("data must be str or bytes-like")

        async with self.send_context():
            # Protect against duplicates if a payload is explicitly set.
            if data in self.pong_waiters:
                raise ConcurrencyError("already waiting for a pong with the same data")

            # Generate a unique random payload otherwise.
            while data is None or data in self.pong_waiters:
                data = struct.pack("!I", random.getrandbits(32))

            pong_waiter = self.loop.create_future()
            # The event loop's default clock is time.monotonic(). Its resolution
            # is a bit low on Windows (~16ms). This is improved in Python 3.13.
            self.pong_waiters[data] = (pong_waiter, self.loop.time())
            self.protocol.send_ping(data)
            return pong_waiter

    async def pong(self, data: Data = b"") -> None:
        """
        Send a Pong_.

        .. _Pong: https://datatracker.ietf.org/doc/html/rfc6455#section-5.5.3

        An unsolicited pong may serve as a unidirectional heartbeat.

        Args:
            data: Payload of the pong. A :class:`str` will be encoded to UTF-8.

        Raises:
            ConnectionClosed: When the connection is closed.

        """
        if isinstance(data, BytesLike):
            data = bytes(data)
        elif isinstance(data, str):
            data = data.encode()
        else:
            raise TypeError("data must be str or bytes-like")

        async with self.send_context():
            self.protocol.send_pong(data)

    # Private methods

    def process_event(self, event: Event) -> None:
        """
        Process one incoming event.

        This method is overridden in subclasses to handle the handshake.

        """
        assert isinstance(event, Frame)
        if event.opcode in DATA_OPCODES:
            self.recv_messages.put(event)

        if event.opcode is Opcode.PONG:
            self.acknowledge_pings(bytes(event.data))

    def acknowledge_pings(self, data: bytes) -> None:
        """
        Acknowledge pings when receiving a pong.

        """
        # Ignore unsolicited pong.
        if data not in self.pong_waiters:
            return

        pong_timestamp = self.loop.time()

        # Sending a pong for only the most recent ping is legal.
        # Acknowledge all previous pings too in that case.
        ping_id = None
        ping_ids = []
        for ping_id, (pong_waiter, ping_timestamp) in self.pong_waiters.items():
            ping_ids.append(ping_id)
            latency = pong_timestamp - ping_timestamp
            if not pong_waiter.done():
                pong_waiter.set_result(latency)
            if ping_id == data:
                self.latency = latency
                break
        else:
            raise AssertionError("solicited pong not found in pings")

        # Remove acknowledged pings from self.pong_waiters.
        for ping_id in ping_ids:
            del self.pong_waiters[ping_id]

    def abort_pings(self) -> None:
        """
        Raise ConnectionClosed in pending pings.

        They'll never receive a pong once the connection is closed.

        """
        assert self.protocol.state is CLOSED
        exc = self.protocol.close_exc

        for pong_waiter, _ping_timestamp in self.pong_waiters.values():
            if not pong_waiter.done():
                pong_waiter.set_exception(exc)
            # If the exception is never retrieved, it will be logged when ping
            # is garbage-collected. This is confusing for users.
            # Given that ping is done (with an exception), canceling it does
            # nothing, but it prevents logging the exception.
            pong_waiter.cancel()

        self.pong_waiters.clear()

    async def keepalive(self) -> None:
        """
        Send a Ping frame and wait for a Pong frame at regular intervals.

        """
        assert self.ping_interval is not None
        latency = 0.0
        try:
            while True:
                # If self.ping_timeout > latency > self.ping_interval,
                # pings will be sent immediately after receiving pongs.
                # The period will be longer than self.ping_interval.
                await asyncio.sleep(self.ping_interval - latency)

                # This cannot raise ConnectionClosed when the connection is
                # closing because ping(), via send_context(), waits for the
                # connection to be closed before raising ConnectionClosed.
                # However, connection_lost() cancels keepalive_task before
                # it gets a chance to resume excuting.
                pong_waiter = await self.ping()
                if self.debug:
                    self.logger.debug("% sent keepalive ping")

                if self.ping_timeout is not None:
                    try:
                        async with asyncio_timeout(self.ping_timeout):
                            # connection_lost cancels keepalive immediately
                            # after setting a ConnectionClosed exception on
                            # pong_waiter. A CancelledError is raised here,
                            # not a ConnectionClosed exception.
                            latency = await pong_waiter
                        self.logger.debug("% received keepalive pong")
                    except asyncio.TimeoutError:
                        if self.debug:
                            self.logger.debug("- timed out waiting for keepalive pong")
                        async with self.send_context():
                            self.protocol.fail(
                                CloseCode.INTERNAL_ERROR,
                                "keepalive ping timeout",
                            )
                        raise AssertionError(
                            "send_context() should wait for connection_lost(), "
                            "which cancels keepalive()"
                        )
        except Exception:
            self.logger.error("keepalive ping failed", exc_info=True)

    def start_keepalive(self) -> None:
        """
        Run :meth:`keepalive` in a task, unless keepalive is disabled.

        """
        if self.ping_interval is not None:
            self.keepalive_task = self.loop.create_task(self.keepalive())

    @contextlib.asynccontextmanager
    async def send_context(
        self,
        *,
        expected_state: State = OPEN,  # CONNECTING during the opening handshake
    ) -> AsyncIterator[None]:
        """
        Create a context for writing to the connection from user code.

        On entry, :meth:`send_context` checks that the connection is open; on
        exit, it writes outgoing data to the socket::

            async with self.send_context():
                self.protocol.send_text(message.encode())

        When the connection isn't open on entry, when the connection is expected
        to close on exit, or when an unexpected error happens, terminating the
        connection, :meth:`send_context` waits until the connection is closed
        then raises :exc:`~websockets.exceptions.ConnectionClosed`.

        """
        # Should we wait until the connection is closed?
        wait_for_close = False
        # Should we close the transport and raise ConnectionClosed?
        raise_close_exc = False
        # What exception should we chain ConnectionClosed to?
        original_exc: BaseException | None = None

        if self.protocol.state is expected_state:
            # Let the caller interact with the protocol.
            try:
                yield
            except (ProtocolError, ConcurrencyError):
                # The protocol state wasn't changed. Exit immediately.
                raise
            except Exception as exc:
                self.logger.error("unexpected internal error", exc_info=True)
                # This branch should never run. It's a safety net in case of
                # bugs. Since we don't know what happened, we will close the
                # connection and raise the exception to the caller.
                wait_for_close = False
                raise_close_exc = True
                original_exc = exc
            else:
                # Check if the connection is expected to close soon.
                if self.protocol.close_expected():
                    wait_for_close = True
                    # If the connection is expected to close soon, set the
                    # close deadline based on the close timeout.
                    # Since we tested earlier that protocol.state was OPEN
                    # (or CONNECTING), self.close_deadline is still None.
                    if self.close_timeout is not None:
                        assert self.close_deadline is None
                        self.close_deadline = self.loop.time() + self.close_timeout
                # Write outgoing data to the socket and enforce flow control.
                try:
                    self.send_data()
                    await self.drain()
                except Exception as exc:
                    if self.debug:
                        self.logger.debug("! error while sending data", exc_info=True)
                    # While the only expected exception here is OSError,
                    # other exceptions would be treated identically.
                    wait_for_close = False
                    raise_close_exc = True
                    original_exc = exc

        else:  # self.protocol.state is not expected_state
            # Minor layering violation: we assume that the connection
            # will be closing soon if it isn't in the expected state.
            wait_for_close = True
            # Calculate close_deadline if it wasn't set yet.
            if self.close_timeout is not None:
                if self.close_deadline is None:
                    self.close_deadline = self.loop.time() + self.close_timeout
            raise_close_exc = True

        # If the connection is expected to close soon and the close timeout
        # elapses, close the socket to terminate the connection.
        if wait_for_close:
            try:
                async with asyncio_timeout_at(self.close_deadline):
                    await asyncio.shield(self.connection_lost_waiter)
            except TimeoutError:
                # There's no risk to overwrite another error because
                # original_exc is never set when wait_for_close is True.
                assert original_exc is None
                original_exc = TimeoutError("timed out while closing connection")
                # Set recv_exc before closing the transport in order to get
                # proper exception reporting.
                raise_close_exc = True
                self.set_recv_exc(original_exc)

        # If an error occurred, close the transport to terminate the connection and
        # raise an exception.
        if raise_close_exc:
            self.transport.abort()
            # Wait for the protocol state to be CLOSED before accessing close_exc.
            await asyncio.shield(self.connection_lost_waiter)
            raise self.protocol.close_exc from original_exc

    def send_data(self) -> None:
        """
        Send outgoing data.

        Raises:
            OSError: When a socket operations fails.

        """
        for data in self.protocol.data_to_send():
            if data:
                self.transport.write(data)
            else:
                # Half-close the TCP connection when possible i.e. no TLS.
                if self.transport.can_write_eof():
                    if self.debug:
                        self.logger.debug("x half-closing TCP connection")
                    # write_eof() doesn't document which exceptions it raises.
                    # OSError is plausible. uvloop can raise RuntimeError here.
                    try:
                        self.transport.write_eof()
                    except (OSError, RuntimeError):  # pragma: no cover
                        pass
                # Else, close the TCP connection.
                else:  # pragma: no cover
                    if self.debug:
                        self.logger.debug("x closing TCP connection")
                    self.transport.close()

    def set_recv_exc(self, exc: BaseException | None) -> None:
        """
        Set recv_exc, if not set yet.

        """
        if self.recv_exc is None:
            self.recv_exc = exc

    # asyncio.Protocol methods

    # Connection callbacks

    def connection_made(self, transport: asyncio.BaseTransport) -> None:
        transport = cast(asyncio.Transport, transport)
        self.recv_messages = Assembler(
            *self.max_queue,
            pause=transport.pause_reading,
            resume=transport.resume_reading,
        )
        transport.set_write_buffer_limits(*self.write_limit)
        self.transport = transport

    def connection_lost(self, exc: Exception | None) -> None:
        # Calling protocol.receive_eof() is safe because it's idempotent.
        # This guarantees that the protocol state becomes CLOSED.
        self.protocol.receive_eof()
        assert self.protocol.state is CLOSED

        self.set_recv_exc(exc)

        # Abort recv() and pending pings with a ConnectionClosed exception.
        self.recv_messages.close()
        self.abort_pings()

        if self.keepalive_task is not None:
            self.keepalive_task.cancel()

        # If self.connection_lost_waiter isn't pending, that's a bug, because:
        # - it's set only here in connection_lost() which is called only once;
        # - it must never be canceled.
        self.connection_lost_waiter.set_result(None)

        # Adapted from asyncio.streams.FlowControlMixin
        if self.paused:  # pragma: no cover
            self.paused = False
            for waiter in self.drain_waiters:
                if not waiter.done():
                    if exc is None:
                        waiter.set_result(None)
                    else:
                        waiter.set_exception(exc)

    # Flow control callbacks

    def pause_writing(self) -> None:  # pragma: no cover
        # Adapted from asyncio.streams.FlowControlMixin
        assert not self.paused
        self.paused = True

    def resume_writing(self) -> None:  # pragma: no cover
        # Adapted from asyncio.streams.FlowControlMixin
        assert self.paused
        self.paused = False
        for waiter in self.drain_waiters:
            if not waiter.done():
                waiter.set_result(None)

    async def drain(self) -> None:  # pragma: no cover
        # We don't check if the connection is closed because we call drain()
        # immediately after write() and write() would fail in that case.

        # Adapted from asyncio.streams.StreamWriter
        # Yield to the event loop so that connection_lost() may be called.
        if self.transport.is_closing():
            await asyncio.sleep(0)

        # Adapted from asyncio.streams.FlowControlMixin
        if self.paused:
            waiter = self.loop.create_future()
            self.drain_waiters.append(waiter)
            try:
                await waiter
            finally:
                self.drain_waiters.remove(waiter)

    # Streaming protocol callbacks

    def data_received(self, data: bytes) -> None:
        # Feed incoming data to the protocol.
        self.protocol.receive_data(data)

        # This isn't expected to raise an exception.
        events = self.protocol.events_received()

        # Write outgoing data to the transport.
        try:
            self.send_data()
        except Exception as exc:
            if self.debug:
                self.logger.debug("! error while sending data", exc_info=True)
            self.set_recv_exc(exc)

        if self.protocol.close_expected():
            # If the connection is expected to close soon, set the
            # close deadline based on the close timeout.
            if self.close_timeout is not None:
                if self.close_deadline is None:
                    self.close_deadline = self.loop.time() + self.close_timeout

        for event in events:
            # This isn't expected to raise an exception.
            self.process_event(event)

    def eof_received(self) -> None:
        # Feed the end of the data stream to the connection.
        self.protocol.receive_eof()

        # This isn't expected to raise an exception.
        events = self.protocol.events_received()

        # There is no error handling because send_data() can only write
        # the end of the data stream here and it shouldn't raise errors.
        self.send_data()

        # This code path is triggered when receiving an HTTP response
        # without a Content-Length header. This is the only case where
        # reading until EOF generates an event; all other events have
        # a known length. Ignore for coverage measurement because tests
        # are in test_client.py rather than test_connection.py.
        for event in events:  # pragma: no cover
            # This isn't expected to raise an exception.
            self.process_event(event)

        # The WebSocket protocol has its own closing handshake: endpoints close
        # the TCP or TLS connection after sending and receiving a close frame.
        # As a consequence, they never need to write after receiving EOF, so
        # there's no reason to keep the transport open by returning True.
        # Besides, that doesn't work on TLS connections.


# broadcast() is defined in the connection module even though it's primarily
# used by servers and documented in the server module because it works with
# client connections too and because it's easier to test together with the
# Connection class.


def broadcast(
    connections: Iterable[Connection],
    message: Data,
    raise_exceptions: bool = False,
) -> None:
    """
    Broadcast a message to several WebSocket connections.

    A string (:class:`str`) is sent as a Text_ frame. A bytestring or bytes-like
    object (:class:`bytes`, :class:`bytearray`, or :class:`memoryview`) is sent
    as a Binary_ frame.

    .. _Text: https://datatracker.ietf.org/doc/html/rfc6455#section-5.6
    .. _Binary: https://datatracker.ietf.org/doc/html/rfc6455#section-5.6

    :func:`broadcast` pushes the message synchronously to all connections even
    if their write buffers are overflowing. There's no backpressure.

    If you broadcast messages faster than a connection can handle them, messages
    will pile up in its write buffer until the connection times out. Keep
    ``ping_interval`` and ``ping_timeout`` low to prevent excessive memory usage
    from slow connections.

    Unlike :meth:`~websockets.asyncio.connection.Connection.send`,
    :func:`broadcast` doesn't support sending fragmented messages. Indeed,
    fragmentation is useful for sending large messages without buffering them in
    memory, while :func:`broadcast` buffers one copy per connection as fast as
    possible.

    :func:`broadcast` skips connections that aren't open in order to avoid
    errors on connections where the closing handshake is in progress.

    :func:`broadcast` ignores failures to write the message on some connections.
    It continues writing to other connections. On Python 3.11 and above, you may
    set ``raise_exceptions`` to :obj:`True` to record failures and raise all
    exceptions in a :pep:`654` :exc:`ExceptionGroup`.

    While :func:`broadcast` makes more sense for servers, it works identically
    with clients, if you have a use case for opening connections to many servers
    and broadcasting a message to them.

    Args:
        websockets: WebSocket connections to which the message will be sent.
        message: Message to send.
        raise_exceptions: Whether to raise an exception in case of failures.

    Raises:
        TypeError: If ``message`` doesn't have a supported type.

    """
    if isinstance(message, str):
        send_method = "send_text"
        message = message.encode()
    elif isinstance(message, BytesLike):
        send_method = "send_binary"
    else:
        raise TypeError("data must be str or bytes")

    if raise_exceptions:
        if sys.version_info[:2] < (3, 11):  # pragma: no cover
            raise ValueError("raise_exceptions requires at least Python 3.11")
        exceptions: list[Exception] = []

    for connection in connections:
        exception: Exception

        if connection.protocol.state is not OPEN:
            continue

        if connection.fragmented_send_waiter is not None:
            if raise_exceptions:
                exception = ConcurrencyError("sending a fragmented message")
                exceptions.append(exception)
            else:
                connection.logger.warning(
                    "skipped broadcast: sending a fragmented message",
                )
            continue

        try:
            # Call connection.protocol.send_text or send_binary.
            # Either way, message is already converted to bytes.
            getattr(connection.protocol, send_method)(message)
            connection.send_data()
        except Exception as write_exception:
            if raise_exceptions:
                exception = RuntimeError("failed to write message")
                exception.__cause__ = write_exception
                exceptions.append(exception)
            else:
                connection.logger.warning(
                    "skipped broadcast: failed to write message: %s",
                    traceback.format_exception_only(
                        # Remove first argument when dropping Python 3.9.
                        type(write_exception),
                        write_exception,
                    )[0].strip(),
                )

    if raise_exceptions and exceptions:
        raise ExceptionGroup("skipped broadcast", exceptions)


# Pretend that broadcast is actually defined in the server module.
broadcast.__module__ = "websockets.asyncio.server"