#
# Copyright 2009 Facebook
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.

"""An I/O event loop for non-blocking sockets.

In Tornado 6.0, `.IOLoop` is a wrapper around the `asyncio` event loop, with a
slightly different interface. The `.IOLoop` interface is now provided primarily
for backwards compatibility; new code should generally use the `asyncio` event
loop interface directly. The `IOLoop.current` class method provides the
`IOLoop` instance corresponding to the running `asyncio` event loop.

"""

import asyncio
import concurrent.futures
import datetime
import functools
import numbers
import os
import sys
import time
import math
import random
import warnings
from inspect import isawaitable

from tornado.concurrent import (
    Future,
    is_future,
    chain_future,
    future_set_exc_info,
    future_add_done_callback,
)
from tornado.log import app_log
from tornado.util import Configurable, TimeoutError, import_object

import typing
from typing import Union, Any, Type, Optional, Callable, TypeVar, Tuple, Awaitable

if typing.TYPE_CHECKING:
    from typing import Dict, List  # noqa: F401

    from typing_extensions import Protocol
else:
    Protocol = object


class _Selectable(Protocol):
    def fileno(self) -> int:
        pass

    def close(self) -> None:
        pass


_T = TypeVar("_T")
_S = TypeVar("_S", bound=_Selectable)


class IOLoop(Configurable):
    """An I/O event loop.

    As of Tornado 6.0, `IOLoop` is a wrapper around the `asyncio` event loop.

    Example usage for a simple TCP server:

    .. testcode::

        import asyncio
        import errno
        import functools
        import socket

        import tornado
        from tornado.iostream import IOStream

        async def handle_connection(connection, address):
            stream = IOStream(connection)
            message = await stream.read_until_close()
            print("message from client:", message.decode().strip())

        def connection_ready(sock, fd, events):
            while True:
                try:
                    connection, address = sock.accept()
                except BlockingIOError:
                    return
                connection.setblocking(0)
                io_loop = tornado.ioloop.IOLoop.current()
                io_loop.spawn_callback(handle_connection, connection, address)

        async def main():
            sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM, 0)
            sock.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
            sock.setblocking(0)
            sock.bind(("", 8888))
            sock.listen(128)

            io_loop = tornado.ioloop.IOLoop.current()
            callback = functools.partial(connection_ready, sock)
            io_loop.add_handler(sock.fileno(), callback, io_loop.READ)
            await asyncio.Event().wait()

        if __name__ == "__main__":
            asyncio.run(main())

    .. testoutput::
       :hide:

    Most applications should not attempt to construct an `IOLoop` directly,
    and instead initialize the `asyncio` event loop and use `IOLoop.current()`.
    In some cases, such as in test frameworks when initializing an `IOLoop`
    to be run in a secondary thread, it may be appropriate to construct
    an `IOLoop` with ``IOLoop(make_current=False)``.

    In general, an `IOLoop` cannot survive a fork or be shared across processes
    in any way. When multiple processes are being used, each process should
    create its own `IOLoop`, which also implies that any objects which depend on
    the `IOLoop` (such as `.AsyncHTTPClient`) must also be created in the child
    processes. As a guideline, anything that starts processes (including the
    `tornado.process` and `multiprocessing` modules) should do so as early as
    possible, ideally the first thing the application does after loading its
    configuration, and *before* any calls to `.IOLoop.start` or `asyncio.run`.

    .. versionchanged:: 4.2
       Added the ``make_current`` keyword argument to the `IOLoop`
       constructor.

    .. versionchanged:: 5.0

       Uses the `asyncio` event loop by default. The ``IOLoop.configure`` method
       cannot be used on Python 3 except to redundantly specify the `asyncio`
       event loop.

    .. versionchanged:: 6.3
       ``make_current=True`` is now the default when creating an IOLoop -
       previously the default was to make the event loop current if there wasn't
       already a current one.
    """

    # These constants were originally based on constants from the epoll module.
    NONE = 0
    READ = 0x001
    WRITE = 0x004
    ERROR = 0x018

    # In Python 3, _ioloop_for_asyncio maps from asyncio loops to IOLoops.
    _ioloop_for_asyncio = dict()  # type: Dict[asyncio.AbstractEventLoop, IOLoop]

    @classmethod
    def configure(
        cls, impl: "Union[None, str, Type[Configurable]]", **kwargs: Any
    ) -> None:
        from tornado.platform.asyncio import BaseAsyncIOLoop

        if isinstance(impl, str):
            impl = import_object(impl)
        if isinstance(impl, type) and not issubclass(impl, BaseAsyncIOLoop):
            raise RuntimeError("only AsyncIOLoop is allowed when asyncio is available")
        super(IOLoop, cls).configure(impl, **kwargs)

    @staticmethod
    def instance() -> "IOLoop":
        """Deprecated alias for `IOLoop.current()`.

        .. versionchanged:: 5.0

           Previously, this method returned a global singleton
           `IOLoop`, in contrast with the per-thread `IOLoop` returned
           by `current()`. In nearly all cases the two were the same
           (when they differed, it was generally used from non-Tornado
           threads to communicate back to the main thread's `IOLoop`).
           This distinction is not present in `asyncio`, so in order
           to facilitate integration with that package `instance()`
           was changed to be an alias to `current()`. Applications
           using the cross-thread communications aspect of
           `instance()` should instead set their own global variable
           to point to the `IOLoop` they want to use.

        .. deprecated:: 5.0
        """
        return IOLoop.current()

    def install(self) -> None:
        """Deprecated alias for `make_current()`.

        .. versionchanged:: 5.0

           Previously, this method would set this `IOLoop` as the
           global singleton used by `IOLoop.instance()`. Now that
           `instance()` is an alias for `current()`, `install()`
           is an alias for `make_current()`.

        .. deprecated:: 5.0
        """
        self.make_current()

    @staticmethod
    def clear_instance() -> None:
        """Deprecated alias for `clear_current()`.

        .. versionchanged:: 5.0

           Previously, this method would clear the `IOLoop` used as
           the global singleton by `IOLoop.instance()`. Now that
           `instance()` is an alias for `current()`,
           `clear_instance()` is an alias for `clear_current()`.

        .. deprecated:: 5.0

        """
        IOLoop.clear_current()

    @typing.overload
    @staticmethod
    def current() -> "IOLoop":
        pass

    @typing.overload
    @staticmethod
    def current(instance: bool = True) -> Optional["IOLoop"]:  # noqa: F811
        pass

    @staticmethod
    def current(instance: bool = True) -> Optional["IOLoop"]:  # noqa: F811
        """Returns the current thread's `IOLoop`.

        If an `IOLoop` is currently running or has been marked as
        current by `make_current`, returns that instance.  If there is
        no current `IOLoop` and ``instance`` is true, creates one.

        .. versionchanged:: 4.1
           Added ``instance`` argument to control the fallback to
           `IOLoop.instance()`.
        .. versionchanged:: 5.0
           On Python 3, control of the current `IOLoop` is delegated
           to `asyncio`, with this and other methods as pass-through accessors.
           The ``instance`` argument now controls whether an `IOLoop`
           is created automatically when there is none, instead of
           whether we fall back to `IOLoop.instance()` (which is now
           an alias for this method). ``instance=False`` is deprecated,
           since even if we do not create an `IOLoop`, this method
           may initialize the asyncio loop.

        .. deprecated:: 6.2
           It is deprecated to call ``IOLoop.current()`` when no `asyncio`
           event loop is running.
        """
        try:
            loop = asyncio.get_event_loop()
        except RuntimeError:
            if not instance:
                return None
            # Create a new asyncio event loop for this thread.
            loop = asyncio.new_event_loop()
            asyncio.set_event_loop(loop)

        try:
            return IOLoop._ioloop_for_asyncio[loop]
        except KeyError:
            if instance:
                from tornado.platform.asyncio import AsyncIOMainLoop

                current = AsyncIOMainLoop()  # type: Optional[IOLoop]
            else:
                current = None
        return current

    def make_current(self) -> None:
        """Makes this the `IOLoop` for the current thread.

        An `IOLoop` automatically becomes current for its thread
        when it is started, but it is sometimes useful to call
        `make_current` explicitly before starting the `IOLoop`,
        so that code run at startup time can find the right
        instance.

        .. versionchanged:: 4.1
           An `IOLoop` created while there is no current `IOLoop`
           will automatically become current.

        .. versionchanged:: 5.0
           This method also sets the current `asyncio` event loop.

        .. deprecated:: 6.2
           Setting and clearing the current event loop through Tornado is
           deprecated. Use ``asyncio.set_event_loop`` instead if you need this.
        """
        warnings.warn(
            "make_current is deprecated; start the event loop first",
            DeprecationWarning,
            stacklevel=2,
        )
        self._make_current()

    def _make_current(self) -> None:
        # The asyncio event loops override this method.
        raise NotImplementedError()

    @staticmethod
    def clear_current() -> None:
        """Clears the `IOLoop` for the current thread.

        Intended primarily for use by test frameworks in between tests.

        .. versionchanged:: 5.0
           This method also clears the current `asyncio` event loop.
        .. deprecated:: 6.2
        """
        warnings.warn(
            "clear_current is deprecated",
            DeprecationWarning,
            stacklevel=2,
        )
        IOLoop._clear_current()

    @staticmethod
    def _clear_current() -> None:
        old = IOLoop.current(instance=False)
        if old is not None:
            old._clear_current_hook()

    def _clear_current_hook(self) -> None:
        """Instance method called when an IOLoop ceases to be current.

        May be overridden by subclasses as a counterpart to make_current.
        """
        pass

    @classmethod
    def configurable_base(cls) -> Type[Configurable]:
        return IOLoop

    @classmethod
    def configurable_default(cls) -> Type[Configurable]:
        from tornado.platform.asyncio import AsyncIOLoop

        return AsyncIOLoop

    def initialize(self, make_current: bool = True) -> None:
        if make_current:
            self._make_current()

    def close(self, all_fds: bool = False) -> None:
        """Closes the `IOLoop`, freeing any resources used.

        If ``all_fds`` is true, all file descriptors registered on the
        IOLoop will be closed (not just the ones created by the
        `IOLoop` itself).

        Many applications will only use a single `IOLoop` that runs for the
        entire lifetime of the process.  In that case closing the `IOLoop`
        is not necessary since everything will be cleaned up when the
        process exits.  `IOLoop.close` is provided mainly for scenarios
        such as unit tests, which create and destroy a large number of
        ``IOLoops``.

        An `IOLoop` must be completely stopped before it can be closed.  This
        means that `IOLoop.stop()` must be called *and* `IOLoop.start()` must
        be allowed to return before attempting to call `IOLoop.close()`.
        Therefore the call to `close` will usually appear just after
        the call to `start` rather than near the call to `stop`.

        .. versionchanged:: 3.1
           If the `IOLoop` implementation supports non-integer objects
           for "file descriptors", those objects will have their
           ``close`` method when ``all_fds`` is true.
        """
        raise NotImplementedError()

    @typing.overload
    def add_handler(
        self, fd: int, handler: Callable[[int, int], None], events: int
    ) -> None:
        pass

    @typing.overload  # noqa: F811
    def add_handler(
        self, fd: _S, handler: Callable[[_S, int], None], events: int
    ) -> None:
        pass

    def add_handler(  # noqa: F811
        self, fd: Union[int, _Selectable], handler: Callable[..., None], events: int
    ) -> None:
        """Registers the given handler to receive the given events for ``fd``.

        The ``fd`` argument may either be an integer file descriptor or
        a file-like object with a ``fileno()`` and ``close()`` method.

        The ``events`` argument is a bitwise or of the constants
        ``IOLoop.READ``, ``IOLoop.WRITE``, and ``IOLoop.ERROR``.

        When an event occurs, ``handler(fd, events)`` will be run.

        .. versionchanged:: 4.0
           Added the ability to pass file-like objects in addition to
           raw file descriptors.
        """
        raise NotImplementedError()

    def update_handler(self, fd: Union[int, _Selectable], events: int) -> None:
        """Changes the events we listen for ``fd``.

        .. versionchanged:: 4.0
           Added the ability to pass file-like objects in addition to
           raw file descriptors.
        """
        raise NotImplementedError()

    def remove_handler(self, fd: Union[int, _Selectable]) -> None:
        """Stop listening for events on ``fd``.

        .. versionchanged:: 4.0
           Added the ability to pass file-like objects in addition to
           raw file descriptors.
        """
        raise NotImplementedError()

    def start(self) -> None:
        """Starts the I/O loop.

        The loop will run until one of the callbacks calls `stop()`, which
        will make the loop stop after the current event iteration completes.
        """
        raise NotImplementedError()

    def stop(self) -> None:
        """Stop the I/O loop.

        If the event loop is not currently running, the next call to `start()`
        will return immediately.

        Note that even after `stop` has been called, the `IOLoop` is not
        completely stopped until `IOLoop.start` has also returned.
        Some work that was scheduled before the call to `stop` may still
        be run before the `IOLoop` shuts down.
        """
        raise NotImplementedError()

    def run_sync(self, func: Callable, timeout: Optional[float] = None) -> Any:
        """Starts the `IOLoop`, runs the given function, and stops the loop.

        The function must return either an awaitable object or
        ``None``. If the function returns an awaitable object, the
        `IOLoop` will run until the awaitable is resolved (and
        `run_sync()` will return the awaitable's result). If it raises
        an exception, the `IOLoop` will stop and the exception will be
        re-raised to the caller.

        The keyword-only argument ``timeout`` may be used to set
        a maximum duration for the function.  If the timeout expires,
        a `asyncio.TimeoutError` is raised.

        This method is useful to allow asynchronous calls in a
        ``main()`` function::

            async def main():
                # do stuff...

            if __name__ == '__main__':
                IOLoop.current().run_sync(main)

        .. versionchanged:: 4.3
           Returning a non-``None``, non-awaitable value is now an error.

        .. versionchanged:: 5.0
           If a timeout occurs, the ``func`` coroutine will be cancelled.

        .. versionchanged:: 6.2
           ``tornado.util.TimeoutError`` is now an alias to ``asyncio.TimeoutError``.
        """
        future_cell = [None]  # type: List[Optional[Future]]

        def run() -> None:
            try:
                result = func()
                if result is not None:
                    from tornado.gen import convert_yielded

                    result = convert_yielded(result)
            except Exception:
                fut = Future()  # type: Future[Any]
                future_cell[0] = fut
                future_set_exc_info(fut, sys.exc_info())
            else:
                if is_future(result):
                    future_cell[0] = result
                else:
                    fut = Future()
                    future_cell[0] = fut
                    fut.set_result(result)
            assert future_cell[0] is not None
            self.add_future(future_cell[0], lambda future: self.stop())

        self.add_callback(run)
        if timeout is not None:

            def timeout_callback() -> None:
                # If we can cancel the future, do so and wait on it. If not,
                # Just stop the loop and return with the task still pending.
                # (If we neither cancel nor wait for the task, a warning
                # will be logged).
                assert future_cell[0] is not None
                if not future_cell[0].cancel():
                    self.stop()

            timeout_handle = self.add_timeout(self.time() + timeout, timeout_callback)
        self.start()
        if timeout is not None:
            self.remove_timeout(timeout_handle)
        assert future_cell[0] is not None
        if future_cell[0].cancelled() or not future_cell[0].done():
            raise TimeoutError("Operation timed out after %s seconds" % timeout)
        return future_cell[0].result()

    def time(self) -> float:
        """Returns the current time according to the `IOLoop`'s clock.

        The return value is a floating-point number relative to an
        unspecified time in the past.

        Historically, the IOLoop could be customized to use e.g.
        `time.monotonic` instead of `time.time`, but this is not
        currently supported and so this method is equivalent to
        `time.time`.

        """
        return time.time()

    def add_timeout(
        self,
        deadline: Union[float, datetime.timedelta],
        callback: Callable,
        *args: Any,
        **kwargs: Any
    ) -> object:
        """Runs the ``callback`` at the time ``deadline`` from the I/O loop.

        Returns an opaque handle that may be passed to
        `remove_timeout` to cancel.

        ``deadline`` may be a number denoting a time (on the same
        scale as `IOLoop.time`, normally `time.time`), or a
        `datetime.timedelta` object for a deadline relative to the
        current time.  Since Tornado 4.0, `call_later` is a more
        convenient alternative for the relative case since it does not
        require a timedelta object.

        Note that it is not safe to call `add_timeout` from other threads.
        Instead, you must use `add_callback` to transfer control to the
        `IOLoop`'s thread, and then call `add_timeout` from there.

        Subclasses of IOLoop must implement either `add_timeout` or
        `call_at`; the default implementations of each will call
        the other.  `call_at` is usually easier to implement, but
        subclasses that wish to maintain compatibility with Tornado
        versions prior to 4.0 must use `add_timeout` instead.

        .. versionchanged:: 4.0
           Now passes through ``*args`` and ``**kwargs`` to the callback.
        """
        if isinstance(deadline, numbers.Real):
            return self.call_at(deadline, callback, *args, **kwargs)
        elif isinstance(deadline, datetime.timedelta):
            return self.call_at(
                self.time() + deadline.total_seconds(), callback, *args, **kwargs
            )
        else:
            raise TypeError("Unsupported deadline %r" % deadline)

    def call_later(
        self, delay: float, callback: Callable, *args: Any, **kwargs: Any
    ) -> object:
        """Runs the ``callback`` after ``delay`` seconds have passed.

        Returns an opaque handle that may be passed to `remove_timeout`
        to cancel.  Note that unlike the `asyncio` method of the same
        name, the returned object does not have a ``cancel()`` method.

        See `add_timeout` for comments on thread-safety and subclassing.

        .. versionadded:: 4.0
        """
        return self.call_at(self.time() + delay, callback, *args, **kwargs)

    def call_at(
        self, when: float, callback: Callable, *args: Any, **kwargs: Any
    ) -> object:
        """Runs the ``callback`` at the absolute time designated by ``when``.

        ``when`` must be a number using the same reference point as
        `IOLoop.time`.

        Returns an opaque handle that may be passed to `remove_timeout`
        to cancel.  Note that unlike the `asyncio` method of the same
        name, the returned object does not have a ``cancel()`` method.

        See `add_timeout` for comments on thread-safety and subclassing.

        .. versionadded:: 4.0
        """
        return self.add_timeout(when, callback, *args, **kwargs)

    def remove_timeout(self, timeout: object) -> None:
        """Cancels a pending timeout.

        The argument is a handle as returned by `add_timeout`.  It is
        safe to call `remove_timeout` even if the callback has already
        been run.
        """
        raise NotImplementedError()

    def add_callback(self, callback: Callable, *args: Any, **kwargs: Any) -> None:
        """Calls the given callback on the next I/O loop iteration.

        It is safe to call this method from any thread at any time,
        except from a signal handler.  Note that this is the **only**
        method in `IOLoop` that makes this thread-safety guarantee; all
        other interaction with the `IOLoop` must be done from that
        `IOLoop`'s thread.  `add_callback()` may be used to transfer
        control from other threads to the `IOLoop`'s thread.

        To add a callback from a signal handler, see
        `add_callback_from_signal`.
        """
        raise NotImplementedError()

    def add_callback_from_signal(
        self, callback: Callable, *args: Any, **kwargs: Any
    ) -> None:
        """Calls the given callback on the next I/O loop iteration.

        Safe for use from a Python signal handler; should not be used
        otherwise.
        """
        raise NotImplementedError()

    def spawn_callback(self, callback: Callable, *args: Any, **kwargs: Any) -> None:
        """Calls the given callback on the next IOLoop iteration.

        As of Tornado 6.0, this method is equivalent to `add_callback`.

        .. versionadded:: 4.0
        """
        self.add_callback(callback, *args, **kwargs)

    def add_future(
        self,
        future: "Union[Future[_T], concurrent.futures.Future[_T]]",
        callback: Callable[["Future[_T]"], None],
    ) -> None:
        """Schedules a callback on the ``IOLoop`` when the given
        `.Future` is finished.

        The callback is invoked with one argument, the
        `.Future`.

        This method only accepts `.Future` objects and not other
        awaitables (unlike most of Tornado where the two are
        interchangeable).
        """
        if isinstance(future, Future):
            # Note that we specifically do not want the inline behavior of
            # tornado.concurrent.future_add_done_callback. We always want
            # this callback scheduled on the next IOLoop iteration (which
            # asyncio.Future always does).
            #
            # Wrap the callback in self._run_callback so we control
            # the error logging (i.e. it goes to tornado.log.app_log
            # instead of asyncio's log).
            future.add_done_callback(
                lambda f: self._run_callback(functools.partial(callback, future))
            )
        else:
            assert is_future(future)
            # For concurrent futures, we use self.add_callback, so
            # it's fine if future_add_done_callback inlines that call.
            future_add_done_callback(
                future, lambda f: self.add_callback(callback, future)
            )

    def run_in_executor(
        self,
        executor: Optional[concurrent.futures.Executor],
        func: Callable[..., _T],
        *args: Any
    ) -> Awaitable[_T]:
        """Runs a function in a ``concurrent.futures.Executor``. If
        ``executor`` is ``None``, the IO loop's default executor will be used.

        Use `functools.partial` to pass keyword arguments to ``func``.

        .. versionadded:: 5.0
        """
        if executor is None:
            if not hasattr(self, "_executor"):
                from tornado.process import cpu_count

                self._executor = concurrent.futures.ThreadPoolExecutor(
                    max_workers=(cpu_count() * 5)
                )  # type: concurrent.futures.Executor
            executor = self._executor
        c_future = executor.submit(func, *args)
        # Concurrent Futures are not usable with await. Wrap this in a
        # Tornado Future instead, using self.add_future for thread-safety.
        t_future = Future()  # type: Future[_T]
        self.add_future(c_future, lambda f: chain_future(f, t_future))
        return t_future

    def set_default_executor(self, executor: concurrent.futures.Executor) -> None:
        """Sets the default executor to use with :meth:`run_in_executor`.

        .. versionadded:: 5.0
        """
        self._executor = executor

    def _run_callback(self, callback: Callable[[], Any]) -> None:
        """Runs a callback with error handling.

        .. versionchanged:: 6.0

           CancelledErrors are no longer logged.
        """
        try:
            ret = callback()
            if ret is not None:
                from tornado import gen

                # Functions that return Futures typically swallow all
                # exceptions and store them in the Future.  If a Future
                # makes it out to the IOLoop, ensure its exception (if any)
                # gets logged too.
                try:
                    ret = gen.convert_yielded(ret)
                except gen.BadYieldError:
                    # It's not unusual for add_callback to be used with
                    # methods returning a non-None and non-yieldable
                    # result, which should just be ignored.
                    pass
                else:
                    self.add_future(ret, self._discard_future_result)
        except asyncio.CancelledError:
            pass
        except Exception:
            app_log.error("Exception in callback %r", callback, exc_info=True)

    def _discard_future_result(self, future: Future) -> None:
        """Avoid unhandled-exception warnings from spawned coroutines."""
        future.result()

    def split_fd(
        self, fd: Union[int, _Selectable]
    ) -> Tuple[int, Union[int, _Selectable]]:
        # """Returns an (fd, obj) pair from an ``fd`` parameter.

        # We accept both raw file descriptors and file-like objects as
        # input to `add_handler` and related methods.  When a file-like
        # object is passed, we must retain the object itself so we can
        # close it correctly when the `IOLoop` shuts down, but the
        # poller interfaces favor file descriptors (they will accept
        # file-like objects and call ``fileno()`` for you, but they
        # always return the descriptor itself).

        # This method is provided for use by `IOLoop` subclasses and should
        # not generally be used by application code.

        # .. versionadded:: 4.0
        # """
        if isinstance(fd, int):
            return fd, fd
        return fd.fileno(), fd

    def close_fd(self, fd: Union[int, _Selectable]) -> None:
        # """Utility method to close an ``fd``.

        # If ``fd`` is a file-like object, we close it directly; otherwise
        # we use `os.close`.

        # This method is provided for use by `IOLoop` subclasses (in
        # implementations of ``IOLoop.close(all_fds=True)`` and should
        # not generally be used by application code.

        # .. versionadded:: 4.0
        # """
        try:
            if isinstance(fd, int):
                os.close(fd)
            else:
                fd.close()
        except OSError:
            pass


class _Timeout(object):
    """An IOLoop timeout, a UNIX timestamp and a callback"""

    # Reduce memory overhead when there are lots of pending callbacks
    __slots__ = ["deadline", "callback", "tdeadline"]

    def __init__(
        self, deadline: float, callback: Callable[[], None], io_loop: IOLoop
    ) -> None:
        if not isinstance(deadline, numbers.Real):
            raise TypeError("Unsupported deadline %r" % deadline)
        self.deadline = deadline
        self.callback = callback
        self.tdeadline = (
            deadline,
            next(io_loop._timeout_counter),
        )  # type: Tuple[float, int]

    # Comparison methods to sort by deadline, with object id as a tiebreaker
    # to guarantee a consistent ordering.  The heapq module uses __le__
    # in python2.5, and __lt__ in 2.6+ (sort() and most other comparisons
    # use __lt__).
    def __lt__(self, other: "_Timeout") -> bool:
        return self.tdeadline < other.tdeadline

    def __le__(self, other: "_Timeout") -> bool:
        return self.tdeadline <= other.tdeadline


class PeriodicCallback(object):
    """Schedules the given callback to be called periodically.

    The callback is called every ``callback_time`` milliseconds when
    ``callback_time`` is a float. Note that the timeout is given in
    milliseconds, while most other time-related functions in Tornado use
    seconds. ``callback_time`` may alternatively be given as a
    `datetime.timedelta` object.

    If ``jitter`` is specified, each callback time will be randomly selected
    within a window of ``jitter * callback_time`` milliseconds.
    Jitter can be used to reduce alignment of events with similar periods.
    A jitter of 0.1 means allowing a 10% variation in callback time.
    The window is centered on ``callback_time`` so the total number of calls
    within a given interval should not be significantly affected by adding
    jitter.

    If the callback runs for longer than ``callback_time`` milliseconds,
    subsequent invocations will be skipped to get back on schedule.

    `start` must be called after the `PeriodicCallback` is created.

    .. versionchanged:: 5.0
       The ``io_loop`` argument (deprecated since version 4.1) has been removed.

    .. versionchanged:: 5.1
       The ``jitter`` argument is added.

    .. versionchanged:: 6.2
       If the ``callback`` argument is a coroutine, and a callback runs for
       longer than ``callback_time``, subsequent invocations will be skipped.
       Previously this was only true for regular functions, not coroutines,
       which were "fire-and-forget" for `PeriodicCallback`.

       The ``callback_time`` argument now accepts `datetime.timedelta` objects,
       in addition to the previous numeric milliseconds.
    """

    def __init__(
        self,
        callback: Callable[[], Optional[Awaitable]],
        callback_time: Union[datetime.timedelta, float],
        jitter: float = 0,
    ) -> None:
        self.callback = callback
        if isinstance(callback_time, datetime.timedelta):
            self.callback_time = callback_time / datetime.timedelta(milliseconds=1)
        else:
            if callback_time <= 0:
                raise ValueError("Periodic callback must have a positive callback_time")
            self.callback_time = callback_time
        self.jitter = jitter
        self._running = False
        self._timeout = None  # type: object

    def start(self) -> None:
        """Starts the timer."""
        # Looking up the IOLoop here allows to first instantiate the
        # PeriodicCallback in another thread, then start it using
        # IOLoop.add_callback().
        self.io_loop = IOLoop.current()
        self._running = True
        self._next_timeout = self.io_loop.time()
        self._schedule_next()

    def stop(self) -> None:
        """Stops the timer."""
        self._running = False
        if self._timeout is not None:
            self.io_loop.remove_timeout(self._timeout)
            self._timeout = None

    def is_running(self) -> bool:
        """Returns ``True`` if this `.PeriodicCallback` has been started.

        .. versionadded:: 4.1
        """
        return self._running

    async def _run(self) -> None:
        if not self._running:
            return
        try:
            val = self.callback()
            if val is not None and isawaitable(val):
                await val
        except Exception:
            app_log.error("Exception in callback %r", self.callback, exc_info=True)
        finally:
            self._schedule_next()

    def _schedule_next(self) -> None:
        if self._running:
            self._update_next(self.io_loop.time())
            self._timeout = self.io_loop.add_timeout(self._next_timeout, self._run)

    def _update_next(self, current_time: float) -> None:
        callback_time_sec = self.callback_time / 1000.0
        if self.jitter:
            # apply jitter fraction
            callback_time_sec *= 1 + (self.jitter * (random.random() - 0.5))
        if self._next_timeout <= current_time:
            # The period should be measured from the start of one call
            # to the start of the next. If one call takes too long,
            # skip cycles to get back to a multiple of the original
            # schedule.
            self._next_timeout += (
                math.floor((current_time - self._next_timeout) / callback_time_sec) + 1
            ) * callback_time_sec
        else:
            # If the clock moved backwards, ensure we advance the next
            # timeout instead of recomputing the same value again.
            # This may result in long gaps between callbacks if the
            # clock jumps backwards by a lot, but the far more common
            # scenario is a small NTP adjustment that should just be
            # ignored.
            #
            # Note that on some systems if time.time() runs slower
            # than time.monotonic() (most common on windows), we
            # effectively experience a small backwards time jump on
            # every iteration because PeriodicCallback uses
            # time.time() while asyncio schedules callbacks using
            # time.monotonic().
            # https://github.com/tornadoweb/tornado/issues/2333
            self._next_timeout += callback_time_sec