Skip to content

Commit

Permalink
Define a CURRENT_JUPYTER_HANDLER context var (#1251)
Browse files Browse the repository at this point in the history
* [Enhancement] Define a CURRENT_JUPYTER_HANDLER context var
* add type to context var
* Introduce CallContext class
* Add CallContext to API docs
* Alphabetize submodules
* Unit test contextvar in the kernel shutdown flow
* Update tests/services/sessions/test_call_context.py

Co-authored-by: Kevin Bates <[email protected]>

* revert unit test back to using kernel_model
* Relocate to base package
* Update location in docs as well

---------

Co-authored-by: Kevin Bates <[email protected]>
  • Loading branch information
Zsailer and kevin-bates authored Apr 11, 2023
1 parent 87b2158 commit ca4b062
Show file tree
Hide file tree
Showing 5 changed files with 208 additions and 0 deletions.
6 changes: 6 additions & 0 deletions docs/source/api/jupyter_server.base.rst
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,12 @@ Submodules
----------


.. automodule:: jupyter_server.base.call_context
:members:
:undoc-members:
:show-inheritance:


.. automodule:: jupyter_server.base.handlers
:members:
:undoc-members:
Expand Down
1 change: 1 addition & 0 deletions jupyter_server/__init__.py
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,7 @@
del os

from ._version import __version__, version_info # noqa
from .base.call_context import CallContext # noqa


def _cleanup():
Expand Down
88 changes: 88 additions & 0 deletions jupyter_server/base/call_context.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,88 @@
"""Provides access to variables pertaining to specific call contexts."""
# Copyright (c) Jupyter Development Team.
# Distributed under the terms of the Modified BSD License.

from contextvars import Context, ContextVar, copy_context
from typing import Any, Dict, List


class CallContext:
"""CallContext essentially acts as a namespace for managing context variables.
Although not required, it is recommended that any "file-spanning" context variable
names (i.e., variables that will be set or retrieved from multiple files or services) be
added as constants to this class definition.
"""

# Add well-known (file-spanning) names here.
#: Provides access to the current request handler once set.
JUPYTER_HANDLER: str = "JUPYTER_HANDLER"

# A map of variable name to value is maintained as the single ContextVar. This also enables
# easier management over maintaining a set of ContextVar instances, since the Context is a
# map of ContextVar instances to their values, and the "name" is no longer a lookup key.
_NAME_VALUE_MAP = "_name_value_map"
_name_value_map: ContextVar[Dict[str, Any]] = ContextVar(_NAME_VALUE_MAP)

@classmethod
def get(cls, name: str) -> Any:
"""Returns the value corresponding the named variable relative to this context.
If the named variable doesn't exist, None will be returned.
Parameters
----------
name : str
The name of the variable to get from the call context
Returns
-------
value: Any
The value associated with the named variable for this call context
"""
name_value_map = CallContext._get_map()

if name in name_value_map:
return name_value_map[name]
return None # TODO - should this raise `LookupError` (or a custom error derived from said)

@classmethod
def set(cls, name: str, value: Any) -> None:
"""Sets the named variable to the specified value in the current call context.
Parameters
----------
name : str
The name of the variable to store into the call context
value : Any
The value of the variable to store into the call context
Returns
-------
None
"""
name_value_map = CallContext._get_map()
name_value_map[name] = value

@classmethod
def context_variable_names(cls) -> List[str]:
"""Returns a list of variable names set for this call context.
Returns
-------
names: List[str]
A list of variable names set for this call context.
"""
name_value_map = CallContext._get_map()
return list(name_value_map.keys())

@classmethod
def _get_map(cls) -> Dict[str, Any]:
"""Get the map of names to their values from the _NAME_VALUE_MAP context var.
If the map does not exist in the current context, an empty map is created and returned.
"""
ctx: Context = copy_context()
if CallContext._name_value_map not in ctx:
CallContext._name_value_map.set({})
return CallContext._name_value_map.get()
4 changes: 4 additions & 0 deletions jupyter_server/base/handlers.py
Original file line number Diff line number Diff line change
Expand Up @@ -26,6 +26,7 @@
from traitlets.config import Application

import jupyter_server
from jupyter_server import CallContext
from jupyter_server._sysinfo import get_sys_info
from jupyter_server._tz import utcnow
from jupyter_server.auth import authorized
Expand Down Expand Up @@ -582,6 +583,9 @@ def check_host(self):

async def prepare(self):
"""Pepare a response."""
# Set the current Jupyter Handler context variable.
CallContext.set(CallContext.JUPYTER_HANDLER, self)

if not self.check_host():
self.current_user = self._jupyter_current_user = None
raise web.HTTPError(403)
Expand Down
109 changes: 109 additions & 0 deletions tests/base/test_call_context.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,109 @@
import asyncio

from jupyter_server import CallContext
from jupyter_server.auth.utils import get_anonymous_username
from jupyter_server.base.handlers import JupyterHandler
from jupyter_server.services.kernels.kernelmanager import AsyncMappingKernelManager


async def test_jupyter_handler_contextvar(jp_fetch, monkeypatch):
# Create some mock kernel Ids
kernel1 = "x-x-x-x-x"
kernel2 = "y-y-y-y-y"

# We'll use this dictionary to track the current user within each request.
context_tracker = {
kernel1: {"started": "no user yet", "ended": "still no user", "user": None},
kernel2: {"started": "no user yet", "ended": "still no user", "user": None},
}

# Monkeypatch the get_current_user method in Tornado's
# request handler to return a random user name for
# each request
async def get_current_user(self):
return get_anonymous_username()

monkeypatch.setattr(JupyterHandler, "get_current_user", get_current_user)

# Monkeypatch the kernel_model method to show that
# the current context variable is truly local and
# not contaminated by other asynchronous parallel requests.
# Note that even though the current implementation of `kernel_model()`
# is synchronous, we can convert this into an async method because the
# kernel handler wraps the call to `kernel_model()` in `ensure_async()`.
async def kernel_model(self, kernel_id):
# Get the Jupyter Handler from the current context.
current: JupyterHandler = CallContext.get(CallContext.JUPYTER_HANDLER)
# Get the current user
context_tracker[kernel_id]["user"] = current.current_user
context_tracker[kernel_id]["started"] = current.current_user
await asyncio.sleep(1.0)
# Track the current user a few seconds later. We'll
# verify that this user was unaffected by other parallel
# requests.
context_tracker[kernel_id]["ended"] = current.current_user
return {"id": kernel_id, "name": "blah"}

monkeypatch.setattr(AsyncMappingKernelManager, "kernel_model", kernel_model)

# Make two requests in parallel.
await asyncio.gather(
jp_fetch("api", "kernels", kernel1),
jp_fetch("api", "kernels", kernel2),
)

# Assert that the two requests had different users
assert context_tracker[kernel1]["user"] != context_tracker[kernel2]["user"]
# Assert that the first request started+ended with the same user
assert context_tracker[kernel1]["started"] == context_tracker[kernel1]["ended"]
# Assert that the second request started+ended with the same user
assert context_tracker[kernel2]["started"] == context_tracker[kernel2]["ended"]


async def test_context_variable_names():
CallContext.set("foo", "bar")
CallContext.set("foo2", "bar2")
names = CallContext.context_variable_names()
assert len(names) == 2
assert set(names) == {"foo", "foo2"}


async def test_same_context_operations():
CallContext.set("foo", "bar")
CallContext.set("foo2", "bar2")

foo = CallContext.get("foo")
assert foo == "bar"

CallContext.set("foo", "bar2")
assert CallContext.get("foo") == CallContext.get("foo2")


async def test_multi_context_operations():
async def context1():
"""The "slower" context. This ensures that, following the sleep, the
context variable set prior to the sleep is still the expected value.
If contexts are not managed properly, we should find that context2() has
corrupted context1().
"""
CallContext.set("foo", "bar1")
await asyncio.sleep(1.0)
assert CallContext.get("foo") == "bar1"
context1_names = CallContext.context_variable_names()
assert len(context1_names) == 1

async def context2():
"""The "faster" context. This ensures that CallContext reflects the
appropriate values of THIS context.
"""
CallContext.set("foo", "bar2")
assert CallContext.get("foo") == "bar2"
CallContext.set("foo2", "bar2")
context2_names = CallContext.context_variable_names()
assert len(context2_names) == 2

await asyncio.gather(context1(), context2())

# Assert that THIS context doesn't have any variables defined.
names = CallContext.context_variable_names()
assert len(names) == 0

0 comments on commit ca4b062

Please sign in to comment.