deer-flow/backend/tests/test_extension_gateway_wiring.py
Nan Gao c542185a7f
feat(extensions): add gateway contribution points and packaged extension management (#4780)
* feat(extensions): add gateway services and routers

* feat(extensions): add standalone reference extension

* fix(extensions): harden contributed gateway routes

* docs(extensions): document gateway contribution points

* feat(extensions): add operator CLI for packaged extension management

Add `deerflow extensions install/list/enable/disable/remove` plus the root
`make extension-*` wrappers, backed by an `ExtensionManager` that owns one
transaction over backend/pyproject.toml, backend/uv.lock, the managed source
snapshot, the uv environment, and the `plugins:` block in config.yaml.

Install accepts a package requirement, a public HTTPS Git URL, or a local
directory. Local directories are copied to backend/extensions/sources/ as
deployable snapshots rather than editable installs, and the root .dockerignore
re-includes that tree so snapshots reach the backend builder. Remote sources are
HTTPS-only; SSH Git, file:// and local wheels are rejected because the stock
Docker builder cannot reproduce them.

Because environment configuration can still resolve a plain package name to a
local wheel (a UV_FIND_LINKS wheelhouse, say), every uv add/remove is followed
by an audit of the new lock: any local reference the stock image build cannot
reproduce rolls back the whole transaction. A config carrying duplicate
top-level `plugins:` keys is rejected outright rather than managed against one
block while the Gateway reads another.

Dependency synchronization now has one lock authority. The `extensions`
dependency group joins [tool.uv].default-groups, every startup path syncs the
same lock with --locked and launches with --no-sync, and the Docker images move
to uv 0.11.1 for the --no-workspace boundary the manager needs.

Loader gains `enabled`, `name` and `package` fields so a disabled extension is
skipped before resolution and import.

Co-authored-by: Codex <codex@openai.com>

* fix(extensions): stop the managed plugins rewrite from destroying config

Two data-safety defects in the managed `plugins:` block writer.

The "next top-level key" boundary was a regex matching only
`[A-Za-z_][A-Za-z0-9_-]*` or a quoted key. `AppConfig` is `extra="allow"`, so a
config may legally carry any top-level key, and a key the pattern cannot
recognize did not fail loudly — it read as "no next section", and the rewrite
replaced that neighbour and its entire subtree with the managed block. `my.key`,
`2fa`, `$schema`, `my key` and non-ASCII keys were all silently deleted by a
plain `extension-enable`/`disable`. Both boundaries now come from the YAML
parser's node marks, so key shape is irrelevant.

The file-final branch never consulted the trailing-comment scan the has-next-key
branch used, so any comment below the block was dropped. Since the manager
appends `plugins:` at end of file, that is the steady-state shape for most
installs: an operator note below the block was destroyed on the next toggle.

Separately, every managed install wrote `required: true` while the loader
defaults to false. That turned any later load failure — broken wheel, missing
native library, deleted snapshot — into a Gateway startup abort recoverable only
with shell access. New records are now written `required: false`, with an
explicit `install --required` opt-in; adopting an existing hand-written record
still preserves the operator's own choice.

* fix(extensions): harden the manager transaction and correct its docs

Follow-up hardening on the extension package manager.

Security posture, which the docs already claimed:
- Scrub `UV_PYTHON`, `UV_INSECURE_HOST`, `UV_CONSTRAINT` and
  `UV_NO_BUILD_ISOLATION` from the controlled uv environment. `UV_PYTHON` swaps
  the interpreter that the entry-point probe then imports and calls, and every
  later `uv run --no-sync` startup uses; `UV_INSECURE_HOST` removes the TLS
  validation the HTTPS-only source rule depends on. Neither is an index, proxy,
  cache or credential-provider setting, so neither was covered by the carve-out.
- Recognize run-together and all-caps secret query parameters (`accesstoken`,
  `ACCESSTOKEN`, `key`, `pw`, `sas`, `code`). The camel-case splitter only fires
  on case transitions, so only the separated spellings were caught. Short
  generic words stay boundary-anchored, so `?keyword=` remains installable.
- Validate the config before running any uv command. `uv add`/`uv sync` execute
  the package's build backend, so a config the manager could never write to must
  fail before that code runs rather than afterwards through rollback.

Transaction integrity:
- Run the second dependency-file restore from a `finally`. The recovery sync
  runs without `--locked` when the checkout had no lock, so uv writes one while
  resolving; if that sync then failed, the restore was skipped and the operator
  kept a lock file they never had. A failing recovery sync now also reports the
  original failure instead of replacing it.
- Skip the recovery sync on cancellation. Answering Ctrl-C with a full
  dependency resolve invites a second interrupt that escapes the handler and
  strands the checkout mid-transaction; the declarations are already restored
  and the next locked startup sync reconciles the environment.
- Retry a non-blocking lock on Windows instead of using `msvcrt.LK_LOCK`, which
  gives up after ~10s — far shorter than a real `uv add` plus `uv sync`, so
  contention surfaced as `Permission denied` rather than serializing.
- Locate the entry-point probe's JSON payload instead of parsing stdout's first
  line, so a `sitecustomize`/`.pth` banner cannot roll back a good install.
- Warn when the lock records a loopback source. `127.0.0.1` inside the image
  builder is a different machine, but unlike an environment-driven wheelhouse
  resolution this is a source the operator typed deliberately, so it is reported
  rather than rolled back. Private-network indexes are untouched: a builder on
  that network can reach them.

Docs: the blanket claim that failed operations restore the config file was
wrong — the conflict branches deliberately preserve a concurrent external edit
and leave `remove` deactivated. Document that, the `required: false` default,
the config preflight, the interrupt behaviour, and where the plugins-block
boundaries come from.

* test(gateway): pin the request-path projection agreement

`get_request_route_path()` imports the private
`starlette._utils.get_route_path` so the auth and CSRF predicates classify
the exact string Starlette's router matches on. Its requirement is not
"strip root_path correctly" but "return what the dispatcher is matching",
so delegating to the router's own implementation keeps the two in lockstep
by construction. Keep the private import rather than vendoring a copy: an
import that disappears fails loudly at startup, while a stale copy diverges
silently at a security boundary.

Cover the property directly instead of the mechanism, so the tests survive
a future reimplementation:

- projection edge cases, including the segment-boundary guard that keeps
  root_path="/api" from slicing "/apifoo/models" into a string the router
  would never match
- agreement with the router under nested mounts
- the two bypasses these predicates exist to prevent: a protected route
  mounted under the "/health" public prefix must still 401, and a POST
  mounted under "/api/webhooks" must still require a CSRF token

Both are verified to fail when the projection is reverted to
`request.url.path` (9/13 red) and when a plausible vendored copy omits the
boundary guard (the 2 boundary cases red).

Declare starlette as a bounded direct dependency so a bump — which is
security-relevant here — shows up in review rather than arriving silently
through FastAPI.

* ci: pin uv to the version production ships

ExtensionManager is not a consumer of uv the build tool -- it is a program
whose whole job is driving `uv` as a subprocess, depending on its CLI
behavior (`--no-workspace`, `--no-sync`, what `uv add` writes into
`[dependency-groups] extensions`) and on the `uv.lock` serialization format.
uv is closer to a runtime dependency with a contract than to incidental
tooling.

backend/Dockerfile pins that binary to 0.11.1, but all eight
astral-sh/setup-uv steps installed whatever was latest at run time, so CI
exercised the manager against a uv that is not the uv production runs. The
sharpest failure that allows: a newer uv bumps uv.lock's `revision`, CI
stays green because the same uv reads back what it wrote, and the pinned uv
in the production image cannot read the committed lock. `uv lock --check`
is version-sensitive for the same reason -- it verifies the lock is what
*this* uv would produce, and two versions can emit equivalent but
non-identical output.

Pin every step to 0.11.1 and lift the one lingering setup-uv@v3 to v7 so
the steps share input and caching behavior.

Pinning alone drifts apart again on the next bump, so add a constraint test
in the style of test_compose_default_bind_host.py: the Dockerfile's
UV_IMAGE tag is the single source of truth, and both compose defaults plus
every setup-uv step must match it. Verified to fail when a pin drifts, when
a step omits `version`, and -- the real scenario -- when the Dockerfile is
bumped alone, which lights up the workflows and both compose files at once.

* fix(gateway): state the extension route auth limit and abort a failed dev sync

Two scoped review follow-ups.

README: contributed routers cannot enter the host's reserved public prefixes,
which makes every extension endpoint session-authenticated -- there is no way
to expose an unauthenticated route. The rejection rule was documented but its
consequence was not, so inbound provider webhooks and public status endpoints
read as merely undocumented rather than out of scope for this release.

docker/dev-entrypoint.sh: the self-heal retry reuses `--locked`, so it repairs
a corrupt .venv but never a lock that disagrees with pyproject.toml. `set -e`
already stopped the script there -- uvicorn was not being started against a
stale environment -- but it exited on a bare uv exit code with no indication of
what to do. Abort explicitly with the cause and the fix.

Tests slice the sync block out of the real script and run it against a stub uv,
so they exercise the shipped code rather than a copy of it (/app/backend only
exists inside the container). They cover the success path, the retry that
recovers, the abort, and the guidance. Verified against the pre-fix script:
only the guidance case goes red, confirming the abort itself was already
correct.

* fix(extensions): point Git SSH shorthand at the HTTPS correction

Git's SCP-like shorthand carries no URL scheme, so `git+git@host:org/repo.git`
reached the scheme rules looking like a bare path and was rejected with
"local path references are not deployable; pass a local directory so DeerFlow
can snapshot it". The operator asked for a remote source, so that guidance
points at the wrong fix. Detect the shorthand ahead of the scheme rules and
report the public-HTTPS correction instead.

The bare `git@host:org/repo.git` spelling took a different wrong turn: packaging
parses it as a direct reference named `git`, leaving `host:org/repo.git`, whose
`host` reads as a URL scheme and produced the generic HTTPS message. Both
spellings now share one message, as does the PEP 508 named form.

* docs: keep the root extension summary within its new budget

#4799 split the depth out of the module guides and added a size gate; the root
file's job is now orientation, and this branch had pushed it 192 bytes past the
soft limit. The manager transaction, source rules, and lock discipline are
already stated in full in the extensions guide, so the root keeps the one-line
orientation and points there instead of restating them.

---------

Co-authored-by: Codex <codex@openai.com>
2026-08-13 23:55:30 +08:00

986 lines
32 KiB
Python

"""Gateway binding tests for app-scoped extension contributions."""
from __future__ import annotations
import asyncio
from contextlib import asynccontextmanager
from types import SimpleNamespace
import pytest
from deerflow.extensions.registry import ExtensionRegistry
class _Service:
def __init__(
self,
name: str,
events: list[str],
*,
fail_start: bool = False,
fail_stop: bool = False,
) -> None:
self.name = name
self.events = events
self.fail_start = fail_start
self.fail_stop = fail_stop
self.deps = None
async def start(self, deps) -> None:
self.events.append(f"start:{self.name}")
self.deps = deps
if self.fail_start:
raise RuntimeError(f"{self.name} failed")
async def stop(self) -> None:
self.events.append(f"stop:{self.name}")
if self.fail_stop:
raise RuntimeError(f"{self.name} failed")
@pytest.mark.asyncio
async def test_services_start_in_order_with_narrow_deps_and_fail_open():
from deerflow.extensions.gateway import start_services
events: list[str] = []
first = _Service("first", events, fail_start=True)
second = _Service("second", events)
registry = ExtensionRegistry()
with registry.attributed_to("first:install"):
registry.service(first)
with registry.attributed_to("second:install"):
registry.service(second)
extensions = registry.build()
session_factory = object()
config = SimpleNamespace(
token_budget=SimpleNamespace(enabled=False, max_tokens=999),
subagents=SimpleNamespace(max_total_per_run=4),
)
diagnostics = await start_services(extensions, config, session_factory)
assert events == ["start:first", "start:second"]
assert first.deps is second.deps
assert second.deps.app_store is extensions.app_store
assert second.deps.session_factory is session_factory
assert second.deps.policy.token_budget_enabled is False
assert second.deps.policy.max_total_tokens is None
assert second.deps.policy.max_subagents_per_run == 4
assert [(diagnostic.source, diagnostic.level) for diagnostic in diagnostics] == [("first:install", "error")]
@pytest.mark.asyncio
async def test_service_originated_cancelled_error_does_not_abort_start_batch():
from deerflow.extensions.gateway import start_services
events: list[str] = []
class _CancelsItself(_Service):
async def start(self, deps) -> None:
self.events.append(f"start:{self.name}")
raise asyncio.CancelledError()
first = _CancelsItself("first", events)
second = _Service("second", events)
registry = ExtensionRegistry()
with registry.attributed_to("first:install"):
registry.service(first)
with registry.attributed_to("second:install"):
registry.service(second)
diagnostics = await start_services(registry.build(), SimpleNamespace(), None)
assert events == ["start:first", "start:second"]
assert len(diagnostics) == 1
assert diagnostics[0].source == "first:install"
assert "CancelledError" in diagnostics[0].message
@pytest.mark.asyncio
async def test_services_stop_in_reverse_order_and_fail_open():
from deerflow.extensions.gateway import stop_services
events: list[str] = []
first = _Service("first", events)
second = _Service("second", events, fail_stop=True)
registry = ExtensionRegistry()
with registry.attributed_to("first:install"):
registry.service(first)
with registry.attributed_to("second:install"):
registry.service(second)
diagnostics = await stop_services(registry.build())
assert events == ["stop:second", "stop:first"]
assert [(diagnostic.source, diagnostic.level) for diagnostic in diagnostics] == [("second:install", "error")]
@pytest.mark.asyncio
async def test_service_originated_cancelled_error_does_not_abort_stop_batch():
from deerflow.extensions.gateway import stop_services
events: list[str] = []
class _CancelsItself(_Service):
async def stop(self) -> None:
self.events.append(f"stop:{self.name}")
raise asyncio.CancelledError()
first = _Service("first", events)
second = _CancelsItself("second", events)
registry = ExtensionRegistry()
with registry.attributed_to("first:install"):
registry.service(first)
with registry.attributed_to("second:install"):
registry.service(second)
diagnostics = await stop_services(registry.build())
assert events == ["stop:second", "stop:first"]
assert len(diagnostics) == 1
assert diagnostics[0].source == "second:install"
assert "CancelledError" in diagnostics[0].message
@pytest.mark.asyncio
async def test_each_service_stop_has_its_own_timeout_budget():
from deerflow.extensions.gateway import stop_services
events: list[str] = []
class _HangingService(_Service):
async def stop(self) -> None:
self.events.append(f"stop:{self.name}")
await asyncio.Event().wait()
first = _Service("first", events)
second = _HangingService("second", events)
registry = ExtensionRegistry()
with registry.attributed_to("first:install"):
registry.service(first)
with registry.attributed_to("second:install"):
registry.service(second)
diagnostics = await stop_services(registry.build(), timeout_seconds=0.01)
assert events == ["stop:second", "stop:first"]
assert len(diagnostics) == 1
assert diagnostics[0].source == "second:install"
assert "timed out" in diagnostics[0].message
@pytest.mark.asyncio
async def test_service_originated_timeout_error_is_reported_as_failure_not_budget_expiry():
from deerflow.extensions.gateway import stop_services
events: list[str] = []
class _RaisesTimeout(_Service):
async def stop(self) -> None:
self.events.append(f"stop:{self.name}")
raise TimeoutError("extension deadline")
first = _Service("first", events)
second = _RaisesTimeout("second", events)
registry = ExtensionRegistry()
with registry.attributed_to("first:install"):
registry.service(first)
with registry.attributed_to("second:install"):
registry.service(second)
diagnostics = await stop_services(registry.build(), timeout_seconds=1.0)
assert events == ["stop:second", "stop:first"]
assert diagnostics[0].source == "second:install"
assert "failed" in diagnostics[0].message
assert "timed out" not in diagnostics[0].message
@pytest.mark.parametrize(
("owner_path", "owner_protocol", "candidate_path", "candidate_protocol", "rejected"),
[
("/exact", "GET", "/exact", "GET", True),
("/items/{item_id}", "GET", "/items/{id}", "GET", True),
("/items/{item_id}", "GET", "/items/new", "GET", True),
("/items/{item_id}", "GET", "/items/prefix-{id}", "GET", True),
("/pre{tenant}", "GET", "/prefoo{id}", "GET", True),
("/items/new", "GET", "/items/{id}", "GET", False),
("/records/{value}", "GET", "/records/{id:int}", "GET", True),
("/records/{value:int}", "GET", "/records/0{id:int}", "GET", True),
("/records/{value:int}", "GET", "/records/new", "GET", False),
("/files/{rest:path}", "GET", "/files/{id:int}", "GET", True),
("/files/{rest:path}", "GET", "/files/{id}", "GET", False),
("/x/{rest:path}/tail", "GET", "/x/a/{id}/tail", "GET", False),
("/items/{item_id}", "GET", "/items/{id}", "POST", False),
("/live/{item_id}", "WS", "/live/{id}", "GET", False),
],
ids=[
"exact",
"renamed-parameter",
"dynamic-shadows-static",
"dynamic-shadows-compound",
"compound-trailing-str-shadows-narrower-compound",
"static-does-not-shadow-dynamic",
"str-covers-int",
"int-shadows-digit-compound",
"int-does-not-cover-static",
"path-covers-descendant",
"path-does-not-cover-newline-capable-str",
"nonterminal-path-does-not-cover-newline-capable-str",
"disjoint-http-methods",
"websocket-does-not-shadow-http",
],
)
def test_router_conflicts_follow_starlette_dispatch_order(
owner_path,
owner_protocol,
candidate_path,
candidate_protocol,
rejected,
):
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
async def endpoint():
return {"ok": True}
app = FastAPI()
if owner_protocol == "WS":
app.add_api_websocket_route(owner_path, endpoint)
else:
app.add_api_route(owner_path, endpoint, methods=[owner_protocol])
router = APIRouter()
if candidate_protocol == "WS":
router.add_api_websocket_route(candidate_path, endpoint)
else:
router.add_api_route(candidate_path, endpoint, methods=[candidate_protocol])
registry = ExtensionRegistry()
with registry.attributed_to("candidate:install"):
registry.routers((router,))
diagnostics = include_contributed_routers(app, registry.build())
assert bool(diagnostics) is rejected
if rejected:
assert diagnostics[0].source == "candidate:install"
assert "host" in diagnostics[0].message
assert candidate_path in diagnostics[0].message
else:
assert any(getattr(route, "path", None) == candidate_path for route in app.routes)
@pytest.mark.parametrize("convertor_name", ["flip", "int"])
def test_re_registered_convertor_does_not_create_a_false_shadow(
monkeypatch,
convertor_name,
):
from fastapi import APIRouter, FastAPI
from starlette.convertors import CONVERTOR_TYPES, Convertor
from starlette.routing import Match
from deerflow.extensions.gateway import include_contributed_routers
class DigitsConvertor(Convertor[str]):
regex = "[0-9]+"
def convert(self, value: str) -> str:
return value
def to_string(self, value: str) -> str:
return value
class LettersConvertor(Convertor[str]):
regex = "[A-Z]+"
def convert(self, value: str) -> str:
return value
def to_string(self, value: str) -> str:
return value
async def endpoint(value: str):
return {"value": value}
monkeypatch.setitem(CONVERTOR_TYPES, convertor_name, DigitsConvertor())
route_path = f"/owned/{{value:{convertor_name}}}"
app = FastAPI()
app.add_api_route(route_path, endpoint, methods=["GET"])
owner = app.routes[-1]
monkeypatch.setitem(CONVERTOR_TYPES, convertor_name, LettersConvertor())
router = APIRouter()
router.add_api_route(route_path, endpoint, methods=["GET"])
registry = ExtensionRegistry()
with registry.attributed_to("candidate:install"):
registry.routers((router,))
diagnostics = include_contributed_routers(app, registry.build())
assert diagnostics == []
candidate = app.routes[-1]
scope = {
"type": "http",
"path": "/owned/A",
"method": "GET",
"root_path": "",
}
assert owner.matches(scope)[0] is Match.NONE
assert candidate.matches(scope)[0] is Match.FULL
def test_router_claim_uses_converter_semantics_at_include_time(monkeypatch):
from fastapi import APIRouter, FastAPI
from starlette.convertors import CONVERTOR_TYPES, Convertor
from starlette.routing import Match
from deerflow.extensions.gateway import include_contributed_routers
class DigitsConvertor(Convertor[str]):
regex = "[0-9]+"
def convert(self, value: str) -> str:
return value
def to_string(self, value: str) -> str:
return value
class LettersConvertor(Convertor[str]):
regex = "[A-Z]+"
def convert(self, value: str) -> str:
return value
def to_string(self, value: str) -> str:
return value
async def endpoint(value: str):
return {"value": value}
monkeypatch.setitem(CONVERTOR_TYPES, "flip", DigitsConvertor())
route_path = "/owned/{value:flip}"
app = FastAPI()
app.add_api_route(route_path, endpoint, methods=["GET"])
owner = app.routes[-1]
router = APIRouter()
router.add_api_route(route_path, endpoint, methods=["GET"])
monkeypatch.setitem(CONVERTOR_TYPES, "flip", LettersConvertor())
registry = ExtensionRegistry()
with registry.attributed_to("candidate:install"):
registry.routers((router,))
diagnostics = include_contributed_routers(app, registry.build())
assert diagnostics == []
candidate = app.routes[-1]
scope = {
"type": "http",
"path": "/owned/A",
"method": "GET",
"root_path": "",
}
assert owner.matches(scope)[0] is Match.NONE
assert candidate.matches(scope)[0] is Match.FULL
def test_recompiled_converter_cannot_enter_a_public_namespace(monkeypatch):
from fastapi import APIRouter, FastAPI
from starlette.convertors import CONVERTOR_TYPES, Convertor
from deerflow.extensions.gateway import include_contributed_routers
class PublicPathConvertor(Convertor[str]):
regex = r"webhooks/.+"
def convert(self, value: str) -> str:
return value
def to_string(self, value: str) -> str:
return value
async def endpoint(value: str):
return {"value": value}
router = APIRouter()
router.add_api_route("/api/{value:int}", endpoint, methods=["GET"])
monkeypatch.setitem(CONVERTOR_TYPES, "int", PublicPathConvertor())
registry = ExtensionRegistry()
with registry.attributed_to("candidate:install"):
registry.routers((router,))
app = FastAPI(docs_url=None, redoc_url=None, openapi_url=None)
diagnostics = include_contributed_routers(app, registry.build())
assert [diagnostic.source for diagnostic in diagnostics] == ["candidate:install"]
assert "public namespace" in diagnostics[0].message
assert not any(getattr(route, "path", None) == "/api/{value:int}" for route in app.routes)
def test_host_mount_claims_descendant_http_paths():
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
app = FastAPI()
app.mount("/assets", FastAPI())
router = APIRouter()
@router.get("/assets/{name:int}")
async def asset(name: int):
return {"name": name}
registry = ExtensionRegistry()
with registry.attributed_to("assets:install"):
registry.routers((router,))
diagnostics = include_contributed_routers(app, registry.build())
assert len(diagnostics) == 1
assert diagnostics[0].source == "assets:install"
assert "host" in diagnostics[0].message
def test_dynamic_host_mount_claims_matching_descendants():
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
app = FastAPI()
app.mount("/pre{tenant}", FastAPI())
router = APIRouter()
@router.get("/prefoo/{item_id:int}")
async def item(item_id: int):
return {"item_id": item_id}
registry = ExtensionRegistry()
with registry.attributed_to("mount:install"):
registry.routers((router,))
diagnostics = include_contributed_routers(app, registry.build())
assert [diagnostic.source for diagnostic in diagnostics] == ["mount:install"]
assert "host" in diagnostics[0].message
@pytest.mark.parametrize(
("mount_path", "candidate_path", "witness"),
[
("/assets", "/assets/{name}", "/assets/a\nb"),
("/pre{tenant}", "/prefoo{id}/{child}", "/prefoo1/a\nb"),
],
)
def test_host_mount_does_not_claim_newline_capable_str_descendants(
mount_path,
candidate_path,
witness,
):
from fastapi import APIRouter, FastAPI
from starlette.routing import Match, Mount
from deerflow.extensions.gateway import include_contributed_routers
async def endpoint():
return {"ok": True}
app = FastAPI()
app.mount(mount_path, FastAPI())
host_mount = next(route for route in app.routes if isinstance(route, Mount) and route.path == mount_path)
router = APIRouter()
router.add_api_route(candidate_path, endpoint, methods=["GET"])
registry = ExtensionRegistry()
with registry.attributed_to("mount:install"):
registry.routers((router,))
diagnostics = include_contributed_routers(app, registry.build())
assert diagnostics == []
candidate_route = next(route for route in app.routes if getattr(route, "path", None) == candidate_path)
scope = {
"type": "http",
"path": witness,
"method": "GET",
"root_path": "",
}
assert host_mount.matches(scope)[0] is Match.NONE
assert candidate_route.matches(scope)[0] is Match.FULL
def test_contributed_websocket_route_is_rejected_until_host_auth_wraps_it():
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
async def websocket_endpoint(websocket):
await websocket.close()
router = APIRouter()
router.add_api_websocket_route("/extension-ws", websocket_endpoint)
registry = ExtensionRegistry()
with registry.attributed_to("websocket:install"):
registry.routers((router,))
app = FastAPI()
diagnostics = include_contributed_routers(app, registry.build())
assert [diagnostic.source for diagnostic in diagnostics] == ["websocket:install"]
assert "WebSocket" in diagnostics[0].message
assert not any(getattr(route, "path", None) == "/extension-ws" for route in app.routes)
@pytest.mark.parametrize(
"path",
[
"/health-extension",
"/docs-private",
"/redoc-private",
"/api/webhooks/extension",
"/api/{rest:path}",
"/api/{section}/extension",
],
)
def test_extension_routes_cannot_enter_host_public_namespaces(path):
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
async def endpoint():
return {"ok": True}
router = APIRouter()
router.add_api_route(path, endpoint, methods=["GET"])
registry = ExtensionRegistry()
with registry.attributed_to("public:install"):
registry.routers((router,))
app = FastAPI(docs_url=None, redoc_url=None, openapi_url=None)
diagnostics = include_contributed_routers(app, registry.build())
assert [diagnostic.source for diagnostic in diagnostics] == ["public:install"]
assert "public namespace" in diagnostics[0].message
assert not any(getattr(route, "path", None) == path for route in app.routes)
@pytest.mark.parametrize(
"path",
[
"/api",
"/heal",
"/api/{item_id:int}",
"/api/{item_id}",
],
)
def test_extension_routes_that_cannot_enter_a_public_namespace_are_allowed(path):
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
async def endpoint():
return {"ok": True}
router = APIRouter()
router.add_api_route(path, endpoint, methods=["GET"])
registry = ExtensionRegistry()
with registry.attributed_to("private:install"):
registry.routers((router,))
app = FastAPI(docs_url=None, redoc_url=None, openapi_url=None)
diagnostics = include_contributed_routers(app, registry.build())
assert diagnostics == []
assert any(getattr(route, "path", None) == path for route in app.routes)
def test_unknown_convertor_near_a_public_namespace_fails_closed(monkeypatch):
from fastapi import APIRouter, FastAPI
from starlette.convertors import CONVERTOR_TYPES, Convertor
from deerflow.extensions.gateway import include_contributed_routers
class UppercaseConvertor(Convertor[str]):
regex = "[A-Z]+"
def convert(self, value: str) -> str:
return value
def to_string(self, value: str) -> str:
return value
monkeypatch.setitem(CONVERTOR_TYPES, "uppercase", UppercaseConvertor())
router = APIRouter()
router.add_api_route(
"/api/{value:uppercase}",
lambda: {"ok": True},
methods=["GET"],
)
registry = ExtensionRegistry()
with registry.attributed_to("custom-public:install"):
registry.routers((router,))
diagnostics = include_contributed_routers(
FastAPI(docs_url=None, redoc_url=None, openapi_url=None),
registry.build(),
)
assert [diagnostic.source for diagnostic in diagnostics] == ["custom-public:install"]
assert "public namespace" in diagnostics[0].message
def test_private_custom_convertor_with_named_backreference_is_allowed(monkeypatch):
from fastapi import APIRouter, FastAPI
from starlette.convertors import CONVERTOR_TYPES, Convertor
from deerflow.extensions.gateway import include_contributed_routers
class DoubledLetterConvertor(Convertor[str]):
regex = r"(?P<char>[A-Z])(?P=char)"
def convert(self, value: str) -> str:
return value
def to_string(self, value: str) -> str:
return value
monkeypatch.setitem(CONVERTOR_TYPES, "doubled", DoubledLetterConvertor())
router = APIRouter()
router.add_api_route(
"/private/{value:doubled}",
lambda value: {"value": value},
methods=["GET"],
)
registry = ExtensionRegistry()
with registry.attributed_to("custom-private:install"):
registry.routers((router,))
app = FastAPI(docs_url=None, redoc_url=None, openapi_url=None)
diagnostics = include_contributed_routers(app, registry.build())
assert diagnostics == []
assert any(getattr(route, "path", None) == "/private/{value:doubled}" for route in app.routes)
def test_extension_public_paths_track_auth_middleware_public_paths():
from app.gateway.auth_middleware import (
_PUBLIC_EXACT_PATHS,
_PUBLIC_PATH_PREFIXES,
_is_public,
)
from deerflow.extensions.gateway import (
_HOST_PUBLIC_EXACT_PATHS,
_HOST_PUBLIC_PATH_PREFIXES,
)
assert _HOST_PUBLIC_PATH_PREFIXES == _PUBLIC_PATH_PREFIXES
assert _HOST_PUBLIC_EXACT_PATHS == _PUBLIC_EXACT_PATHS
assert all(_is_public(f"{path}//") for path in _HOST_PUBLIC_EXACT_PATHS)
@pytest.mark.parametrize(
("host_path", "host_method", "candidate_path", "candidate_method"),
[
(
"/api/v1/auth/login/local",
"POST",
"/api/v1/auth/login/local",
"GET",
),
(
"/api/v1/auth/login/local",
"POST",
"/api/v1/auth/login/local/",
"GET",
),
(
"/api/v1/auth/login/local",
"POST",
"/api/v1/auth/login/local//",
"GET",
),
("/api/v1/auth/me", "GET", "/api/v1/auth/me", "POST"),
("/api/v1/auth/me", "GET", "/api/v1/auth/me/", "POST"),
],
)
def test_extension_routes_cannot_claim_reserved_exact_paths_with_a_disjoint_method(
host_path,
host_method,
candidate_path,
candidate_method,
):
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
async def endpoint():
return {"ok": True}
app = FastAPI()
app.add_api_route(host_path, endpoint, methods=[host_method])
router = APIRouter()
router.add_api_route(candidate_path, endpoint, methods=[candidate_method])
registry = ExtensionRegistry()
with registry.attributed_to("public-exact:install"):
registry.routers((router,))
diagnostics = include_contributed_routers(app, registry.build())
assert [diagnostic.source for diagnostic in diagnostics] == ["public-exact:install"]
assert "reserved" in diagnostics[0].message
assert not any(getattr(route, "path", None) == candidate_path and getattr(route, "methods", set()) == {candidate_method} for route in app.routes)
@pytest.mark.parametrize(
"path",
[
"/api/v1/auth/me",
"/api/v1/auth/me/",
"/api/v1/auth/me//",
],
)
def test_extension_csrf_reserved_exact_paths_track_csrf_exemption(
monkeypatch,
path,
):
from starlette.requests import Request
from app.gateway import csrf_middleware
from deerflow.extensions.gateway import (
_CSRF_STATE_CHANGING_METHODS,
_HOST_CSRF_EXEMPT_EXACT_PATHS,
_HOST_PUBLIC_EXACT_PATHS,
)
monkeypatch.setattr(csrf_middleware, "is_auth_disabled", lambda: False)
request = Request(
{
"type": "http",
"method": "POST",
"scheme": "http",
"path": path,
"raw_path": path.encode(),
"query_string": b"",
"headers": [],
"server": ("testserver", 80),
}
)
assert path.rstrip("/") in _HOST_CSRF_EXEMPT_EXACT_PATHS
assert _HOST_CSRF_EXEMPT_EXACT_PATHS == csrf_middleware._CSRF_EXEMPT_EXACT_PATHS
assert _CSRF_STATE_CHANGING_METHODS == csrf_middleware._CSRF_STATE_CHANGING_METHODS
assert csrf_middleware._AUTH_EXEMPT_PATHS <= _HOST_PUBLIC_EXACT_PATHS
assert csrf_middleware.should_check_csrf(request) is False
def test_safe_method_at_csrf_exempt_exact_path_is_not_reserved():
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
router = APIRouter()
router.add_api_route(
"/api/v1/auth/me",
lambda: {"ok": True},
methods=["GET"],
)
registry = ExtensionRegistry()
with registry.attributed_to("safe-csrf:install"):
registry.routers((router,))
app = FastAPI(docs_url=None, redoc_url=None, openapi_url=None)
diagnostics = include_contributed_routers(app, registry.build())
assert diagnostics == []
assert any(getattr(route, "path", None) == "/api/v1/auth/me" for route in app.routes)
def test_router_with_one_conflict_is_rejected_atomically_and_names_first_owner():
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
async def endpoint():
return {"ok": True}
first = APIRouter()
first.add_api_route("/shared", endpoint, methods=["GET"])
second = APIRouter()
second.add_api_route("/would-have-been-reachable", endpoint, methods=["GET"])
second.add_api_route("/shared", endpoint, methods=["GET"])
registry = ExtensionRegistry()
with registry.attributed_to("first:install"):
registry.routers((first,))
with registry.attributed_to("second:install"):
registry.routers((second,))
app = FastAPI()
diagnostics = include_contributed_routers(app, registry.build())
paths = [getattr(route, "path", None) for route in app.routes]
assert "/shared" in paths
assert "/would-have-been-reachable" not in paths
assert len(diagnostics) == 1
assert diagnostics[0].source == "second:install"
assert "first:install" in diagnostics[0].message
def test_router_is_rejected_when_its_own_earlier_route_shadows_a_later_one():
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
async def endpoint():
return {"ok": True}
router = APIRouter()
router.add_api_route("/same/{value}", endpoint, methods=["GET"])
router.add_api_route("/same/fixed", endpoint, methods=["GET"])
registry = ExtensionRegistry()
with registry.attributed_to("self-shadow:install"):
registry.routers((router,))
app = FastAPI()
diagnostics = include_contributed_routers(app, registry.build())
assert [diagnostic.source for diagnostic in diagnostics] == ["self-shadow:install"]
assert "self-shadow:install" in diagnostics[0].message
assert not any(getattr(route, "path", "").startswith("/same/") for route in app.routes)
def test_contributed_mount_is_rejected_but_does_not_starve_later_router():
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
bad = APIRouter()
bad.mount("/nested", FastAPI())
good = APIRouter()
@good.get("/extension-good")
async def extension_good():
return {"ok": True}
registry = ExtensionRegistry()
with registry.attributed_to("bad:install"):
registry.routers((bad,))
with registry.attributed_to("good:install"):
registry.routers((good,))
app = FastAPI()
diagnostics = include_contributed_routers(app, registry.build())
assert [diagnostic.source for diagnostic in diagnostics] == ["bad:install"]
assert "Mount" in diagnostics[0].message
assert any(getattr(route, "path", None) == "/extension-good" for route in app.routes)
def test_router_with_unsupported_route_item_is_rejected_atomically():
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
async def endpoint():
return {"ok": True}
router = APIRouter()
router.add_api_route("/otherwise-valid", endpoint, methods=["GET"])
router.routes.append(object())
registry = ExtensionRegistry()
with registry.attributed_to("unsupported:install"):
registry.routers((router,))
app = FastAPI()
diagnostics = include_contributed_routers(app, registry.build())
assert [diagnostic.source for diagnostic in diagnostics] == ["unsupported:install"]
assert "unsupported route" in diagnostics[0].message
assert not any(getattr(route, "path", None) == "/otherwise-valid" for route in app.routes)
def test_include_router_failure_rolls_back_partial_routes_before_continuing():
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
async def endpoint():
return {"ok": True}
broken = APIRouter()
broken.add_api_route("/partial", endpoint, methods=["GET"])
broken.add_api_route("/explodes", endpoint, methods=["GET"])
broken.routes[-1].endpoint = None
later = APIRouter()
later.add_api_route("/partial", endpoint, methods=["GET"])
registry = ExtensionRegistry()
with registry.attributed_to("broken:install"):
registry.routers((broken,))
with registry.attributed_to("later:install"):
registry.routers((later,))
app = FastAPI()
diagnostics = include_contributed_routers(app, registry.build())
partial_routes = [route for route in app.routes if getattr(route, "path", None) == "/partial"]
assert [diagnostic.source for diagnostic in diagnostics] == ["broken:install"]
assert len(partial_routes) == 1
def test_router_lifecycle_hooks_are_rejected_in_favor_of_extension_service():
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
async def endpoint():
return {"ok": True}
async def startup_hook():
raise RuntimeError("must never be installed")
bad = APIRouter()
bad.add_api_route("/has-lifecycle", endpoint, methods=["GET"])
bad.add_event_handler("startup", startup_hook)
good = APIRouter()
good.add_api_route("/after-lifecycle", endpoint, methods=["GET"])
registry = ExtensionRegistry()
with registry.attributed_to("lifecycle:install"):
registry.routers((bad,))
with registry.attributed_to("good:install"):
registry.routers((good,))
app = FastAPI()
diagnostics = include_contributed_routers(app, registry.build())
paths = [getattr(route, "path", None) for route in app.routes]
assert [diagnostic.source for diagnostic in diagnostics] == ["lifecycle:install"]
assert "ExtensionService" in diagnostics[0].message
assert "/has-lifecycle" not in paths
assert "/after-lifecycle" in paths
assert startup_hook not in app.router.on_startup
def test_router_custom_lifespan_is_rejected_in_favor_of_extension_service():
from fastapi import APIRouter, FastAPI
from deerflow.extensions.gateway import include_contributed_routers
@asynccontextmanager
async def lifespan(_app):
yield
async def endpoint():
return {"ok": True}
router = APIRouter(lifespan=lifespan)
router.add_api_route("/custom-lifespan", endpoint, methods=["GET"])
registry = ExtensionRegistry()
with registry.attributed_to("lifespan:install"):
registry.routers((router,))
app = FastAPI()
diagnostics = include_contributed_routers(app, registry.build())
assert [diagnostic.source for diagnostic in diagnostics] == ["lifespan:install"]
assert "ExtensionService" in diagnostics[0].message
assert not any(getattr(route, "path", None) == "/custom-lifespan" for route in app.routes)