feat(cloud): add targeted task governance foundation
Tests / Test passed: 659

This commit is contained in:
2026-07-13 22:21:12 +08:00
parent a3ba94be04
commit 2cd314b183
41 changed files with 2099 additions and 15 deletions
+51 -2
View File
@@ -22,6 +22,7 @@ from cloud.auth import (
NullAuthProvider,
Principal,
)
from cloud.governance import TaskSubmissionPolicyError, enforce_user_submission_policy
from cloud.sdk.models import (
DeviceResponse,
ErrorResponse,
@@ -94,8 +95,21 @@ def create_cloud_router(
payload: TaskSubmissionRequest,
request: Request,
) -> TaskSubmissionResponse:
_authorize(request, TASKS_SUBMIT_SCOPE)
task_constraints = _build_constraints(payload.constraints)
principal = _authorize(request, TASKS_SUBMIT_SCOPE)
try:
task_constraints = _build_constraints(payload.constraints)
_validate_task_target(pool, task_constraints)
_enforce_user_policy(principal, scheduler.store, task_constraints)
except TaskSubmissionPolicyError as exc:
raise HTTPException(
status_code=status.HTTP_403_FORBIDDEN,
detail=str(exc),
) from exc
except ValueError as exc:
raise HTTPException(
status_code=status.HTTP_400_BAD_REQUEST,
detail=str(exc),
) from exc
try:
task_id = scheduler.submit(
goal=payload.goal,
@@ -128,6 +142,8 @@ def create_cloud_router(
attempt_count=task.attempt_count,
lease_expires_at=task.lease_expires_at,
failure_reason=task.failure_reason,
target_host_id=task.constraints.target_host_id,
target_device_id=task.constraints.target_device_id,
)
@router.get("/tasks", response_model=TaskListResponse)
@@ -158,6 +174,8 @@ def create_cloud_router(
assigned_host_id=task.assigned_host_id,
attempt_count=task.attempt_count,
failure_reason=task.failure_reason,
target_host_id=task.constraints.target_host_id,
target_device_id=task.constraints.target_device_id,
created_at=task.created_at,
)
for task in tasks
@@ -303,4 +321,35 @@ def _build_constraints(model):
return TaskConstraints(
driver_type=model.driver_type,
capability_tags=list(model.capability_tags),
target_host_id=model.target_host_id,
target_device_id=model.target_device_id,
)
def _validate_task_target(pool, constraints) -> None:
if constraints.target_device_id and not constraints.target_host_id:
raise ValueError("target_device_id requires target_host_id")
if constraints.target_host_id is None:
return
if not any(host.host_id == constraints.target_host_id for host in pool.list_hosts()):
raise ValueError(f"target host {constraints.target_host_id!r} is not known")
if constraints.target_device_id is not None and not any(
device.host_id == constraints.target_host_id
and device.device_id == constraints.target_device_id
for device in pool.list_devices()
):
raise ValueError(
f"target device {constraints.target_device_id!r} is not owned by "
f"host {constraints.target_host_id!r}"
)
def _enforce_user_policy(principal, store, constraints) -> None:
if not principal.id.startswith("user:"):
return
policy = store.get_user_submission_policy(principal.id.removeprefix("user:"))
enforce_user_submission_policy(
policy,
target_host_id=constraints.target_host_id,
target_device_id=constraints.target_device_id,
)
+32 -1
View File
@@ -72,15 +72,24 @@ class CloudClient:
workflow_definition_id: str | None = None,
driver_type: str | None = None,
capability_tags: list[str] | None = None,
target_host_id: str | None = None,
target_device_id: str | None = None,
) -> dict[str, Any]:
payload: dict[str, Any] = {
"goal": goal,
"workflow_definition_id": workflow_definition_id,
}
if driver_type is not None or capability_tags is not None:
if (
driver_type is not None
or capability_tags is not None
or target_host_id is not None
or target_device_id is not None
):
payload["constraints"] = {
"driver_type": driver_type,
"capability_tags": list(capability_tags or []),
"target_host_id": target_host_id,
"target_device_id": target_device_id,
}
resp = self._request("POST", "/tasks", json=payload)
return resp.json()
@@ -209,6 +218,28 @@ class CloudClient:
def revoke_user_sessions(self, user_id: str) -> None:
self._request("DELETE", f"/users/{user_id}/sessions")
# ------------------------------------------------------------- governance
def get_user_submission_policy(self, user_id: str) -> dict[str, Any]:
return self._request("GET", f"/users/{user_id}/submission-policy").json()
def update_user_submission_policy(
self, user_id: str, **policy: Any
) -> dict[str, Any]:
return self._request(
"PUT", f"/users/{user_id}/submission-policy", json=policy
).json()
def get_host_governance_policy(self, host_id: str) -> dict[str, Any]:
return self._request("GET", f"/hosts/{host_id}/governance-policy").json()
def update_host_governance_policy(
self, host_id: str, **policy: Any
) -> dict[str, Any]:
return self._request(
"PUT", f"/hosts/{host_id}/governance-policy", json=policy
).json()
# ------------------------------------------------------------------ helpers
def _url(self, path: str) -> str:
@@ -0,0 +1,169 @@
from __future__ import annotations
from uuid import uuid4
from fastapi import APIRouter, HTTPException, Request, status
from cloud.auth import AuthProvider, GOVERNANCE_ADMIN_SCOPE, GOVERNANCE_READ_SCOPE
from cloud.observability import current_correlation_id
from cloud.sdk.models import (
HostGovernancePolicyRequest,
HostGovernancePolicyResponse,
UserSubmissionPolicyRequest,
UserSubmissionPolicyResponse,
)
from cloud.user_auth import AuthAuditEvent
from core.models import utc_now
def create_governance_router(*, repository, auth_provider: AuthProvider) -> APIRouter:
router = APIRouter(prefix="/v1", tags=["cloud-governance"])
def authorize(request: Request, required_scope: str):
principal = auth_provider.authenticate(request)
if principal is None:
raise HTTPException(
status_code=status.HTTP_401_UNAUTHORIZED,
detail="unauthorized",
headers={"WWW-Authenticate": "Bearer"},
)
if principal.must_change_password or not principal.has_scope(required_scope):
raise HTTPException(
status_code=status.HTTP_403_FORBIDDEN,
detail=f"missing required scope: {required_scope}",
)
return principal
@router.get(
"/users/{user_id}/submission-policy",
response_model=UserSubmissionPolicyResponse,
)
def get_user_policy(user_id: str, request: Request) -> UserSubmissionPolicyResponse:
authorize(request, GOVERNANCE_READ_SCOPE)
policy = repository.get_user_submission_policy(user_id)
if policy is None:
raise HTTPException(status_code=404, detail="submission policy not found")
return UserSubmissionPolicyResponse(
user_id=policy.user_id,
revision=policy.revision,
submission_enabled=policy.submission_enabled,
allowed_host_ids=list(policy.allowed_host_ids)
if policy.allowed_host_ids is not None
else None,
allowed_device_targets=[
{"host_id": host_id, "device_id": device_id}
for host_id, device_id in policy.allowed_device_targets or ()
]
if policy.allowed_device_targets is not None
else None,
updated_at=policy.updated_at,
)
@router.put(
"/users/{user_id}/submission-policy",
response_model=UserSubmissionPolicyResponse,
)
def put_user_policy(
user_id: str,
payload: UserSubmissionPolicyRequest,
request: Request,
) -> UserSubmissionPolicyResponse:
principal = authorize(request, GOVERNANCE_ADMIN_SCOPE)
try:
policy = repository.upsert_user_submission_policy(
user_id=user_id,
submission_enabled=payload.submission_enabled,
allowed_host_ids=_unique_strings(payload.allowed_host_ids),
allowed_device_targets=(
tuple((item.host_id, item.device_id) for item in payload.allowed_device_targets)
if payload.allowed_device_targets is not None
else None
),
updated_at=utc_now(),
)
except KeyError as exc:
raise HTTPException(status_code=404, detail="user not found") from exc
_audit(repository, principal.id, user_id, "user_submission_policy_update")
return UserSubmissionPolicyResponse(
user_id=policy.user_id,
revision=policy.revision,
submission_enabled=policy.submission_enabled,
allowed_host_ids=list(policy.allowed_host_ids)
if policy.allowed_host_ids is not None
else None,
allowed_device_targets=[
{"host_id": host_id, "device_id": device_id}
for host_id, device_id in policy.allowed_device_targets or ()
]
if policy.allowed_device_targets is not None
else None,
updated_at=policy.updated_at,
)
@router.get(
"/hosts/{host_id}/governance-policy",
response_model=HostGovernancePolicyResponse,
)
def get_host_policy(host_id: str, request: Request) -> HostGovernancePolicyResponse:
authorize(request, GOVERNANCE_READ_SCOPE)
policy = repository.get_host_governance_policy(host_id)
if policy is None:
raise HTTPException(status_code=404, detail="Host policy not found")
return _host_response(policy)
@router.put(
"/hosts/{host_id}/governance-policy",
response_model=HostGovernancePolicyResponse,
)
def put_host_policy(
host_id: str,
payload: HostGovernancePolicyRequest,
request: Request,
) -> HostGovernancePolicyResponse:
principal = authorize(request, GOVERNANCE_ADMIN_SCOPE)
try:
policy = repository.upsert_host_governance_policy(
host_id=host_id,
self_submission_enabled=payload.self_submission_enabled,
max_active_tasks=payload.max_active_tasks,
daily_token_budget=payload.daily_token_budget,
updated_at=utc_now(),
)
except KeyError as exc:
raise HTTPException(status_code=404, detail="Host not found") from exc
_audit(repository, principal.id, host_id, "host_governance_policy_update")
return _host_response(policy)
return router
def _unique_strings(values: list[str] | None) -> tuple[str, ...] | None:
if values is None:
return None
return tuple(dict.fromkeys(value for value in values if value))
def _host_response(policy) -> HostGovernancePolicyResponse:
return HostGovernancePolicyResponse(
host_id=policy.host_id,
revision=policy.revision,
self_submission_enabled=policy.self_submission_enabled,
max_active_tasks=policy.max_active_tasks,
daily_token_budget=policy.daily_token_budget,
updated_at=policy.updated_at,
)
def _audit(repository, actor_id: str, target: str, action: str) -> None:
repository.record_auth_audit(
AuthAuditEvent(
id=uuid4().hex,
occurred_at=utc_now(),
actor_principal_id=actor_id,
target_user_id=target if action.startswith("user_") else None,
action=action,
outcome="success",
correlation_id=current_correlation_id(),
metadata={"target": target},
)
)
@@ -11,6 +11,8 @@ from pydantic import BaseModel, Field
class TaskConstraintsModel(BaseModel):
driver_type: str | None = None
capability_tags: list[str] = Field(default_factory=list)
target_host_id: str | None = Field(default=None, min_length=1)
target_device_id: str | None = Field(default=None, min_length=1)
class TaskSubmissionRequest(BaseModel):
@@ -33,6 +35,8 @@ class TaskStatusResponse(BaseModel):
attempt_count: int = 0
lease_expires_at: datetime | None = None
failure_reason: str | None = None
target_host_id: str | None = None
target_device_id: str | None = None
class TaskListItem(BaseModel):
@@ -44,6 +48,8 @@ class TaskListItem(BaseModel):
assigned_host_id: str | None = None
attempt_count: int = 0
failure_reason: str | None = None
target_host_id: str | None = None
target_device_id: str | None = None
created_at: datetime
@@ -145,3 +151,32 @@ class PasswordResetRequest(BaseModel):
class ErrorResponse(BaseModel):
detail: str
class DeviceTargetModel(BaseModel):
host_id: str = Field(min_length=1)
device_id: str = Field(min_length=1)
class UserSubmissionPolicyRequest(BaseModel):
submission_enabled: bool = True
allowed_host_ids: list[str] | None = None
allowed_device_targets: list[DeviceTargetModel] | None = None
class UserSubmissionPolicyResponse(UserSubmissionPolicyRequest):
user_id: str
revision: int
updated_at: datetime
class HostGovernancePolicyRequest(BaseModel):
self_submission_enabled: bool = True
max_active_tasks: int | None = Field(default=None, ge=1)
daily_token_budget: int | None = Field(default=None, ge=1)
class HostGovernancePolicyResponse(HostGovernancePolicyRequest):
host_id: str
revision: int
updated_at: datetime