mirror of
https://github.com/fawney19/Aether.git
synced 2026-09-02 01:10:23 +08:00
feat: 引入 Rust executor/gateway sidecar 及 Python 侧双后端适配
- 新增 Rust workspace crates: aether-contracts, aether-executor, aether-gateway - aether-executor: 支持 Unix Socket/TCP 双传输模式，处理同步/流式上游请求 - aether-gateway: 作为本地主入口代理，集成 /api/internal/gateway/resolve 认证预解析 - Python 侧新增 ExecutionPlan 契约和 RustExecutorClient，各 handler 支持 executor_backend=rust 时将可序列化请求转发给 Rust executor 执行 - 重构 dev.sh 支持 executor/gateway 进程编排与生命周期管理 - 新增 internal gateway 路由，提供 resolve/passthrough 端点 - handler 层(chat/cli/video/endpoint_checker 等)全面适配 Rust executor 回退逻辑 - pipeline 层支持 trusted auth context 跳过重复认证 - 新增 Rust CI workflow 及对应测试用例
This commit is contained in:
fawney19
565
tests/api/handlers/base/test_chat_rust_executor.py
Normal file
565
tests/api/handlers/base/test_chat_rust_executor.py
Normal file
@@ -0,0 +1,565 @@
|
||||
from __future__ import annotations
|
||||
|
||||
from types import SimpleNamespace
|
||||
|
||||
import httpx
|
||||
import pytest
|
||||
|
||||
import src.api.handlers.base.chat_sync_executor as chat_sync_mod
|
||||
from src.api.handlers.base.chat_sync_executor import ChatSyncExecutor
|
||||
from src.core.exceptions import EmbeddedErrorException
|
||||
from src.services.request.executor_plan import (
|
||||
ExecutionPlan,
|
||||
ExecutionPlanBody,
|
||||
PreparedExecutionPlan,
|
||||
)
|
||||
from src.services.request.rust_executor_client import (
|
||||
RustExecutorClientError,
|
||||
RustExecutorSyncResult,
|
||||
)
|
||||
|
||||
|
||||
class _FakeEnvelope:
|
||||
name = "fake-envelope"
|
||||
|
||||
def __init__(self) -> None:
|
||||
self.status_codes: list[int] = []
|
||||
self.postprocessed_payloads: list[dict[str, object]] = []
|
||||
|
||||
def on_http_status(self, *, base_url: str | None, status_code: int) -> None:
|
||||
self.status_codes.append(status_code)
|
||||
|
||||
def on_connection_error(self, *, base_url: str | None, exc: Exception) -> None:
|
||||
raise AssertionError("connection error hook should not be used in this test")
|
||||
|
||||
def unwrap_response(self, data: dict[str, object]) -> dict[str, object]:
|
||||
return dict(data["payload"]) # type: ignore[index]
|
||||
|
||||
def postprocess_unwrapped_response(self, *, model: str, data: dict[str, object]) -> None:
|
||||
self.postprocessed_payloads.append(dict(data))
|
||||
|
||||
|
||||
class _FakeNormalizer:
|
||||
def response_from_internal(
|
||||
self,
|
||||
internal_resp: object,
|
||||
*,
|
||||
requested_model: str,
|
||||
) -> dict[str, object]:
|
||||
return {
|
||||
"aggregated": True,
|
||||
"requested_model": requested_model,
|
||||
"internal_id": getattr(internal_resp, "id", "missing"),
|
||||
}
|
||||
|
||||
|
||||
def _make_prepared_plan() -> PreparedExecutionPlan:
|
||||
return PreparedExecutionPlan(
|
||||
contract=ExecutionPlan(
|
||||
request_id="req-test",
|
||||
candidate_id=None,
|
||||
provider_name="openai",
|
||||
provider_id="prov-1",
|
||||
endpoint_id="ep-1",
|
||||
key_id="key-1",
|
||||
method="POST",
|
||||
url="https://example.com/v1/chat/completions",
|
||||
headers={"content-type": "application/json"},
|
||||
body=ExecutionPlanBody(json_body={"model": "gpt-4.1"}),
|
||||
stream=False,
|
||||
provider_api_format="openai:chat",
|
||||
client_api_format="openai:chat",
|
||||
model_name="gpt-4.1",
|
||||
),
|
||||
payload={"model": "gpt-4.1"},
|
||||
headers={"content-type": "application/json"},
|
||||
upstream_is_stream=False,
|
||||
needs_conversion=False,
|
||||
provider_type="openai",
|
||||
request_timeout=30.0,
|
||||
)
|
||||
|
||||
|
||||
def _make_proxy_prepared_plan() -> PreparedExecutionPlan:
|
||||
prepared = _make_prepared_plan()
|
||||
prepared.contract.proxy = chat_sync_mod.ExecutionProxySnapshot(
|
||||
enabled=True,
|
||||
mode="http",
|
||||
label="proxy.internal",
|
||||
url="http://proxy.internal:8080",
|
||||
)
|
||||
prepared.proxy_config = {"url": "http://proxy.internal:8080"}
|
||||
return prepared
|
||||
|
||||
|
||||
def _make_tunnel_prepared_plan() -> PreparedExecutionPlan:
|
||||
prepared = _make_prepared_plan()
|
||||
prepared.contract.proxy = chat_sync_mod.ExecutionProxySnapshot(
|
||||
enabled=True,
|
||||
mode="tunnel",
|
||||
node_id="node-1",
|
||||
label="relay-node",
|
||||
)
|
||||
prepared.delegate_config = {"tunnel": True, "node_id": "node-1"}
|
||||
prepared.proxy_config = {"node_id": "node-1"}
|
||||
return prepared
|
||||
|
||||
|
||||
def _make_upstream_stream_prepared_plan() -> PreparedExecutionPlan:
|
||||
prepared = _make_prepared_plan()
|
||||
prepared.contract.stream = True
|
||||
prepared.upstream_is_stream = True
|
||||
return prepared
|
||||
|
||||
|
||||
def _make_tls_prepared_plan() -> PreparedExecutionPlan:
|
||||
prepared = _make_prepared_plan()
|
||||
prepared.contract.tls_profile = "claude_code_nodejs"
|
||||
prepared.provider_type = "claude_code"
|
||||
return prepared
|
||||
|
||||
|
||||
def _make_executor() -> ChatSyncExecutor:
|
||||
handler = SimpleNamespace(request_id="req-test")
|
||||
executor = ChatSyncExecutor(handler)
|
||||
executor._ctx.provider_api_format_for_error = "openai:chat"
|
||||
executor._ctx.client_api_format_for_error = "openai:chat"
|
||||
executor._ctx.needs_conversion_for_error = False
|
||||
return executor
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_sync_plan_uses_rust_executor_when_available(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
executor = _make_executor()
|
||||
prepared_plan = _make_prepared_plan()
|
||||
|
||||
monkeypatch.setattr(chat_sync_mod.config, "executor_backend", "rust")
|
||||
|
||||
async def _fake_execute_sync_json(self: object, plan: ExecutionPlan) -> RustExecutorSyncResult:
|
||||
assert plan.request_id == "req-test"
|
||||
return RustExecutorSyncResult(
|
||||
status_code=200,
|
||||
response_json={"id": "chatcmpl-1"},
|
||||
headers={"content-type": "application/json"},
|
||||
)
|
||||
|
||||
async def _should_not_fallback(**kwargs: object) -> dict[str, object]:
|
||||
raise AssertionError("local execution should not be used")
|
||||
|
||||
monkeypatch.setattr(
|
||||
chat_sync_mod.RustExecutorClient,
|
||||
"execute_sync_json",
|
||||
_fake_execute_sync_json,
|
||||
)
|
||||
monkeypatch.setattr(executor, "_execute_sync_plan_locally", _should_not_fallback)
|
||||
|
||||
response = await executor._execute_sync_plan(
|
||||
prepared_plan=prepared_plan,
|
||||
provider=SimpleNamespace(name="provider"),
|
||||
model="gpt-4.1",
|
||||
)
|
||||
|
||||
assert response == {"id": "chatcmpl-1"}
|
||||
assert executor._ctx.status_code == 200
|
||||
assert executor._ctx.response_json == {"id": "chatcmpl-1"}
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_sync_plan_allows_supported_proxy_urls_for_rust(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
executor = _make_executor()
|
||||
prepared_plan = _make_proxy_prepared_plan()
|
||||
|
||||
monkeypatch.setattr(chat_sync_mod.config, "executor_backend", "rust")
|
||||
|
||||
async def _fake_execute_sync_json(self: object, plan: ExecutionPlan) -> RustExecutorSyncResult:
|
||||
assert plan.proxy is not None
|
||||
assert plan.proxy.url == "http://proxy.internal:8080"
|
||||
return RustExecutorSyncResult(
|
||||
status_code=200,
|
||||
response_json={"id": "chatcmpl-proxy"},
|
||||
headers={"content-type": "application/json"},
|
||||
)
|
||||
|
||||
async def _should_not_fallback(**kwargs: object) -> dict[str, object]:
|
||||
raise AssertionError("local execution should not be used")
|
||||
|
||||
monkeypatch.setattr(
|
||||
chat_sync_mod.RustExecutorClient,
|
||||
"execute_sync_json",
|
||||
_fake_execute_sync_json,
|
||||
)
|
||||
monkeypatch.setattr(executor, "_execute_sync_plan_locally", _should_not_fallback)
|
||||
|
||||
response = await executor._execute_sync_plan(
|
||||
prepared_plan=prepared_plan,
|
||||
provider=SimpleNamespace(name="provider"),
|
||||
model="gpt-4.1",
|
||||
)
|
||||
|
||||
assert response == {"id": "chatcmpl-proxy"}
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_sync_plan_allows_tunnel_delegate_for_rust(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
executor = _make_executor()
|
||||
prepared_plan = _make_tunnel_prepared_plan()
|
||||
|
||||
monkeypatch.setattr(chat_sync_mod.config, "executor_backend", "rust")
|
||||
|
||||
async def _fake_execute_sync_json(self: object, plan: ExecutionPlan) -> RustExecutorSyncResult:
|
||||
assert plan.proxy is not None
|
||||
assert plan.proxy.mode == "tunnel"
|
||||
assert plan.proxy.node_id == "node-1"
|
||||
return RustExecutorSyncResult(
|
||||
status_code=200,
|
||||
response_json={"id": "chatcmpl-tunnel"},
|
||||
headers={"content-type": "application/json"},
|
||||
)
|
||||
|
||||
async def _should_not_fallback(**kwargs: object) -> dict[str, object]:
|
||||
raise AssertionError("local execution should not be used")
|
||||
|
||||
monkeypatch.setattr(
|
||||
chat_sync_mod.RustExecutorClient,
|
||||
"execute_sync_json",
|
||||
_fake_execute_sync_json,
|
||||
)
|
||||
monkeypatch.setattr(executor, "_execute_sync_plan_locally", _should_not_fallback)
|
||||
|
||||
response = await executor._execute_sync_plan(
|
||||
prepared_plan=prepared_plan,
|
||||
provider=SimpleNamespace(name="provider"),
|
||||
model="gpt-4.1",
|
||||
)
|
||||
|
||||
assert response == {"id": "chatcmpl-tunnel"}
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_sync_plan_allows_tls_profile_for_rust(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
executor = _make_executor()
|
||||
prepared_plan = _make_tls_prepared_plan()
|
||||
|
||||
monkeypatch.setattr(chat_sync_mod.config, "executor_backend", "rust")
|
||||
|
||||
async def _fake_execute_sync_json(self: object, plan: ExecutionPlan) -> RustExecutorSyncResult:
|
||||
assert plan.tls_profile == "claude_code_nodejs"
|
||||
return RustExecutorSyncResult(
|
||||
status_code=200,
|
||||
response_json={"id": "chatcmpl-tls"},
|
||||
headers={"content-type": "application/json"},
|
||||
)
|
||||
|
||||
async def _should_not_fallback(**kwargs: object) -> dict[str, object]:
|
||||
raise AssertionError("local execution should not be used")
|
||||
|
||||
monkeypatch.setattr(
|
||||
chat_sync_mod.RustExecutorClient,
|
||||
"execute_sync_json",
|
||||
_fake_execute_sync_json,
|
||||
)
|
||||
monkeypatch.setattr(executor, "_execute_sync_plan_locally", _should_not_fallback)
|
||||
|
||||
response = await executor._execute_sync_plan(
|
||||
prepared_plan=prepared_plan,
|
||||
provider=SimpleNamespace(name="provider"),
|
||||
model="claude-3.7-sonnet",
|
||||
)
|
||||
|
||||
assert response == {"id": "chatcmpl-tls"}
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_sync_plan_applies_envelope_postprocessing_after_rust(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
executor = _make_executor()
|
||||
prepared_plan = _make_prepared_plan()
|
||||
prepared_plan.envelope = _FakeEnvelope()
|
||||
|
||||
monkeypatch.setattr(chat_sync_mod.config, "executor_backend", "rust")
|
||||
|
||||
async def _fake_execute_sync_json(self: object, plan: ExecutionPlan) -> RustExecutorSyncResult:
|
||||
return RustExecutorSyncResult(
|
||||
status_code=200,
|
||||
response_json={"payload": {"id": "wrapped-1", "message": "ok"}},
|
||||
headers={"content-type": "application/json"},
|
||||
)
|
||||
|
||||
async def _should_not_fallback(**kwargs: object) -> dict[str, object]:
|
||||
raise AssertionError("local execution should not be used")
|
||||
|
||||
monkeypatch.setattr(
|
||||
chat_sync_mod.RustExecutorClient,
|
||||
"execute_sync_json",
|
||||
_fake_execute_sync_json,
|
||||
)
|
||||
monkeypatch.setattr(executor, "_execute_sync_plan_locally", _should_not_fallback)
|
||||
|
||||
response = await executor._execute_sync_plan(
|
||||
prepared_plan=prepared_plan,
|
||||
provider=SimpleNamespace(name="provider"),
|
||||
model="gpt-4.1",
|
||||
)
|
||||
|
||||
assert response == {"id": "wrapped-1", "message": "ok"}
|
||||
assert prepared_plan.envelope.status_codes == [200]
|
||||
assert prepared_plan.envelope.postprocessed_payloads == [
|
||||
{"id": "wrapped-1", "message": "ok"}
|
||||
]
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_sync_plan_applies_format_conversion_after_rust(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
executor = _make_executor()
|
||||
prepared_plan = _make_prepared_plan()
|
||||
prepared_plan.needs_conversion = True
|
||||
prepared_plan.contract.provider_api_format = "gemini:chat"
|
||||
prepared_plan.contract.client_api_format = "openai:chat"
|
||||
executor._ctx.provider_api_format_for_error = "gemini:chat"
|
||||
executor._ctx.client_api_format_for_error = "openai:chat"
|
||||
executor._ctx.needs_conversion_for_error = True
|
||||
|
||||
monkeypatch.setattr(chat_sync_mod.config, "executor_backend", "rust")
|
||||
|
||||
class _FakeRegistry:
|
||||
def convert_response(
|
||||
self,
|
||||
response_json: dict[str, object],
|
||||
source_format: str,
|
||||
target_format: str,
|
||||
*,
|
||||
requested_model: str,
|
||||
) -> dict[str, object]:
|
||||
assert source_format == "gemini:chat"
|
||||
assert target_format == "openai:chat"
|
||||
assert requested_model == "gpt-4.1"
|
||||
return {
|
||||
"converted": True,
|
||||
"source_id": response_json["provider_id"],
|
||||
}
|
||||
|
||||
async def _fake_execute_sync_json(self: object, plan: ExecutionPlan) -> RustExecutorSyncResult:
|
||||
assert plan.provider_api_format == "gemini:chat"
|
||||
return RustExecutorSyncResult(
|
||||
status_code=200,
|
||||
response_json={"provider_id": "gemini-1"},
|
||||
headers={"content-type": "application/json"},
|
||||
)
|
||||
|
||||
async def _should_not_fallback(**kwargs: object) -> dict[str, object]:
|
||||
raise AssertionError("local execution should not be used")
|
||||
|
||||
monkeypatch.setattr(
|
||||
chat_sync_mod.RustExecutorClient,
|
||||
"execute_sync_json",
|
||||
_fake_execute_sync_json,
|
||||
)
|
||||
monkeypatch.setattr(chat_sync_mod, "get_format_converter_registry", lambda: _FakeRegistry())
|
||||
monkeypatch.setattr(executor, "_execute_sync_plan_locally", _should_not_fallback)
|
||||
|
||||
response = await executor._execute_sync_plan(
|
||||
prepared_plan=prepared_plan,
|
||||
provider=SimpleNamespace(name="provider"),
|
||||
model="gpt-4.1",
|
||||
)
|
||||
|
||||
assert response == {"converted": True, "source_id": "gemini-1"}
|
||||
assert executor._ctx.provider_response_json == {"provider_id": "gemini-1"}
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_sync_plan_aggregates_upstream_stream_after_rust(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
executor = _make_executor()
|
||||
prepared_plan = _make_upstream_stream_prepared_plan()
|
||||
|
||||
monkeypatch.setattr(chat_sync_mod.config, "executor_backend", "rust")
|
||||
|
||||
class _FakeRegistry:
|
||||
def get_normalizer(self, format_id: str) -> _FakeNormalizer:
|
||||
assert format_id == "openai:chat"
|
||||
return _FakeNormalizer()
|
||||
|
||||
captured_chunks: list[bytes] = []
|
||||
|
||||
async def _fake_aggregate(
|
||||
byte_iter: object,
|
||||
*,
|
||||
provider_api_format: str,
|
||||
provider_name: str,
|
||||
model: str,
|
||||
request_id: str,
|
||||
envelope: object = None,
|
||||
provider_parser: object = None,
|
||||
) -> object:
|
||||
async for chunk in byte_iter: # type: ignore[attr-defined]
|
||||
captured_chunks.append(chunk)
|
||||
assert provider_api_format == "openai:chat"
|
||||
assert provider_name == "provider"
|
||||
assert model == "gpt-4.1"
|
||||
assert request_id == "req-test"
|
||||
return SimpleNamespace(id="agg-1")
|
||||
|
||||
async def _fake_execute_sync_json(self: object, plan: ExecutionPlan) -> RustExecutorSyncResult:
|
||||
assert plan.stream is True
|
||||
return RustExecutorSyncResult(
|
||||
status_code=200,
|
||||
response_body_bytes=b"data: {\"id\":\"chunk-1\"}\n\ndata: [DONE]\n\n",
|
||||
headers={"content-type": "text/event-stream"},
|
||||
)
|
||||
|
||||
async def _should_not_fallback(**kwargs: object) -> dict[str, object]:
|
||||
raise AssertionError("local execution should not be used")
|
||||
|
||||
monkeypatch.setattr(
|
||||
chat_sync_mod.RustExecutorClient,
|
||||
"execute_sync_json",
|
||||
_fake_execute_sync_json,
|
||||
)
|
||||
monkeypatch.setattr(chat_sync_mod, "get_format_converter_registry", lambda: _FakeRegistry())
|
||||
monkeypatch.setattr(
|
||||
"src.api.handlers.base.upstream_stream_bridge.aggregate_upstream_stream_to_internal_response",
|
||||
_fake_aggregate,
|
||||
)
|
||||
monkeypatch.setattr(executor, "_execute_sync_plan_locally", _should_not_fallback)
|
||||
|
||||
response = await executor._execute_sync_plan(
|
||||
prepared_plan=prepared_plan,
|
||||
provider=SimpleNamespace(name="provider"),
|
||||
model="gpt-4.1",
|
||||
)
|
||||
|
||||
assert response == {
|
||||
"aggregated": True,
|
||||
"requested_model": "gpt-4.1",
|
||||
"internal_id": "agg-1",
|
||||
}
|
||||
assert captured_chunks == [b"data: {\"id\":\"chunk-1\"}\n\ndata: [DONE]\n\n"]
|
||||
assert executor._ctx.status_code == 200
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_sync_plan_turns_rust_http_error_into_httpx_status_error(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
executor = _make_executor()
|
||||
prepared_plan = _make_prepared_plan()
|
||||
|
||||
monkeypatch.setattr(chat_sync_mod.config, "executor_backend", "rust")
|
||||
|
||||
async def _fake_execute_sync_json(self: object, plan: ExecutionPlan) -> RustExecutorSyncResult:
|
||||
assert plan.url.endswith("/chat/completions")
|
||||
return RustExecutorSyncResult(
|
||||
status_code=429,
|
||||
response_json={"error": {"message": "slow down"}},
|
||||
headers={"content-type": "application/json"},
|
||||
)
|
||||
|
||||
async def _should_not_fallback(**kwargs: object) -> dict[str, object]:
|
||||
raise AssertionError("local execution should not be used")
|
||||
|
||||
monkeypatch.setattr(
|
||||
chat_sync_mod.RustExecutorClient,
|
||||
"execute_sync_json",
|
||||
_fake_execute_sync_json,
|
||||
)
|
||||
monkeypatch.setattr(executor, "_execute_sync_plan_locally", _should_not_fallback)
|
||||
|
||||
with pytest.raises(httpx.HTTPStatusError) as exc_info:
|
||||
await executor._execute_sync_plan(
|
||||
prepared_plan=prepared_plan,
|
||||
provider=SimpleNamespace(name="provider"),
|
||||
model="gpt-4.1",
|
||||
)
|
||||
|
||||
assert exc_info.value.response.status_code == 429
|
||||
assert '"message": "slow down"' in exc_info.value.upstream_response # type: ignore[attr-defined]
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_sync_plan_preserves_embedded_error_semantics_from_rust(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
executor = _make_executor()
|
||||
prepared_plan = _make_prepared_plan()
|
||||
|
||||
monkeypatch.setattr(chat_sync_mod.config, "executor_backend", "rust")
|
||||
|
||||
async def _fake_execute_sync_json(self: object, plan: ExecutionPlan) -> RustExecutorSyncResult:
|
||||
assert plan.provider_api_format == "openai:chat"
|
||||
return RustExecutorSyncResult(
|
||||
status_code=200,
|
||||
response_json={
|
||||
"error": {
|
||||
"message": "bad request",
|
||||
"type": "invalid_request_error",
|
||||
"code": 400,
|
||||
}
|
||||
},
|
||||
headers={"content-type": "application/json"},
|
||||
)
|
||||
|
||||
monkeypatch.setattr(
|
||||
chat_sync_mod.RustExecutorClient,
|
||||
"execute_sync_json",
|
||||
_fake_execute_sync_json,
|
||||
)
|
||||
|
||||
with pytest.raises(EmbeddedErrorException) as exc_info:
|
||||
await executor._execute_sync_plan(
|
||||
prepared_plan=prepared_plan,
|
||||
provider=SimpleNamespace(name="provider"),
|
||||
model="gpt-4.1",
|
||||
)
|
||||
|
||||
assert exc_info.value.error_message == "bad request"
|
||||
assert exc_info.value.error_code == 400
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_sync_plan_falls_back_to_local_when_rust_unavailable(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
executor = _make_executor()
|
||||
prepared_plan = _make_prepared_plan()
|
||||
|
||||
monkeypatch.setattr(chat_sync_mod.config, "executor_backend", "rust")
|
||||
|
||||
async def _fake_execute_sync_json(self: object, plan: ExecutionPlan) -> RustExecutorSyncResult:
|
||||
assert plan.request_id == "req-test"
|
||||
raise RustExecutorClientError("executor down")
|
||||
|
||||
fallback_called = False
|
||||
|
||||
async def _fake_local_execute(**kwargs: object) -> dict[str, object]:
|
||||
nonlocal fallback_called
|
||||
fallback_called = True
|
||||
return {"id": "local-fallback"}
|
||||
|
||||
monkeypatch.setattr(
|
||||
chat_sync_mod.RustExecutorClient,
|
||||
"execute_sync_json",
|
||||
_fake_execute_sync_json,
|
||||
)
|
||||
monkeypatch.setattr(executor, "_execute_sync_plan_locally", _fake_local_execute)
|
||||
|
||||
response = await executor._execute_sync_plan(
|
||||
prepared_plan=prepared_plan,
|
||||
provider=SimpleNamespace(name="provider"),
|
||||
model="gpt-4.1",
|
||||
)
|
||||
|
||||
assert fallback_called is True
|
||||
assert response == {"id": "local-fallback"}
|
||||
553
tests/api/handlers/base/test_chat_rust_stream.py
Normal file
553
tests/api/handlers/base/test_chat_rust_stream.py
Normal file
@@ -0,0 +1,553 @@
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import AsyncGenerator
|
||||
from types import SimpleNamespace
|
||||
from typing import Any
|
||||
|
||||
import httpx
|
||||
import pytest
|
||||
|
||||
import src.api.handlers.base.chat_handler_base as chatmod
|
||||
import src.services.proxy_node.resolver as proxymod
|
||||
from src.api.handlers.base.chat_handler_base import ChatHandlerBase
|
||||
from src.api.handlers.base.stream_context import StreamContext
|
||||
from src.services.request.rust_executor_client import (
|
||||
RustExecutorClientError,
|
||||
RustExecutorStreamResult,
|
||||
)
|
||||
|
||||
|
||||
class _DummyAuthInfo:
|
||||
auth_header = "authorization"
|
||||
auth_value = "Bearer test"
|
||||
decrypted_auth_config = None
|
||||
|
||||
def as_tuple(self) -> tuple[str, str]:
|
||||
return self.auth_header, self.auth_value
|
||||
|
||||
|
||||
class _PassBuilder:
|
||||
def build(self, request_body: dict[str, Any], *args: Any, **kwargs: Any) -> Any:
|
||||
return request_body, {"content-type": "application/json"}
|
||||
|
||||
|
||||
class _DummyStreamResponseCtx:
|
||||
def __init__(self) -> None:
|
||||
self.closed = False
|
||||
|
||||
async def __aexit__(self, exc_type: object, exc: object, tb: object) -> None:
|
||||
self.closed = True
|
||||
|
||||
|
||||
class _FakeStreamProcessor:
|
||||
def __init__(self) -> None:
|
||||
self.prefetched_chunks: list[bytes] | None = None
|
||||
self.response_ctx: _DummyStreamResponseCtx | None = None
|
||||
|
||||
async def prefetch_and_check_error(
|
||||
self,
|
||||
byte_iterator: Any,
|
||||
provider: Any,
|
||||
endpoint: Any,
|
||||
ctx: Any,
|
||||
max_prefetch_lines: int = 5,
|
||||
max_prefetch_bytes: int = 65536,
|
||||
) -> list[bytes]:
|
||||
del provider, endpoint, ctx, max_prefetch_lines, max_prefetch_bytes
|
||||
first = await anext(byte_iterator)
|
||||
self.prefetched_chunks = [first]
|
||||
return self.prefetched_chunks
|
||||
|
||||
async def create_response_stream(
|
||||
self,
|
||||
ctx: Any,
|
||||
byte_iterator: Any,
|
||||
response_ctx: _DummyStreamResponseCtx,
|
||||
prefetched_chunks: list[bytes] | None = None,
|
||||
*,
|
||||
start_time: float | None = None,
|
||||
) -> AsyncGenerator[bytes]:
|
||||
del ctx, start_time
|
||||
self.response_ctx = response_ctx
|
||||
try:
|
||||
for chunk in prefetched_chunks or []:
|
||||
yield chunk
|
||||
async for chunk in byte_iterator:
|
||||
yield chunk
|
||||
finally:
|
||||
await response_ctx.__aexit__(None, None, None)
|
||||
|
||||
|
||||
class _DummyChatHandler(ChatHandlerBase):
|
||||
FORMAT_ID = "openai:chat"
|
||||
|
||||
def __init__(self) -> None:
|
||||
self.request_id = "req-test"
|
||||
self.api_key = SimpleNamespace(id="user-key-1")
|
||||
self._request_builder = _PassBuilder()
|
||||
self.allowed_api_formats = ["openai:chat"]
|
||||
self.api_family = None
|
||||
self.endpoint_kind = None
|
||||
self.start_time = 0.0
|
||||
|
||||
async def _convert_request(self, request: Any) -> Any:
|
||||
return request
|
||||
|
||||
def _extract_usage(self, response: dict) -> dict[str, int]:
|
||||
return {}
|
||||
|
||||
async def _get_mapped_model(
|
||||
self,
|
||||
source_model: str,
|
||||
provider_id: str,
|
||||
api_format: str | None = None,
|
||||
) -> str | None:
|
||||
del source_model, provider_id, api_format
|
||||
return None
|
||||
|
||||
def apply_mapped_model(self, request_body: dict[str, Any], mapped_model: str) -> dict[str, Any]:
|
||||
out = dict(request_body)
|
||||
out["model"] = mapped_model
|
||||
return out
|
||||
|
||||
def prepare_provider_request_body(self, request_body: dict[str, Any]) -> dict[str, Any]:
|
||||
return dict(request_body)
|
||||
|
||||
def finalize_provider_request(
|
||||
self,
|
||||
request_body: dict[str, Any],
|
||||
*,
|
||||
mapped_model: str | None,
|
||||
provider_api_format: str | None,
|
||||
) -> dict[str, Any]:
|
||||
del mapped_model, provider_api_format
|
||||
return dict(request_body)
|
||||
|
||||
def get_model_for_url(
|
||||
self,
|
||||
request_body: dict[str, Any],
|
||||
mapped_model: str | None,
|
||||
) -> str | None:
|
||||
return mapped_model or str(request_body.get("model") or "")
|
||||
|
||||
|
||||
def _patch_stream_setup(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
*,
|
||||
proxy_info: dict[str, Any] | None = None,
|
||||
delegate_config: dict[str, Any] | None = None,
|
||||
) -> None:
|
||||
async def _fake_get_provider_auth(endpoint: Any, key: Any) -> _DummyAuthInfo:
|
||||
del endpoint, key
|
||||
return _DummyAuthInfo()
|
||||
|
||||
async def _fake_resolve_proxy_info(proxy_config: Any) -> Any:
|
||||
del proxy_config
|
||||
return proxy_info
|
||||
|
||||
async def _fake_resolve_delegate(proxy_config: Any) -> Any:
|
||||
del proxy_config
|
||||
return delegate_config
|
||||
|
||||
async def _fake_get_system_proxy() -> None:
|
||||
return None
|
||||
|
||||
monkeypatch.setattr(chatmod, "get_provider_auth", _fake_get_provider_auth)
|
||||
monkeypatch.setattr(
|
||||
chatmod,
|
||||
"get_provider_behavior",
|
||||
lambda **kwargs: SimpleNamespace(
|
||||
envelope=None,
|
||||
same_format_variant=None,
|
||||
cross_format_variant=None,
|
||||
),
|
||||
)
|
||||
monkeypatch.setattr(chatmod, "build_provider_url", lambda *args, **kwargs: "https://upstream.test/v1/chat/completions")
|
||||
monkeypatch.setattr(chatmod, "get_upstream_stream_policy", lambda *args, **kwargs: None)
|
||||
monkeypatch.setattr(
|
||||
chatmod,
|
||||
"resolve_upstream_is_stream",
|
||||
lambda *, client_is_stream, policy: client_is_stream,
|
||||
)
|
||||
monkeypatch.setattr(chatmod, "enforce_stream_mode_for_upstream", lambda *args, **kwargs: None)
|
||||
monkeypatch.setattr(
|
||||
chatmod,
|
||||
"maybe_patch_request_with_prompt_cache_key",
|
||||
lambda request_body, **kwargs: request_body,
|
||||
)
|
||||
monkeypatch.setattr(proxymod, "resolve_effective_proxy", lambda provider_proxy, key_proxy=None: None)
|
||||
monkeypatch.setattr(proxymod, "resolve_proxy_info_async", _fake_resolve_proxy_info)
|
||||
monkeypatch.setattr(proxymod, "get_proxy_label", lambda proxy_info: "direct")
|
||||
monkeypatch.setattr(proxymod, "resolve_delegate_config_async", _fake_resolve_delegate)
|
||||
monkeypatch.setattr(proxymod, "get_system_proxy_config_async", _fake_get_system_proxy)
|
||||
monkeypatch.setattr(proxymod, "build_proxy_url_async", _fake_get_system_proxy)
|
||||
|
||||
|
||||
async def _iter_chunks(chunks: list[bytes]) -> AsyncGenerator[bytes]:
|
||||
for chunk in chunks:
|
||||
yield chunk
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_stream_request_uses_rust_executor_when_available(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
_patch_stream_setup(monkeypatch)
|
||||
monkeypatch.setattr(chatmod.config, "executor_backend", "rust")
|
||||
|
||||
handler = _DummyChatHandler()
|
||||
stream_processor = _FakeStreamProcessor()
|
||||
ctx = StreamContext(model="gpt-test", api_format="openai:chat")
|
||||
ctx.client_api_format = "openai:chat"
|
||||
|
||||
provider = SimpleNamespace(name="provider", id="provider-1", provider_type="", proxy=None)
|
||||
endpoint = SimpleNamespace(id="endpoint-1", api_format="openai:chat", base_url="https://x")
|
||||
key = SimpleNamespace(id="key-1", proxy=None)
|
||||
candidate = SimpleNamespace(
|
||||
request_candidate_id="cand-1",
|
||||
mapping_matched_model=None,
|
||||
needs_conversion=False,
|
||||
output_limit=None,
|
||||
)
|
||||
|
||||
dummy_ctx = _DummyStreamResponseCtx()
|
||||
|
||||
async def _fake_execute_stream(self: object, plan: object) -> RustExecutorStreamResult:
|
||||
assert getattr(plan, "stream") is True
|
||||
return RustExecutorStreamResult(
|
||||
status_code=200,
|
||||
headers={"content-type": "text/event-stream", "x-upstream-test": "true"},
|
||||
byte_iterator=_iter_chunks(
|
||||
[
|
||||
b"data: {\"id\":\"chunk-1\"}\n\n",
|
||||
b"data: [DONE]\n\n",
|
||||
]
|
||||
),
|
||||
response_ctx=dummy_ctx,
|
||||
)
|
||||
|
||||
monkeypatch.setattr(chatmod.RustExecutorClient, "execute_stream", _fake_execute_stream)
|
||||
|
||||
stream = await handler._execute_stream_request(
|
||||
ctx,
|
||||
stream_processor,
|
||||
provider,
|
||||
endpoint,
|
||||
key,
|
||||
{"model": "gpt-test", "messages": [{"role": "user", "content": "hello"}]},
|
||||
{},
|
||||
candidate=candidate,
|
||||
)
|
||||
|
||||
received = [chunk async for chunk in stream]
|
||||
|
||||
assert received == [
|
||||
b"data: {\"id\":\"chunk-1\"}\n\n",
|
||||
b"data: [DONE]\n\n",
|
||||
]
|
||||
assert ctx.status_code == 200
|
||||
assert ctx.response_headers["x-upstream-test"] == "true"
|
||||
assert stream_processor.prefetched_chunks == [b"data: {\"id\":\"chunk-1\"}\n\n"]
|
||||
assert dummy_ctx.closed is True
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_stream_request_accepts_async_generator_stream_processor(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
_patch_stream_setup(monkeypatch)
|
||||
monkeypatch.setattr(chatmod.config, "executor_backend", "rust")
|
||||
|
||||
handler = _DummyChatHandler()
|
||||
stream_processor = _FakeStreamProcessor()
|
||||
ctx = StreamContext(model="gpt-test", api_format="openai:chat")
|
||||
ctx.client_api_format = "openai:chat"
|
||||
|
||||
provider = SimpleNamespace(name="provider", id="provider-1", provider_type="", proxy=None)
|
||||
endpoint = SimpleNamespace(id="endpoint-1", api_format="openai:chat", base_url="https://x")
|
||||
key = SimpleNamespace(id="key-1", proxy=None)
|
||||
candidate = SimpleNamespace(
|
||||
request_candidate_id="cand-1",
|
||||
mapping_matched_model=None,
|
||||
needs_conversion=False,
|
||||
output_limit=None,
|
||||
)
|
||||
|
||||
dummy_ctx = _DummyStreamResponseCtx()
|
||||
|
||||
async def _fake_execute_stream(self: object, plan: object) -> RustExecutorStreamResult:
|
||||
assert getattr(plan, "stream") is True
|
||||
return RustExecutorStreamResult(
|
||||
status_code=200,
|
||||
headers={"content-type": "text/event-stream"},
|
||||
byte_iterator=_iter_chunks(
|
||||
[
|
||||
b"data: {\"id\":\"chunk-1\"}\n\n",
|
||||
b"data: [DONE]\n\n",
|
||||
]
|
||||
),
|
||||
response_ctx=dummy_ctx,
|
||||
)
|
||||
|
||||
monkeypatch.setattr(chatmod.RustExecutorClient, "execute_stream", _fake_execute_stream)
|
||||
|
||||
stream = await handler._execute_stream_request(
|
||||
ctx,
|
||||
stream_processor,
|
||||
provider,
|
||||
endpoint,
|
||||
key,
|
||||
{"model": "gpt-test", "messages": [{"role": "user", "content": "hello"}]},
|
||||
{},
|
||||
candidate=candidate,
|
||||
)
|
||||
|
||||
received = [chunk async for chunk in stream]
|
||||
|
||||
assert received == [
|
||||
b"data: {\"id\":\"chunk-1\"}\n\n",
|
||||
b"data: [DONE]\n\n",
|
||||
]
|
||||
assert dummy_ctx.closed is True
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_stream_request_allows_tunnel_delegate_for_rust(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
_patch_stream_setup(
|
||||
monkeypatch,
|
||||
proxy_info={"node_id": "node-1", "node_name": "relay-node", "mode": "tunnel"},
|
||||
delegate_config={"tunnel": True, "node_id": "node-1"},
|
||||
)
|
||||
monkeypatch.setattr(chatmod.config, "executor_backend", "rust")
|
||||
|
||||
handler = _DummyChatHandler()
|
||||
stream_processor = _FakeStreamProcessor()
|
||||
ctx = StreamContext(model="gpt-test", api_format="openai:chat")
|
||||
ctx.client_api_format = "openai:chat"
|
||||
|
||||
provider = SimpleNamespace(
|
||||
name="provider",
|
||||
id="provider-1",
|
||||
provider_type="",
|
||||
proxy={"enabled": True, "node_id": "node-1"},
|
||||
)
|
||||
endpoint = SimpleNamespace(id="endpoint-1", api_format="openai:chat", base_url="https://x")
|
||||
key = SimpleNamespace(id="key-1", proxy=None)
|
||||
candidate = SimpleNamespace(
|
||||
request_candidate_id="cand-1",
|
||||
mapping_matched_model=None,
|
||||
needs_conversion=False,
|
||||
output_limit=None,
|
||||
)
|
||||
|
||||
dummy_ctx = _DummyStreamResponseCtx()
|
||||
|
||||
async def _fake_execute_stream(self: object, plan: object) -> RustExecutorStreamResult:
|
||||
assert getattr(plan, "proxy") is not None
|
||||
assert getattr(plan.proxy, "mode") == "tunnel"
|
||||
assert getattr(plan.proxy, "node_id") == "node-1"
|
||||
return RustExecutorStreamResult(
|
||||
status_code=200,
|
||||
headers={"content-type": "text/event-stream"},
|
||||
byte_iterator=_iter_chunks([b"data: [DONE]\n\n"]),
|
||||
response_ctx=dummy_ctx,
|
||||
)
|
||||
|
||||
monkeypatch.setattr(chatmod.RustExecutorClient, "execute_stream", _fake_execute_stream)
|
||||
|
||||
stream = await handler._execute_stream_request(
|
||||
ctx,
|
||||
stream_processor,
|
||||
provider,
|
||||
endpoint,
|
||||
key,
|
||||
{"model": "gpt-test", "messages": [{"role": "user", "content": "hello"}]},
|
||||
{},
|
||||
candidate=candidate,
|
||||
)
|
||||
|
||||
received = [chunk async for chunk in stream]
|
||||
|
||||
assert received == [b"data: [DONE]\n\n"]
|
||||
assert dummy_ctx.closed is True
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_stream_request_allows_tls_profile_for_rust(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
_patch_stream_setup(monkeypatch)
|
||||
monkeypatch.setattr(chatmod.config, "executor_backend", "rust")
|
||||
|
||||
handler = _DummyChatHandler()
|
||||
stream_processor = _FakeStreamProcessor()
|
||||
ctx = StreamContext(model="gpt-test", api_format="openai:chat")
|
||||
ctx.client_api_format = "openai:chat"
|
||||
|
||||
provider = SimpleNamespace(name="provider", id="provider-1", provider_type="", proxy=None)
|
||||
endpoint = SimpleNamespace(id="endpoint-1", api_format="openai:chat", base_url="https://x")
|
||||
key = SimpleNamespace(id="key-1", proxy=None)
|
||||
candidate = SimpleNamespace(
|
||||
request_candidate_id="cand-1",
|
||||
mapping_matched_model=None,
|
||||
needs_conversion=False,
|
||||
output_limit=None,
|
||||
)
|
||||
|
||||
dummy_ctx = _DummyStreamResponseCtx()
|
||||
|
||||
async def _fake_prepare_provider_request(self: object, **kwargs: Any) -> object:
|
||||
del self, kwargs
|
||||
return chatmod.ProviderRequestResult(
|
||||
request_body={"model": "gpt-test", "messages": [{"role": "user", "content": "hello"}]},
|
||||
url_model="gpt-test",
|
||||
mapped_model=None,
|
||||
envelope=None,
|
||||
extra_headers={},
|
||||
upstream_is_stream=True,
|
||||
needs_conversion=False,
|
||||
provider_api_format="openai:chat",
|
||||
client_api_format="openai:chat",
|
||||
auth_info=_DummyAuthInfo(),
|
||||
tls_profile="claude_code_nodejs",
|
||||
)
|
||||
|
||||
monkeypatch.setattr(
|
||||
_DummyChatHandler,
|
||||
"_prepare_provider_request",
|
||||
_fake_prepare_provider_request,
|
||||
)
|
||||
|
||||
async def _fake_execute_stream(self: object, plan: object) -> RustExecutorStreamResult:
|
||||
assert getattr(plan, "tls_profile") == "claude_code_nodejs"
|
||||
return RustExecutorStreamResult(
|
||||
status_code=200,
|
||||
headers={"content-type": "text/event-stream"},
|
||||
byte_iterator=_iter_chunks([b"data: [DONE]\n\n"]),
|
||||
response_ctx=dummy_ctx,
|
||||
)
|
||||
|
||||
monkeypatch.setattr(chatmod.RustExecutorClient, "execute_stream", _fake_execute_stream)
|
||||
|
||||
stream = await handler._execute_stream_request(
|
||||
ctx,
|
||||
stream_processor,
|
||||
provider,
|
||||
endpoint,
|
||||
key,
|
||||
{"model": "gpt-test", "messages": [{"role": "user", "content": "hello"}]},
|
||||
{},
|
||||
candidate=candidate,
|
||||
)
|
||||
|
||||
received = [chunk async for chunk in stream]
|
||||
|
||||
assert received == [b"data: [DONE]\n\n"]
|
||||
assert dummy_ctx.closed is True
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_stream_request_turns_rust_upstream_error_into_http_status_error(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
_patch_stream_setup(monkeypatch)
|
||||
monkeypatch.setattr(chatmod.config, "executor_backend", "rust")
|
||||
|
||||
handler = _DummyChatHandler()
|
||||
stream_processor = _FakeStreamProcessor()
|
||||
ctx = StreamContext(model="gpt-test", api_format="openai:chat")
|
||||
ctx.client_api_format = "openai:chat"
|
||||
|
||||
provider = SimpleNamespace(name="provider", id="provider-1", provider_type="", proxy=None)
|
||||
endpoint = SimpleNamespace(id="endpoint-1", api_format="openai:chat", base_url="https://x")
|
||||
key = SimpleNamespace(id="key-1", proxy=None)
|
||||
candidate = SimpleNamespace(
|
||||
request_candidate_id="cand-1",
|
||||
mapping_matched_model=None,
|
||||
needs_conversion=False,
|
||||
output_limit=None,
|
||||
)
|
||||
|
||||
dummy_ctx = _DummyStreamResponseCtx()
|
||||
|
||||
async def _fake_execute_stream(self: object, plan: object) -> RustExecutorStreamResult:
|
||||
assert getattr(plan, "stream") is True
|
||||
return RustExecutorStreamResult(
|
||||
status_code=429,
|
||||
headers={"content-type": "application/json"},
|
||||
byte_iterator=_iter_chunks([b'{"error":{"message":"slow down"}}']),
|
||||
response_ctx=dummy_ctx,
|
||||
)
|
||||
|
||||
monkeypatch.setattr(chatmod.RustExecutorClient, "execute_stream", _fake_execute_stream)
|
||||
|
||||
with pytest.raises(httpx.HTTPStatusError) as exc_info:
|
||||
await handler._execute_stream_request(
|
||||
ctx,
|
||||
stream_processor,
|
||||
provider,
|
||||
endpoint,
|
||||
key,
|
||||
{"model": "gpt-test", "messages": [{"role": "user", "content": "hello"}]},
|
||||
{},
|
||||
candidate=candidate,
|
||||
)
|
||||
|
||||
assert exc_info.value.response.status_code == 429
|
||||
assert "slow down" in exc_info.value.upstream_response # type: ignore[attr-defined]
|
||||
assert dummy_ctx.closed is True
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_execute_stream_request_falls_back_to_python_when_rust_unavailable(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
_patch_stream_setup(monkeypatch)
|
||||
monkeypatch.setattr(chatmod.config, "executor_backend", "rust")
|
||||
|
||||
handler = _DummyChatHandler()
|
||||
ctx = StreamContext(model="gpt-test", api_format="openai:chat")
|
||||
ctx.client_api_format = "openai:chat"
|
||||
|
||||
provider = SimpleNamespace(name="provider", id="provider-1", provider_type="", proxy=None)
|
||||
endpoint = SimpleNamespace(id="endpoint-1", api_format="openai:chat", base_url="https://x")
|
||||
key = SimpleNamespace(id="key-1", proxy=None)
|
||||
candidate = SimpleNamespace(
|
||||
request_candidate_id="cand-1",
|
||||
mapping_matched_model=None,
|
||||
needs_conversion=False,
|
||||
output_limit=None,
|
||||
)
|
||||
|
||||
async def _fake_execute_stream(self: object, plan: object) -> RustExecutorStreamResult:
|
||||
del plan
|
||||
raise RustExecutorClientError("executor down")
|
||||
|
||||
class _FakeHTTPClient:
|
||||
def stream(self, **kwargs: Any) -> Any:
|
||||
raise RuntimeError("local-http-client-used")
|
||||
|
||||
async def _fake_get_upstream_client(*args: Any, **kwargs: Any) -> _FakeHTTPClient:
|
||||
return _FakeHTTPClient()
|
||||
|
||||
monkeypatch.setattr(chatmod.RustExecutorClient, "execute_stream", _fake_execute_stream)
|
||||
monkeypatch.setattr(
|
||||
"src.clients.http_client.HTTPClientPool.get_upstream_client",
|
||||
_fake_get_upstream_client,
|
||||
)
|
||||
|
||||
with pytest.raises(RuntimeError) as exc_info:
|
||||
await handler._execute_stream_request(
|
||||
ctx,
|
||||
object(),
|
||||
provider,
|
||||
endpoint,
|
||||
key,
|
||||
{"model": "gpt-test", "messages": [{"role": "user", "content": "hello"}]},
|
||||
{},
|
||||
candidate=candidate,
|
||||
)
|
||||
|
||||
assert "local-http-client-used" in str(exc_info.value)
|
||||
427
tests/api/handlers/base/test_cli_rust_executor.py
Normal file
427
tests/api/handlers/base/test_cli_rust_executor.py
Normal file
@@ -0,0 +1,427 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import json
|
||||
from collections.abc import AsyncGenerator
|
||||
from types import SimpleNamespace
|
||||
from typing import Any
|
||||
|
||||
import pytest
|
||||
|
||||
import src.api.handlers.base.cli_stream_mixin as cli_stream_mod
|
||||
import src.api.handlers.base.cli_sync_mixin as cli_sync_mod
|
||||
import src.services.proxy_node.resolver as proxymod
|
||||
import src.services.task as taskmod
|
||||
from src.api.handlers.base.cli_stream_mixin import CliStreamMixin
|
||||
from src.api.handlers.base.cli_sync_mixin import CliSyncMixin
|
||||
from src.api.handlers.base.stream_context import StreamContext
|
||||
from src.services.request.rust_executor_client import (
|
||||
RustExecutorStreamResult,
|
||||
RustExecutorSyncResult,
|
||||
)
|
||||
|
||||
|
||||
class _DummyParser:
|
||||
def extract_usage_from_response(self, response: dict[str, Any]) -> dict[str, int]:
|
||||
del response
|
||||
return {
|
||||
"input_tokens": 0,
|
||||
"output_tokens": 0,
|
||||
"cache_read_tokens": 0,
|
||||
"cache_creation_tokens": 0,
|
||||
}
|
||||
|
||||
def extract_text_content(self, response: dict[str, Any]) -> str:
|
||||
return str(response.get("id") or "")
|
||||
|
||||
|
||||
class _DummyTelemetry:
|
||||
async def record_success(self, **kwargs: Any) -> int:
|
||||
del kwargs
|
||||
return 0
|
||||
|
||||
async def record_failure(self, **kwargs: Any) -> None:
|
||||
del kwargs
|
||||
|
||||
|
||||
class _DummySyncHandler(CliSyncMixin):
|
||||
FORMAT_ID = "openai:cli"
|
||||
|
||||
def __init__(self) -> None:
|
||||
self.db = None
|
||||
self.redis = None
|
||||
self.user = SimpleNamespace(id="user-1")
|
||||
self.api_key = SimpleNamespace(id="user-key-1")
|
||||
self.request_id = "req-cli-sync"
|
||||
self.client_ip = "127.0.0.1"
|
||||
self.user_agent = "pytest"
|
||||
self.start_time = 0.0
|
||||
self.allowed_api_formats = ["openai:cli"]
|
||||
self.primary_api_format = "openai:cli"
|
||||
self.api_family = None
|
||||
self.endpoint_kind = None
|
||||
self.telemetry = _DummyTelemetry()
|
||||
self.perf_metrics = None
|
||||
self._parser = _DummyParser()
|
||||
|
||||
@property
|
||||
def parser(self) -> _DummyParser:
|
||||
return self._parser
|
||||
|
||||
def _create_pending_usage(self, **kwargs: object) -> bool:
|
||||
del kwargs
|
||||
return True
|
||||
|
||||
def _build_request_metadata(self, http_request: Any | None = None) -> dict[str, Any]:
|
||||
del http_request
|
||||
return {}
|
||||
|
||||
def _merge_scheduling_metadata(
|
||||
self,
|
||||
request_metadata: dict[str, Any] | None,
|
||||
**kwargs: Any,
|
||||
) -> dict[str, Any]:
|
||||
del kwargs
|
||||
return dict(request_metadata or {})
|
||||
|
||||
def _resolve_capability_requirements(
|
||||
self,
|
||||
model_name: str,
|
||||
request_headers: dict[str, str] | None = None,
|
||||
request_body: dict[str, Any] | None = None,
|
||||
) -> dict[str, bool]:
|
||||
del model_name, request_headers, request_body
|
||||
return {}
|
||||
|
||||
async def _resolve_preferred_key_ids(
|
||||
self,
|
||||
model_name: str,
|
||||
request_body: dict[str, Any] | None = None,
|
||||
) -> list[str] | None:
|
||||
del model_name, request_body
|
||||
return None
|
||||
|
||||
def extract_model_from_request(
|
||||
self,
|
||||
request_body: dict[str, Any],
|
||||
path_params: dict[str, Any] | None = None,
|
||||
) -> str:
|
||||
del path_params
|
||||
return str(request_body.get("model") or "unknown")
|
||||
|
||||
async def _get_mapped_model(self, source_model: str, provider_id: str) -> str | None:
|
||||
del source_model, provider_id
|
||||
return None
|
||||
|
||||
async def _build_upstream_request(self, **kwargs: Any) -> Any:
|
||||
payload = dict(kwargs["request_body"])
|
||||
return SimpleNamespace(
|
||||
payload=payload,
|
||||
headers={"content-type": "application/json"},
|
||||
url="https://upstream.test/v1/responses",
|
||||
url_model=str(payload.get("model") or ""),
|
||||
envelope=None,
|
||||
upstream_is_stream=False,
|
||||
tls_profile=None,
|
||||
selected_base_url=None,
|
||||
)
|
||||
|
||||
def _extract_response_metadata(self, response_json: dict[str, Any]) -> dict[str, Any]:
|
||||
return {"id": response_json.get("id")}
|
||||
|
||||
|
||||
class _DummyStreamResponseCtx:
|
||||
def __init__(self) -> None:
|
||||
self.closed = False
|
||||
|
||||
async def __aexit__(self, exc_type: object, exc: object, tb: object) -> None:
|
||||
self.closed = True
|
||||
|
||||
|
||||
class _DummyCliStreamHandler(CliStreamMixin):
|
||||
FORMAT_ID = "openai:cli"
|
||||
|
||||
def __init__(self, *, upstream_is_stream: bool) -> None:
|
||||
self.request_id = "req-cli-stream"
|
||||
self.api_key = SimpleNamespace(id="user-key-1")
|
||||
self._upstream_is_stream = upstream_is_stream
|
||||
|
||||
async def _get_mapped_model(self, source_model: str, provider_id: str) -> str | None:
|
||||
del source_model, provider_id
|
||||
return None
|
||||
|
||||
async def _build_upstream_request(self, **kwargs: Any) -> Any:
|
||||
payload = dict(kwargs["request_body"])
|
||||
return SimpleNamespace(
|
||||
payload=payload,
|
||||
headers={"content-type": "application/json"},
|
||||
url="https://upstream.test/v1/responses",
|
||||
url_model=str(payload.get("model") or ""),
|
||||
envelope=None,
|
||||
upstream_is_stream=self._upstream_is_stream,
|
||||
tls_profile=None,
|
||||
selected_base_url=None,
|
||||
)
|
||||
|
||||
def apply_mapped_model(self, request_body: dict[str, Any], mapped_model: str) -> dict[str, Any]:
|
||||
out = dict(request_body)
|
||||
out["model"] = mapped_model
|
||||
return out
|
||||
|
||||
def _extract_response_metadata(self, response_json: dict[str, Any]) -> dict[str, Any]:
|
||||
return {"id": response_json.get("id")}
|
||||
|
||||
def _record_converted_chunks(self, ctx: Any, converted_events: Any) -> None:
|
||||
del ctx, converted_events
|
||||
|
||||
def _mark_first_output(self, ctx: Any, output_state: dict[str, Any]) -> None:
|
||||
del ctx
|
||||
output_state["first_yield"] = False
|
||||
|
||||
async def _prefetch_and_check_embedded_error(
|
||||
self,
|
||||
byte_iterator: Any,
|
||||
provider: Any,
|
||||
endpoint: Any,
|
||||
ctx: Any,
|
||||
) -> list[bytes]:
|
||||
del provider, endpoint, ctx
|
||||
first = await anext(byte_iterator)
|
||||
return [first]
|
||||
|
||||
async def _create_response_stream_with_prefetch(
|
||||
self,
|
||||
ctx: Any,
|
||||
byte_iterator: Any,
|
||||
response_ctx: _DummyStreamResponseCtx,
|
||||
prefetched_chunks: list[bytes],
|
||||
) -> AsyncGenerator[bytes]:
|
||||
del ctx
|
||||
|
||||
async def _gen() -> AsyncGenerator[bytes]:
|
||||
try:
|
||||
for chunk in prefetched_chunks:
|
||||
yield chunk
|
||||
async for chunk in byte_iterator:
|
||||
yield chunk
|
||||
finally:
|
||||
await response_ctx.__aexit__(None, None, None)
|
||||
|
||||
return _gen()
|
||||
|
||||
|
||||
async def _iter_chunks(chunks: list[bytes]) -> AsyncGenerator[bytes]:
|
||||
for chunk in chunks:
|
||||
yield chunk
|
||||
|
||||
|
||||
def _patch_proxy_resolver(monkeypatch: pytest.MonkeyPatch) -> None:
|
||||
monkeypatch.setattr(proxymod, "resolve_effective_proxy", lambda provider_proxy, key_proxy=None: None)
|
||||
monkeypatch.setattr(proxymod, "get_proxy_label", lambda proxy_info: "direct")
|
||||
|
||||
async def _fake_resolve_proxy_info(proxy_config: Any) -> Any:
|
||||
del proxy_config
|
||||
return None
|
||||
|
||||
async def _fake_resolve_delegate(proxy_config: Any) -> Any:
|
||||
del proxy_config
|
||||
return None
|
||||
|
||||
async def _fake_build_proxy_url(proxy_config: Any) -> Any:
|
||||
del proxy_config
|
||||
return None
|
||||
|
||||
monkeypatch.setattr(proxymod, "resolve_proxy_info_async", _fake_resolve_proxy_info)
|
||||
monkeypatch.setattr(proxymod, "resolve_delegate_config_async", _fake_resolve_delegate)
|
||||
monkeypatch.setattr(proxymod, "build_proxy_url_async", _fake_build_proxy_url)
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_cli_process_sync_uses_rust_executor_when_available(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
handler = _DummySyncHandler()
|
||||
monkeypatch.setattr(cli_sync_mod.config, "executor_backend", "rust")
|
||||
_patch_proxy_resolver(monkeypatch)
|
||||
|
||||
class _FakeTaskService:
|
||||
def __init__(self, db: Any, redis: Any) -> None:
|
||||
del db, redis
|
||||
|
||||
async def execute(self, **kwargs: Any) -> Any:
|
||||
candidate = SimpleNamespace(
|
||||
request_candidate_id="cand-1",
|
||||
mapping_matched_model=None,
|
||||
needs_conversion=False,
|
||||
output_limit=None,
|
||||
)
|
||||
provider = SimpleNamespace(
|
||||
name="provider",
|
||||
id="provider-1",
|
||||
provider_type="",
|
||||
proxy=None,
|
||||
request_timeout=None,
|
||||
stream_first_byte_timeout=None,
|
||||
)
|
||||
endpoint = SimpleNamespace(id="endpoint-1", api_format="openai:cli")
|
||||
key = SimpleNamespace(id="key-1", api_key="sk-test", proxy=None)
|
||||
response = await kwargs["request_func"](provider, endpoint, key, candidate)
|
||||
return SimpleNamespace(
|
||||
response=response,
|
||||
provider_name="provider",
|
||||
request_candidate_id="cand-1",
|
||||
provider_id="provider-1",
|
||||
endpoint_id="endpoint-1",
|
||||
key_id="key-1",
|
||||
pool_summary=None,
|
||||
)
|
||||
|
||||
async def _fake_execute_sync_json(self: object, plan: object) -> RustExecutorSyncResult:
|
||||
assert getattr(plan, "provider_api_format") == "openai:cli"
|
||||
return RustExecutorSyncResult(
|
||||
status_code=200,
|
||||
response_json={"id": "resp-rust-cli"},
|
||||
headers={"content-type": "application/json"},
|
||||
)
|
||||
|
||||
monkeypatch.setattr(taskmod, "TaskService", _FakeTaskService)
|
||||
monkeypatch.setattr(
|
||||
cli_sync_mod.RustExecutorClient,
|
||||
"execute_sync_json",
|
||||
_fake_execute_sync_json,
|
||||
)
|
||||
|
||||
response = await handler.process_sync(
|
||||
original_request_body={"model": "gpt-4.1", "input": "hello"},
|
||||
original_headers={},
|
||||
)
|
||||
|
||||
assert response.status_code == 200
|
||||
assert json.loads(response.body) == {"id": "resp-rust-cli"}
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_cli_execute_stream_request_uses_rust_sync_bridge(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
handler = _DummyCliStreamHandler(upstream_is_stream=False)
|
||||
ctx = StreamContext(model="gpt-test", api_format="openai:cli")
|
||||
ctx.client_api_format = "openai:cli"
|
||||
|
||||
monkeypatch.setattr(cli_stream_mod.config, "executor_backend", "rust")
|
||||
_patch_proxy_resolver(monkeypatch)
|
||||
|
||||
async def _fake_execute_sync_json(self: object, plan: object) -> RustExecutorSyncResult:
|
||||
assert getattr(plan, "stream") is False
|
||||
return RustExecutorSyncResult(
|
||||
status_code=200,
|
||||
response_json={"id": "sync-bridge-rust"},
|
||||
headers={"content-type": "application/json"},
|
||||
)
|
||||
|
||||
async def _fake_streamify(**kwargs: Any) -> AsyncGenerator[bytes]:
|
||||
assert kwargs["response_json"] == {"id": "sync-bridge-rust"}
|
||||
yield b"data: cli-bridge\n\n"
|
||||
|
||||
monkeypatch.setattr(
|
||||
cli_stream_mod.RustExecutorClient,
|
||||
"execute_sync_json",
|
||||
_fake_execute_sync_json,
|
||||
)
|
||||
monkeypatch.setattr(handler, "_streamify_sync_response", _fake_streamify)
|
||||
|
||||
provider = SimpleNamespace(
|
||||
name="provider",
|
||||
id="provider-1",
|
||||
provider_type="",
|
||||
proxy=None,
|
||||
request_timeout=None,
|
||||
stream_first_byte_timeout=None,
|
||||
)
|
||||
endpoint = SimpleNamespace(id="endpoint-1", api_format="openai:cli", base_url="https://x")
|
||||
key = SimpleNamespace(id="key-1", proxy=None, auth_type="", api_key="sk-test")
|
||||
candidate = SimpleNamespace(
|
||||
request_candidate_id="cand-1",
|
||||
mapping_matched_model=None,
|
||||
needs_conversion=False,
|
||||
output_limit=None,
|
||||
)
|
||||
|
||||
stream = await handler._execute_stream_request(
|
||||
ctx,
|
||||
provider,
|
||||
endpoint,
|
||||
key,
|
||||
{"model": "gpt-test", "input": "hello"},
|
||||
{},
|
||||
candidate=candidate,
|
||||
)
|
||||
chunks = [chunk async for chunk in stream]
|
||||
|
||||
assert chunks == [b"data: cli-bridge\n\n"]
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_cli_execute_stream_request_uses_rust_native_stream(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
handler = _DummyCliStreamHandler(upstream_is_stream=True)
|
||||
ctx = StreamContext(model="gpt-test", api_format="openai:cli")
|
||||
ctx.client_api_format = "openai:cli"
|
||||
|
||||
monkeypatch.setattr(cli_stream_mod.config, "executor_backend", "rust")
|
||||
_patch_proxy_resolver(monkeypatch)
|
||||
|
||||
async def _fake_execute_stream(self: object, plan: object) -> RustExecutorStreamResult:
|
||||
assert getattr(plan, "stream") is True
|
||||
return RustExecutorStreamResult(
|
||||
status_code=200,
|
||||
headers={"content-type": "text/event-stream", "x-upstream-test": "true"},
|
||||
byte_iterator=_iter_chunks(
|
||||
[
|
||||
b"data: {\"id\":\"chunk-1\"}\n\n",
|
||||
b"data: [DONE]\n\n",
|
||||
]
|
||||
),
|
||||
response_ctx=_DummyStreamResponseCtx(),
|
||||
)
|
||||
|
||||
monkeypatch.setattr(
|
||||
cli_stream_mod.RustExecutorClient,
|
||||
"execute_stream",
|
||||
_fake_execute_stream,
|
||||
)
|
||||
|
||||
provider = SimpleNamespace(
|
||||
name="provider",
|
||||
id="provider-1",
|
||||
provider_type="",
|
||||
proxy=None,
|
||||
request_timeout=None,
|
||||
stream_first_byte_timeout=None,
|
||||
)
|
||||
endpoint = SimpleNamespace(id="endpoint-1", api_format="openai:cli", base_url="https://x")
|
||||
key = SimpleNamespace(id="key-1", proxy=None, api_key="sk-test")
|
||||
candidate = SimpleNamespace(
|
||||
request_candidate_id="cand-1",
|
||||
mapping_matched_model=None,
|
||||
needs_conversion=False,
|
||||
output_limit=None,
|
||||
)
|
||||
|
||||
stream = await handler._execute_stream_request(
|
||||
ctx,
|
||||
provider,
|
||||
endpoint,
|
||||
key,
|
||||
{"model": "gpt-test", "input": "hello"},
|
||||
{},
|
||||
candidate=candidate,
|
||||
)
|
||||
chunks = [chunk async for chunk in stream]
|
||||
|
||||
assert chunks == [
|
||||
b"data: {\"id\":\"chunk-1\"}\n\n",
|
||||
b"data: [DONE]\n\n",
|
||||
]
|
||||
assert ctx.status_code == 200
|
||||
assert ctx.response_headers["x-upstream-test"] == "true"
|
||||
168
tests/api/handlers/base/test_endpoint_checker_rust.py
Normal file
168
tests/api/handlers/base/test_endpoint_checker_rust.py
Normal file
@@ -0,0 +1,168 @@
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
from unittest.mock import AsyncMock
|
||||
|
||||
import pytest
|
||||
|
||||
from src.api.handlers.base.endpoint_checker import EndpointCheckRequest, HttpRequestExecutor
|
||||
from src.services.request.rust_executor_client import (
|
||||
RustExecutorStreamResult,
|
||||
RustExecutorSyncResult,
|
||||
)
|
||||
|
||||
|
||||
class _DummyStreamContext:
|
||||
def __init__(self) -> None:
|
||||
self.closed = False
|
||||
|
||||
async def __aexit__(self, exc_type: object, exc: object, tb: object) -> None:
|
||||
self.closed = True
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_endpoint_checker_sync_prefers_rust_executor(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
from src.api.handlers.base import endpoint_checker as mod
|
||||
from src.services.request import rust_executor_client as rust_mod
|
||||
|
||||
monkeypatch.setattr(mod.config, "executor_backend", "rust")
|
||||
|
||||
executor = HttpRequestExecutor(timeout=15.0)
|
||||
proxy_snapshot = object()
|
||||
monkeypatch.setattr(
|
||||
executor,
|
||||
"_build_rust_proxy_snapshot",
|
||||
AsyncMock(return_value=proxy_snapshot),
|
||||
)
|
||||
|
||||
captured: dict[str, Any] = {}
|
||||
|
||||
async def _fake_execute_sync_json(
|
||||
self: object,
|
||||
plan: Any,
|
||||
) -> RustExecutorSyncResult:
|
||||
captured["plan"] = plan
|
||||
return RustExecutorSyncResult(
|
||||
status_code=200,
|
||||
response_json={"id": "resp_1", "usage": {"prompt_tokens": 1, "completion_tokens": 2}},
|
||||
headers={"content-type": "application/json"},
|
||||
)
|
||||
|
||||
monkeypatch.setattr(rust_mod.RustExecutorClient, "execute_sync_json", _fake_execute_sync_json)
|
||||
|
||||
result = await executor.execute(
|
||||
EndpointCheckRequest(
|
||||
url="https://upstream.test/v1/chat/completions",
|
||||
headers={"authorization": "Bearer test"},
|
||||
json_body={"model": "gpt-test", "messages": [{"role": "user", "content": "hi"}]},
|
||||
api_format="openai:chat",
|
||||
provider_name="openai",
|
||||
model_name="gpt-test",
|
||||
api_key_id="key_1",
|
||||
provider_id="provider_1",
|
||||
)
|
||||
)
|
||||
|
||||
assert result.status_code == 200
|
||||
assert result.response_data == {
|
||||
"id": "resp_1",
|
||||
"usage": {"prompt_tokens": 1, "completion_tokens": 2},
|
||||
}
|
||||
assert captured["plan"].proxy is proxy_snapshot
|
||||
assert captured["plan"].method == "POST"
|
||||
assert captured["plan"].url == "https://upstream.test/v1/chat/completions"
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_endpoint_checker_stream_prefers_rust_executor(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
from src.api.handlers.base import endpoint_checker as mod
|
||||
from src.services.request import rust_executor_client as rust_mod
|
||||
|
||||
monkeypatch.setattr(mod.config, "executor_backend", "rust")
|
||||
|
||||
executor = HttpRequestExecutor(timeout=15.0)
|
||||
monkeypatch.setattr(
|
||||
executor,
|
||||
"_build_rust_proxy_snapshot",
|
||||
AsyncMock(return_value=None),
|
||||
)
|
||||
|
||||
async def _byte_iter() -> Any:
|
||||
yield b'data: {"choices":[{"delta":{"content":"Hel'
|
||||
yield b'lo"}}]}\n\n'
|
||||
yield b'data: {"choices":[{"delta":{"content":" world"},"finish_reason":"stop"}]}\n\n'
|
||||
|
||||
stream_ctx = _DummyStreamContext()
|
||||
|
||||
async def _fake_execute_stream(self: object, plan: Any) -> RustExecutorStreamResult:
|
||||
return RustExecutorStreamResult(
|
||||
status_code=200,
|
||||
headers={"content-type": "text/event-stream"},
|
||||
byte_iterator=_byte_iter(),
|
||||
response_ctx=stream_ctx,
|
||||
)
|
||||
|
||||
monkeypatch.setattr(rust_mod.RustExecutorClient, "execute_stream", _fake_execute_stream)
|
||||
|
||||
result = await executor.execute(
|
||||
EndpointCheckRequest(
|
||||
url="https://upstream.test/v1/chat/completions",
|
||||
headers={"authorization": "Bearer test"},
|
||||
json_body={
|
||||
"model": "gpt-test",
|
||||
"messages": [{"role": "user", "content": "hi"}],
|
||||
"stream": True,
|
||||
},
|
||||
api_format="openai:chat",
|
||||
provider_name="openai",
|
||||
model_name="gpt-test",
|
||||
is_stream=True,
|
||||
)
|
||||
)
|
||||
|
||||
assert result.status_code == 200
|
||||
assert result.response_data == {
|
||||
"choices": [{"delta": {"content": " world"}, "finish_reason": "stop"}]
|
||||
}
|
||||
assert stream_ctx.closed is True
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_endpoint_checker_proxy_snapshot_falls_back_to_system_proxy(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
from src.services.proxy_node import resolver as resolver_mod
|
||||
|
||||
executor = HttpRequestExecutor()
|
||||
|
||||
monkeypatch.setattr(
|
||||
resolver_mod,
|
||||
"get_system_proxy_config_async",
|
||||
AsyncMock(return_value={"enabled": True, "url": "http://system-proxy.test:8080"}),
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
resolver_mod,
|
||||
"resolve_delegate_config_async",
|
||||
AsyncMock(return_value=None),
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
resolver_mod,
|
||||
"build_proxy_url_async",
|
||||
AsyncMock(return_value="http://system-proxy.test:8080"),
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
resolver_mod,
|
||||
"resolve_proxy_info_async",
|
||||
AsyncMock(return_value={"mode": "http", "label": "system-proxy"}),
|
||||
)
|
||||
|
||||
snapshot = await executor._build_rust_proxy_snapshot(None)
|
||||
|
||||
assert snapshot is not None
|
||||
assert snapshot.enabled is True
|
||||
assert snapshot.url == "http://system-proxy.test:8080"
|
||||
assert snapshot.mode == "http"
|
||||
226
tests/api/handlers/gemini/test_video_rust_handler.py
Normal file
226
tests/api/handlers/gemini/test_video_rust_handler.py
Normal file
@@ -0,0 +1,226 @@
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import AsyncGenerator
|
||||
from datetime import datetime, timezone
|
||||
from types import SimpleNamespace
|
||||
from unittest.mock import AsyncMock
|
||||
|
||||
import httpx
|
||||
import pytest
|
||||
from fastapi.responses import JSONResponse, StreamingResponse
|
||||
|
||||
import src.api.handlers.gemini.video_handler as video_mod
|
||||
import src.services.proxy_node.resolver as resolver_mod
|
||||
import src.services.request.rust_executor_client as rust_client_mod
|
||||
from src.api.handlers.gemini.video_handler import GeminiVeoHandler
|
||||
from src.core.api_format.conversion.internal_video import VideoStatus
|
||||
from src.services.request.rust_executor_client import RustExecutorStreamResult
|
||||
|
||||
|
||||
class _DummyStreamResponseCtx:
|
||||
def __init__(self) -> None:
|
||||
self.closed = False
|
||||
|
||||
async def __aexit__(self, exc_type: object, exc: object, tb: object) -> None:
|
||||
self.closed = True
|
||||
|
||||
|
||||
async def _iter_chunks(chunks: list[bytes]) -> AsyncGenerator[bytes]:
|
||||
for chunk in chunks:
|
||||
yield chunk
|
||||
|
||||
|
||||
def _make_handler() -> GeminiVeoHandler:
|
||||
return GeminiVeoHandler(
|
||||
db=SimpleNamespace(),
|
||||
user=SimpleNamespace(id="user-1"),
|
||||
api_key=SimpleNamespace(id="api-key-1"),
|
||||
request_id="req-gemini-video-test",
|
||||
client_ip="127.0.0.1",
|
||||
user_agent="pytest",
|
||||
start_time=0.0,
|
||||
)
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_handle_create_task_uses_rust_sync_helper(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
handler = _make_handler()
|
||||
monkeypatch.setattr(video_mod.UsageService, "create_pending_usage", lambda *args, **kwargs: None)
|
||||
monkeypatch.setattr(
|
||||
handler._normalizer,
|
||||
"video_request_to_internal",
|
||||
lambda body: SimpleNamespace(model=str(body.get("model") or "veo-3")),
|
||||
)
|
||||
|
||||
candidate = SimpleNamespace(provider=SimpleNamespace(name="provider-1", id="prov-1"))
|
||||
endpoint = SimpleNamespace(
|
||||
id="ep-1",
|
||||
provider_id="prov-1",
|
||||
api_family="gemini",
|
||||
endpoint_kind="video",
|
||||
base_url="https://generativelanguage.googleapis.com",
|
||||
body_rules=None,
|
||||
)
|
||||
provider_key = SimpleNamespace(id="key-1")
|
||||
|
||||
monkeypatch.setattr(
|
||||
handler,
|
||||
"_resolve_upstream_key",
|
||||
AsyncMock(return_value=("upstream-key", endpoint, provider_key, None)),
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
handler,
|
||||
"_build_upstream_url",
|
||||
lambda base_url, model: f"https://generativelanguage.googleapis.com/v1beta/models/{model}:predictLongRunning",
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
handler,
|
||||
"_build_upstream_headers",
|
||||
lambda original_headers, upstream_key, endpoint, auth_info, **kwargs: {
|
||||
"x-goog-api-key": upstream_key
|
||||
},
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
video_mod.HTTPClientPool,
|
||||
"get_default_client_async",
|
||||
AsyncMock(side_effect=AssertionError("python fallback should not run")),
|
||||
)
|
||||
|
||||
async def _fake_rust_sync(**kwargs: object) -> httpx.Response:
|
||||
assert kwargs["method"] == "POST"
|
||||
assert kwargs["provider_id"] == "prov-1"
|
||||
assert kwargs["endpoint_id"] == "ep-1"
|
||||
assert kwargs["key_id"] == "key-1"
|
||||
assert kwargs["body"] == {"model": "veo-3", "prompt": "hello"}
|
||||
return httpx.Response(
|
||||
200,
|
||||
request=httpx.Request("POST", str(kwargs["url"])),
|
||||
json={"name": "operations/ext-1"},
|
||||
)
|
||||
|
||||
monkeypatch.setattr(handler, "_try_rust_sync_http_response", _fake_rust_sync)
|
||||
|
||||
async def _fake_submit_with_failover(**kwargs: object) -> JSONResponse:
|
||||
response = await kwargs["submit_func"](candidate)
|
||||
assert response.status_code == 200
|
||||
assert response.json()["name"] == "operations/ext-1"
|
||||
return JSONResponse(status_code=400, content={"error": {"message": "stop"}})
|
||||
|
||||
monkeypatch.setattr(handler, "_submit_with_failover", _fake_submit_with_failover)
|
||||
|
||||
response = await handler.handle_create_task(
|
||||
http_request=SimpleNamespace(
|
||||
headers={},
|
||||
url=SimpleNamespace(scheme="https", netloc="example.com"),
|
||||
),
|
||||
original_headers={},
|
||||
original_request_body={"model": "veo-3", "prompt": "hello"},
|
||||
)
|
||||
|
||||
assert response.status_code == 400
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_handle_download_content_uses_rust_executor_with_proxy_snapshot(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
monkeypatch.setattr(video_mod.config, "executor_backend", "rust")
|
||||
handler = _make_handler()
|
||||
dummy_ctx = _DummyStreamResponseCtx()
|
||||
|
||||
monkeypatch.setattr(
|
||||
handler,
|
||||
"_get_task_by_external_id",
|
||||
lambda task_id: SimpleNamespace(
|
||||
id=task_id,
|
||||
status=VideoStatus.COMPLETED.value,
|
||||
video_url="https://storage.example.com/video.mp4",
|
||||
video_expires_at=datetime.now(timezone.utc).replace(year=2099),
|
||||
model="veo-3",
|
||||
),
|
||||
)
|
||||
|
||||
endpoint = SimpleNamespace(
|
||||
id="ep-1",
|
||||
provider_id="prov-1",
|
||||
api_family="gemini",
|
||||
endpoint_kind="video",
|
||||
base_url="https://generativelanguage.googleapis.com",
|
||||
proxy={"enabled": True, "url": "http://proxy.local:8080"},
|
||||
)
|
||||
key = SimpleNamespace(id="key-1", api_key="encrypted", proxy=None)
|
||||
monkeypatch.setattr(handler, "_get_endpoint_and_key", lambda task: (endpoint, key))
|
||||
monkeypatch.setattr(video_mod.crypto_service, "decrypt", lambda _: "upstream-key")
|
||||
monkeypatch.setattr(
|
||||
video_mod,
|
||||
"get_provider_auth",
|
||||
AsyncMock(return_value=None),
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
video_mod,
|
||||
"resolve_provider_proxy",
|
||||
lambda endpoint, key: {"enabled": True, "url": "http://proxy.local:8080"},
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
resolver_mod,
|
||||
"resolve_effective_proxy",
|
||||
lambda provider_proxy, key_proxy=None: provider_proxy,
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
resolver_mod,
|
||||
"get_system_proxy_config_async",
|
||||
AsyncMock(return_value=None),
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
resolver_mod,
|
||||
"resolve_delegate_config_async",
|
||||
AsyncMock(return_value=None),
|
||||
)
|
||||
|
||||
async def _fake_build_proxy_url_async(proxy_config: object) -> str:
|
||||
assert proxy_config == {"enabled": True, "url": "http://proxy.local:8080"}
|
||||
return "http://proxy.local:8080"
|
||||
|
||||
async def _fake_resolve_proxy_info_async(proxy_config: object) -> dict[str, str]:
|
||||
assert proxy_config == {"enabled": True, "url": "http://proxy.local:8080"}
|
||||
return {"url": "http://proxy.local:8080"}
|
||||
|
||||
monkeypatch.setattr(resolver_mod, "build_proxy_url_async", _fake_build_proxy_url_async)
|
||||
monkeypatch.setattr(
|
||||
resolver_mod,
|
||||
"resolve_proxy_info_async",
|
||||
_fake_resolve_proxy_info_async,
|
||||
)
|
||||
|
||||
async def _fake_execute_stream(self: object, plan: object) -> RustExecutorStreamResult:
|
||||
assert getattr(plan, "method") == "GET"
|
||||
assert getattr(plan, "url") == "https://storage.example.com/video.mp4"
|
||||
assert getattr(plan, "headers") == {"x-goog-api-key": "upstream-key"}
|
||||
assert getattr(plan, "proxy").url == "http://proxy.local:8080"
|
||||
return RustExecutorStreamResult(
|
||||
status_code=200,
|
||||
headers={"content-type": "video/mp4", "x-rust-download": "true"},
|
||||
byte_iterator=_iter_chunks([b"gemini-", b"video"]),
|
||||
response_ctx=dummy_ctx,
|
||||
)
|
||||
|
||||
monkeypatch.setattr(rust_client_mod.RustExecutorClient, "execute_stream", _fake_execute_stream)
|
||||
monkeypatch.setattr(
|
||||
video_mod.HTTPClientPool,
|
||||
"get_default_client_async",
|
||||
AsyncMock(side_effect=AssertionError("python fallback should not run")),
|
||||
)
|
||||
|
||||
response = await handler.handle_download_content(
|
||||
task_id="operations/ext-1",
|
||||
http_request=SimpleNamespace(),
|
||||
original_headers={},
|
||||
query_params=None,
|
||||
)
|
||||
|
||||
assert isinstance(response, StreamingResponse)
|
||||
body = b"".join([chunk async for chunk in response.body_iterator])
|
||||
assert body == b"gemini-video"
|
||||
assert dummy_ctx.closed is True
|
||||
198
tests/api/handlers/openai/test_video_rust_download.py
Normal file
198
tests/api/handlers/openai/test_video_rust_download.py
Normal file
@@ -0,0 +1,198 @@
|
||||
from __future__ import annotations
|
||||
|
||||
from collections.abc import AsyncGenerator
|
||||
from types import SimpleNamespace
|
||||
from typing import Any
|
||||
|
||||
import pytest
|
||||
from fastapi.responses import Response, StreamingResponse
|
||||
|
||||
import src.api.handlers.openai.video_handler as video_mod
|
||||
from src.api.handlers.openai.video_handler import OpenAIVideoHandler
|
||||
from src.core.api_format.conversion.internal_video import VideoStatus
|
||||
from src.services.request.rust_executor_client import (
|
||||
RustExecutorClientError,
|
||||
RustExecutorStreamResult,
|
||||
)
|
||||
|
||||
|
||||
class _DummyStreamResponseCtx:
|
||||
def __init__(self) -> None:
|
||||
self.closed = False
|
||||
|
||||
async def __aexit__(self, exc_type: object, exc: object, tb: object) -> None:
|
||||
self.closed = True
|
||||
|
||||
|
||||
async def _iter_chunks(chunks: list[bytes]) -> AsyncGenerator[bytes]:
|
||||
for chunk in chunks:
|
||||
yield chunk
|
||||
|
||||
|
||||
def _make_handler() -> OpenAIVideoHandler:
|
||||
return OpenAIVideoHandler(
|
||||
db=SimpleNamespace(),
|
||||
user=SimpleNamespace(id="user-1"),
|
||||
api_key=SimpleNamespace(id="api-key-1"),
|
||||
request_id="req-video-test",
|
||||
client_ip="127.0.0.1",
|
||||
user_agent="pytest",
|
||||
start_time=0.0,
|
||||
)
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_handle_download_content_uses_rust_executor_for_direct_video_url(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
monkeypatch.setattr(video_mod.config, "executor_backend", "rust")
|
||||
handler = _make_handler()
|
||||
dummy_ctx = _DummyStreamResponseCtx()
|
||||
|
||||
monkeypatch.setattr(
|
||||
handler,
|
||||
"_get_task",
|
||||
lambda task_id: SimpleNamespace(
|
||||
id=task_id,
|
||||
status=VideoStatus.COMPLETED.value,
|
||||
video_url="https://cdn.example.com/video.mp4",
|
||||
model="sora-2",
|
||||
),
|
||||
)
|
||||
|
||||
async def _fake_execute_stream(self: object, plan: object) -> RustExecutorStreamResult:
|
||||
assert getattr(plan, "method") == "GET"
|
||||
assert getattr(plan, "url") == "https://cdn.example.com/video.mp4"
|
||||
assert getattr(plan, "body").json_body is None
|
||||
assert getattr(plan, "body").body_bytes_b64 is None
|
||||
return RustExecutorStreamResult(
|
||||
status_code=200,
|
||||
headers={"content-type": "video/mp4", "x-rust-download": "true"},
|
||||
byte_iterator=_iter_chunks([b"video-", b"bytes"]),
|
||||
response_ctx=dummy_ctx,
|
||||
)
|
||||
|
||||
monkeypatch.setattr(video_mod.RustExecutorClient, "execute_stream", _fake_execute_stream)
|
||||
|
||||
response = await handler.handle_download_content(
|
||||
task_id="task-1",
|
||||
http_request=SimpleNamespace(),
|
||||
original_headers={},
|
||||
query_params={"variant": "video"},
|
||||
)
|
||||
|
||||
assert isinstance(response, StreamingResponse)
|
||||
assert response.headers["x-rust-download"] == "true"
|
||||
body = b"".join([chunk async for chunk in response.body_iterator])
|
||||
assert body == b"video-bytes"
|
||||
assert dummy_ctx.closed is True
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_handle_download_content_uses_rust_executor_for_upstream_content_endpoint(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
monkeypatch.setattr(video_mod.config, "executor_backend", "rust")
|
||||
handler = _make_handler()
|
||||
dummy_ctx = _DummyStreamResponseCtx()
|
||||
|
||||
monkeypatch.setattr(
|
||||
handler,
|
||||
"_get_task",
|
||||
lambda task_id: SimpleNamespace(
|
||||
id=task_id,
|
||||
status=VideoStatus.COMPLETED.value,
|
||||
video_url=None,
|
||||
external_task_id="ext-1",
|
||||
model="sora-2",
|
||||
),
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
handler,
|
||||
"_get_endpoint_and_key",
|
||||
lambda task: (
|
||||
SimpleNamespace(id="ep-1", provider_id="prov-1", base_url="https://api.openai.com"),
|
||||
SimpleNamespace(id="key-1", api_key="encrypted"),
|
||||
),
|
||||
)
|
||||
monkeypatch.setattr(video_mod.crypto_service, "decrypt", lambda _: "upstream-key")
|
||||
monkeypatch.setattr(
|
||||
handler,
|
||||
"_build_upstream_url",
|
||||
lambda base_url, suffix=None: "https://api.openai.com/v1/videos/ext-1/content",
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
handler,
|
||||
"_build_upstream_headers",
|
||||
lambda original_headers, upstream_key, endpoint: {"authorization": f"Bearer {upstream_key}"},
|
||||
)
|
||||
|
||||
async def _fake_execute_stream(self: object, plan: object) -> RustExecutorStreamResult:
|
||||
assert getattr(plan, "method") == "GET"
|
||||
assert getattr(plan, "url") == "https://api.openai.com/v1/videos/ext-1/content"
|
||||
assert getattr(plan, "headers") == {"authorization": "Bearer upstream-key"}
|
||||
assert getattr(plan, "provider_id") == "prov-1"
|
||||
assert getattr(plan, "endpoint_id") == "ep-1"
|
||||
assert getattr(plan, "key_id") == "key-1"
|
||||
return RustExecutorStreamResult(
|
||||
status_code=200,
|
||||
headers={"content-type": "video/mp4", "x-rust-download": "true"},
|
||||
byte_iterator=_iter_chunks([b"upstream-", b"video"]),
|
||||
response_ctx=dummy_ctx,
|
||||
)
|
||||
|
||||
monkeypatch.setattr(video_mod.RustExecutorClient, "execute_stream", _fake_execute_stream)
|
||||
|
||||
response = await handler.handle_download_content(
|
||||
task_id="task-1",
|
||||
http_request=SimpleNamespace(),
|
||||
original_headers={},
|
||||
query_params=None,
|
||||
)
|
||||
|
||||
assert isinstance(response, StreamingResponse)
|
||||
body = b"".join([chunk async for chunk in response.body_iterator])
|
||||
assert body == b"upstream-video"
|
||||
assert dummy_ctx.closed is True
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_handle_download_content_falls_back_to_python_proxy_when_rust_unavailable(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
monkeypatch.setattr(video_mod.config, "executor_backend", "rust")
|
||||
handler = _make_handler()
|
||||
|
||||
monkeypatch.setattr(
|
||||
handler,
|
||||
"_get_task",
|
||||
lambda task_id: SimpleNamespace(
|
||||
id=task_id,
|
||||
status=VideoStatus.COMPLETED.value,
|
||||
video_url="https://cdn.example.com/video.mp4",
|
||||
model="sora-2",
|
||||
),
|
||||
)
|
||||
|
||||
async def _failing_execute_stream(self: object, plan: object) -> RustExecutorStreamResult:
|
||||
del self, plan
|
||||
raise RustExecutorClientError("executor down")
|
||||
|
||||
fallback_response = Response(content=b"python-fallback", media_type="video/mp4")
|
||||
|
||||
async def _fake_proxy_direct_url(url: str, task_id: str) -> Response:
|
||||
assert url == "https://cdn.example.com/video.mp4"
|
||||
assert task_id == "task-1"
|
||||
return fallback_response
|
||||
|
||||
monkeypatch.setattr(video_mod.RustExecutorClient, "execute_stream", _failing_execute_stream)
|
||||
monkeypatch.setattr(handler, "_proxy_direct_url", _fake_proxy_direct_url)
|
||||
|
||||
response = await handler.handle_download_content(
|
||||
task_id="task-1",
|
||||
http_request=SimpleNamespace(),
|
||||
original_headers={},
|
||||
query_params={"variant": "video"},
|
||||
)
|
||||
|
||||
assert response is fallback_response
|
||||
103
tests/api/handlers/openai/test_video_rust_sync.py
Normal file
103
tests/api/handlers/openai/test_video_rust_sync.py
Normal file
@@ -0,0 +1,103 @@
|
||||
from __future__ import annotations
|
||||
|
||||
from types import SimpleNamespace
|
||||
from unittest.mock import AsyncMock
|
||||
|
||||
import httpx
|
||||
import pytest
|
||||
from fastapi.responses import JSONResponse
|
||||
|
||||
import src.api.handlers.openai.video_handler as video_mod
|
||||
from src.api.handlers.openai.video_handler import OpenAIVideoHandler
|
||||
|
||||
|
||||
def _make_handler() -> OpenAIVideoHandler:
|
||||
return OpenAIVideoHandler(
|
||||
db=SimpleNamespace(),
|
||||
user=SimpleNamespace(id="user-1"),
|
||||
api_key=SimpleNamespace(id="api-key-1"),
|
||||
request_id="req-video-sync-test",
|
||||
client_ip="127.0.0.1",
|
||||
user_agent="pytest",
|
||||
start_time=0.0,
|
||||
)
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_handle_create_task_uses_rust_sync_helper(
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
handler = _make_handler()
|
||||
monkeypatch.setattr(video_mod.UsageService, "create_pending_usage", lambda *args, **kwargs: None)
|
||||
monkeypatch.setattr(
|
||||
handler._normalizer,
|
||||
"video_request_to_internal",
|
||||
lambda body: SimpleNamespace(model=str(body.get("model") or "sora-2")),
|
||||
)
|
||||
|
||||
candidate = SimpleNamespace(provider=SimpleNamespace(name="provider-1", id="prov-1"))
|
||||
endpoint = SimpleNamespace(
|
||||
id="ep-1",
|
||||
api_family="openai",
|
||||
endpoint_kind="video",
|
||||
base_url="https://api.openai.com",
|
||||
body_rules=None,
|
||||
)
|
||||
provider_key = SimpleNamespace(id="key-1")
|
||||
|
||||
monkeypatch.setattr(
|
||||
handler,
|
||||
"_resolve_upstream_key",
|
||||
AsyncMock(return_value=("upstream-key", endpoint, provider_key)),
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
handler,
|
||||
"_build_upstream_url",
|
||||
lambda base_url: "https://api.openai.com/v1/videos",
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
handler,
|
||||
"_build_upstream_headers",
|
||||
lambda original_headers, upstream_key, endpoint, **kwargs: {
|
||||
"authorization": f"Bearer {upstream_key}"
|
||||
},
|
||||
)
|
||||
monkeypatch.setattr(
|
||||
video_mod.HTTPClientPool,
|
||||
"get_default_client_async",
|
||||
AsyncMock(side_effect=AssertionError("python fallback should not run")),
|
||||
)
|
||||
|
||||
async def _fake_rust_sync(**kwargs: object) -> httpx.Response:
|
||||
assert kwargs["method"] == "POST"
|
||||
assert kwargs["url"] == "https://api.openai.com/v1/videos"
|
||||
assert kwargs["provider_id"] == "prov-1"
|
||||
assert kwargs["endpoint_id"] == "ep-1"
|
||||
assert kwargs["key_id"] == "key-1"
|
||||
assert kwargs["body"] == {"model": "sora-2", "prompt": "hello"}
|
||||
return httpx.Response(
|
||||
200,
|
||||
request=httpx.Request("POST", str(kwargs["url"])),
|
||||
json={"id": "ext-1"},
|
||||
)
|
||||
|
||||
create_failed = AsyncMock()
|
||||
monkeypatch.setattr(handler, "_try_rust_sync_http_response", _fake_rust_sync)
|
||||
monkeypatch.setattr(handler, "_create_failed_task_and_usage", create_failed)
|
||||
|
||||
async def _fake_submit_with_failover(**kwargs: object) -> JSONResponse:
|
||||
response = await kwargs["submit_func"](candidate)
|
||||
assert response.status_code == 200
|
||||
assert response.json()["id"] == "ext-1"
|
||||
return JSONResponse(status_code=400, content={"error": {"message": "stop"}})
|
||||
|
||||
monkeypatch.setattr(handler, "_submit_with_failover", _fake_submit_with_failover)
|
||||
|
||||
response = await handler.handle_create_task(
|
||||
http_request=SimpleNamespace(),
|
||||
original_headers={},
|
||||
original_request_body={"model": "sora-2", "prompt": "hello"},
|
||||
)
|
||||
|
||||
assert response.status_code == 400
|
||||
create_failed.assert_awaited_once()
|
||||
Reference in New Issue
Block a user