sync update from LAN repo (source master @ ca48c4ec)
This commit is contained in:
1 parent
8bc8b0e082
commit
ea9dcff9c0
873 files changed
+107866
-14011
No files matched your search
@@ -2,6 +2,8 @@
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import errno
|
||||
from pathlib import Path
|
||||
from types import SimpleNamespace
|
||||
|
||||
import pytest
|
||||
@@ -26,6 +28,80 @@ from deerflow.persistence.model_capabilities import (
|
||||
)
|
||||
|
||||
|
||||
def test_config_atomic_replace_retries_transient_busy_filesystem_error(
|
||||
tmp_path: Path,
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
config_path = tmp_path / "config.yaml"
|
||||
original = "models:\n - name: old\n"
|
||||
updated = "models:\n - name: new\n"
|
||||
config_path.write_text(original, encoding="utf-8")
|
||||
real_replace = router.os.replace
|
||||
attempts = 0
|
||||
|
||||
def replace_with_transient_busy(source: str, target: Path) -> None:
|
||||
nonlocal attempts
|
||||
attempts += 1
|
||||
if attempts < 3:
|
||||
raise OSError(errno.EBUSY, "resource busy", str(target))
|
||||
real_replace(source, target)
|
||||
|
||||
monkeypatch.setattr(router.os, "replace", replace_with_transient_busy)
|
||||
monkeypatch.setattr(router, "sleep", lambda _delay: None)
|
||||
|
||||
backup = router._backup_and_write_config(config_path, original, updated)
|
||||
|
||||
assert attempts == 3
|
||||
assert config_path.read_text(encoding="utf-8") == updated
|
||||
assert backup.read_text(encoding="utf-8") == original
|
||||
|
||||
|
||||
def test_config_atomic_replace_does_not_retry_permanent_error(
|
||||
tmp_path: Path,
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
config_path = tmp_path / "config.yaml"
|
||||
original = "models:\n - name: old\n"
|
||||
config_path.write_text(original, encoding="utf-8")
|
||||
attempts = 0
|
||||
|
||||
def fail_replace(_source: str, _target: Path) -> None:
|
||||
nonlocal attempts
|
||||
attempts += 1
|
||||
raise OSError(errno.EACCES, "permission denied")
|
||||
|
||||
monkeypatch.setattr(router.os, "replace", fail_replace)
|
||||
|
||||
with pytest.raises(OSError, match="permission denied"):
|
||||
router._backup_and_write_config(config_path, original, "models:\n - name: new\n")
|
||||
|
||||
assert attempts == 1
|
||||
assert config_path.read_text(encoding="utf-8") == original
|
||||
assert not list(tmp_path.glob(".config.yaml.*.tmp"))
|
||||
|
||||
|
||||
def test_config_single_file_mount_is_updated_without_atomic_rename(
|
||||
tmp_path: Path,
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
config_path = tmp_path / "config.yaml"
|
||||
original = "models:\n - name: old\n"
|
||||
updated = "models:\n - name: mounted-new\n"
|
||||
config_path.write_text(original, encoding="utf-8")
|
||||
monkeypatch.setattr(router, "_is_file_mountpoint", lambda _path: True)
|
||||
|
||||
def unexpected_replace(_source: str, _target: Path) -> None:
|
||||
pytest.fail("single-file bind mount must not be replaced by rename")
|
||||
|
||||
monkeypatch.setattr(router.os, "replace", unexpected_replace)
|
||||
|
||||
backup = router._backup_and_write_config(config_path, original, updated)
|
||||
|
||||
assert config_path.read_text(encoding="utf-8") == updated
|
||||
assert backup.read_text(encoding="utf-8") == original
|
||||
assert not list(tmp_path.glob(".config.yaml.*.tmp"))
|
||||
|
||||
|
||||
def test_replace_models_block_keeps_other_sections_and_comments() -> None:
|
||||
raw = """config_version: 1
|
||||
models:
|
||||
@@ -139,6 +215,260 @@ def test_reasoning_capability_route_precedes_generic_model_update_and_accepts_sl
|
||||
]
|
||||
|
||||
|
||||
def test_write_file_transport_route_precedes_generic_model_update() -> None:
|
||||
scope = {
|
||||
"type": "http",
|
||||
"method": "PUT",
|
||||
"path": "/api/admin/models/write-file-transport/zai-org/GLM-5-FP8",
|
||||
}
|
||||
matching_paths = [route.path for route in router.router.routes if route.matches(scope)[0] is Match.FULL]
|
||||
|
||||
assert matching_paths[:2] == [
|
||||
"/api/admin/models/write-file-transport/{model_name:path}",
|
||||
"/api/admin/models/{model_name:path}",
|
||||
]
|
||||
|
||||
|
||||
def _admin_request():
|
||||
return SimpleNamespace(
|
||||
state=SimpleNamespace(user=SimpleNamespace(system_role="admin")),
|
||||
app=SimpleNamespace(state=SimpleNamespace()),
|
||||
)
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_write_file_transport_toggle_persists_native_and_buffered_gateway(
|
||||
tmp_path,
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
config_path = tmp_path / "config.yaml"
|
||||
config_path.write_text(
|
||||
"""models:
|
||||
- name: flash
|
||||
use: langchain_openai:ChatOpenAI
|
||||
model: flash
|
||||
""",
|
||||
encoding="utf-8",
|
||||
)
|
||||
saved: dict[str, object] = {}
|
||||
|
||||
async def save_models(_request, raw_before: str, models: list[dict[str, object]]) -> list[dict[str, object]]:
|
||||
saved["raw_before"] = raw_before
|
||||
saved["models"] = models
|
||||
return models
|
||||
|
||||
monkeypatch.setattr(router, "_resolve_config_path", lambda: config_path)
|
||||
monkeypatch.setattr(router, "_save_models", save_models)
|
||||
|
||||
enabled = await router.update_write_file_transport(
|
||||
"flash",
|
||||
router.WriteFileTransportToggleRequest(write_file_transport="buffered_gateway"),
|
||||
_admin_request(),
|
||||
)
|
||||
|
||||
assert enabled.write_file_transport == "buffered_gateway"
|
||||
assert saved["models"] == [
|
||||
{
|
||||
"name": "flash",
|
||||
"use": "langchain_openai:ChatOpenAI",
|
||||
"model": "flash",
|
||||
"write_file_transport": "buffered_gateway",
|
||||
}
|
||||
]
|
||||
|
||||
disabled = await router.update_write_file_transport(
|
||||
"flash",
|
||||
router.WriteFileTransportToggleRequest(write_file_transport="native"),
|
||||
_admin_request(),
|
||||
)
|
||||
|
||||
assert disabled.write_file_transport == "native"
|
||||
assert saved["models"][0]["write_file_transport"] == "native"
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
@pytest.mark.parametrize(
|
||||
("use", "use_responses_api"),
|
||||
[
|
||||
("langchain_anthropic:ChatAnthropic", None),
|
||||
("langchain_openai:ChatOpenAI", True),
|
||||
],
|
||||
)
|
||||
async def test_write_file_transport_rejects_incompatible_models(
|
||||
use: str,
|
||||
use_responses_api: bool | None,
|
||||
tmp_path,
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
config_path = tmp_path / "config.yaml"
|
||||
original = f"""models:
|
||||
- name: incompatible
|
||||
use: {use}
|
||||
model: incompatible
|
||||
"""
|
||||
if use_responses_api:
|
||||
original = original.replace(
|
||||
" model: incompatible\n",
|
||||
" model: incompatible\n use_responses_api: true\n",
|
||||
)
|
||||
config_path.write_text(original, encoding="utf-8")
|
||||
monkeypatch.setattr(router, "_resolve_config_path", lambda: config_path)
|
||||
|
||||
async def fail_save_models(_request, _raw_before, _models):
|
||||
raise AssertionError("incompatible models must not be persisted")
|
||||
|
||||
monkeypatch.setattr(router, "_save_models", fail_save_models)
|
||||
|
||||
with pytest.raises(HTTPException) as exc_info:
|
||||
await router.update_write_file_transport(
|
||||
"incompatible",
|
||||
router.WriteFileTransportToggleRequest(write_file_transport="buffered_gateway"),
|
||||
_admin_request(),
|
||||
)
|
||||
|
||||
assert exc_info.value.status_code == 422
|
||||
assert config_path.read_text(encoding="utf-8") == original
|
||||
|
||||
|
||||
def test_strict_stream_integrity_route_precedes_generic_model_update() -> None:
|
||||
scope = {
|
||||
"type": "http",
|
||||
"method": "PUT",
|
||||
"path": "/api/admin/models/strict-stream-integrity/zai-org/GLM-5-FP8",
|
||||
}
|
||||
matching_paths = [route.path for route in router.router.routes if route.matches(scope)[0] is Match.FULL]
|
||||
|
||||
assert matching_paths[:2] == [
|
||||
"/api/admin/models/strict-stream-integrity/{model_name:path}",
|
||||
"/api/admin/models/{model_name:path}",
|
||||
]
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_strict_stream_integrity_toggle_persists_explicit_bool(
|
||||
tmp_path,
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
config_path = tmp_path / "config.yaml"
|
||||
config_path.write_text(
|
||||
"""models:
|
||||
- name: flash
|
||||
use: langchain_openai:ChatOpenAI
|
||||
model: flash
|
||||
""",
|
||||
encoding="utf-8",
|
||||
)
|
||||
saved: dict[str, object] = {}
|
||||
|
||||
async def save_models(_request, raw_before: str, models: list[dict[str, object]]) -> list[dict[str, object]]:
|
||||
saved["models"] = models
|
||||
return models
|
||||
|
||||
monkeypatch.setattr(router, "_resolve_config_path", lambda: config_path)
|
||||
monkeypatch.setattr(router, "_save_models", save_models)
|
||||
|
||||
enabled = await router.update_strict_stream_integrity(
|
||||
"flash",
|
||||
router.StrictStreamIntegrityToggleRequest(strict_stream_integrity=True),
|
||||
_admin_request(),
|
||||
)
|
||||
|
||||
assert enabled.strict_stream_integrity is True
|
||||
assert saved["models"] == [
|
||||
{
|
||||
"name": "flash",
|
||||
"use": "langchain_openai:ChatOpenAI",
|
||||
"model": "flash",
|
||||
"strict_stream_integrity": True,
|
||||
}
|
||||
]
|
||||
|
||||
disabled = await router.update_strict_stream_integrity(
|
||||
"flash",
|
||||
router.StrictStreamIntegrityToggleRequest(strict_stream_integrity=False),
|
||||
_admin_request(),
|
||||
)
|
||||
|
||||
assert disabled.strict_stream_integrity is False
|
||||
assert saved["models"][0]["strict_stream_integrity"] is False
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
@pytest.mark.parametrize(
|
||||
("use", "use_responses_api"),
|
||||
[
|
||||
("langchain_anthropic:ChatAnthropic", None),
|
||||
("langchain_openai:ChatOpenAI", True),
|
||||
],
|
||||
)
|
||||
async def test_strict_stream_integrity_rejects_incompatible_models(
|
||||
use: str,
|
||||
use_responses_api: bool | None,
|
||||
tmp_path,
|
||||
monkeypatch: pytest.MonkeyPatch,
|
||||
) -> None:
|
||||
config_path = tmp_path / "config.yaml"
|
||||
original = f"""models:
|
||||
- name: incompatible
|
||||
use: {use}
|
||||
model: incompatible
|
||||
"""
|
||||
if use_responses_api:
|
||||
original = original.replace(
|
||||
" model: incompatible\n",
|
||||
" model: incompatible\n use_responses_api: true\n",
|
||||
)
|
||||
config_path.write_text(original, encoding="utf-8")
|
||||
monkeypatch.setattr(router, "_resolve_config_path", lambda: config_path)
|
||||
|
||||
async def fail_save_models(_request, _raw_before, _models):
|
||||
raise AssertionError("incompatible models must not be persisted")
|
||||
|
||||
monkeypatch.setattr(router, "_save_models", fail_save_models)
|
||||
|
||||
with pytest.raises(HTTPException) as exc_info:
|
||||
await router.update_strict_stream_integrity(
|
||||
"incompatible",
|
||||
router.StrictStreamIntegrityToggleRequest(strict_stream_integrity=True),
|
||||
_admin_request(),
|
||||
)
|
||||
|
||||
assert exc_info.value.status_code == 422
|
||||
assert config_path.read_text(encoding="utf-8") == original
|
||||
|
||||
|
||||
def test_legacy_model_edit_preserves_strict_stream_integrity() -> None:
|
||||
existing = {
|
||||
"name": "relay-model",
|
||||
"use": "langchain_openai:ChatOpenAI",
|
||||
"model": "relay-model",
|
||||
"strict_stream_integrity": True,
|
||||
}
|
||||
|
||||
merged = router._request_to_raw_model(
|
||||
router.ModelUpsertRequest(
|
||||
name="relay-model",
|
||||
use="langchain_openai:ChatOpenAI",
|
||||
model="relay-model-v2",
|
||||
),
|
||||
existing,
|
||||
)
|
||||
|
||||
assert merged["model"] == "relay-model-v2"
|
||||
assert merged["strict_stream_integrity"] is True
|
||||
|
||||
changed = router._request_to_raw_model(
|
||||
router.ModelUpsertRequest(
|
||||
name="relay-model",
|
||||
use="langchain_openai:ChatOpenAI",
|
||||
model="relay-model-v2",
|
||||
strict_stream_integrity=False,
|
||||
),
|
||||
merged,
|
||||
)
|
||||
assert changed["strict_stream_integrity"] is False
|
||||
|
||||
|
||||
def _reasoning_request(store, *, config: AppConfig | None = None):
|
||||
return SimpleNamespace(
|
||||
state=SimpleNamespace(
|
||||
@@ -272,15 +602,21 @@ async def test_reasoning_capability_endpoint_and_model_apis_use_database_not_yam
|
||||
|
||||
updated = await router.update_reasoning_effort_capability(
|
||||
"zai-org/GLM-5-FP8",
|
||||
router.ReasoningEffortCapabilityRequest(supports_reasoning_effort=True),
|
||||
router.ReasoningEffortCapabilityRequest(
|
||||
supports_reasoning_effort=True,
|
||||
default_reasoning_effort="low",
|
||||
),
|
||||
request,
|
||||
)
|
||||
admin_list = await router.list_model_configurations(request)
|
||||
public_list = await public_models.list_models(request, config)
|
||||
|
||||
assert updated.supports_reasoning_effort is True
|
||||
assert updated.default_reasoning_effort == "low"
|
||||
assert admin_list.models[0].supports_reasoning_effort is True
|
||||
assert public_list.models[0].supports_reasoning_effort is True
|
||||
assert public_list.models[0].default_reasoning_effort == "low"
|
||||
assert (await store.get_capability("zai-org/GLM-5-FP8"))["default_reasoning_effort"] == "low"
|
||||
assert config_path.read_text(encoding="utf-8") == original
|
||||
|
||||
|
||||
@@ -605,6 +941,7 @@ def test_final_reasoning_policy_overrides_both_preseeded_runtime_containers() ->
|
||||
config,
|
||||
capability_enabled=True,
|
||||
model_supports_thinking=True,
|
||||
default_reasoning_effort="low",
|
||||
)
|
||||
|
||||
assert normalized == "low"
|
||||
@@ -661,7 +998,7 @@ def test_reasoning_policy_strips_effort_when_selection_is_not_available(
|
||||
assert "reasoning_effort" not in config["context"]
|
||||
|
||||
|
||||
def test_reasoning_policy_rejects_unknown_effort() -> None:
|
||||
def test_reasoning_policy_uses_global_default_instead_of_stale_thread_value() -> None:
|
||||
config = {
|
||||
"configurable": {},
|
||||
"context": {
|
||||
@@ -670,14 +1007,15 @@ def test_reasoning_policy_rejects_unknown_effort() -> None:
|
||||
},
|
||||
}
|
||||
|
||||
with pytest.raises(HTTPException) as exc_info:
|
||||
services.enforce_reasoning_effort_run_config(
|
||||
config,
|
||||
capability_enabled=True,
|
||||
model_supports_thinking=True,
|
||||
)
|
||||
|
||||
assert exc_info.value.status_code == 422
|
||||
normalized = services.enforce_reasoning_effort_run_config(
|
||||
config,
|
||||
capability_enabled=True,
|
||||
model_supports_thinking=True,
|
||||
default_reasoning_effort="low",
|
||||
)
|
||||
assert normalized == "low"
|
||||
assert config["configurable"]["reasoning_effort"] == "low"
|
||||
assert config["context"]["reasoning_effort"] == "low"
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
@@ -692,7 +1030,7 @@ def test_reasoning_policy_rejects_unknown_effort() -> None:
|
||||
"reasoning_effort": "minimal",
|
||||
},
|
||||
"qwen-a",
|
||||
"low",
|
||||
"medium",
|
||||
),
|
||||
(
|
||||
{
|
||||
@@ -732,6 +1070,7 @@ async def test_start_run_applies_database_reasoning_policy_after_final_context_m
|
||||
await store.set_reasoning_effort_support(
|
||||
"qwen-a",
|
||||
supports_reasoning_effort=True,
|
||||
default_reasoning_effort="medium",
|
||||
)
|
||||
|
||||
class ThreadStore:
|
||||
@@ -764,6 +1103,9 @@ async def test_start_run_applies_database_reasoning_policy_after_final_context_m
|
||||
async def set_status(self, *_args, **_kwargs):
|
||||
return None
|
||||
|
||||
def watch_completion(self, _record):
|
||||
return None
|
||||
|
||||
run_context = SimpleNamespace(
|
||||
app_config=app_config,
|
||||
thread_store=ThreadStore(),
|
||||
|
||||
Reference in new issue
Block a user