zlei6/ragflow
1
0
Fork 0
mirror of https://github.com/infiniflow/ragflow.git synced 2026-07-10 21:55:42 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: align model default handling (#16782)

Browse Source
This commit is contained in:
buua436
2026-07-10 10:34:19 +08:00
committed by GitHub
parent d48a5622df
commit 74bbbba3e0
15 changed files with 169 additions and 71 deletions
Download Patch File Download Diff File Expand all files Collapse all files

37
test/testcases/restful_api/test_chats.py
View File

@@ -16,6 +16,7 @@
import asyncio
import importlib.util
import re
import sys
from copy import deepcopy
from concurrent.futures import ThreadPoolExecutor
@@ -775,7 +776,7 @@ def _load_chat_routes_unit_module(monkeypatch):
class _StubTenantService:
@staticmethod
def get_by_id(_tenant_id):
return True, SimpleNamespace(llm_id="glm-4")
return True, SimpleNamespace(llm_id="glm-4", tenant_llm_id="tenant-llm-id")
@staticmethod
def get_joined_tenants_by_user_id(_user_id):
@@ -1142,7 +1143,7 @@ def test_chat_create_accepts_provider_scoped_rerank_id_unit(monkeypatch):
"vector_similarity_weight": 0.25,
},
)
monkeypatch.setattr(module.TenantService, "get_by_id", lambda _tid: (True, SimpleNamespace(llm_id="glm-4@CI@ZHIPU-AI")))
monkeypatch.setattr(module.TenantService, "get_by_id", lambda _tid: (True, SimpleNamespace(llm_id="glm-4@CI@ZHIPU-AI", tenant_llm_id="tenant-llm-id")))
monkeypatch.setattr(module.DialogService, "query", lambda **_kwargs: [])
monkeypatch.setattr(module.KnowledgebaseService, "accessible", lambda **_kwargs: [SimpleNamespace(id="kb-1")])
monkeypatch.setattr(module.KnowledgebaseService, "query", lambda **_kwargs: [_DummyKB()])
@@ -1166,7 +1167,7 @@ def test_chat_create_accepts_provider_scoped_rerank_id_unit(monkeypatch):
assert saved["rerank_id"] == "custom-reranker@OpenAI"
assert {
"tenant_id": "tenant-1",
"model_name": "custom-reranker@OpenAI",
"model_ref": "custom-reranker@OpenAI",
"model_type": "rerank",
} in query_calls
@@ -1176,7 +1177,7 @@ def test_chat_create_allows_default_knowledge_placeholder_without_sources_unit(m
module = _load_chat_routes_unit_module(monkeypatch)
saved = {}
_set_route_unit_request_json(monkeypatch, module, {"name": "chat-a"})
monkeypatch.setattr(module.TenantService, "get_by_id", lambda _tid: (True, SimpleNamespace(llm_id="glm-4")))
monkeypatch.setattr(module.TenantService, "get_by_id", lambda _tid: (True, SimpleNamespace(llm_id="glm-4", tenant_llm_id="tenant-llm-id")))
monkeypatch.setattr(module.DialogService, "query", lambda **_kwargs: [])
monkeypatch.setattr(module, "get_api_key", lambda *_args, **_kwargs: SimpleNamespace(id=1))
@@ -1215,7 +1216,7 @@ def test_chat_create_uses_direct_chat_fields_unit(monkeypatch):
"vector_similarity_weight": 0.25,
},
)
monkeypatch.setattr(module.TenantService, "get_by_id", lambda _tid: (True, SimpleNamespace(llm_id="glm-4")))
monkeypatch.setattr(module.TenantService, "get_by_id", lambda _tid: (True, SimpleNamespace(llm_id="glm-4", tenant_llm_id="tenant-llm-id")))
monkeypatch.setattr(module.DialogService, "query", lambda **_kwargs: [])
monkeypatch.setattr(module.KnowledgebaseService, "accessible", lambda **_kwargs: [SimpleNamespace(id="kb-1")])
monkeypatch.setattr(module.KnowledgebaseService, "query", lambda **_kwargs: [_DummyKB()])
@@ -1371,7 +1372,7 @@ def test_patch_chat_drops_response_only_fields_before_update_unit(monkeypatch):
_set_route_unit_request_json(monkeypatch, module, payload)
monkeypatch.setattr(module.DialogService, "query", lambda **kwargs: [] if "name" in kwargs else [SimpleNamespace(id="chat-1")])
monkeypatch.setattr(module.DialogService, "get_by_id", lambda _id: (True, _DummyDialogRecord(existing)))
monkeypatch.setattr(module.TenantService, "get_by_id", lambda _tid: (True, SimpleNamespace(llm_id="glm-4")))
monkeypatch.setattr(module.TenantService, "get_by_id", lambda _tid: (True, SimpleNamespace(llm_id="glm-4", tenant_llm_id="tenant-llm-id")))
monkeypatch.setattr(module.KnowledgebaseService, "accessible", lambda **_kwargs: [SimpleNamespace(id="kb-1")])
monkeypatch.setattr(module.KnowledgebaseService, "query", lambda **_kwargs: [_DummyKB()])
monkeypatch.setattr(module, "get_api_key", lambda *args, **kwargs: SimpleNamespace(id=1))
@@ -1399,7 +1400,7 @@ def test_patch_chat_merges_prompt_and_llm_settings_unit(monkeypatch):
)
monkeypatch.setattr(module.DialogService, "query", lambda **_kwargs: [SimpleNamespace(id="chat-1")])
monkeypatch.setattr(module.DialogService, "get_by_id", lambda _id: (True, _DummyDialogRecord(existing)))
monkeypatch.setattr(module.TenantService, "get_by_id", lambda _tid: (True, SimpleNamespace(llm_id="glm-4")))
monkeypatch.setattr(module.TenantService, "get_by_id", lambda _tid: (True, SimpleNamespace(llm_id="glm-4", tenant_llm_id="tenant-llm-id")))
def _update(_chat_id, payload):
updated.update(payload)
@@ -1442,7 +1443,7 @@ def test_update_chat_allows_knowledge_placeholder_without_sources_unit(monkeypat
)
monkeypatch.setattr(module.DialogService, "query", lambda **_kwargs: [SimpleNamespace(id="chat-1")])
monkeypatch.setattr(module.DialogService, "get_by_id", lambda _id: (True, _DummyDialogRecord(existing)))
monkeypatch.setattr(module.TenantService, "get_by_id", lambda _tid: (True, SimpleNamespace(llm_id="glm-4")))
monkeypatch.setattr(module.TenantService, "get_by_id", lambda _tid: (True, SimpleNamespace(llm_id="glm-4", tenant_llm_id="tenant-llm-id")))
updated = {}
def _update(_chat_id, payload):
@@ -1540,7 +1541,15 @@ def test_chat_create_llm_contract(rest_client, clear_chats, ensure_parsed_docume
body = res.json()
assert body["code"] == expected_code, (scenario_name, body)
if expected_code == 0:
assert body["data"]["llm_id"] == expected_llm_id, (scenario_name, body)
actual_llm_id = body["data"]["llm_id"]
tenant_llm_id = body["data"].get("tenant_llm_id")
if expected_llm_id == "glm-4-flash@CI@ZHIPU-AI":
if tenant_llm_id:
assert actual_llm_id == tenant_llm_id, (scenario_name, body)
else:
assert re.fullmatch(r"[0-9a-f]{32}", actual_llm_id), (scenario_name, body)
else:
assert actual_llm_id == expected_llm_id, (scenario_name, body)
assert body["data"]["llm_setting"] == expected_llm_setting, (scenario_name, body)
else:
assert body["message"] == expected_message, (scenario_name, body)
@@ -1819,7 +1828,15 @@ def test_chat_update_llm_contract(rest_client, clear_chats, ensure_parsed_docume
get_payload = get_res.json()
assert get_payload["code"] == 0, (scenario_name, get_payload)
assert get_payload["data"]["name"] == updated_name, (scenario_name, get_payload)
assert get_payload["data"]["llm_id"] == expected_llm_id, (scenario_name, get_payload)
actual_llm_id = get_payload["data"]["llm_id"]
tenant_llm_id = get_payload["data"].get("tenant_llm_id")
if expected_llm_id == "glm-4-flash@CI@ZHIPU-AI":
if tenant_llm_id:
assert actual_llm_id == tenant_llm_id, (scenario_name, get_payload)
else:
assert re.fullmatch(r"[0-9a-f]{32}", actual_llm_id), (scenario_name, get_payload)
else:
assert actual_llm_id == expected_llm_id, (scenario_name, get_payload)
assert get_payload["data"]["llm_setting"] == expected_llm_setting, (scenario_name, get_payload)
else:
assert body["message"] == expected_message, (scenario_name, body)

7
test/testcases/restful_api/test_datasets.py
View File

@@ -832,7 +832,7 @@ def test_dataset_update_embedding_model_invalid_and_none_contract(rest_client, c
assert list_res.status_code == 200
list_payload = list_res.json()
assert list_payload["code"] == 0, list_payload
assert list_payload["data"][0]["embedding_model"] == "BAAI/bge-small-en-v1.5@Local@Builtin", list_payload
assert list_payload["data"][0]["embedding_model"].startswith("BAAI/bge-small-en-v1.5"), list_payload
@pytest.mark.p2
@@ -1197,7 +1197,10 @@ def test_dataset_create_embedding_model_contract(rest_client, clear_datasets, na
pytest.xfail(f"Environment has no authorized tenant model for {embedding_model}: {payload}")
assert payload["code"] == expected_code, payload
if expected_embedding_model is not None:
assert payload["data"]["embedding_model"] == expected_embedding_model, payload
if name in {"embedding_model_unset", "embedding_model_none"}:
assert payload["data"]["embedding_model"].startswith("BAAI/bge-small-en-v1.5"), payload
else:
assert payload["data"]["embedding_model"] == expected_embedding_model, payload
if expected_message is not None:
assert payload["message"] == expected_message, payload

4
test/testcases/restful_api/test_user_tenant_routes_unit.py
View File

@@ -1532,7 +1532,7 @@ def _load_chat_routes_unit_module(monkeypatch):
"TenantService",
(),
{
"get_by_id": staticmethod(lambda _tenant_id: (True, SimpleNamespace(llm_id="glm-4"))),
"get_by_id": staticmethod(lambda _tenant_id: (True, SimpleNamespace(llm_id="glm-4", tenant_llm_id="tenant-llm-id"))),
"get_joined_tenants_by_user_id": staticmethod(lambda _user_id: [{"tenant_id": "tenant-1"}, {"tenant_id": "team-tenant-2"}]),
},
)
@@ -1599,7 +1599,7 @@ def test_create_chat_uses_tenant_default_llm_when_llm_id_is_null_unit(monkeypatc
res = _run(module.create.__wrapped__())
assert res["code"] == 0
assert saved["llm_id"] == "glm-4"
assert saved["llm_id"] == "tenant-llm-id"
assert saved["llm_setting"]["temperature"] == 0.8

4
test/testcases/test_sdk_api/test_dataset_mangement/test_create_dataset.py
View File

@@ -237,13 +237,13 @@ class TestDatasetCreate:
def test_embedding_model_unset(self, client):
payload = {"name": "embedding_model_unset"}
dataset = client.create_dataset(**payload)
assert dataset.embedding_model == "BAAI/bge-small-en-v1.5@Local@Builtin", str(dataset)
assert dataset.embedding_model.split("@", 1)[0] == "BAAI/bge-small-en-v1.5", str(dataset)
@pytest.mark.p2
def test_embedding_model_none(self, client):
payload = {"name": "embedding_model_none", "embedding_model": None}
dataset = client.create_dataset(**payload)
assert dataset.embedding_model == "BAAI/bge-small-en-v1.5@Local@Builtin", str(dataset)
assert dataset.embedding_model.split("@", 1)[0] == "BAAI/bge-small-en-v1.5", str(dataset)
@pytest.mark.p2
@pytest.mark.parametrize(

4
test/testcases/test_sdk_api/test_dataset_mangement/test_update_dataset.py
View File

@@ -229,10 +229,10 @@ class TestDatasetUpdate:
def test_embedding_model_none(self, client, add_dataset_func):
dataset = add_dataset_func
dataset.update({"embedding_model": None})
assert dataset.embedding_model == "BAAI/bge-small-en-v1.5@Local@Builtin", str(dataset)
assert dataset.embedding_model.split("@", 1)[0] == "BAAI/bge-small-en-v1.5", str(dataset)
retrieved_dataset = client.get_dataset(name=dataset.name)
assert retrieved_dataset.embedding_model == "BAAI/bge-small-en-v1.5@Local@Builtin", str(retrieved_dataset)
assert retrieved_dataset.embedding_model.split("@", 1)[0] == "BAAI/bge-small-en-v1.5", str(retrieved_dataset)
@pytest.mark.p2
@pytest.mark.parametrize(

23
test/unit_test/api/apps/services/test_delete_datasets.py
View File

@@ -18,6 +18,7 @@
import importlib.util
import sys
from pathlib import Path
from enum import IntEnum
from types import ModuleType, SimpleNamespace
from unittest.mock import MagicMock
@@ -26,6 +27,16 @@ import pytest
pytestmark = pytest.mark.p2
class _StubModelTypeBinary(IntEnum):
CHAT = 1
EMBEDDING = 2
SPEECH2TEXT = 4
IMAGE2TEXT = 8
RERANK = 16
TTS = 32
OCR = 64
def _stub(monkeypatch, name, **attrs):
mod = ModuleType(name)
for key, value in attrs.items():
@@ -122,6 +133,8 @@ def _load_delete_datasets_module(monkeypatch, *, f2d_rows, file_filter_delete):
_stub(
monkeypatch,
"api.db.db_models",
DB=SimpleNamespace(connection_context=lambda: lambda func: func),
TenantModel=SimpleNamespace(),
File=SimpleNamespace(source_type="source_type", id="id", type="type", name="name"),
)
_stub(
@@ -130,8 +143,18 @@ def _load_delete_datasets_module(monkeypatch, *, f2d_rows, file_filter_delete):
PAGERANK_FLD="pagerank",
TAG_FLD="tag",
FileSource=SimpleNamespace(KNOWLEDGEBASE="knowledgebase"),
PipelineTaskType=SimpleNamespace(
PARSE="parse",
DOWNLOAD="download",
RAPTOR="raptor",
GRAPH_RAG="graph_rag",
MINDMAP="mindmap",
ARTIFACT="artifact",
SKILL="skill",
),
StatusEnum=SimpleNamespace(),
LLMType=SimpleNamespace(),
ModelTypeBinary=_StubModelTypeBinary,
)
_stub(
monkeypatch,

2
test/unit_test/api/apps/services/test_models_api_service_list_tenant_added_models.py
View File

@@ -25,6 +25,7 @@ import sys
from pathlib import Path
from enum import IntEnum
from types import ModuleType, SimpleNamespace
from unittest.mock import MagicMock
import pytest
@@ -141,6 +142,7 @@ def _load_module(monkeypatch, *, tenant_model_records, factory_llm_infos=None):
ensure_mineru_from_env=lambda *a, **kw: None,
ensure_paddleocr_from_env=lambda *a, **kw: None,
ensure_opendataloader_from_env=lambda *a, **kw: None,
resolve_model_id=MagicMock(),
)
_stub(
monkeypatch,