pengjunjie/ragflow
1
0
Fork 1
mirror of https://github.com/infiniflow/ragflow.git synced 2025-12-08 20:42:30 +08:00
Code Issues Packages Projects Releases Wiki Activity

Fix: Create dataset performance unmatched between HTTP api and web ui (#10960)

Browse Source 
### What problem does this PR solve?

Fix: Create dataset performance unmatched between HTTP api and web ui
#10925

### Type of change

- [x] Bug Fix (non-breaking change which fixes an issue)
This commit is contained in:
Billy Bao
2025-11-04 13:45:14 +08:00
committed by GitHub
parent 1e45137284
commit 19f71a961a
12 changed files with 201 additions and 222 deletions
Download Patch File Download Diff File Expand all files Collapse all files

62
api/apps/kb_app.py
View File

@ -22,7 +22,6 @@ from flask_login import login_required, current_user
import numpy as np
from api.db import LLMType
from api.db.services import duplicate_name
from api.db.services.llm_service import LLMBundle
from api.db.services.document_service import DocumentService, queue_raptor_o_graphrag_tasks
from api.db.services.file2document_service import File2DocumentService
@ -31,7 +30,6 @@ from api.db.services.pipeline_operation_log_service import PipelineOperationLogS
from api.db.services.task_service import TaskService, GRAPH_RAPTOR_FAKE_DOC_ID
from api.db.services.user_service import TenantService, UserTenantService
from api.utils.api_utils import get_error_data_result, server_error_response, get_data_error_result, validate_request, not_allowed_parameters
from common.misc_utils import get_uuid
from api.db import PipelineTaskType, StatusEnum, FileSource, VALID_FILE_TYPES, VALID_TASK_STATUS
from api.db.services.knowledgebase_service import KnowledgebaseService
from api.db.db_models import File
@ -50,63 +48,17 @@ from rag.utils.doc_store_conn import OrderByExpr
@validate_request("name")
def create():
req = request.json
dataset_name = req["name"]
if not isinstance(dataset_name, str):
return get_data_error_result(message="Dataset name must be string.")
if dataset_name.strip() == "":
return get_data_error_result(message="Dataset name can't be empty.")
if len(dataset_name.encode("utf-8")) > DATASET_NAME_LIMIT:
return get_data_error_result(
message=f"Dataset name length is {len(dataset_name)} which is larger than {DATASET_NAME_LIMIT}")
req = KnowledgebaseService.create_with_name(
name = req.pop("name", None),
tenant_id = current_user.id,
parser_id = req.pop("parser_id", None),
**req
)
dataset_name = dataset_name.strip()
dataset_name = duplicate_name(
KnowledgebaseService.query,
name=dataset_name,
tenant_id=current_user.id,
status=StatusEnum.VALID.value)
try:
req["id"] = get_uuid()
req["name"] = dataset_name
req["tenant_id"] = current_user.id
req["created_by"] = current_user.id
if not req.get("parser_id"):
req["parser_id"] = "naive"
e, t = TenantService.get_by_id(current_user.id)
if not e:
return get_data_error_result(message="Tenant not found.")
req["parser_config"] = {
"layout_recognize": "DeepDOC",
"chunk_token_num": 512,
"delimiter": "\n",
"auto_keywords": 0,
"auto_questions": 0,
"html4excel": False,
"topn_tags": 3,
"raptor": {
"use_raptor": True,
"prompt": "Please summarize the following paragraphs. Be careful with the numbers, do not make things up. Paragraphs as following:\n {cluster_content}\nThe above is the content you need to summarize.",
"max_token": 256,
"threshold": 0.1,
"max_cluster": 64,
"random_seed": 0
},
"graphrag": {
"use_graphrag": True,
"entity_types": [
"organization",
"person",
"geo",
"event",
"category"
],
"method": "light"
}
}
if not KnowledgebaseService.save(**req):
return get_data_error_result()
return get_json_result(data={"kb_id": req["id"]})
return get_json_result(data={"kb_id":req["id"]})
except Exception as e:
return server_error_response(e)

78
api/apps/sdk/dataset.py
View File

@ -28,12 +28,10 @@ from api.db.services.file2document_service import File2DocumentService
from api.db.services.file_service import FileService
from api.db.services.knowledgebase_service import KnowledgebaseService
from api.db.services.user_service import TenantService
from common.misc_utils import get_uuid
from api.utils.api_utils import (
deep_merge,
get_error_argument_result,
get_error_data_result,
get_error_operating_result,
get_error_permission_result,
get_parser_config,
get_result,
@ -80,29 +78,28 @@ def create(tenant_id):
properties:
name:
type: string
description: Name of the dataset.
description: Dataset name (required).
avatar:
type: string
description: Base64 encoding of the avatar.
description: Optional base64-encoded avatar image.
description:
type: string
description: Description of the dataset.
description: Optional dataset description.
embedding_model:
type: string
description: Embedding model Name.
description: Optional embedding model name; if omitted, the tenant's default embedding model is used.
permission:
type: string
enum: ['me', 'team']
description: Dataset permission.
description: Visibility of the dataset (private to me or shared with team).
chunk_method:
type: string
enum: ["naive", "book", "email", "laws", "manual", "one", "paper",
"picture", "presentation", "qa", "table", "tag"
]
description: Chunking method.
"picture", "presentation", "qa", "table", "tag"]
description: Chunking method; if omitted, defaults to "naive".
parser_config:
type: object
description: Parser configuration.
description: Optional parser configuration; server-side defaults will be applied.
responses:
200:
description: Successful operation.
@ -117,44 +114,43 @@ def create(tenant_id):
# |----------------|-------------|
# | embedding_model| embd_id |
# | chunk_method | parser_id |
req, err = validate_and_parse_json_request(request, CreateDatasetReq)
if err is not None:
return get_error_argument_result(err)
req = KnowledgebaseService.create_with_name(
name = req.pop("name", None),
tenant_id = tenant_id,
parser_id = req.pop("parser_id", None),
**req
)
# Insert embedding model(embd id)
ok, t = TenantService.get_by_id(tenant_id)
if not ok:
return get_error_permission_result(message="Tenant not found")
if not req.get("embd_id"):
req["embd_id"] = t.embd_id
else:
ok, err = verify_embedding_availability(req["embd_id"], tenant_id)
if not ok:
return err
try:
if KnowledgebaseService.get_or_none(name=req["name"], tenant_id=tenant_id, status=StatusEnum.VALID.value):
return get_error_operating_result(message=f"Dataset name '{req['name']}' already exists")
req["parser_config"] = get_parser_config(req["parser_id"], req["parser_config"])
req["id"] = get_uuid()
req["tenant_id"] = tenant_id
req["created_by"] = tenant_id
ok, t = TenantService.get_by_id(tenant_id)
if not ok:
return get_error_permission_result(message="Tenant not found")
if not req.get("embd_id"):
req["embd_id"] = t.embd_id
else:
ok, err = verify_embedding_availability(req["embd_id"], tenant_id)
if not ok:
return err
if not KnowledgebaseService.save(**req):
return get_error_data_result(message="Create dataset error.(Database error)")
ok, k = KnowledgebaseService.get_by_id(req["id"])
if not ok:
return get_error_data_result(message="Dataset created failed")
response_data = remap_dictionary_keys(k.to_dict())
return get_result(data=response_data)
except OperationalError as e:
if not KnowledgebaseService.save(**req):
return get_error_data_result()
ok, k = KnowledgebaseService.get_by_id(req["id"])
if not ok:
return get_error_data_result(message="Dataset created failed")
response_data = remap_dictionary_keys(k.to_dict())
return get_result(data=response_data)
except Exception as e:
logging.exception(e)
return get_error_data_result(message="Database operation failed")
@manager.route("/datasets", methods=["DELETE"]) # noqa: F821
@token_required
def delete(tenant_id):

64
api/db/services/knowledgebase_service.py
View File

@ -21,7 +21,11 @@ from api.db import StatusEnum, TenantPermission
from api.db.db_models import DB, Document, Knowledgebase, User, UserTenant, UserCanvas
from api.db.services.common_service import CommonService
from common.time_utils import current_timestamp, datetime_format
from api.db.services import duplicate_name
from api.db.services.user_service import TenantService
from common.misc_utils import get_uuid
from api.constants import DATASET_NAME_LIMIT
from api.utils.api_utils import get_parser_config, get_data_error_result
class KnowledgebaseService(CommonService):
"""Service class for managing knowledge base operations.
@ -363,6 +367,64 @@ class KnowledgebaseService(CommonService):
# List of all knowledge base IDs
return [m["id"] for m in cls.model.select(cls.model.id).dicts()]
@classmethod
@DB.connection_context()
def create_with_name(
cls,
*,
name: str,
tenant_id: str,
parser_id: str | None = None,
**kwargs
):
"""Create a dataset (knowledgebase) by name with kb_app defaults.
This encapsulates the creation logic used in kb_app.create so other callers
(including RESTful endpoints) can reuse the same behavior.
Returns:
(ok: bool, model_or_msg): On success, returns (True, Knowledgebase model instance);
on failure, returns (False, error_message).
"""
# Validate name
if not isinstance(name, str):
return get_data_error_result(message="Dataset name must be string.")
dataset_name = name.strip()
if dataset_name == "":
return get_data_error_result(message="Dataset name can't be empty.")
if len(dataset_name.encode("utf-8")) > DATASET_NAME_LIMIT:
return get_data_error_result(message=f"Dataset name length is {len(dataset_name)} which is larger than {DATASET_NAME_LIMIT}")
# Deduplicate name within tenant
dataset_name = duplicate_name(
cls.query,
name=dataset_name,
tenant_id=tenant_id,
status=StatusEnum.VALID.value,
)
# Verify tenant exists
ok, _t = TenantService.get_by_id(tenant_id)
if not ok:
return False, "Tenant not found."
# Build payload
kb_id = get_uuid()
payload = {
"id": kb_id,
"name": dataset_name,
"tenant_id": tenant_id,
"created_by": tenant_id,
"parser_id": (parser_id or "naive"),
**kwargs
}
# Default parser_config (align with kb_app.create) — do not accept external overrides
payload["parser_config"] = get_parser_config(parser_id, kwargs.get("parser_config"))
return payload
@classmethod
@DB.connection_context()
def get_list(cls, joined_tenant_ids, user_id,

30
api/utils/api_utils.py
View File

@ -298,8 +298,34 @@ def get_parser_config(chunk_method, parser_config):
# Define default configurations for each chunking method
key_mapping = {
"naive": {"chunk_token_num": 512, "delimiter": r"\n", "html4excel": False, "layout_recognize": "DeepDOC",
"raptor": {"use_raptor": False}, "graphrag": {"use_graphrag": False}},
"naive": {
"layout_recognize": "DeepDOC",
"chunk_token_num": 512,
"delimiter": "\n",
"auto_keywords": 0,
"auto_questions": 0,
"html4excel": False,
"topn_tags": 3,
"raptor": {
"use_raptor": True,
"prompt": "Please summarize the following paragraphs. Be careful with the numbers, do not make things up. Paragraphs as following:\n {cluster_content}\nThe above is the content you need to summarize.",
"max_token": 256,
"threshold": 0.1,
"max_cluster": 64,
"random_seed": 0,
},
"graphrag": {
"use_graphrag": True,
"entity_types": [
"organization",
"person",
"geo",
"event",
"category",
],
"method": "light",
},
},
"qa": {"raptor": {"use_raptor": False}, "graphrag": {"use_graphrag": False}},
"tag": None,
"resume": None,

6
sdk/python/test/test_frontend_api/test_dataset.py
View File

@ -101,10 +101,10 @@ def test_invalid_name_dataset(get_auth):
# create dataset
# with pytest.raises(Exception) as e:
res = create_dataset(get_auth, 0)
assert res['code'] == 102
assert res['code'] == 100
res = create_dataset(get_auth, "")
assert res['code'] == 102
assert res['code'] == 100
long_string = ""
@ -112,7 +112,7 @@ def test_invalid_name_dataset(get_auth):
long_string += random.choice(string.ascii_letters + string.digits)
res = create_dataset(get_auth, long_string)
assert res['code'] == 102
assert res['code'] == 100
print(res)

29
test/testcases/configs.py
View File

@ -34,3 +34,32 @@ DATASET_NAME_LIMIT = 128
DOCUMENT_NAME_LIMIT = 255
CHAT_ASSISTANT_NAME_LIMIT = 255
SESSION_WITH_CHAT_NAME_LIMIT = 255
DEFAULT_PARSER_CONFIG = {
"layout_recognize": "DeepDOC",
"chunk_token_num": 512,
"delimiter": "\n",
"auto_keywords": 0,
"auto_questions": 0,
"html4excel": False,
"topn_tags": 3,
"raptor": {
"use_raptor": True,
"prompt": "Please summarize the following paragraphs. Be careful with the numbers, do not make things up. Paragraphs as following:\n {cluster_content}\nThe above is the content you need to summarize.",
"max_token": 256,
"threshold": 0.1,
"max_cluster": 64,
"random_seed": 0,
},
"graphrag": {
"use_graphrag": True,
"entity_types": [
"organization",
"person",
"geo",
"event",
"category",
],
"method": "light",
},
}

29
test/testcases/test_http_api/test_dataset_mangement/test_create_dataset.py
View File

@ -23,7 +23,7 @@ from libs.auth import RAGFlowHttpApiAuth
from utils import encode_avatar
from utils.file_utils import create_image_file
from utils.hypothesis_utils import valid_names
from configs import DEFAULT_PARSER_CONFIG
@pytest.mark.usefixtures("clear_datasets")
class TestAuthorization:
@ -637,42 +637,21 @@ class TestDatasetCreate:
payload = {"name": "parser_config_empty", "parser_config": {}}
res = create_dataset(HttpApiAuth, payload)
assert res["code"] == 0, res
assert res["data"]["parser_config"] == {
"chunk_token_num": 512,
"delimiter": r"\n",
"html4excel": False,
"layout_recognize": "DeepDOC",
"raptor": {"use_raptor": False},
"graphrag": {"use_graphrag": False},
}, res
assert res["data"]["parser_config"] == DEFAULT_PARSER_CONFIG, res
@pytest.mark.p2
def test_parser_config_unset(self, HttpApiAuth):
payload = {"name": "parser_config_unset"}
res = create_dataset(HttpApiAuth, payload)
assert res["code"] == 0, res
assert res["data"]["parser_config"] == {
"chunk_token_num": 512,
"delimiter": r"\n",
"html4excel": False,
"layout_recognize": "DeepDOC",
"raptor": {"use_raptor": False},
"graphrag": {"use_graphrag": False},
}, res
assert res["data"]["parser_config"] == DEFAULT_PARSER_CONFIG, res
@pytest.mark.p3
def test_parser_config_none(self, HttpApiAuth):
payload = {"name": "parser_config_none", "parser_config": None}
res = create_dataset(HttpApiAuth, payload)
assert res["code"] == 0, res
assert res["data"]["parser_config"] == {
"chunk_token_num": 512,
"delimiter": "\\n",
"html4excel": False,
"layout_recognize": "DeepDOC",
"raptor": {"use_raptor": False},
"graphrag": {"use_graphrag": False},
}, res
assert res["data"]["parser_config"] == DEFAULT_PARSER_CONFIG, res
@pytest.mark.p2
@pytest.mark.parametrize(

20
test/testcases/test_http_api/test_dataset_mangement/test_update_dataset.py
View File

@ -25,7 +25,7 @@ from libs.auth import RAGFlowHttpApiAuth
from utils import encode_avatar
from utils.file_utils import create_image_file
from utils.hypothesis_utils import valid_names
from configs import DEFAULT_PARSER_CONFIG
# TODO: Missing scenario for updating embedding_model with chunk_count != 0
@ -748,14 +748,7 @@ class TestDatasetUpdate:
res = list_datasets(HttpApiAuth)
assert res["code"] == 0, res
assert res["data"][0]["parser_config"] == {
"chunk_token_num": 512,
"delimiter": r"\n",
"html4excel": False,
"layout_recognize": "DeepDOC",
"raptor": {"use_raptor": False},
"graphrag": {"use_graphrag": False},
}, res
assert res["data"][0]["parser_config"] == DEFAULT_PARSER_CONFIG, res
@pytest.mark.p3
def test_parser_config_none(self, HttpApiAuth, add_dataset_func):
@ -766,14 +759,7 @@ class TestDatasetUpdate:
res = list_datasets(HttpApiAuth, {"id": dataset_id})
assert res["code"] == 0, res
assert res["data"][0]["parser_config"] == {
"chunk_token_num": 512,
"delimiter": r"\n",
"html4excel": False,
"layout_recognize": "DeepDOC",
"raptor": {"use_raptor": False},
"graphrag": {"use_graphrag": False},
}, res
assert res["data"][0]["parser_config"] == DEFAULT_PARSER_CONFIG, res
@pytest.mark.p3
def test_parser_config_empty_with_chunk_method_change(self, HttpApiAuth, add_dataset_func):

29
test/testcases/test_http_api/test_file_management_within_dataset/test_update_document.py
View File

@ -19,7 +19,7 @@ import pytest
from common import list_documents, update_document
from configs import DOCUMENT_NAME_LIMIT, INVALID_API_TOKEN
from libs.auth import RAGFlowHttpApiAuth
from configs import DEFAULT_PARSER_CONFIG
@pytest.mark.p1
class TestAuthorization:
@ -308,14 +308,7 @@ class TestUpdateDocumentParserConfig:
("naive", {}, 0, ""),
(
"naive",
{
"chunk_token_num": 512,
"layout_recognize": "DeepDOC",
"html4excel": False,
"delimiter": r"\n",
"task_page_size": 12,
"raptor": {"use_raptor": False},
},
DEFAULT_PARSER_CONFIG,
0,
"",
),
@ -419,7 +412,14 @@ class TestUpdateDocumentParserConfig:
"",
marks=pytest.mark.skip(reason="issues/6098"),
),
("naive", {"raptor": {"use_raptor": True}}, 0, ""),
("naive", {"raptor": {"use_raptor": {
"use_raptor": True,
"prompt": "Please summarize the following paragraphs. Be careful with the numbers, do not make things up. Paragraphs as following:\n {cluster_content}\nThe above is the content you need to summarize.",
"max_token": 256,
"threshold": 0.1,
"max_cluster": 64,
"random_seed": 0,
},}}, 0, ""),
("naive", {"raptor": {"use_raptor": False}}, 0, ""),
pytest.param(
"naive",
@ -534,14 +534,7 @@ class TestUpdateDocumentParserConfig:
if expected_code == 0:
res = list_documents(HttpApiAuth, dataset_id, {"id": document_ids[0]})
if parser_config == {}:
assert res["data"]["docs"][0]["parser_config"] == {
"chunk_token_num": 512,
"delimiter": r"\n",
"html4excel": False,
"layout_recognize": "DeepDOC",
"raptor": {"use_raptor": False},
"graphrag": {"use_graphrag": False},
}
assert res["data"]["docs"][0]["parser_config"] == DEFAULT_PARSER_CONFIG
else:
for k, v in parser_config.items():
assert res["data"]["docs"][0]["parser_config"][k] == v

29
test/testcases/test_sdk_api/test_dataset_mangement/test_create_dataset.py
View File

@ -23,7 +23,7 @@ from ragflow_sdk import DataSet, RAGFlow
from utils import encode_avatar
from utils.file_utils import create_image_file
from utils.hypothesis_utils import valid_names
from configs import DEFAULT_PARSER_CONFIG
@pytest.mark.usefixtures("clear_datasets")
class TestAuthorization:
@ -586,14 +586,7 @@ class TestDatasetCreate:
def test_parser_config_empty(self, client):
excepted_value = DataSet.ParserConfig(
client,
{
"chunk_token_num": 512,
"delimiter": r"\n",
"html4excel": False,
"layout_recognize": "DeepDOC",
"raptor": {"use_raptor": False},
"graphrag": {"use_graphrag": False},
},
DEFAULT_PARSER_CONFIG,
)
parser_config_o = DataSet.ParserConfig(client, {})
payload = {"name": "parser_config_empty", "parser_config": parser_config_o}
@ -604,14 +597,7 @@ class TestDatasetCreate:
def test_parser_config_unset(self, client):
excepted_value = DataSet.ParserConfig(
client,
{
"chunk_token_num": 512,
"delimiter": r"\n",
"html4excel": False,
"layout_recognize": "DeepDOC",
"raptor": {"use_raptor": False},
"graphrag": {"use_graphrag": False},
},
DEFAULT_PARSER_CONFIG,
)
payload = {"name": "parser_config_unset"}
dataset = client.create_dataset(**payload)
@ -621,14 +607,7 @@ class TestDatasetCreate:
def test_parser_config_none(self, client):
excepted_value = DataSet.ParserConfig(
client,
{
"chunk_token_num": 512,
"delimiter": r"\n",
"html4excel": False,
"layout_recognize": "DeepDOC",
"raptor": {"use_raptor": False},
"graphrag": {"use_graphrag": False},
},
DEFAULT_PARSER_CONFIG,
)
payload = {"name": "parser_config_empty", "parser_config": None}
dataset = client.create_dataset(**payload)

20
test/testcases/test_sdk_api/test_dataset_mangement/test_update_dataset.py
View File

@ -24,7 +24,7 @@ from ragflow_sdk import DataSet
from utils import encode_avatar
from utils.file_utils import create_image_file
from utils.hypothesis_utils import valid_names
from configs import DEFAULT_PARSER_CONFIG
class TestRquest:
@pytest.mark.p2
@ -634,14 +634,7 @@ class TestDatasetUpdate:
dataset = add_dataset_func
expected_config = DataSet.ParserConfig(
client,
{
"chunk_token_num": 512,
"delimiter": r"\n",
"html4excel": False,
"layout_recognize": "DeepDOC",
"raptor": {"use_raptor": False},
"graphrag": {"use_graphrag": False},
},
DEFAULT_PARSER_CONFIG,
)
dataset.update({"parser_config": {}})
assert str(dataset.parser_config) == str(expected_config), str(dataset)
@ -654,14 +647,7 @@ class TestDatasetUpdate:
dataset = add_dataset_func
expected_config = DataSet.ParserConfig(
client,
{
"chunk_token_num": 512,
"delimiter": r"\n",
"html4excel": False,
"layout_recognize": "DeepDOC",
"raptor": {"use_raptor": False},
"graphrag": {"use_graphrag": False},
},
DEFAULT_PARSER_CONFIG,
)
dataset.update({"parser_config": None})
assert str(dataset.parser_config) == str(expected_config), str(dataset)

27
test/testcases/test_sdk_api/test_file_management_within_dataset/test_update_document.py
View File

@ -17,7 +17,7 @@
import pytest
from configs import DOCUMENT_NAME_LIMIT
from ragflow_sdk import DataSet
from configs import DEFAULT_PARSER_CONFIG
class TestDocumentsUpdated:
@pytest.mark.p1
@ -206,14 +206,7 @@ class TestUpdateDocumentParserConfig:
("naive", {}, ""),
(
"naive",
{
"chunk_token_num": 512,
"layout_recognize": "DeepDOC",
"html4excel": False,
"delimiter": r"\n",
"task_page_size": 12,
"raptor": {"use_raptor": False},
},
DEFAULT_PARSER_CONFIG,
"",
),
pytest.param(
@ -294,7 +287,12 @@ class TestUpdateDocumentParserConfig:
"",
marks=pytest.mark.skip(reason="issues/6098"),
),
("naive", {"raptor": {"use_raptor": True}}, ""),
("naive", {"raptor": {"use_raptor": True,
"prompt": "Please summarize the following paragraphs. Be careful with the numbers, do not make things up. Paragraphs as following:\n {cluster_content}\nThe above is the content you need to summarize.",
"max_token": 256,
"threshold": 0.1,
"max_cluster": 64,
"random_seed": 0,}}, ""),
("naive", {"raptor": {"use_raptor": False}}, ""),
pytest.param(
"naive",
@ -400,13 +398,6 @@ class TestUpdateDocumentParserConfig:
else:
expected_config = DataSet.ParserConfig(
client,
{
"chunk_token_num": 512,
"delimiter": r"\n",
"html4excel": False,
"layout_recognize": "DeepDOC",
"raptor": {"use_raptor": False},
"graphrag": {"use_graphrag": False},
},
DEFAULT_PARSER_CONFIG,
)
assert str(updated_doc.parser_config) == str(expected_config), str(updated_doc)