From c941b948f1fa8eff615985a83110a401ae426ab3 Mon Sep 17 00:00:00 2001
From: zhaoqingang <zhaoqg0118@163.com>
Date: 星期三, 18 十二月 2024 11:03:49 +0800
Subject: [PATCH] 知识库和智能体同步
---
app/task/fetch_agent.py | 115 +++++++++++++++++++++++++++++++++++++++++++++++----------
1 files changed, 94 insertions(+), 21 deletions(-)
diff --git a/app/task/fetch_agent.py b/app/task/fetch_agent.py
index 5d08434..ea9a2e6 100644
--- a/app/task/fetch_agent.py
+++ b/app/task/fetch_agent.py
@@ -7,7 +7,9 @@
from app.config.config import settings
from app.config.const import RAGFLOW, BISHENG, DIFY
-from app.models import DialogModel
+from app.models import KnowledgeModel
+from app.models.dialog_model import DialogModel
+from app.models.user_model import UserAppModel
from app.models.agent_model import AgentModel
from app.models.base_model import SessionLocal, Base
from app.service.v2.app_register import AppRegisterDao
@@ -48,6 +50,16 @@
description = Column(Text, nullable=False)
tenant_id = Column(String(36), nullable=False)
+
+class RgKnowledge(Base):
+ __tablename__ = 'knowledgebase'
+ id = Column(String(36), primary_key=True) # id
+ name = Column(String(128)) # 鍚嶇О
+ permission = Column(String(32), default="me")
+ tenant_id = Column(String(32)) # 鍒涘缓浜篿d
+ description = Column(Text) # 璇存槑
+ status = Column(String(1)) # 鐘舵��
+ doc_num = Column(Integer) # 鏂囨。
# 瑙ｆ瀽鍚嶅瓧
def parse_names(names_str: str) -> List[str]:
@@ -130,15 +142,15 @@
result = db.query(AgentModel).delete()
db.commit() # 鎻愪氦浜嬪姟
initial_agents = [
- ('80ee430a-e396-48c4-a12c-7c7cdf5eda51', 1, '鎶ュ憡鐢熸垚', 'BISHENG', 'report'),
+ # ('80ee430a-e396-48c4-a12c-7c7cdf5eda51', 1, '鎶ュ憡鐢熸垚', 'DIFY', 'report'),
('basic_excel_merge', 2, '鎶ヨ〃鍚堝苟', 'BASIC', 'excelMerge'),
- ('bfd090d589d811efb3630242ac190006', 4, '鏂囨。鏅鸿兘', 'BISHENG', 'report'),
+ ('7638f00638a24c21a68ec6c49b304a35', 4, '鏂囨。鏅鸿兘', 'DIFY', 'documentIa'),
('da3451da89d911efb9490242ac190006', 3, '鐭ヨ瘑闂瓟', 'RAGFLOW', 'knowledgeQA'),
('e96eb7a589db11ef87d20242ac190006', 5, '鏅鸿兘闂瓟', 'RAGFLOW', 'chat'),
('basic_excel_talk', 6, '鏅鸿兘鏁版嵁', 'BASIC', 'excelTalk'),
('basic_question_talk', 7, '鍑洪缁勫嵎', 'BASIC', 'questionTalk'),
('9d75142a-66eb-4e23-b7d4-03efe4584915', 8, '灏忔暟缁樺浘', 'DIFY', 'imageTalk'),
- ('basic_paper_talk', 8, '鏂囨。鍑哄嵎', 'BASIC', 'paperTalk'),
+ ('basic_paper_talk', 9, '鏂囨。鍑哄嵎', 'BASIC', 'paperTalk'),
('basic_report_clean', 10, '鏂囨。鎶ュ憡', 'DIFY', 'reportWorkflow')
]
@@ -157,10 +169,10 @@
def sync_agents():
try:
- bisheng_data = get_data_from_bisheng(BISHENG_NAMES_TO_SYNC)
+ # bisheng_data = get_data_from_bisheng(BISHENG_NAMES_TO_SYNC)
ragflow_data = get_data_from_ragflow(RAGFLOW_NAMES_TO_SYNC)
- update_ids_in_local(bisheng_data)
+ # update_ids_in_local(bisheng_data)
update_ids_in_local(ragflow_data)
print("Agents synchronized successfully")
@@ -185,19 +197,27 @@
finally:
db.close()
+def get_rag_user_id(db, tenant_id, app_type):
+
+ user = db.query(UserAppModel).filter(UserAppModel.app_type==app_type, UserAppModel.app_id==tenant_id).first()
+ if user:
+ return user.user_id
+ return tenant_id
+
+
def get_data_from_bisheng_v2(names: List[str]) -> List[Dict]:
db = SessionBisheng()
try:
if names:
query = db.query(Flow.id, Flow.name, Flow.description, Flow.status, Flow.user_id) \
- .filter(Flow.name.in_(names), Flow.status==2)
+ .filter(Flow.name.in_(names))
else:
- query = db.query(Flow.id, Flow.name, Flow.description, Flow.status, Flow.user_id).filter(Flow.status==2)
+ query = db.query(Flow.id, Flow.name, Flow.description, Flow.status, Flow.user_id)
results = query.all()
# print(f"Executing query: {query}")
# 鏍煎紡鍖杋d涓篣UID
- formatted_results = [{"id":format_uuid(row[0]), "name": row[1], "description": row[2], "status": str(row[3]-1), "user_id": str(row[4])} for row in results]
+ formatted_results = [{"id":format_uuid(row[0]), "name": row[1], "description": row[2], "status": "1" if row[3] ==2 else "0", "user_id": str(row[4])} for row in results]
return formatted_results
finally:
db.close()
@@ -213,7 +233,7 @@
results = query.all()
formatted_results = [
- {"id": format_uuid(row[0]), "name": row[1], "description": row[2], "status": str(row[3]),
+ {"id": format_uuid(row[0]), "name": row[1], "description": row[2], "status": str(row[3]) if row[3] ==1 else "2",
"user_id": str(row[4])} for row in results]
return formatted_results
finally:
@@ -242,23 +262,23 @@
def update_ids_in_local_v2(data: List[Dict], dialog_type:str):
db = SessionLocal()
agent_id_list = []
- print("----------------------------------------")
- print(data)
- print("*********************************************")
+ type_dict = {"1": RAGFLOW,"2": BISHENG,"4": DIFY}
try:
for row in data:
agent_id_list.append(row["id"])
existing_agent = db.query(DialogModel).filter_by(id=row["id"]).first()
if existing_agent:
existing_agent.name = row["name"]
+ existing_agent.status = row["status"]
existing_agent.description = row["description"]
+ # existing_agent.tenant_id = get_rag_user_id(db, row["user_id"], type_dict[dialog_type])
else:
- existing = DialogModel(id=row["id"], name=row["name"], description=row["description"], tenant_id=row["user_id"], dialog_type=dialog_type)
+ existing = DialogModel(id=row["id"], status=row["status"], name=row["name"], description=row["description"], tenant_id=get_rag_user_id(db, row["user_id"], type_dict[dialog_type]), dialog_type=dialog_type)
db.add(existing)
db.commit()
for dialog in db.query(DialogModel).filter_by(dialog_type=dialog_type).all():
if dialog.id not in agent_id_list:
- db.query(DialogModel).filter_by(id=dialog.id).delete()
+ db.query(DialogModel).filter_by(id=dialog.id).update({"status": "2"})
db.commit()
except IntegrityError:
db.rollback()
@@ -269,7 +289,16 @@
def get_data_from_ragflow_knowledge():
- ...
+ db = SessionRagflow()
+ try:
+
+ results = db.query(RgKnowledge.id, RgKnowledge.name, RgKnowledge.description, RgKnowledge.status, RgKnowledge.tenant_id, RgKnowledge.doc_num, RgKnowledge.permission).all()
+ formatted_results = [
+ {"id": row[0], "name": row[1], "description": row[2], "status": str(row[3]),
+ "user_id": str(row[4]), "doc_num": row[5], "permission": row[6]} for row in results]
+ return formatted_results
+ finally:
+ db.close()
def sync_agents_v2():
db = SessionLocal()
@@ -279,17 +308,60 @@
for app in app_register:
if app["id"] == RAGFLOW:
ragflow_data = get_data_from_ragflow_v2([])
- update_ids_in_local_v2(ragflow_data, "1")
+ if ragflow_data:
+ update_ids_in_local_v2(ragflow_data, "1")
elif app["id"] == BISHENG:
bisheng_data = get_data_from_bisheng_v2([])
- update_ids_in_local_v2(bisheng_data, "2")
+ if bisheng_data:
+ update_ids_in_local_v2(bisheng_data, "2")
elif app["id"] == DIFY:
dify_data = get_data_from_dify_v2([])
- update_ids_in_local_v2(dify_data, "4")
+ if dify_data:
+ update_ids_in_local_v2(dify_data, "4")
print("Agents synchronized successfully")
except Exception as e:
print(f"Failed to sync agents: {str(e)}")
+def update_ids_in_local_knowledge(data, klg_type):
+ type_dict = {"1": RAGFLOW, "2": BISHENG, "4": DIFY}
+ db = SessionLocal()
+ agent_id_list = []
+ try:
+ for row in data:
+ agent_id_list.append(row["id"])
+ existing_agent = db.query(KnowledgeModel).filter_by(id=row["id"]).first()
+ if existing_agent:
+ existing_agent.name = row["name"]
+ existing_agent.description = row["description"]
+ # existing_agent.tenant_id = get_rag_user_id(db, row["user_id"], type_dict[klg_type])
+ existing_agent.permission = row["permission"]
+ existing_agent.documents = row["doc_num"]
+ existing_agent.status = row["status"]
+ else:
+ existing = KnowledgeModel(id=row["id"], name=row["name"], description=row["description"],
+ tenant_id=get_rag_user_id(db, row["user_id"], type_dict[klg_type]),status=row["status"],
+ knowledge_type=1, permission=row["permission"], documents=row["doc_num"])
+ db.add(existing)
+ db.commit()
+ for dialog in db.query(KnowledgeModel).filter_by(knowledge_type=type_dict[klg_type]).all():
+ if dialog.id not in agent_id_list:
+ db.query(KnowledgeModel).filter_by(id=dialog.id).delete()
+ db.commit()
+ except IntegrityError:
+ db.rollback()
+ raise
+ finally:
+ db.close()
+
+def get_one_from_ragflow_knowledge(klg_id):
+ db = SessionRagflow()
+ try:
+
+ row = db.query(RgKnowledge.id, RgKnowledge.name, RgKnowledge.description, RgKnowledge.status, RgKnowledge.tenant_id, RgKnowledge.doc_num, RgKnowledge.permission).filter(RgKnowledge.id==klg_id).first()
+ return {"id": row[0], "name": row[1], "description": row[2], "status": str(row[3]),
+ "user_id": str(row[4]), "doc_num": row[5], "permission": row[6]} if row else {}
+ finally:
+ db.close()
def sync_knowledge():
@@ -299,8 +371,9 @@
app_register = AppRegisterDao(db).get_apps()
for app in app_register:
if app["id"] == RAGFLOW:
- ragflow_data = get_data_from_ragflow_knowledge([])
- update_ids_in_local_v2(ragflow_data, "1")
+ ragflow_data = get_data_from_ragflow_knowledge()
+ if ragflow_data:
+ update_ids_in_local_knowledge(ragflow_data, "1")
# elif app["id"] == BISHENG:
# bisheng_data = get_data_from_bisheng_v2([])
# update_ids_in_local_v2(bisheng_data, "2")
--
Gitblit v1.8.0