rag-gateway/app/task/fetch_agent.py

import json
import os
from pickle import PROTO
from typing import Dict, List, Tuple

from sqlalchemy import create_engine, Column, String, Integer, Text
from sqlalchemy.exc import IntegrityError
from sqlalchemy.orm import sessionmaker, Session

from app.config.config import settings
from app.config.const import RAGFLOW, BISHENG, DIFY, ENV_CONF_PATH, Dialog_STATSU_DELETE, Dialog_STATSU_ON
from app.models import KnowledgeModel, ComplexChatDao
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.models.resource_model import ResourceModel, ResourceTypeModel
from app.service.v2.app_register import AppRegisterDao

# 创建数据库引擎和会话工厂
engine_bisheng = create_engine(settings.sgb_db_url)
engine_ragflow = create_engine(settings.fwr_db_url)
engine_dify = create_engine(settings.dify_database_url)

SessionBisheng = sessionmaker(autocommit=False, autoflush=False, bind=engine_bisheng)
SessionRagflow = sessionmaker(autocommit=False, autoflush=False, bind=engine_ragflow)
SessionDify = sessionmaker(autocommit=False, autoflush=False, bind=engine_dify)


class Flow(Base):
    __tablename__ = 'flow'
    id = Column(String(255), primary_key=True)
    name = Column(String(255), nullable=False)
    status = Column(Integer, nullable=False)
    description = Column(String(255), nullable=False)
    user_id = Column(Integer, nullable=False)


class Dialog(Base):
    __tablename__ = 'dialog'
    id = Column(String(255), primary_key=True)
    name = Column(String(255), nullable=False)
    status = Column(String(1), nullable=False)
    description = Column(String(255), nullable=False)
    tenant_id = Column(String(36), nullable=False)
    kb_ids = Column(String(128), nullable=False)


class DfApps(Base):
    __tablename__ = 'apps'
    id = Column(String(36), primary_key=True)
    name = Column(String(255), nullable=False)
    status = Column(String(16), nullable=False)
    description = Column(Text, nullable=False)
    tenant_id = Column(String(36), nullable=False)
    mode = Column(String(36), nullable=False)


class RgKnowledge(Base):
    __tablename__ = 'knowledgebase'
    id = Column(String(36), primary_key=True)  # id
    name = Column(String(128))  # 名称
    permission = Column(String(32), default="me")
    tenant_id = Column(String(32))  # 创建人id
    description = Column(Text)  # 说明
    status = Column(String(1))  # 状态
    doc_num = Column(Integer)  # 文档
    embd_id = Column(String(128))  # 文档


class RgUserTenant(Base):
    __tablename__ = 'user_tenant'
    id = Column(String(36), primary_key=True)  # id
    tenant_id = Column(String(32))  # 名称
    user_id = Column(String(32))
    role = Column(String(32))  # 创建人id


# 解析名字
def parse_names(names_str: str) -> List[str]:
    return [name.strip() for name in names_str.split(',')]


BISHENG_NAMES_TO_SYNC = parse_names(settings.fetch_sgb_agent)
RAGFLOW_NAMES_TO_SYNC = parse_names(settings.fetch_fwr_agent)


def get_data_from_bisheng(names: List[str]) -> List[Tuple]:
    db = SessionBisheng()
    try:
        if names:
            query = db.query(Flow.id, Flow.name) \
                .filter(Flow.status == 2, Flow.name.in_(names))
        else:
            query = db.query(Flow.id, Flow.name) \
                .filter(Flow.status == 2)

        results = query.all()
        print(f"Executing query: {query}")
        # 格式化id为UUID
        formatted_results = [(format_uuid(row[0]), row[1]) for row in results]
        return formatted_results
    finally:
        db.close()


def format_uuid(uuid_str: str) -> str:
    # 确保输入字符串长度为32
    if len(uuid_str) != 32:
        raise ValueError("Input string must be 32 characters long")

    # 插入连字符
    formatted_uuid = f"{uuid_str[:8]}-{uuid_str[8:12]}-{uuid_str[12:16]}-{uuid_str[16:20]}-{uuid_str[20:]}"
    return formatted_uuid


def get_data_from_ragflow(names: List[str]) -> List[Tuple]:
    db = SessionRagflow()
    try:
        if names:
            query = db.query(Dialog.id, Dialog.name) \
                .filter(Dialog.status == 1, Dialog.name.in_(names))
        else:
            query = db.query(Dialog.id, Dialog.name) \
                .filter(Dialog.status == 1)

        results = query.all()
        print(f"Executing query: {query}")
        return results
    finally:
        db.close()


def update_ids_in_local(data: List[Tuple]):
    db = SessionLocal()
    try:
        for row in data:
            name = row[1]
            new_id = row[0]
            existing_agent = db.query(AgentModel).filter_by(name=name).first()
            if existing_agent:
                existing_agent.id = new_id
                db.add(existing_agent)
        db.commit()
    except IntegrityError:
        db.rollback()
        raise
    finally:
        db.close()


def initialize_agents():
    db = SessionLocal()
    try:
        count = db.query(AgentModel).count()
        if count > 0:
            result = db.query(AgentModel).delete()
            db.commit()  # 提交事务
        initial_agents = [
            # ('80ee430a-e396-48c4-a12c-7c7cdf5eda51', 1, '报告生成', 'DIFY', 'report'),
            ('basic_excel_merge', 2, '报表合并', 'BASIC', 'excelMerge'),
            ('7638f00638a24c21a68ec6c49b304a35', 4, '文档智能', 'DIFY', 'documentIa'),
            ('da3451da89d911efb9490242ac190006', 3, '知识问答', 'RAGFLOW', 'knowledgeQA'),
            ('e96eb7a589db11ef87d20242ac190006', 5, '智能问答', 'RAGFLOW', 'chat'),
            ('basic_excel_talk', 6, '智能数据', 'BASIC', 'excelTalk'),
            ('basic_question_talk', 7, '出题组卷', 'BASIC', 'questionTalk'),
            ('9d75142a-66eb-4e23-b7d4-03efe4584915', 8, '小数绘图', 'DIFY', 'imageTalk'),
            ('2f6ddf93-7ba6-4b2d-b991-d96421404600', 9, '文档出卷', 'DIFY', 'paperTalk'),
            ('basic_report_clean', 10, '文档报告', 'DIFY', 'reportWorkflow')
        ]

        for agent in initial_agents:
            agent_id = format_uuid(agent[0]) if len(agent[0]) == 32 else agent[0]
            db.add(AgentModel(id=agent_id, sort=agent[1], name=agent[2], agent_type=agent[3], type=agent[4]))

        db.commit()
        print("Initial agents inserted successfully")
    except IntegrityError:
        db.rollback()
        raise
    finally:
        db.close()


def sync_agents():
    try:
        # 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(ragflow_data)

        print("Agents synchronized successfully")
    except Exception as e:
        print(f"Failed to sync agents: {str(e)}")


def update_ids_in_local(data: List[Tuple]):
    db = SessionLocal()
    try:
        for row in data:
            name = row[1]
            new_id = row[0]
            existing_agent = db.query(AgentModel).filter_by(name=name).first()
            if existing_agent:
                existing_agent.id = new_id
                db.add(existing_agent)
        db.commit()
    except IntegrityError:
        db.rollback()
        raise
    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 == "1")
        else:
            query = db.query(Flow.id, Flow.name, Flow.description, Flow.status, Flow.user_id).filter(Flow.status == "1")

        results = query.all()
        # print(f"Executing query: {query}")
        # 格式化id为UUID
        formatted_results = [
            {"id": row[0], "name": row[1], "description": row[2], "status": row[3], "user_id": str(row[4]),
             "mode": "agent-dialog"} for row in results]
        return formatted_results
    finally:
        db.close()


def get_data_from_ragflow_v2(base_db, names: List[str], tenant_id) -> List[Dict]:
    db = SessionRagflow()
    para = {
        "user_input_form": [],
        "retriever_resource": {
            "enabled": True
        },
        "file_upload": {
            "enabled": False
        }
    }
    try:
        chat_ids = ComplexChatDao(base_db).get_complex_chat_ids()
        # print(chat_ids)
        if names:
            query = db.query(Dialog.id, Dialog.name, Dialog.description, Dialog.status, Dialog.tenant_id) \
                .filter(Dialog.name.in_(names), Dialog.status == "1")
        else:
            query = db.query(Dialog.id, Dialog.name, Dialog.description, Dialog.status, Dialog.tenant_id, Dialog.kb_ids).filter(
                Dialog.status == "1", Dialog.tenant_id == tenant_id)

        results = query.all()
        formatted_results = [
            {"id": row[0], "name": row[1], "description": row[2], "status": "1" if row[3] == "1" else "2",
             "user_id": str(row[4]), "mode": "agent-dialog", "parameters": para, "kb_ids": row[5]} for row in results if row[0] not in chat_ids]
        return formatted_results
    finally:
        db.close()


def get_data_from_dy_v2(base_db, names: List[str]) -> List[Dict]:
    db = SessionDify()
    try:
        chat_ids = ComplexChatDao(base_db).get_complex_chat_ids()
        # print(chat_ids)
        if names:
            query = db.query(DfApps.id, DfApps.name, DfApps.description, DfApps.status, DfApps.tenant_id, DfApps.mode) \
                .filter(DfApps.name.in_(names))
        else:
            query = db.query(DfApps.id, DfApps.name, DfApps.description, DfApps.status, DfApps.tenant_id, DfApps.mode)

        results = query.all()
        formatted_results = [
            {"id": str(row[0]), "name": row[1], "description": row[2], "status": "1",
             "user_id": str(row[4]), "mode": row[5], "parameters": {}} for row in results if str(row[0]) not in chat_ids]
        return formatted_results
    finally:
        db.close()


def update_ids_in_local_v2(data: List[Dict], dialog_type: str):
    db = SessionLocal()
    agent_id_list = []
    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.description = row["description"]
                existing_agent.mode = row["mode"]
                existing_agent.kb_ids = row.get("kb_ids", "")
                if existing_agent.status == Dialog_STATSU_DELETE:
                    existing_agent.status = Dialog_STATSU_ON
                if row["parameters"]:
                    existing_agent.parameters = json.dumps(row["parameters"])
            else:
                existing = DialogModel(id=row["id"], status=row["status"], name=row["name"],
                                       description=row["description"], kb_ids=row.get("kb_ids", ""),
                                       tenant_id=get_rag_user_id(db, row["user_id"], type_dict[dialog_type]),
                                       dialog_type=dialog_type, mode=row["mode"], parameters=json.dumps(row["parameters"]))
                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:
                # print(dialog.id)
                db.query(DialogModel).filter_by(id=dialog.id).update({"status": Dialog_STATSU_DELETE})
                db.commit()
    except IntegrityError:
        db.rollback()
        raise
    finally:
        db.close()


def get_data_from_ragflow_knowledge(tenant_id):
    db = SessionRagflow()
    try:

        results = db.query(RgKnowledge.id, RgKnowledge.name, RgKnowledge.description, RgKnowledge.status,
                           RgKnowledge.tenant_id, RgKnowledge.doc_num, RgKnowledge.permission).filter(RgKnowledge.tenant_id==tenant_id).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()

    try:
        app_register = AppRegisterDao(db).get_apps()
        for app in app_register:
            try:
                if app["id"] == RAGFLOW:
                    ragflow_data = get_data_from_ragflow_v2(db, [], app["name"])
                    if ragflow_data:
                        update_ids_in_local_v2(ragflow_data, "1")
                elif app["id"] == DIFY:
                    dify_data = get_data_from_dy_v2(db, [])
                    if dify_data:
                        update_ids_in_local_v2(dify_data, "4")
            except Exception as e:
                print(f"Failed to sync agents: {str(e)}")
        print("v2 Agents synchronized successfully")
    except Exception as e:
        print(f"v2 Failed to sync agents: {str(e)}")
    finally:
        db.close()


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=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, RgKnowledge.embd_id).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], "embd_id": row[7]} if row else {}
    finally:
        db.close()


def get_one_from_ragflow_dialog(dialog_id):
    db = SessionRagflow()
    try:
        row = db.query(Dialog.id, Dialog.name, Dialog.description, Dialog.status, Dialog.tenant_id, Dialog.kb_ids) \
            .filter(Dialog.id==dialog_id).first()
        return {"id": row[0], "name": row[1], "description": row[2], "status": str(row[3]),
                "user_id": str(row[4]), "kb_ids": row[5]} if row else {}
    finally:
        db.close()


def sync_knowledge():
    db = SessionLocal()

    try:
        app_register = AppRegisterDao(db).get_apps()
        for app in app_register:
            if app["id"] == RAGFLOW:
                ragflow_data = get_data_from_ragflow_knowledge(app["name"])
                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")
            # elif app["id"] == DIFY:
            #     dify_data = get_data_from_dify_v2([])
            #     update_ids_in_local_v2(dify_data, "4")
        print("sync knowledge successfully")
    except Exception as e:
        print(f"Failed to sync knowledge: {str(e)}")
    finally:
        db.close()


def update_ragflow_user_tenant(user_id: str):
    db = SessionRagflow()
    try:
        if user_id:
            db.query(RgUserTenant).filter(RgUserTenant.user_id == user_id, RgUserTenant.role == "invite").update(
                {"role": "normal"})
            db.query(RgUserTenant).filter(RgUserTenant.tenant_id == user_id, RgUserTenant.role == "invite").update(
                {"role": "normal"})
        else:
            db.query(RgUserTenant).filter(RgUserTenant.role == "invite").update({"role": "normal"})
        db.commit()
    finally:
        db.close()


def import_type_table(session: Session, node: dict, parent=None):
    resource_type = ResourceTypeModel(
        id=node['id'],
        name=node['name'],
        description=node.get('description')
    )
    if parent:
        resource_type.parent = parent
    session.add(resource_type)
    session.commit()


def import_tree(session: Session, node: dict, parent=None):
    resource = ResourceModel(
        id=node['id'],
        name=node['name'],
        url=node['url'],
        path=node.get('path'),
        perms=node['perms'],
        description=node.get('description'),
        icon=node.get('icon'),
        seq=node['seq'],
        target=node.get('target'),
        canbdeeleted=node.get('canbdeeleted'),
        resource_type_id=node['resource_type_id'],
        resource_id=node.get('resource_id'),
        status=node['status'],
        hidden=node.get('hidden')
    )
    if parent:
        resource.parent = parent
    session.add(resource)
    if 'children' in node:
        for child in node['children']:
            import_tree(session, child, parent=resource)
    session.commit()


def sync_resources_from_json():
    db = SessionLocal()
    try:
        if db.query(ResourceTypeModel).count() == 0:
            with open(os.path.join(ENV_CONF_PATH, "resource_type.json"), 'r', encoding='utf-8') as file:
                type_json_data = json.load(file)

            db.query(ResourceTypeModel).delete()
            db.commit()

            for node in type_json_data:
                import_type_table(db, node)
            print("add resourceType record successfully")
        else:
            print("sync resourcesType successfully")
        if db.query(ResourceModel).count() == 0:
            with open(os.path.join(ENV_CONF_PATH, "resource.json"), 'r', encoding='utf-8') as file:
                json_data = json.load(file)

            db.query(ResourceModel).delete()
            db.commit()

            for node in json_data:
                import_tree(db, node)
            print("add resources record successfully")
        else:
            print("sync resources successfully")
    except Exception as e:
        print(f"Failed to sync resources or resource type: {str(e)}")
    finally:
        db.close()


if __name__ == "__main__":
    # a = get_data_from_dify_v2([])
    # print(a)
    update_ragflow_user_tenant("")