Feat/models manage (#86)

* feat/add model management

* fix:model add and update

* feat:add model provider qwen

backend/app/core/db.py

@@ -1,12 +1,13 @@
 import os
-
-from sqlalchemy import text
+import importlib
+from sqlalchemy import text, func
 from sqlmodel import Session, create_engine, select
 
 from app.core.config import settings
 from app.core.tools import managed_tools
 from app.curd import users
-from app.models import Skill, User, UserCreate
+from app.models import Skill, User, UserCreate, ModelProvider, Models
+from app.core.model_providers.model_provider_manager import model_provider_manager
 
 
 def get_url():
@@ -25,7 +26,20 @@ def get_url():
 # make sure all SQLModel models are imported (app.models) before initializing DB
 # otherwise, SQLModel might fail to initialize relationships properly
 # for more details: https://github.com/tiangolo/full-stack-fastapi-template/issues/28
-
+def print_skills_info(session: Session) -> None:
+        print("\nSkills Information:")
+        skills = session.exec(select(Skill).order_by(Skill.id)).all()
+        for skill in skills:
+            print(f"Skill: {skill.name} (ID: {skill.id})")
+            print(f"  Display Name: {skill.display_name}")
+            print(f"  Description: {skill.description}")
+            print(f"  Managed: {'Yes' if skill.managed else 'No'}")
+            print(f"  Owner ID: {skill.owner_id}")
+            if skill.input_parameters:
+                print("  Input Parameters:")
+                for param, param_type in skill.input_parameters.items():
+                    print(f"    - {param}: {param_type}")
+            print()
 
 def init_db(session: Session) -> None:
     # Tables should be created with Alembic migrations
@@ -61,12 +75,15 @@ def init_db(session: Session) -> None:
             if (
                 existing_skill.description != skill_info.description
                 or existing_skill.display_name != skill_info.display_name
-                or existing_skill.input_parameters != skill_info.input_parameters  # 检查输入参数是否变化
+                or existing_skill.input_parameters
+                != skill_info.input_parameters  # 检查输入参数是否变化
             ):
                 # Update the existing skill's description and input parameters
                 existing_skill.description = skill_info.description
                 existing_skill.display_name = skill_info.display_name
-                existing_skill.input_parameters = skill_info.input_parameters  # 更新输入参数
+                existing_skill.input_parameters = (
+                    skill_info.input_parameters
+                )  # 更新输入参数
                 session.add(existing_skill)  # Mark the modified object for saving
         else:
             new_skill = Skill(
@@ -90,39 +107,64 @@ def init_db(session: Session) -> None:
 
     session.commit()
 
+        # 打印 skills 信息
+    print_skills_info(session)
+
+
 
 def init_modelprovider_model_db(session: Session) -> None:
-    # Insert or update ModelProvider data
-    model_provider_sql = """
-    INSERT INTO ModelProvider (id, provider_name, base_url, api_key, icon, description)
-    VALUES
-        (4, 'openai', 'fakeurl', 'fakeapikey', 'string', 'open ai'),
-        (1, 'Ollama', 'fakeurl', 'fakeapikey', 'string', 'string fake'),
-        (2, 'Siliconflow', 'fakeurl', 'fakeapikey', 'string', 'siliconflow'),
-        (3, 'zhipuai', 'https://open.bigmodel.cn/api/paas/v4', 'fakeapikey', 'zhipuai', '智谱AI')
-    ON CONFLICT (id) DO NOTHING;
-    """
-
-    # Insert Models data
-    models_sql = """
-    INSERT INTO Models (id, ai_model_name, provider_id)
-    VALUES
-        (1, 'gpt4', 4),
-        (2, 'gpt4o', 4),
-        (3, 'gpt4o-mini', 4),
-        (4, 'llama3.1:8b', 1),
-        (5, 'Qwen/Qwen2-7B-Instruct', 2),
-        (6, 'glm-4-alltools', 3),
-        (7, 'glm-4-flash', 3),
-        (8, 'glm-4-0520', 3),
-        (9, 'glm-4-plus', 3),
-        (10, 'glm-4v-plus', 3),
-        (11, 'glm-4', 3),
-        (12, 'glm-4v', 3)
-    ON CONFLICT (id) DO NOTHING;
-    """
-
-    # Execute the SQL statements
-    session.exec(text(model_provider_sql))
-    session.exec(text(models_sql))
-    session.commit()
+    # 获取所有提供商配置
+    providers = model_provider_manager.get_all_providers()
+
+    # 按照提供商名称排序，确保处理顺序一致
+    for provider_name in sorted(providers.keys()):
+        provider_data = providers[provider_name]
+
+        # 查找现有的提供商记录
+        db_provider = session.exec(select(ModelProvider).where(ModelProvider.provider_name == provider_data['provider_name'])).first()
+
+        if db_provider:
+            # 更新提供商信息，但保留现有的 API 密钥和基础 URL
+            db_provider.icon = provider_data['icon']
+            db_provider.description = provider_data['description']
+            # 注意：我们不更新 api_key 和 base_url，因为它们可能已被用户修改
+        else:
+            # 如果提供商不存在，创建新记录
+            db_provider = ModelProvider(
+                provider_name=provider_data['provider_name'],
+                base_url=provider_data['base_url'],
+                api_key=provider_data['api_key'],
+                icon=provider_data['icon'],
+                description=provider_data['description']
+            )
+            session.add(db_provider)
+
+        session.flush()  # 确保 provider_id 已生成
+
+        # 获取该提供商支持的模型
+        supported_models = set(model_provider_manager.get_supported_models(provider_name))
+
+        # 获取数据库中该提供商现有的模型
+        existing_models = set(model.ai_model_name for model in session.exec(select(Models).where(Models.provider_id == db_provider.id)))
+
+        # 添加新模型
+        for model_name in sorted(supported_models - existing_models):
+            new_model = Models(ai_model_name=model_name, provider_id=db_provider.id)
+            session.add(new_model)
+
+        # 删除不再支持的模型
+        for model_name in sorted(existing_models - supported_models):
+            session.exec(select(Models).where(Models.ai_model_name == model_name, Models.provider_id == db_provider.id)).delete()
+
+    session.commit()
+
+    # 打印当前数据库状态，用于验证
+    providers = session.exec(select(ModelProvider).order_by(ModelProvider.id)).all()
+    for provider in providers:
+        print(f"Provider: {provider.provider_name} (ID: {provider.id})")
+        print(f"  Base URL: {provider.base_url}")
+        print(f"  API Key: {'*' * len(provider.api_key)}")  # 出于安全考虑，不打印实际的 API 密钥
+        models = session.exec(select(Models).where(Models.provider_id == provider.id).order_by(Models.id)).all()
+        for model in models:
+            print(f"  - Model: {model.ai_model_name} (ID: {model.id})")
+

backend/app/core/model_providers/__init__.py

@@ -0,0 +1,8 @@
+import importlib
+
+def get_provider_config(provider_name: str):
+    try:
+        module = importlib.import_module(f'app.core.model_providers.{provider_name}.config')
+        return module.PROVIDER_CONFIG
+    except ImportError:
+        raise ValueError(f"No configuration found for provider: {provider_name}")

backend/app/core/model_providers/model_provider_manager.py

@@ -0,0 +1,48 @@
+import os
+import importlib
+from typing import Dict, Any, List, Callable
+
+class ModelProviderManager:
+    def __init__(self):
+        self.providers: Dict[str, Dict[str, Any]] = {}
+        self.models: Dict[str, List[str]] = {}
+        self.init_functions: Dict[str, Callable] = {}
+        self.load_providers()
+
+    def load_providers(self):
+        providers_dir = os.path.dirname(os.path.abspath(__file__))
+        for item in os.listdir(providers_dir):
+            if os.path.isdir(os.path.join(providers_dir, item)) and not item.startswith("__"):
+                try:
+                    module = importlib.import_module(f".{item}.config", package="app.core.model_providers")
+                    provider_config = getattr(module, 'PROVIDER_CONFIG', None)
+                    supported_models = getattr(module, 'SUPPORTED_MODELS', [])
+                    init_function = getattr(module, 'init_model', None)
+
+                    if provider_config and init_function:
+                        self.providers[item] = provider_config
+                        self.models[item] = supported_models
+                        self.init_functions[item] = init_function
+                except ImportError as e:
+                    print(f"Failed to load provider config for {item}: {e}")
+
+    def get_provider_config(self, provider_name: str) -> Dict[str, Any]:
+        return self.providers.get(provider_name, {})
+
+    def get_supported_models(self, provider_name: str) -> List[str]:
+        return self.models.get(provider_name, [])
+
+    def get_all_providers(self) -> Dict[str, Dict[str, Any]]:
+        return self.providers
+
+    def get_all_models(self) -> Dict[str, List[str]]:
+        return self.models
+
+    def init_model(self, provider_name: str, model: str, temperature: float, **kwargs):
+        init_function = self.init_functions.get(provider_name)
+        if init_function:
+            return init_function(model, temperature, **kwargs)
+        else:
+            raise ValueError(f"No initialization function found for provider: {provider_name}")
+
+model_provider_manager = ModelProviderManager()

backend/app/core/model_providers/ollama/config.py

@@ -0,0 +1,21 @@
+from langchain_ollama import ChatOllama
+
+PROVIDER_CONFIG = {
+    'provider_name': 'Ollama',
+    'base_url': 'http://host.docker.internal:11434',
+    'api_key': 'fakeapikey',
+    'icon': 'ollama_icon',
+    'description': 'Ollama API provider'
+}
+
+SUPPORTED_MODELS = [
+    'llama3.1:8b',
+]
+
+def init_model(model: str, temperature: float, **kwargs):
+    return ChatOllama(
+        model=model,
+        temperature=temperature,
+        base_url=PROVIDER_CONFIG['base_url'],
+        **kwargs
+    )

backend/app/core/model_providers/openai/config.py

@@ -0,0 +1,29 @@
+from langchain_openai import ChatOpenAI
+
+PROVIDER_CONFIG = {
+    "provider_name": "openai",
+    "base_url": "https://api.openai.com/v1",
+    "api_key": "your_api_key_here",
+    "icon": "openai_icon",
+    "description": "OpenAI API provider",
+}
+
+SUPPORTED_MODELS = [
+    "gpt-4",
+    "gpt-4-0314",
+    "gpt-4-32k",
+    "gpt-4-32k-0314",
+    "gpt-3.5-turbo",
+    "gpt-3.5-turbo-16k",
+    "gpt-4o-mini",
+]
+
+
+def init_model(model: str, temperature: float, **kwargs):
+    return ChatOpenAI(
+        model=model,
+        temperature=temperature,
+        openai_api_key=PROVIDER_CONFIG["api_key"],
+        openai_api_base=PROVIDER_CONFIG["base_url"],
+        **kwargs
+    )

backend/app/core/model_providers/qwen/config.py

@@ -0,0 +1,24 @@
+from langchain_openai import ChatOpenAI
+
+PROVIDER_CONFIG = {
+    "provider_name": "Qwen",
+    "base_url": "fakeurl",
+    "api_key": "fakeapikey",
+    "icon": "qwen_icon",
+    "description": "qwen  API provider",
+}
+
+SUPPORTED_MODELS = [
+    "Qwen2-7B-Instruct",
+    "Qwen2.5-70B",
+]
+
+
+def init_model(model: str, temperature: float, **kwargs):
+    return ChatOpenAI(
+        model=model,
+        temperature=temperature,
+        openai_api_key=PROVIDER_CONFIG["api_key"],
+        openai_api_base=PROVIDER_CONFIG["base_url"],
+        **kwargs
+    )

backend/app/core/model_providers/siliconflow/config.py

@@ -0,0 +1,22 @@
+from langchain_openai import ChatOpenAI
+
+PROVIDER_CONFIG = {
+    'provider_name': 'Siliconflow',
+    'base_url': 'fakeurl',
+    'api_key': 'fakeapikey',
+    'icon': 'siliconflow_icon',
+    'description': 'Siliconflow API provider'
+}
+
+SUPPORTED_MODELS = [
+    'Qwen/Qwen2-7B-Instruct',
+]
+
+def init_model(model: str, temperature: float, **kwargs):
+    return ChatOpenAI(
+        model=model,
+        temperature=temperature,
+        openai_api_key=PROVIDER_CONFIG['api_key'],
+        openai_api_base=PROVIDER_CONFIG['base_url'],
+        **kwargs
+    )

backend/app/core/model_providers/zhipuai/config.py

@@ -0,0 +1,29 @@
+from langchain_openai import ChatOpenAI
+
+PROVIDER_CONFIG = {
+    "provider_name": "zhipuai",
+    "base_url": "https://open.bigmodel.cn/api/paas/v4",
+    "api_key": "5e867fc4396cff20bc3431d39e8c240f.d5Y8YBqIawDigP46",
+    "icon": "zhipuai_icon",
+    "description": "智谱AI",
+}
+
+SUPPORTED_MODELS = [
+    "glm-4-alltools",
+    "glm-4-flash",
+    "glm-4-0520",
+    "glm-4-plus",
+    "glm-4v-plus",
+    "glm-4",
+    "glm-4v",
+]
+
+
+def init_model(model: str, temperature: float, **kwargs):
+    return ChatOpenAI(
+        model=model,
+        temperature=temperature,
+        openai_api_key=PROVIDER_CONFIG["api_key"],
+        openai_api_base=PROVIDER_CONFIG["base_url"],
+        **kwargs
+    )