540 lines
16 KiB
Markdown
540 lines
16 KiB
Markdown
# 扩展性设计
|
||
|
||
本章详细描述系统的扩展性设计，包括模块可插拔、数据库驱动可切换、中间件可插拔三大扩展维度。
|
||
|
||
:::tip 扩展性设计原则
|
||
遵循开闭原则（OCP）：对扩展开放，对修改关闭。新功能通过添加新模块实现，而非修改现有代码。核心框架通过基类、装饰器、中间件等机制提供扩展点，业务模块只需声明差异即可接入。
|
||
:::
|
||
|
||
## 一、模块可插拔设计
|
||
|
||
### 模块标准结构
|
||
|
||
每个业务模块遵循固定的四文件结构，新增模块只需创建这四个文件即可接入系统：
|
||
|
||
```
|
||
src/modules/{group}/{name}/
|
||
├── models.py # ORM 模型
|
||
├── repository.py # 数据访问层
|
||
├── schemas.py # 请求校验
|
||
└── service.py # 业务逻辑层
|
||
```
|
||
|
||
### 新增模块步骤
|
||
|
||
以新增「培训管理（Training）」模块为例。
|
||
|
||
**步骤一：创建模块目录和文件**
|
||
|
||
创建 `src/modules/training/` 目录，包含以下四个文件：
|
||
|
||
`models.py` — ORM 模型：
|
||
|
||
```python
|
||
# src/modules/training/models.py
|
||
from sqlalchemy import Column, String, Integer, text
|
||
from core.base_model import BaseModel
|
||
from core.base_model import base_model
|
||
from core.config import DB_PREFIX
|
||
|
||
|
||
class Training(BaseModel):
|
||
__tablename__ = DB_PREFIX + "training"
|
||
__table_comment__ = "培训记录表"
|
||
|
||
title = Column(String(200), nullable=False, comment="培训标题")
|
||
content = Column(String(2000), nullable=True, comment="培训内容")
|
||
trainer = Column(String(50), nullable=True, comment="培训讲师")
|
||
status = Column(Integer, default=1, server_default=text('1'), comment="状态")
|
||
sort = Column(Integer, default=0, server_default=text('0'), comment="排序")
|
||
```
|
||
|
||
`repository.py` — 数据访问层：
|
||
|
||
```python
|
||
# src/modules/training/repository.py
|
||
from training.models import Training
|
||
from core.base_repository import BaseRepository
|
||
|
||
|
||
class TrainingRepository(BaseRepository(Training]):
|
||
pass
|
||
|
||
|
||
training_repo = TrainingRepository(Training)
|
||
```
|
||
|
||
`schemas.py` — 请求校验：
|
||
|
||
```python
|
||
# src/modules/training/schemas.py
|
||
from pydantic import BaseModel, Field
|
||
from core.base_schema import BaseSchema
|
||
|
||
|
||
class TrainingForm(BaseSchema):
|
||
title: str = Field(..., min_length=1, max_length=200, description="培训标题")
|
||
content: str = Field(None, max_length=2000, description="培训内容")
|
||
trainer: str = Field(None, max_length=50, description="培训讲师")
|
||
status: int = Field(..., ge=1, le=2, description="状态")
|
||
sort: int = Field(..., ge=0, le=99999, description="排序")
|
||
|
||
|
||
class TrainingStatusForm(BaseModel):
|
||
id: int = Field(..., gt=0, description="培训ID")
|
||
status: int = Field(..., ge=1, le=2, description="状态")
|
||
```
|
||
|
||
`service.py` — 业务逻辑层：
|
||
|
||
```python
|
||
# src/modules/training/service.py
|
||
from core.base_service import BaseService
|
||
from training.models import Training
|
||
from training.repository import training_repo
|
||
|
||
|
||
class TrainingService(BaseService(Training]):
|
||
repo = training_repo
|
||
model = Training
|
||
page_like_fields = ('title', 'trainer')
|
||
page_eq_fields = ('status',)
|
||
page_order_by = (('sort', 'asc'),)
|
||
unique_fields = {'title': '培训标题不能重复'}
|
||
|
||
|
||
training_service = TrainingService()
|
||
```
|
||
|
||
**步骤二：创建 Handler**
|
||
|
||
创建 `src/modules/training/handlers.py`：
|
||
|
||
```python
|
||
# src/modules/training/handlers.py
|
||
from common.middleware.access_decorators import check_demo, permission_required
|
||
from common.middleware.operation_log import operation_log
|
||
from core import response as R
|
||
from core.base_handler import BaseHandler
|
||
from modules.training.service import training_service
|
||
|
||
|
||
class TrainingPageHandler(BaseHandler):
|
||
@permission_required("sys:training:page")
|
||
async def get(self):
|
||
return await training_service.get_page(self)
|
||
|
||
|
||
class TrainingDetailHandler(BaseHandler):
|
||
@permission_required("sys:training:detail")
|
||
async def get(self, id):
|
||
return R.ok(self, data=training_service.get_detail(id))
|
||
|
||
|
||
class TrainingAddHandler(BaseHandler):
|
||
@permission_required("sys:training:add")
|
||
@check_demo
|
||
@operation_log("培训管理", "添加")
|
||
async def post(self):
|
||
return await training_service.add(self)
|
||
|
||
|
||
class TrainingUpdateHandler(BaseHandler):
|
||
@permission_required("sys:training:update")
|
||
@check_demo
|
||
@operation_log("培训管理", "更新")
|
||
async def put(self):
|
||
return await training_service.update(self)
|
||
|
||
|
||
class TrainingDeleteHandler(BaseHandler):
|
||
@permission_required("sys:training:delete")
|
||
@check_demo
|
||
@operation_log("培训管理", "删除")
|
||
async def delete(self, id):
|
||
return await training_service.delete(self, id)
|
||
|
||
|
||
class TrainingStatusHandler(BaseHandler):
|
||
@permission_required("sys:training:status")
|
||
@check_demo
|
||
@operation_log("培训管理", "设置状态")
|
||
async def put(self):
|
||
return await training_service.update_status(self)
|
||
|
||
|
||
class TrainingBatchDeleteHandler(BaseHandler):
|
||
@permission_required("sys:training:delete")
|
||
@check_demo
|
||
@operation_log("培训管理", "批量删除")
|
||
async def post(self):
|
||
return await training_service.batch_delete(self)
|
||
```
|
||
|
||
**步骤三：注册路由**
|
||
|
||
在 `src/api/v1/` 下创建路由文件，并在 `router.py` 中注册：
|
||
|
||
```python
|
||
# src/api/v1/training.py
|
||
from modules.training.handlers import (
|
||
TrainingAddHandler, TrainingBatchDeleteHandler, TrainingDeleteHandler,
|
||
TrainingDetailHandler, TrainingPageHandler, TrainingStatusHandler,
|
||
TrainingUpdateHandler,
|
||
)
|
||
|
||
routes = [
|
||
(r"/training/page", TrainingPageHandler),
|
||
(r"/training/detail/(\d+)", TrainingDetailHandler),
|
||
(r"/training/add", TrainingAddHandler),
|
||
(r"/training/update", TrainingUpdateHandler),
|
||
(r"/training/delete/(\d+)", TrainingDeleteHandler),
|
||
(r"/training/status", TrainingStatusHandler),
|
||
(r"/training/batchDelete", TrainingBatchDeleteHandler),
|
||
]
|
||
```
|
||
|
||
```python
|
||
# src/api/v1/router.py
|
||
from api.v1 import training
|
||
|
||
_RESOURCE_MODULES = [
|
||
# ... 其他模块 ...
|
||
training,
|
||
]
|
||
```
|
||
|
||
**步骤四：配置菜单权限**
|
||
|
||
在数据库 `tornado_menu` 表中插入对应的菜单和权限节点记录。权限标识格式为 `sys:{module}:{action}`，如 `sys:training:add`、`sys:training:page`。
|
||
|
||
### 模块解耦
|
||
|
||
模块之间通过以下方式解耦：
|
||
|
||
| 解耦方式 | 说明 | 示例 |
|
||
|:---|:---|:---|
|
||
| **Repository 单例** | 模块级单例，通过 import 引用 | `from modules.user.repository import UserRepository` |
|
||
| **Service 单例** | 模块级单例，通过 import 引用 | `from user.service import user_service` |
|
||
| **跨模块引用** | 通过 Repository 引用其他模块的模型 | PositionService 引用 User 模型检查引用 |
|
||
| **事件机制** | 暂未实现（未来可通过事件总线解耦） | — |
|
||
|
||
### 复杂模块扩展
|
||
|
||
对于业务逻辑复杂的模块（如 User、Role、Menu），在继承 `BaseService` 的基础上添加自定义方法：
|
||
|
||
```python
|
||
# src/modules/user/service.py
|
||
class UserService(BaseService(User]):
|
||
repo = user_repo
|
||
model = User
|
||
page_like_fields = ('username', 'realname')
|
||
page_eq_fields = ('status', 'dept_id')
|
||
|
||
# 自定义方法：用户详情（含角色列表）
|
||
def get_user_detail(self, user_id):
|
||
user = self.repo.get_by_id(user_id)
|
||
if not user:
|
||
return None
|
||
data = user.to_dict()
|
||
data['roleIds'] = get_user_role_ids(user_id)
|
||
data['roleNames'] = get_user_role_names(user_id)
|
||
return data
|
||
|
||
# 自定义方法：修改密码
|
||
def change_password(self, user_id, old_pwd, new_pwd):
|
||
user = self.repo.get_by_id(user_id)
|
||
if not verify_password(old_pwd, user.password):
|
||
return R.failed("原密码错误")
|
||
user.password = hash_password(new_pwd)
|
||
user_repo.create(user)
|
||
return R.ok(msg="密码修改成功")
|
||
|
||
# 覆盖钩子：新增前哈希密码
|
||
def _before_add(self, request, data):
|
||
data.password = hash_password(data.password or DEFAULT_PASSWORD)
|
||
|
||
# 覆盖钩子：删除前检查
|
||
def _before_delete(self, ids):
|
||
if 1 in parse_id_list(str(ids)):
|
||
return "超级管理员不能删除"
|
||
return None
|
||
```
|
||
|
||
## 二、数据库驱动可切换
|
||
|
||
### 驱动注册机制
|
||
|
||
项目通过 `_SUPPORTED_DRIVERS` 声明支持的驱动，`_normalize_driver()` 处理别名映射和校验，`build_database_url()` 按驱动构建连接串：
|
||
|
||
```python
|
||
# src/config/database.py（简化）
|
||
import os
|
||
from urllib.parse import quote_plus
|
||
|
||
# 支持的数据库驱动
|
||
_SUPPORTED_DRIVERS = ("mysql", "postgresql", "mssql", "sqlite", "oracle")
|
||
|
||
|
||
def _normalize_driver(driver: str) -> str:
|
||
"""归一化 DB_DRIVER：别名映射 + 校验"""
|
||
name = str(driver).strip().lower()
|
||
alias_map = {
|
||
'postgres': 'postgresql',
|
||
'pg': 'postgresql',
|
||
'sqlserver': 'mssql',
|
||
'sql_server': 'mssql',
|
||
'sqlite3': 'sqlite',
|
||
}
|
||
name = alias_map.get(name, name)
|
||
if name not in _SUPPORTED_DRIVERS:
|
||
raise ValueError(f"不支持的 DB_DRIVER: {driver!r}")
|
||
return name
|
||
|
||
|
||
DB_DRIVER = _normalize_driver(os.getenv('DB_DRIVER', 'mysql'))
|
||
DB_HOST = os.getenv('DB_HOST', '127.0.0.1')
|
||
DB_PORT = int(os.getenv('DB_PORT', '3306'))
|
||
DB_DATABASE = os.getenv('DB_DATABASE', 'djangoadmin.tornado.elevue')
|
||
DB_USERNAME = os.getenv('DB_USERNAME', 'root')
|
||
DB_PASSWORD = os.getenv('DB_PASSWORD', '')
|
||
|
||
|
||
def build_database_url(driver: str = None) -> str:
|
||
"""按驱动构建 SQLAlchemy 连接串"""
|
||
drv = _normalize_driver(driver if driver is not None else DB_DRIVER)
|
||
|
||
if drv == 'sqlite':
|
||
return 'sqlite:///./' + DB_DATABASE
|
||
|
||
base = quote_plus(DB_USERNAME) + ':' + quote_plus(DB_PASSWORD) + '@' + DB_HOST + ':' + str(DB_PORT)
|
||
|
||
if drv == 'mysql':
|
||
return 'mysql+pymysql://' + base + '/' + DB_DATABASE + '?charset=utf8mb4'
|
||
if drv == 'postgresql':
|
||
return 'postgresql+psycopg://' + base + '/' + DB_DATABASE
|
||
if drv == 'oracle':
|
||
return 'oracle+oracledb://' + base + '/?service_name=' + quote_plus(DB_DATABASE)
|
||
# mssql
|
||
return 'mssql+pymssql://' + base + '/' + DB_DATABASE + '?charset=utf8'
|
||
```
|
||
|
||
各驱动的连接串格式：
|
||
|
||
| 驱动 | 连接串格式 | Python 包 |
|
||
|:----|:----|:----|
|
||
| mysql | `mysql+pymysql://user:pass@host:port/db?charset=utf8mb4` | PyMySQL |
|
||
| postgresql | `postgresql+psycopg://user:pass@host:port/db` | psycopg[binary] |
|
||
| mssql | `mssql+pymssql://user:pass@host:port/db?charset=utf8` | pymssql |
|
||
| oracle | `oracle+oracledb://user:pass@host:port/?service_name=db` | oracledb |
|
||
| sqlite | `sqlite:///./path/to/db` | 内置 |
|
||
|
||
### 切换方式
|
||
|
||
只需修改 `.env` 文件中的 `DB_DRIVER` 和对应的连接参数：
|
||
|
||
```bash
|
||
# MySQL（默认）
|
||
DB_DRIVER=mysql
|
||
DB_HOST=127.0.0.1
|
||
DB_PORT=3306
|
||
DB_NAME=djangoadmin
|
||
DB_USER=root
|
||
DB_PASSWORD=your_password
|
||
|
||
# PostgreSQL
|
||
DB_DRIVER=postgresql
|
||
DB_HOST=127.0.0.1
|
||
DB_PORT=5432
|
||
DB_NAME=djangoadmin
|
||
DB_USER=postgres
|
||
DB_PASSWORD=your_password
|
||
|
||
# SQLite
|
||
DB_DRIVER=sqlite
|
||
DB_DATABASE=./data.db
|
||
|
||
# SQL Server
|
||
DB_DRIVER=mssql
|
||
DB_HOST=127.0.0.1
|
||
DB_PORT=1433
|
||
DB_NAME=djangoadmin
|
||
DB_USER=sa
|
||
DB_PASSWORD=your_password
|
||
```
|
||
|
||
:::warning 驱动切换注意事项
|
||
1. 切换驱动后需安装对应的 Python 包（如 `psycopg[binary]` for PostgreSQL）
|
||
2. 数据迁移使用 `scripts/migrate_db.py` 工具
|
||
3. SQLite 不支持并发写入，仅适用于开发/测试环境
|
||
4. Oracle 需要额外的客户端库配置
|
||
:::
|
||
|
||
### 跨库数据迁移
|
||
|
||
```bash
|
||
# 干跑：核对迁移计划
|
||
python scripts/migrate_db.py --src-driver mysql --src-host ... --dst-driver postgresql --dst-host ... --dry-run
|
||
|
||
# 执行迁移
|
||
python scripts/migrate_db.py --src-driver mysql --src-host ... --dst-driver postgresql --dst-host ...
|
||
|
||
# 幂等重跑（先删除目标库数据）
|
||
python scripts/migrate_db.py --src-driver mysql --src-host ... --dst-driver postgresql --dst-host ... --drop-target-first
|
||
```
|
||
|
||
## 三、横切关注点扩展
|
||
|
||
### Handler 生命周期
|
||
|
||
Tornado 通过 Handler 生命周期方法实现横切关注点：
|
||
|
||
```python
|
||
# src/core/base_handler.py
|
||
class BaseHandler(tornado.web.RequestHandler):
|
||
def set_default_headers(self):
|
||
"""设置响应头：CORS + trace_id"""
|
||
cros_required(self)
|
||
apply_trace_id(self)
|
||
|
||
@login_required
|
||
def prepare(self):
|
||
"""请求预处理：JWT 认证"""
|
||
pass
|
||
|
||
def on_finish(self):
|
||
"""请求结束后：释放数据库会话"""
|
||
db.close()
|
||
```
|
||
|
||
### 自定义装饰器
|
||
|
||
通过 `permission_required` 和 `check_demo` 的模式，可以创建自定义装饰器：
|
||
|
||
```python
|
||
# src/common/middleware/operation_log.py
|
||
def operation_log(module: str, action: str):
|
||
"""操作日志记录装饰器"""
|
||
def decorator(func):
|
||
@wraps(func)
|
||
async def wrapper(self, *args, **kwargs):
|
||
start = time.time()
|
||
result = await func(self, *args, **kwargs)
|
||
duration = time.time() - start
|
||
save_log(module, action, duration)
|
||
return result
|
||
return wrapper
|
||
return decorator
|
||
```
|
||
|
||
使用示例：
|
||
|
||
```python
|
||
class TrainingAddHandler(BaseHandler):
|
||
@permission_required("sys:training:add")
|
||
@check_demo
|
||
@operation_log("培训管理", "添加")
|
||
async def post(self):
|
||
return await training_service.add(self)
|
||
```
|
||
|
||
装饰器执行顺序（从外到内）：`@permission_required` → `@check_demo` → `@operation_log` → 方法体。
|
||
|
||
## 四、装饰器扩展
|
||
|
||
### 自定义装饰器
|
||
|
||
通过 `permission_required` 和 `check_demo` 的模式，可以创建自定义装饰器：
|
||
|
||
```python
|
||
# src/common/decorators.py
|
||
|
||
# 自定义装饰器示例：操作日志记录
|
||
def log_operation(module: str, action: str):
|
||
"""
|
||
操作日志记录装饰器
|
||
自动记录接口调用的模块、操作类型、执行耗时
|
||
|
||
Args:
|
||
module: 所属模块名称，如 "培训管理"
|
||
action: 操作类型，如 "新增"、"编辑"、"删除"
|
||
"""
|
||
def decorator(func):
|
||
@wraps(func)
|
||
async def wrapper(*args, **kwargs):
|
||
# 记录开始时间，用于计算接口执行耗时
|
||
start = time.time()
|
||
# 执行原业务函数
|
||
result = await func(*args, **kwargs)
|
||
# 计算耗时（秒）
|
||
duration = time.time() - start
|
||
# 异步记录操作日志到数据库（包含模块、操作、耗时等信息）
|
||
save_log(module, action, duration)
|
||
return result
|
||
return wrapper
|
||
return decorator
|
||
|
||
|
||
# src/modules/training/handlers.py
|
||
|
||
# 使用示例：多个装饰器叠加使用
|
||
# 执行顺序：从外到内（先 permission_required，再 check_demo，再 operation_log）
|
||
class TrainingAddHandler(BaseHandler):
|
||
@permission_required("sys:training:add") # 权限校验
|
||
@check_demo # 演示环境拦截
|
||
@operation_log("培训管理", "添加") # 操作日志记录
|
||
async def post(self):
|
||
"""新增培训记录"""
|
||
return await training_service.add(self)
|
||
```
|
||
|
||
## 五、配置扩展
|
||
|
||
### 环境变量驱动
|
||
|
||
所有可配置项通过 `.env` 文件管理，新增配置项只需：
|
||
|
||
1. 在 `.env` 文件中添加变量
|
||
2. 在 `src/config/` 对应模块中读取
|
||
3. 在业务代码中引用
|
||
|
||
```bash
|
||
# .env
|
||
CUSTOM_FEATURE_ENABLED=true
|
||
CUSTOM_TIMEOUT=30
|
||
```
|
||
|
||
```python
|
||
# src/config/custom.py
|
||
CUSTOM_FEATURE_ENABLED = os.getenv("CUSTOM_FEATURE_ENABLED", "false").lower() == "true"
|
||
CUSTOM_TIMEOUT = int(os.getenv("CUSTOM_TIMEOUT", "30"))
|
||
```
|
||
|
||
## 六、前端扩展
|
||
|
||
### 新增页面
|
||
|
||
1. 在 `ui/src/api/` 创建 API 文件
|
||
2. 在 `ui/src/views/` 创建页面组件
|
||
3. 在数据库 `tornado_menu` 表中配置菜单路由
|
||
|
||
### 新增组件
|
||
|
||
公共组件放在 `ui/src/components/` 目录，遵循 Vue3 组件规范。
|
||
|
||
## 扩展性矩阵
|
||
|
||
| 扩展维度 | 扩展方式 | 修改范围 | 复杂度 |
|
||
|:---|:---|:---|:---|
|
||
| 新增 CRUD 模块 | 创建 4 文件 + 注册路由 | 新增文件，不改现有代码 | 低 |
|
||
| 新增复杂模块 | 继承基类 + 自定义方法 | 新增文件，不改现有代码 | 中 |
|
||
| 切换数据库驱动 | 修改 .env 配置 | 仅改配置 | 低 |
|
||
| 新增中间件 | 创建中间件 + 注册 | 新增文件 + 改 app.py | 低 |
|
||
| 新增装饰器 | 创建装饰器 + 应用 | 新增文件 | 低 |
|
||
| 新增配置项 | .env + config 模块 | 新增配置 | 低 |
|
||
| 新增前端页面 | API + Views + Menu | 新增文件 + 改数据库 | 中 |
|
||
|
||
## 总结
|
||
|
||
通过分层架构、基类模板方法、装饰器、中间件、环境变量配置等机制，实现了高度的可扩展性。新增 CRUD 模块只需 4 个标准文件 + 1 行路由注册，不修改任何现有代码。数据库驱动通过环境变量切换，中间件通过注册机制插拔。这种设计使得系统在功能持续增长的同时，保持代码结构的清晰和稳定。
|