Cherrypy 在 _cp_dispatch/popargs 中使用插件

Question

我将 cherrypy 与 sqlalchemy 结合使用，以构建无头（仅 CLI 客户端）restful 服务器。

我使用以下 receipe 将 sqlalchemy 绑定到 cherrypy 引擎：https://bitbucket.org/Lawouach/cherrypy-recipes/src/c8290261eefb/web/database/sql_alchemy/ 收据略有修改，以便在不存在的情况下构建数据库。

服务器公开了几个 uri，例如客户端、文章、商店...

GET /clients 获取客户端列表
GET /clients/1 获取客户端 ID #1
GET /clients/foo 获取名为 foo
PUT /clients/1 更新客户端#1
DELETE /client/foo 删除名为 foo 的客户端 ...

我打算使用装饰器 popargs 和 _cp_dispatch 以便提前将我的资源名称转换为它们的 ID处理。我使用 cherrypy.dispatch.MethodDispatcher 作为调度程序（所以我可以编写 GET/POST/PUT/DELETE 方法）

我可以通过任何 GET/POST/PUT/DELETE 方法访问该插件，但我无法通过 _cp_dispatch.

访问它

知道如何在进入 GET/POST/PUT/DELETE 方法之前将资源名称转换为它们的 ID 吗？

这是我的问题的重现者

$ tree
.
├── __init__.py
├── models.py
├── my.db
├── root.py
├── saplugin.py
├── saplugin.py.original
└── satool.py

我的 sqlalchemy 模型（我有几个，重现问题只需要一个）

$ cat models.py
# -*- coding: utf-8 -*-
from sqlalchemy.ext.declarative import declarative_base
from sqlalchemy import Column, Integer, String
Base = declarative_base()

class User(Base):

    __tablename__ = 'users'

    id = Column(Integer, primary_key=True)
    name = Column(String)

这里是服务器的主脚本

$ cat root.py
# -*- coding: utf-8 -*-

import cherrypy
from models import User
from saplugin import SAEnginePlugin
from satool import SATool

def name_to_id(param, table):
    if not param.isdigit():
        param = cherrypy.request.db.query(table.id).\
                filter(table.name == param).one()
    return param

class Root(object):

    exposed = True

    def __init__(self):
        self.users = Users()

    def GET(self):
        # Get the SQLAlchemy session associated
        # with this request.
        # It'll be released once the request
        # processing terminates
        return "Hello World"


class Users(object):

    exposed = True

    @cherrypy.popargs('user_id')
    def GET(self, user_id=None, **kwargs):
        user_id = name_to_id(user_id, User)
        return "I'm resource %s" % user_id

if __name__ == '__main__':
    # Register the SQLAlchemy plugin
    SAEnginePlugin(cherrypy.engine).subscribe()

    # Register the SQLAlchemy tool
    cherrypy.tools.db = SATool()

    cherrypy.quickstart(Root(), '', {'/': {'tools.db.on': True,
                                        "request.dispatch": cherrypy.dispatch.MethodDispatcher()}})

这是应用于 sqlalchemy 插件 receipe 的修改

$ diff -up saplugin.py.original saplugin.py
--- saplugin.py.original    2015-06-15 18:14:45.469706863 +0200
+++ saplugin.py 2015-06-15 18:14:37.042741785 +0200
@@ -3,6 +3,7 @@ import cherrypy
from cherrypy.process import wspbus, plugins
from sqlalchemy import create_engine
from sqlalchemy.orm import scoped_session, sessionmaker
+from models import User, Base

__all__ = ['SAEnginePlugin']

@@ -26,7 +27,16 @@ class SAEnginePlugin(plugins.SimplePlugi
        self.sa_engine = create_engine('sqlite:///my.db', echo=False)
        self.bus.subscribe("bind-session", self.bind)
        self.bus.subscribe("commit-session", self.commit)
- 
+        # Creating the database
+        self.bus.log('Creating database')
+        self.bus.subscribe("create-all", self.create_all)
+        try:
+            self.create_all()
+        except Exception as err:
+            logging.error("Can't start")
+            logging.error(err.message)
+            sys.exit(1)
+
    def stop(self):
        self.bus.log('Stopping down DB access')
        self.bus.unsubscribe("bind-session", self.bind)
@@ -59,4 +69,8 @@ class SAEnginePlugin(plugins.SimplePlugi
            raise
        finally:
            self.session.remove()
-    
+
+    def create_all(self):
+        """ create database structure """
+        self.bus.log('Creating database')
+        Base.metadata.create_all(self.sa_engine)

这是服务器的日志，当我在 GET 方法的开头转换名称时。

$ python root.py 
[15/Jun/2015:18:16:23] ENGINE Listening for SIGHUP.
[15/Jun/2015:18:16:23] ENGINE Listening for SIGTERM.
[15/Jun/2015:18:16:23] ENGINE Listening for SIGUSR1.
[15/Jun/2015:18:16:23] ENGINE Bus STARTING
[15/Jun/2015:18:16:23] ENGINE Starting up DB access
[15/Jun/2015:18:16:23] ENGINE Creating database
[15/Jun/2015:18:16:23] ENGINE Creating database
[15/Jun/2015:18:16:23] ENGINE Started monitor thread 'Autoreloader'.
[15/Jun/2015:18:16:23] ENGINE Started monitor thread '_TimeoutMonitor'.
[15/Jun/2015:18:16:23] ENGINE Serving on 127.0.0.1:8080
[15/Jun/2015:18:16:23] ENGINE Bus STARTED
127.0.0.1 - - [15/Jun/2015:18:16:26] "GET /users/1 HTTP/1.1" 200 14 "" "curl/7.29.0"
127.0.0.1 - - [15/Jun/2015:18:16:28] "GET /users/foo HTTP/1.1" 200 14 "" "curl/7.29.0"

这里是查询之前的日志

$ curl 127.0.0.1:8080/users/1; echo
I'm resource 1
$ curl 127.0.0.1:8080/users/foo; echo
I'm resource 1

这里是数据库的内容

$ sqlite3 my.db 
sqlite> .schema
CREATE TABLE users (
    id INTEGER NOT NULL, 
    name VARCHAR, 
    PRIMARY KEY (id)
);
sqlite> select * from users;
1|foo
2|bar

我不能同时使用装饰器 popargs 和 _cp_dispatch 方法。我还没有找到如何使用装饰器处理我的 url 段。当我尝试只使用 _cp_dispatch 方法时，我遇到了以下错误：

File "/usr/lib/python2.7/site-packages/cherrypy/_cprequest.py", line 628, in respond
    self.get_resource(path_info)
File "/usr/lib/python2.7/site-packages/cherrypy/_cprequest.py", line 744, in get_resource
    dispatch(path)
File "/usr/lib/python2.7/site-packages/cherrypy/_cpdispatch.py", line 423, in __call__
    resource, vpath = self.find_handler(path_info)
File "/usr/lib/python2.7/site-packages/cherrypy/_cpdispatch.py", line 311, in find_handler
    subnode = dispatch(vpath=iternames)
File "root.py", line 49, in _cp_dispatch
    vpath[0] = name_to_id(vpath[0], Users)
File "root.py", line 10, in name_to_id
    param = cherrypy.request.db.query(table.id).\
File "/usr/lib/python2.7/site-packages/cherrypy/__init__.py", line 208, in __getattr__
    return getattr(child, name)
AttributeError: 'Request' object has no attribute 'db'

这是应用于用户的修改 class:

class Users(object):

    # [....]

    def _cp_dispatch(self, vpath):
        if len(vpath) > 1:
            raise
        vpath[0] = name_to_id(vpath[0], Users)
        return vpath

我正在使用以下版本（我正在 Centos 7 环境中工作，无法使用 pip 中的那些进行更改）： $ rpm -qa | egrep "cherrypy|sqlalchemy" python-cherrypy-3.2.2-4.el7.noarch python-sqlalchemy-0.9.7-3.el7.x86_64

非常感谢您的帮助！！！！

Answer 1

documentation.

中描述了此问题的解决方案

我已将以下工具添加到根脚本中：

class UserManager(cherrypy.Tool):
    def __init__(self):
        cherrypy.Tool.__init__(self, 'before_handler',
                            self.load, priority=10)

    def load(self, **kwargs):
        req = cherrypy.request

        # let's assume we have a db session
        # attached to the request somehow 
        db = req.db

        # retrieve the user id and remove it
        # from the request parameters
        if 'user' in req.params:
            user = req.params.pop('user')
            if not user.isdigit():
                req.params['user'] = db.query(User).\
                        filter(User.name == user).one()
            else:
                req.params['user'] = db.query(User).\
                        filter(User.id == user).one()

cherrypy.tools.user = UserManager()

并重写用户 class，如下所示：

@cherrypy.popargs("user")
class Users(object):

    exposed = True

    @cherrypy.tools.user()
    def GET(self, user=None, **kwargs):
        if user:
            return "I'm resource %s" % user.id
        else:
            return "All users"

装饰器 @cherrypy.popargs("user") 告诉 cherrypy 获取 url 中的第一段并将其存储到键 "user"。装饰器 @cherrypy.tools.user() 告诉 cherrypy 在进入 GET 方法之前调用工具用户。在 UserManager class 的 load 方法中，我将用户 ID 或名称转换为 sqlalchemy 模型。

Cherrypy 在 _cp_dispatch/popargs 中使用插件

Cherrypy use plugin in _cp_dispatch/popargs

python

sqlalchemy

cherrypy