例子中涉及到SQLAlchemy 得相关操作,可以参考 上一随笔
Openstack 中规定,扩展openstack得api有两种方式
- 创建新的WSGI 资源
- 扩展原有得WSGI资源得控制器(我得理解是,接受到API请求后,具体得响应逻辑)
这两种方式中,都要求写一个新的模块来声明控制器类去处理请求和实现扩展。
在一个API模块中,可以有一个或多个得资源和扩展控制器。
根据osapi_compute_extension 得配置, ExtensionManager 由nova/api/openstack/compute/contrib/ 下的__init__.py 文件加载标准的或者新的扩展。
所以扩展的api统一写在nova/api/openstack/compute/contrib/ 目录下
如本例子中得 nova/api/openstack/compute/contrib/documents.py
扩展API流程
- 实现控制器,完成对资源的基本操作,如增删改查和其他一些用户自定义的RESTful资源操作;
- 实现一个extensions.ExtensionDescriptor的子类, 并实现get_resources 或者(/以及)get_controller_extensions,来建立新的资源或扩展资源控制器(即改写原有的业务逻辑)。具体实现哪个方法这取决于是否要修改原有的RESTFul业务逻辑, 或者说两个功能都需要;
- 将控制器和扩展的资源类,写如新创建的的资源中; 规范是资源类是模块(问家名)首字母大写,这样做的目的是使nova.api.openstack.extensions.load_standard_extensions这个类能够是别该新资源,并予以加载;
- 当添加新的资源(extensions.ResourceExtension的子类)的时候,如果想要去除掉 {tenent_id} 链接,则需要编写自定义的路由访问规则(本例子中没有涉及)
documents.py 实现
1 # vim: tabstop=4 shiftwidth=4 softtabstop=4 2 3 # Author: JiangYiTao 4 # Email: willierjyt@gmail.com 5 6 import webob 7 from webob import exc 8 9 from nova import db 10 from nova import exception 11 from nova.api.openstack import extensions 12 authorize = extensions.extension_authorizer('compute', 'documents') 13 14 # 请求控制器, 即处理对资源的请求,予以响应 15 class DocumentsController(): 16 """the Documents API Controller declearation""" 17 18 def index(self, req): 19 import pdb; pdb.set_trace() 20 documents = {} 21 context = req.environ['nova.context'] 22 authorize(context) 23 24 documents["key"] = "helloworld" 25 return documents 26 27 def create(self, req): 28 documents = {} 29 context = req.environ['nova.context'] 30 authorize(context) 31 32 documents["key"] = "helloworld" 33 return documents 34 35 def show(self, req, id):
37 documents = {} 38 context = req.environ['nova.context'] 39 authorize(context) 40 41 try: 42 document = db.document_get(context, id) 43 except : 44 raise webob.exc.HTTPNotFound(explanation="Document not found") 45 46 documents["document"] = document 47 return documents 48 49 def update(self, req): 50 documents = {} 51 context = req.environ['nova.context'] 52 authorize(context) 53 54 documents["key"] = "helloworld" 55 return documents 56 57 def delete(self, req, id): 58 return webob.Response(status_int=202) 59 # 根据命名规范, 模块(python源文件)中的类名是模块名的首字母大写 60 class Documents(extensions.ExtensionDescriptor): 61 """Documents ExtensionDescriptor implementation""" 62 63 name = "documents" 64 alias = "os-documents" 65 namespace = "www.www.com" 66 updated = "2013-05-19T00:00:00+00:00" 67 68 def get_resources(self): 69 """register the new Documents Restful resource""" 70 71 resources = [extensions.ResourceExtension('os-documents', 72 DocumentsController()) 73 ] 74 75 return resources
在之后可以由以下几种方式来操作documents 资源
GET v2/{tenant_id}/ os-documents POST v2/{tenant_id}/ os-documents GET v2/{tenant_id}/ os-documents/{document_id} PUT v2/{tenant_id}/ os-documents/{document_id} DELETE v2/{tenant_id}/ os-documents/{document_id}
扩展api时所修改的文件
1 nova/db/api.py 2 nova/db/sqlalchemy/api.py 3 nova/db/sqlalchemy/models.py
nova/db/api.py 文件内容
#数据操作API提供的方法,由Nova API 根据请求进行相应的操作, 由上面的请求控制器进行调用
1 def document_get(context, document_id): 2 """Get a document or raise if it does not exist.""" 3 return IMPL.document_get(context, document_id)
nova/db/sqlalchemy/api.py 文件内容
# 完成通过由SQLAlchemy操作数据库
1 @require_admin_context 2 def document_get(context, document_id): 4 session = get_session() 5 with session.begin(): 6 query = model_query(context, models.Document, session=session, read_deleted="yes").filter_by(id=document_id) 7 8 result = query.first() 9 10 if not result or not query: 11 raise Exception() 12 13 return result
SQLAlchemy 中定义的资源
nova/db/sqlalchemy/models.py(具体使用见上一 篇日志)
1 class Document(BASE, NovaBase): 2 """Represents a document of customized extension.""" 3 4 __tablename__ = 'documents' 5 id = Column(Integer, primary_key=True) 6 title = Column(String(255))
至此,添加添加新的nova API功能完成
重起api服务
调用 curl -v -X GET -H 'X-Auth-Token: 8e5971b3ce0a4f039b895681e7c29361' http://127.0.0.1:8774/v2/9b5903dd2d3443d8bb75ddffac27239a/extensions/os-documents | python -mjson.tool命令,
可以看到返回,扩展添加成功
1 { 2 "extension": { 3 "alias": "os-documents", 4 "description": "Documents ExtensionDescriptor implementation", 5 "links": [], 6 "name": "documents", 7 "namespace": "www.www.com", 8 "updated": "2013-05-19T00:00:00+00:00" 9 } 10 }
数据库添加表documents, 结构如下
mysql> desc documents; +------------+--------------+------+-----+---------+----------------+ | Field | Type | Null | Key | Default | Extra | +------------+--------------+------+-----+---------+----------------+ | id | int(11) | NO | PRI | NULL | auto_increment | | title | varchar(255) | NO | | NULL | | | created_at | datetime | YES | | NULL | | | updated_at | datetime | YES | | NULL | | | deleted_at | datetime | YES | | NULL | | | deleted | int(11) | YES | | NULL | | +------------+--------------+------+-----+---------+----------------+ 6 rows in set (0.05 sec)
mysql> select * from documents; +----+----------------+------------+------------+------------+---------+ | id | title | created_at | updated_at | deleted_at | deleted | +----+----------------+------------+------------+------------+---------+ | 1 | abcdefgiifeife | NULL | NULL | NULL | NULL | | 2 | 1qaz2wsx | NULL | NULL | NULL | NULL | +----+----------------+------------+------------+------------+---------+ 2 rows in set (0.03 sec)
调用命令
1 curl -X GET -H 'X-Auth-Token: 8e5971b3ce0a4f039b895681e7c29361' http://127.0.0.1:8774/v2/9b5903dd2d3443d8bb75ddffac27239a/os-documents/1 | python -mjson.tool 2 返回结果 3 { 4 "document": { 5 "created_at": null, 6 "deleted": null, 7 "deleted_at": null, 8 "id": 1, 9 "title": "abcdefgiifeife", 10 "updated_at": null 11 } 12 }
至此新添加的资源,API 扩展成功, 可以在此基础上进行进一步的修改,完成需求
参考文档:
https://www.ibm.com/developerworks/community/blogs/e93514d3-c4f0-4aa0-8844-497f370090f5/entry/openstack_nova_api?lang=zh
http://docs.openstack.org/developer/nova/devref/addmethod.openstackapi.html