应用案例

from http.server import HTTPServer, BaseHTTPRequestHandler  

IP = '127.0.0.1'  
PORT = 8000  

class Handler(BaseHTTPRequestHandler):  
    def do_GET(self):  
        self.send_response(200)  
        self.send_header('Content-type', 'text/html')  
        self.end_headers()  
        message = "Hello, World!"  

        self.wfile.write(bytes(message, "utf8"))  

with HTTPServer((IP, PORT), Handler) as httpd:  
    print("serving at port", PORT)  
    httpd.serve_forever()

以上是使用内置模块 http.server 实现的一个最简单的 http 服务器,能处理 http GET 请求。

python 内置的 http server 主要集中在两个代码文件上,分别是 socketserver.pyhttp/server.pysocketserver.py 提供 socket 通信能力的 Server 封装并预留了用户自定义请求处理的接口;http/server.py 基于前者做进一步封装,用得比较多的是 HTTP 的封装。

从开头的例子出发阅读代码(python 3.10.1),大致梳理出以下代码结构,图画得很随意无规范可言,只是为了更具象化解释。

问题一:实现一个 HTTP 服务器大致需要什么要素

先看图 1,左边 BaseServer 一列是类,从上往下是父类到子类;右边 server_forever() 一列是方法,从上往下是逐步深入的调用链。

                从父类到子类                                 主线流程

            +----------------+                      +------------------+
            |                |                      |                  |
            |   BaseServer   +--------------------->| serve_forever()  |
            |                |                      |                  |
            +--------+-------+                      +--------=+--------+
                     |                                        |
                     |                                        |
                     |                                        |
                     V                                        V
            +----------------+                  +----------------------------+
            |                |                  |                            |
            |   TCPServer    |                  | _handle_request_noblock()  |
            |                |                  |                            |
            +--------+-------+                  +-------------+--------------+
                     |                                        |
         +-----------+------------+                           |
         |                        |                           |
         V                        V                           V
+----------------+       +----------------+         +------------------+
|                |       |                |         |                  |
|   HTTPServer   |       |   UDPServer    |         | process_request()|
|                |       |                |         |                  |
+----------------+       +----------------+         +---------+--------+
                                                              |
                                                              |
                                                              |
                                                              V
                                                    +------------------+
                                                    |                  |
                                                    | finish_request() |
                                                    |                  |
                                                    +------------------+

图 1

例子中使用了 HTTPServer 这个类,字面意思,这个类就是一个 HTTP 服务器,顺着继承链看到 HTTPServerTCPServer 的子类,符合 HTTP 报文是基于 TCP 协议传输的认知,HTTPServer 类其实没什么内容,代码如下:

class HTTPServer(socketserver.TCPServer):  

    allow_reuse_address = 1 # Seems to make sense in testing environment  

     def server_bind(self):  
         """Override server_bind to store the server name."""  
         socketserver.TCPServer.server_bind(self)
         host, port = self.server_address[:2]
         self.server_name = socket.getfqdn(host)
         self.server_port = port

TCPServer 的源码实现得益于父类的预留接口,只需要 TCP socket 走一遍 bindlisten
acceptclose 流程(子类 UDPServer 同理)。

重点关注 BaseServer,这里是网络请求处理核心流程的实现,文章最开头的例子中 serve_forever() 这个入口方法就是在此类被实现,我在源码上加了些简单的注释:

def serve_forever(self, poll_interval=0.5):
    """Handle one request at a time until shutdown.

    Polls for shutdown every poll_interval seconds. Ignores
    self.timeout. If you need to do periodic tasks, do them in
    another thread. 
    """
    self.__is_shut_down.clear()  
    try:  
        # XXX: Consider using another file descriptor or connecting to the  
        # socket to wake this up instead of polling. Polling reduces our 
        # responsiveness to a shutdown request and wastes cpu at all other 
        # times. with _ServerSelector() as selector:  
        selector.register(self, selectors.EVENT_READ)  # 注册Server描述符并监听I/O读事件

        while not self.__shutdown_request:  
            ready = selector.select(poll_interval)  # 超时时长poll_interval避免长时间阻塞,在while循环下实现轮询
            # bpo-35017: shutdown() called during select(), exit immediately.  
            if self.__shutdown_request:  
                break  
            if ready:  
                self._handle_request_noblock()  # 请求过来,I/O读事件准备好,开始处理请求

            self.service_actions()  
     finally:  
         self.__shutdown_request = False  
         self.__is_shut_down.set()

_handle_request_noblock() 中看到,一个网络请求的处理流程无非就是 verify_request()process_request()shoutdown_request() 加上些许异常处理逻辑,比较简明。在 finish_request() 中出现 RequestHandlerClass 的类对象创建,这里其实就是用户自定义的 RequestHandler(在 BaseServer__int__() 中被初始化)。源码如下,较好理解:

def _handle_request_noblock(self):  
    """Handle one request, without blocking.
    I assume that selector.select() has returned that the socket is
    readable before this function was called, so there should be no risk of
    blocking in get_request(). 
    """
    try:  
        request, client_address = self.get_request()  
    except OSError:  
        return  
    if self.verify_request(request, client_address):  # 从这里开始就是网络请求的处理流程
        try:  
            self.process_request(request, client_address)  
        except Exception:  
            self.handle_error(request, client_address)  
            self.shutdown_request(request)  
        except:  
            self.shutdown_request(request)  
            raise  
    else:  
        self.shutdown_request(request)


def process_request(self, request, client_address):  
    """Call finish_request.  

    Overridden by ForkingMixIn and ThreadingMixIn.  
    """
    self.finish_request(request, client_address)  
    self.shutdown_request(request)


def finish_request(self, request, client_address):  
    """Finish one request by instantiating RequestHandlerClass."""  
    self.RequestHandlerClass(request, client_address, self)


def shutdown_request(self, request):  
    """Called to shutdown and close an individual request."""  
    self.close_request(request)

总结:要实现一个 HTTP 服务器,需要包含 TCP socket 实现,网络请求流程大致抽象为 verify_request()process_request()shoutdown_request() 。如果考虑支持用户自定义请求处理,还需要预留接口提供扩展性。当然如何要支持处理 HTTP 协议,还需要具备解析 HTTP 报文的能力,下文继续探讨。

问题二:python 内置的 HTTP Server 是怎么实现的

前文介绍了内置一个网络请求的处理流程(等价于 HTTP Server 的运行流程),一定程度上解释了本节的问题,但欠缺一点细节,没有体现 HTTP 报文的解析逻辑在哪里实现。其实内置的 HTTP Server 的把 HTTP 协议解析的工作解耦出去,单独做成 BaseHTTPRequestHandler 类,这样允许用户自行实现任意应用层的协议解析工作,参考下面图 2:


                              +----------------------+        +----------------+
                              |                      |        |                |
                              |  BaseRequestHandler  +------->|   __init__()   |
                              |                      |        |                |
                              +-----------+----------+        +----------------+
                                          |
                                          |
                                          |                   +----------------+
                                          |                   |                |
                                          V              +--->|     setup()    |
                              +----------------------+   |    |                |
                              |                      |   |    +----------------+
                              | StreamRequestHandler +---+
                              |                      |   |
                              +-----------+----------+   |    +----------------+
                                          |              |    |                |
                                          |              +---->    finish()    |
                                          V                   |                |
+------------------------+    +----------------------+        +----------------+
|                        |    |                      |
|SimpleHTTPRequestHandler|<---+BaseHTTPRequestHandler|
|                        |    |                      |
+------------------------+    +-----------+----------+
                                          |
                                          |
                                          |
                                          V
                                +------------------+
                                |                  |
                                |     handler()    |
                                |                  |
                                +---------+--------+          +----------------+
                                          |                   |                |
                                          |              +--->| parse_request()|
                                          |              |    |                |
                                          V              |    +----------------+
                              +----------------------+   |
                              |                      |   |
                              | handler_one_request()+---+
                              |                      |   |    +----------------+
                              +----------------------+   |    |                |
                                                         +--->|    do_XXX()    |
                                                              |                |
                                                              +----------------+

图 2

图 2 中,但凡带括号的都是方法,不带括号的是类,从上往下也是父类到子类。本着代码最大化复用的原则,父类 BaseRequestHandler__init__() 中将工作流程确定下来,分别是 setup()handler()finish() 的先后调用顺序。setup()finish() 在子类 StreamRequestHandler 被实现,最后在 BaseHTTPRequestHandler 类中实现 HTTP 协议解析功能,以及用 HTTP method 来决定调用哪个用户自定义的 do_XXX() 方法,如 do_GET()do_POST() 等。代码如下:

class BaseRequestHandler:  
    """Base class for request handler classes.  
    ......
    """  
    def __init__(self, request, client_address, server):  
        self.request = request  
        self.client_address = client_address  
        self.server = server  
        self.setup()  
        try:  
            self.handle()  
        finally:  
            self.finish()  

    def setup(self):  
        pass  

    def handle(self):  
        pass  

    def finish(self):  
        pass


class StreamRequestHandler(BaseRequestHandler):  
    """Define self.rfile and self.wfile for stream sockets."""  

    # 省略代码

    def setup(self):  
        # 设置链接超时时长、nagle算法、读写缓冲区
        self.connection = self.request  
        if self.timeout is not None:  
            self.connection.settimeout(self.timeout)  
        if self.disable_nagle_algorithm:  
            self.connection.setsockopt(socket.IPPROTO_TCP, socket.TCP_NODELAY, True)  
        self.rfile = self.connection.makefile('rb', self.rbufsize)  
        if self.wbufsize == 0:  
            self.wfile = _SocketWriter(self.connection)  
        else:  
            self.wfile = self.connection.makefile('wb', self.wbufsize)  

    def finish(self):  
        if not self.wfile.closed:  
            try:  
                self.wfile.flush()  
            except socket.error:  
                # A final socket error may have occurred here, such as  
                # the local error ECONNABORTED. 
                pass  
        self.wfile.close()  
        self.rfile.close()

HTTP 协议解析关注 parse_request() 方法,由于代码较多不单独贴过来,思路如下:

  1. 解析 HTTP 协议版本号,确定版本解析是否支持(1.1 <= version < 2.0)
  2. 获取 HTTP method
  3. 解析 HTTP header 解析完 HTTP 协议后,根据所获取的 HTTP method,调用用户自定义的对应方法,至此结束。

总结

python 内置的 HTTP Server 实现比较简洁,功能相对简单。如果要自行从零实现一个 HTTP Server,设计上参考 python 的实现,应该具备以下要素:

  1. TCP socket 通信
  2. HTTP 协议的报文解析
  3. 用户自定义的 RequestHandler 调用(设计上需要引入拓展)


↙↙↙阅读原文可查看相关链接,并与作者交流