非同步網路據說能極大的提高網路server的連線速度,所以打算寫一個專題,來學習和了解非同步網路.因為Python有個非常出名的非同步Lib:Twisted,所以就用Python來完成.
OK ,首先寫一個pythone socket的server段,對開放三個端口:10000,10001,10002.krondo的例子中是每個server綁定一個端口,測試的時候需要分別開3個shell,分別運行.這太麻煩了,就分別用三個Thread來運行這些services.
import optparse import os import socket import time from threading import Thread import StringIO txt = '''1111 2222 3333 4444 ''' def server(listen_socket): while True: buf = StringIO.StringIO(txt) sock, addr = listen_socket.accept() print 'Somebody at %s wants poetry!' % (addr,) while True: try: line = buf.readline().strip() if not line: sock.close() break sock.sendall(line) # this is a blocking call print 'send bytes to client:%s' % line #sock.close() except socket.error: sock.close() break time.sleep(1) #server和client连接后,server会故意每发送一个单词后等待一秒钟后再发送另一个单词 def main(): ports = [10000, 10001, 10002] for port in ports: listen_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM) listen_socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1) addres = (str('127.0.0.1'), port) listen_socket.bind(addres) listen_socket.listen(5) print "start listen at:%s" % (port,) worker = Thread(target = server, args = [listen_socket]) worker.setDaemon(True) worker.start() if __name__ == '__main__': main() while True: time.sleep(0.1) #如果不sleep的话,CPU会被Python完全占用了 pass
下面是一個client,沒有才用異步網絡,連接這個三個端口的server:
import socket if __name__ == '__main__': ports = [10000, 10001, 10002] for port in ports: address = (str('127.0.0.1'), port) sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM) sock.connect(address) poem = '' while True: data = sock.recv(4) if not data: sock.close() break poem += data print poem
下面用異步的client來讀取,代碼如下:
import datetime, errno, optparse, select, socket def connect(port): """Connect to the given server and return a non-blocking socket.""" address = (str('127.0.0.1'), port) sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM) sock.connect(address) sock.setblocking(0) return sock def format_address(address): host, port = address return '%s:%s' % (host or '127.0.0.1', port) if __name__ == '__main__': ports = [10000, 10001, 10002] start = datetime.datetime.now() sockets = map(connect, ports) poems = dict.fromkeys(sockets, '') # socket -> accumulated poem # socket -> task numbers sock2task = dict([(s, i + 1) for i, s in enumerate(sockets)]) sockets = list(sockets) # make a copy while sockets: #运用select来确保那些可读取的异步socket可以立即开始读取IO #OS不停的搜索目前可以read的socket,有的话就返回rlist rlist, _, _ = select.select(sockets, [], []) for sock in rlist: data = '' while True: try: new_data = sock.recv(1024) except socket.error, e: if e.args[0] == errno.EWOULDBLOCK: break raise else: if not new_data: break else: print new_data data += new_data task_num = sock2task[sock] if not data: sockets.remove(sock) sock.close() print 'Task %d finished' % task_num else: addr_fmt = format_address(sock.getpeername()) msg = 'Task %d: got %d bytes of poetry from %s' print msg % (task_num, len(data), addr_fmt) poems[sock] += data elapsed = datetime.datetime.now() - start print 'Got poems in %s' % elapsed
結果只需要4秒就完成了讀取任務。效率是剛才同步socket的三倍。在客戶端的非同步改造主要有兩點:
同步模式下,客戶端分別創建socket;而在非同步模式下,client開始就創建了所有的socket。
透過「sock.setblocking(0)」設定socket為非同步模式。
透過Unix系統的select倆回傳可讀取IO
最為核心的是26行和29行。尤其是29行的select操作傳回待讀取socket的列表。