1 Web静态服务器
在“HTTP协议、http服务器的实现-1”中已经完成了“根据用户的需求返回相应的页面”的任务,接下分别使用多进程、多线程、协程gevent,来完成http服务器。
1.1 使用多进程完成http服务器
import socket
import re
import multiprocessing
def service_client(new_socket):
"""为这个客户端返回数据"""
# 1. 接收浏览器发送过来的请求 ,即http请求
# GET / HTTP/1.1
# .....
request = new_socket.recv(1024).decode("utf-8")
# print(">>>"*50)
# print(request)
request_lines = request.splitlines()
print("")
print(">" * 20)
print(request_lines)
# GET /index.html HTTP/1.1
# get post put del
file_name = ""
ret = re.match(r"[^/]+(/[^ ]*)", request_lines[0])
if ret:
file_name = ret.group(1)
# print("*"*50, file_name)
if file_name == "/":
file_name = "/index.html"
# 2. 返回http格式的数据,给浏览器
try:
f = open("./html" + file_name, "rb")
except:
response = "HTTP/1.1 404 NOT FOUND\r\n"
response += "\r\n"
response += "------file not found-----"
new_socket.send(response.encode("utf-8"))
else:
html_content = f.read()
f.close()
# 2.1 准备发送给浏览器的数据---header
response = "HTTP/1.1 200 OK\r\n"
response += "\r\n"
# 2.2 准备发送给浏览器的数据---boy
# response += "hahahhah"
# 将response header发送给浏览器
new_socket.send(response.encode("utf-8"))
# 将response body发送给浏览器
new_socket.send(html_content)
# 关闭套接
new_socket.close()
def main():
"""用来完成整体的控制"""
# 1. 创建套接字
tcp_server_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
tcp_server_socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
# 2. 绑定
tcp_server_socket.bind(("", 7890))
# 3. 变为监听套接字
tcp_server_socket.listen(128)
while True:
# 4. 等待新客户端的链接
new_socket, client_addr = tcp_server_socket.accept()
# 5. 为这个客户端服务
p = multiprocessing.Process(target=service_client, args=(new_socket,))
p.start()
# 关闭主进程中的新套接字,否则浏览器一直转圈
new_socket.close()
# 关闭监听套接字
tcp_server_socket.close()
if __name__ == "__main__":
main()
说明:
创建子进程会复制一份资源,会造成有两个套接字同时指向一个文件描述符(fd),相当于以前学的硬链接,关闭子进程中的套接字,还会有主进程中的套接字没有关闭,所以主进程必须关闭套接字,否则造成浏览器一直等待下一个数据的到来(一直转圈圈) 。(如下草图所示)
1.2 使用多线程来完成http服务器
import socket
import re
import threading
def service_client(new_socket):
"""为这个客户端返回数据"""
# 1. 接收浏览器发送过来的请求 ,即http请求
# GET / HTTP/1.1
# .....
request = new_socket.recv(1024).decode("utf-8")
# print(">>>"*50)
# print(request)
request_lines = request.splitlines()
print("")
print(">" * 20)
print(request_lines)
# GET /index.html HTTP/1.1
# get post put del
file_name = ""
ret = re.match(r"[^/]+(/[^ ]*)", request_lines[0])
if ret:
file_name = ret.group(1)
# print("*"*50, file_name)
if file_name == "/":
file_name = "/index.html"
# 2. 返回http格式的数据,给浏览器
try:
f = open("./html" + file_name, "rb")
except:
response = "HTTP/1.1 404 NOT FOUND\r\n"
response += "\r\n"
response += "------file not found-----"
new_socket.send(response.encode("utf-8"))
else:
html_content = f.read()
f.close()
# 2.1 准备发送给浏览器的数据---header
response = "HTTP/1.1 200 OK\r\n"
response += "\r\n"
# 2.2 准备发送给浏览器的数据---boy
# response += "hahahhah"
# 将response header发送给浏览器
new_socket.send(response.encode("utf-8"))
# 将response body发送给浏览器
new_socket.send(html_content)
# 关闭套接
new_socket.close()
def main():
"""用来完成整体的控制"""
# 1. 创建套接字
tcp_server_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
tcp_server_socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
# 2. 绑定
tcp_server_socket.bind(("", 7890))
# 3. 变为监听套接字
tcp_server_socket.listen(128)
while True:
# 4. 等待新客户端的链接
new_socket, client_addr = tcp_server_socket.accept()
# 5. 为这个客户端服务
t = threading.Thread(target=service_client, args=(new_socket,))
t.start()
# new_socket.close()
# 关闭监听套接字
tcp_server_socket.close()
if __name__ == "__main__":
main()
说明:
只需将 multiprocessing模块 改为 threading 模块,Process 改为 Thread 即可。还需要注意, 创建子线程不需要复制资源,所以也不需要关闭主进程的新套接字(切记,否则程序会出错)
1.3 使用gevent来实现http服务器
import socket
import re
import gevent
from gevent import monkey
monkey.patch_all()
def service_client(new_socket):
"""为这个客户端返回数据"""
# 1. 接收浏览器发送过来的请求 ,即http请求
# GET / HTTP/1.1
# .....
request = new_socket.recv(1024).decode("utf-8")
# print(">>>"*50)
# print(request)
request_lines = request.splitlines()
print("")
print(">" * 20)
print(request_lines)
# GET /index.html HTTP/1.1
# get post put del
file_name = ""
ret = re.match(r"[^/]+(/[^ ]*)", request_lines[0])
if ret:
file_name = ret.group(1)
# print("*"*50, file_name)
if file_name == "/":
file_name = "/index.html"
# 2. 返回http格式的数据,给浏览器
try:
f = open("./html" + file_name, "rb")
except:
response = "HTTP/1.1 404 NOT FOUND\r\n"
response += "\r\n"
response += "------file not found-----"
new_socket.send(response.encode("utf-8"))
else:
html_content = f.read()
f.close()
# 2.1 准备发送给浏览器的数据---header
response = "HTTP/1.1 200 OK\r\n"
response += "\r\n"
# 2.2 准备发送给浏览器的数据---boy
# response += "hahahhah"
# 将response header发送给浏览器
new_socket.send(response.encode("utf-8"))
# 将response body发送给浏览器
new_socket.send(html_content)
# 关闭套接
new_socket.close()
def main():
"""用来完成整体的控制"""
# 1. 创建套接字
tcp_server_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
tcp_server_socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
# 2. 绑定
tcp_server_socket.bind(("", 7890))
# 3. 变为监听套接字
tcp_server_socket.listen(128)
while True:
# 4. 等待新客户端的链接
new_socket, client_addr = tcp_server_socket.accept()
# 5. 为这个客户端服务
gevent.spawn(service_client, new_socket)
# new_socket.close()
# 关闭监听套接字
tcp_server_socket.close()
if __name__ == "__main__":
main()
2 Web静态服务器 - 非堵塞模式
2.1 单进程-单线程-非堵塞
import socket
import time
def main():
tcp_server_tcp = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
tcp_server_tcp.bind(("", 7899))
tcp_server_tcp.listen(128)
tcp_server_tcp.setblocking(False) # 设置套接字为非堵塞的方式
client_socket_list = list() # 创建一个新列表,存储为浏览器服务的新套接字
while True:
# time.sleep(0.5) # 程序调试的时候使用
try:
new_socket, new_addr = tcp_server_tcp.accept()
except Exception as ret:
print("---没有新的客户端到来---")
else:
print("---只要没有产生异常,那么也就意味着 来了一个新的客户端----")
new_socket.setblocking(False) # 设置套接字为非堵塞的方式
client_socket_list.append(new_socket) # 将新的套接字放入列表中
for client_socket in client_socket_list:
try:
# client_socket.setblocking(False) # 设置套接字为非堵塞的方式
recv_data = client_socket.recv(1024)
except Exception as ret:
print(ret)
print("----这个客户端没有发送过来数据----")
else:
print("-----没有异常-----")
print(recv_data)
if recv_data:
# 对方发送过来数据
print("----客户端发送过来了数据-----")
else:
# 对方调用close 导致了 recv返回
client_socket.close()
client_socket_list.remove(client_socket)
print("---客户端已经关闭----")
if __name__ == "__main__":
main()
2.2 单进程-单线程-非堵塞-长链接
import socket
import re
def service_client(new_socket, request):
"""为这个客户端返回数据"""
# 1. 接收浏览器发送过来的请求 ,即http请求
# GET / HTTP/1.1
# .....
# request = new_socket.recv(1024).decode("utf-8")
# print(">>>"*50)
# print(request)
request_lines = request.splitlines()
print("")
print(">" * 20)
print(request_lines)
# GET /index.html HTTP/1.1
# get post put del
file_name = ""
ret = re.match(r"[^/]+(/[^ ]*)", request_lines[0])
if ret:
file_name = ret.group(1)
# print("*"*50, file_name)
if file_name == "/":
file_name = "/index.html"
# 2. 返回http格式的数据,给浏览器
try:
f = open("./html" + file_name, "rb")
except:
response = "HTTP/1.1 404 NOT FOUND\r\n"
response += "\r\n"
response += "------file not found-----"
new_socket.send(response.encode("utf-8"))
else:
html_content = f.read()
f.close()
response_body = html_content
response_header = "HTTP/1.1 200 OK\r\n"
response_header += "Content-Length:%d\r\n" % len(response_body)
response_header += "\r\n"
response = response_header.encode("utf-8") + response_body
new_socket.send(response)
def main():
"""用来完成整体的控制"""
# 1. 创建套接字
tcp_server_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
tcp_server_socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
# 2. 绑定
tcp_server_socket.bind(("", 7890))
# 3. 变为监听套接字
tcp_server_socket.listen(128)
tcp_server_socket.setblocking(False) # 将套接字变为非堵塞
client_socket_list = list()
while True:
# 4. 等待新客户端的链接
try:
new_socket, client_addr = tcp_server_socket.accept()
except Exception as ret:
pass
else:
new_socket.setblocking(False)
client_socket_list.append(new_socket)
for client_socket in client_socket_list:
try:
recv_data = client_socket.recv(1024).decode("utf-8")
except Exception as ret:
pass
else:
if recv_data:
service_client(client_socket, recv_data)
else:
client_socket.close()
client_socket_list.remove(client_socket)
# 关闭监听套接字
tcp_server_socket.close()
if __name__ == "__main__":
main()
说明:
HTTP/1.1 版本是长链接方式,但是由于我们发送完信息就调用 close(),强制转换为短链接,所以要想变为长链接就不能调用 close()。结果浏览器会一直等待我们发送数据,所以我们应该在响应头中用
response_header += "Content-Length:%d\r\n" % len(response_body)
来告知浏览器,我们发送数据的长度。
3 Web静态服务器 - epoll
epoll简单模型
epoll 的优势(重点):
epoll是Linux内核为处理大批量文件描述符而作了改进的poll,是Linux下多路复用IO接口select/poll的增强版本,它能显著提高程序在大量并发连接中只有少量活跃的情况下的系统CPU利用率。
另一点原因就是获取事件的时候,它无须遍历整个被侦听的描述符集,只要遍历那些被内核IO事件异步唤醒而加入Ready队列的描述符集合就行了。
epoll除了提供select/poll那种IO事件的水平触发(Level Triggered)外,还提供了边缘触发(Edge Triggered),这就使得用户空间程序有可能缓存IO状态,减少epoll_wait/epoll_pwait的调用,提高应用程序效率。
代码如下:
import socket
import re
import select
def service_client(new_socket, request):
"""为这个客户端返回数据"""
# 1. 接收浏览器发送过来的请求 ,即http请求
# GET / HTTP/1.1
# .....
# request = new_socket.recv(1024).decode("utf-8")
# print(">>>"*50)
# print(request)
request_lines = request.splitlines()
print("")
print(">" * 20)
print(request_lines)
# GET /index.html HTTP/1.1
# get post put del
file_name = ""
ret = re.match(r"[^/]+(/[^ ]*)", request_lines[0])
if ret:
file_name = ret.group(1)
# print("*"*50, file_name)
if file_name == "/":
file_name = "/index.html"
# 2. 返回http格式的数据,给浏览器
try:
f = open("./html" + file_name, "rb")
except:
response = "HTTP/1.1 404 NOT FOUND\r\n"
response += "\r\n"
response += "------file not found-----"
new_socket.send(response.encode("utf-8"))
else:
html_content = f.read()
f.close()
response_body = html_content
response_header = "HTTP/1.1 200 OK\r\n"
response_header += "Content-Length:%d\r\n" % len(response_body)
response_header += "\r\n"
response = response_header.encode("utf-8") + response_body
new_socket.send(response)
def main():
"""用来完成整体的控制"""
# 1. 创建套接字
tcp_server_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
tcp_server_socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
# 2. 绑定
tcp_server_socket.bind(("", 7890))
# 3. 变为监听套接字
tcp_server_socket.listen(128)
tcp_server_socket.setblocking(False) # 将套接字变为非堵塞
# 创建一个epoll对象
epl = select.epoll()
# 将监听套接字对应的fd注册到epoll中
epl.register(tcp_server_socket.fileno(), select.EPOLLIN)
fd_event_dict = dict()
while True:
fd_event_list = epl.poll() # 默认会堵塞,直到 os监测到数据到来 通过事件通知方式 告诉这个程序,此时才会解堵塞
# [(fd, event), (套接字对应的文件描述符, 这个文件描述符到底是什么事件 例如 可以调用recv接收等)]
for fd, event in fd_event_list:
# 等待新客户端的链接
if fd == tcp_server_socket.fileno():
new_socket, client_addr = tcp_server_socket.accept()
epl.register(new_socket.fileno(), select.EPOLLIN)
fd_event_dict[new_socket.fileno()] = new_socket
elif event == select.EPOLLIN:
# 判断已经链接的客户端是否有数据发送过来
recv_data = fd_event_dict[fd].recv(1024).decode("utf-8")
if recv_data:
service_client(fd_event_dict[fd], recv_data)
else:
fd_event_dict[fd].close()
epl.unregister(fd)
del fd_event_dict[fd]
# 关闭监听套接字
tcp_server_socket.close()
if __name__ == "__main__":
main()
说明:
本代码只需要看懂即可,不需要自己写,思想很重要,代码不是很重要。
代码说明:
- EPOLLIN (可读)
- EPOLLOUT (可写)
- EPOLLET (ET模式)
epoll对文件描述符的操作有两种模式:LT(level trigger)和ET(edge trigger)。LT模式是默认模式,LT模式与ET模式的区别如下:
LT模式:当epoll检测到描述符事件发生并将此事件通知应用程序,应用程序可以不立即处理该事件。下次调用epoll时,会再次响应应用程序并通知此事件。
ET模式:当epoll检测到描述符事件发生并将此事件通知应用程序,应用程序必须立即处理该事件。如果不处理,下次调用epoll时,不会再次响应应用程序并通知此事件。