我正在实现我自己的http服务器：
import socket
import threading
import queue
import ssl
from manipulator.parser import LineBuffer,LoggableHttpRequest
class SocketServer:
"""
Basic Socket Server in python
"""
def __init__(self,host,port,max_threads,ssl_context:ssl.SSLContext=None):
print("Create Server For Http")
self.host = host
self.port = port
self.server_socket = self.initSocket()
self.max_threads = max_threads
self.request_queue = queue.Queue()
self.ssl_context=None
if(ssl_context != None):
print("Initialise SSL context")
self.ssl_context = ssl_context
def initSocket(self):
return socket.socket(socket.AF_INET, socket.SOCK_STREAM)
def __accept(self):
self.server_socket.listen(5)
while True:
try:
client_socket, client_address = self.server_socket.accept()
if self.ssl_context is not None :
print(self.ssl_context)
client_socket = self.ssl_context.wrap_socket(client_socket, server_side=True)
self.request_queue.put((client_socket, client_address))
except:
print("Error Occured")
def __handle(self):
while True:
client_socket, address = self.request_queue.get()
print("Address",address)
try:
# Read HTTP Request
# Log Http Request
# Manipulate Http Request
# Forward or respond
buffer = LineBuffer()
request = HttpRequest(self.db)
buffer.pushData(client_socket.recv(2048))
line = buffer.getLine()
if(line is not None):
request.parse(line)
content = '<html><body>Hello World</body></html>\r\n'.encode()
headers = f'HTTP/1.1 200 OK\r\nContent-Length: {len(content)}\r\nContent-Type: text/html\r\n\r\n'.encode()
client_socket.sendall(headers + content)
finally:
client_socket.shutdown(socket.SHUT_RDWR)
client_socket.close()
self.request_queue.task_done()
def __initThreads(self):
for _ in range(self.max_threads):
threading.Thread(target=self.__handle, daemon=True).start()
def start(self):
self.server_socket.bind((self.host, self.port))
self.__initThreads()
self.__accept()
我这样做的原因是我想尽快记录和分析传入的 httpo 请求。另外，许多第 3 方库确实需要我想避免的 C 绑定。
到目前为止，我制作了一个行分块器，将请求分成 :
class LineBuffer:
def __init__(self):
self.buffer = b''
def pushData(self,line):
self.buffer += str.encode(line)
def getLine(self):
if b'\r\n' in self.buffer:
line,sep,self.buffer = self.buffer.partition(b'\r\n')
return line+sep
return None
我想解析每一行并将其序列化为一个表示 http 请求的对象，我可以以流式传输方式进一步传输它：
class HttpRequest:
def __init__(self,db):
self.headers={} #ParsedHeaderrs
self.body="" #Http Body
self.version=None
self.method=None
self.id=None
self.raw=""
class HttpParser:
def __init__(self,db):
self.db = db
self.currentRequest=None
def parse(line):
# do parsing here
return
我最担心的是一个客户端会发送2个请求的场景：
要求1：
GET / HTTP/1.1\r\n
HOST lala1.com \r\n
要求2：
POST /file HTTP/1.1\r\n
HOST lala2.com \r\n
\r\n
Qm9QUVM5NDMuLnEvXVN7O2E=
fDMpQjcpOlFodClgOGUzYQ==
NVgvNipmU1d3YFgtLFUhQiM=
MiZwSk0zKno9TkVxNyZFL3s=
NEhGJXZ7OGciOE8mYF5JNA==
dVlJLzpdKlUjXl4tcEpufQ==
XVgiXCdjQyckMjY/Ikt6Rw==
alksJlZ+XHFzQSYqaHlHIztt
YiRnPjdye0gvanV3ZGxaZkI=
MjgwTX0uYHw6M295RS52UDM=
YU0yQ2dQLmJUQVpCNS89PWJB
Ti10MHJBTjAqUFUlIU0sMyRN
但是我的服务器收到的顺序是：
GET / HTTP/1.1\r\n
POST /file HTTP/1.1\r\n
HOST lala1.com \r\n
\r\n\r\nQm9QUVM5ND
HOST lala2.com \r\n
MuLnEvXVN7O2E=
fDMpQjcpOlFodClgOGUzYQ==
NVgvNipmU1d3YFgtLFUhQiM=
MiZwSk0zKno9TkVxNyZFL3s=
NEhGJXZ7OGciOE8mYF5JNA==
dVlJLzpdKlUjXl4tcEpufQ==
XVgiXCdjQyckMjY/Ikt6Rw==
alksJlZ+XHFzQSYqaHlHIztt
YiRnPjdye0gvanV3ZGxaZkI=
MjgwTX0uYHw6M295RS52UDM=
YU0yQ2dQLmJUQVpCNS89PWJB
Ti10MHJBTjAqUFUlIU0sMyRN
\r\n
我的情况可行吗？还是tcp套接字自己处理数据顺序？
HTTP/1 请求和响应是序列化的，即同一 TCP 连接内不存在多个请求或响应的交错，并且响应必须与请求具有相同的顺序并位于同一 TCP 连接上。对于 HTTP/2，这是不同的，但您的代码只需要 HTTP/1。