- android - RelativeLayout 背景可绘制重叠内容
- android - 如何链接 cpufeatures lib 以获取 native android 库?
- java - OnItemClickListener 不起作用,但 OnLongItemClickListener 在自定义 ListView 中起作用
- java - Android 文件转字符串
我想在服务器和客户端之间发送/接收多个 TCP 消息。例如:
服务器:
ip=""
port=8888
buffersize=1024
s=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.bind((myIP, myPort))
s.listen(1)
(cl, adress) = s.accept()
cl.send("hi!".encoded())
msg=cl.recv(bufferSize).decode()
msg=msg[0]
print(" Received message is '{}'".format(msg))
客户:
ip=""
port=8888
s=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
s.connect(ip,port)
msg=s.recv(1024).decode()
print(" {} ".format(msg))
#sending to server
s.send("OK!".encode())
他们可以互相交流。我想发送第二条消息,我想循环接收第二条消息。
最佳答案
我写了一段代码来证明这一点。文章在这里:https://aknirala.github.io/maintaining/python3TCP.html代码在这里:https://github.com/aknirala/python3TCP
我发现人们(无情地)删除了我之前发布的链接(即:https://realpython.com/python-sockets/)说:如果链接更改,则答案将失效。但是我不明白为什么需要删除,因为在那之后,没有答案!!留下评论来解释事情就好了。无论如何,这里是代码(服务器想要执行 1000 个任务,每个任务 10 次。
服务器端:
#!/usr/bin/env python
# coding: utf-8
# In[ ]:
import socket #For TCP communication.
import types #For types.SimpleNamespace
import selectors #For selectors.DefaultSelector()
import time
import random
# In[ ]:
totalTasks = 1000
repAllTasks = 10
d = {}
for i in range(totalTasks):
d[i] = 0
msgStart = "crrrazy_msg_START"
msgEnd = "crrrazy_msg_END" #Hoping that this will not be in the message.
verbose = False
#Instead of having start and end tokens, a better way is to have headers which will
#have size of the message.
# In[ ]:
sel = selectors.DefaultSelector()
# In[ ]:
def accept_wrapper(sock):
'''
This function will accept the new clients, and will register them.
'''
conn, addr = sock.accept()
print("\n\n accepted connection from", addr)
conn.setblocking(False) #If we want multiple clients we need to do this.
#Below we are assigning three things to each clinet connection, if we want more details
#like, when was the last time clinet communicated we need to pass one more argument here
# and later, ensure that it gets updated.
data = types.SimpleNamespace(addr=addr, inb="", outb=b"")
#Above notice that inb is not binary. This is becasue our code will be simpler
#if we'll keep it string (as a single message will be recieved over multiple
#iterations). But we need to convert it from binary stream before appending it there.
#In server we have made a loop and are senidng data till all is send, so outb can
#be removed, as it is never used.
events = selectors.EVENT_READ | selectors.EVENT_WRITE
sel.register(conn, events, data=data)
# In[ ]:
host, port = "<<<ENTER YOUR MACHINE IP>>>", 65432#sys.argv[1], int(sys.argv[2]) #TODO: Chnage to your machine IP
lsock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
lsock.bind((host, port))
lsock.listen()
print("listening on", (host, port))
lsock.setblocking(False)
sel.register(lsock, selectors.EVENT_READ, data=None)
# In[ ]:
underProcess = {}
for i in range(totalTasks):
underProcess[i] = 0
ctr = 0
lastRetCtr = 0
def getNextTask(tD):
'''
tD is a dictionary which has, key values of taskID and units which has been completed.
This function will update the d, accordingly and will then return the next taskID to be done
if tD is None, we are just checking what is the next value to be returned, so no actual
changes are made to any variable.
'''
global d
global underProcess
global repAllTasks
global ctr
global lastRetCtr
#print('Got tD as: ',tD)
#Updating the task dictionary d, and also removing that from underProcess
if tD is not None:
for k in tD:
underProcess[k] -= 1
d[k] += tD[k]
lastRetCtr += 1
ctr = lastRetCtr
for k in d:
if d[k] + underProcess[k] < repAllTasks:
if tD is not None:
underProcess[k] += 1
return k
for k in underProcess:
if underProcess[k] > 0:
#print('k:',k,'underProcess[k]: ',underProcess[k], ' ctr: ',ctr, 'lastRetCtr: ',lastRetCtr)
if tD is not None:
ctr += 1
if ctr - lastRetCtr > 10: #This is just a simple mechanism to resend stuff
# when some unprocess task has not been done.
# If till 10 iterations no client returns then it will
# resend the task (not an ideal mechanism)
if ctr - lastRetCtr > 15:
ctr = lastRetCtr
print('Returning: ',k)
return k
else:
if tD is not None:
time.sleep(1)
return -2
return -1
# In[ ]:
tmpGlobalMsg = "" #These two are defined global coz I don't know how to work with
tmpGlobalTD = None #exec with local variables. :-( Let me know how to get rid of them
ctr = 0
def service_connection(key, mask):
#print('In service_connection')
global d
global msgStart
global msgEnd
global tmpGlobalMsg
global tmpGlobalTD
global ctr
global verbose
sock = key.fileobj
data = key.data
if mask & selectors.EVENT_READ:
recv_data = sock.recv(1024)
if recv_data:
data.inb += recv_data.decode()
eI = data.inb.find(msgEnd) #Is the end str found? If so we have recievd entire message
if eI >= 0:
sI = data.inb.find(msgStart)
if sI < 0 or sI > eI: #Ideally throw an exception here.
sI = 0
else:
sI += len(msgStart)
msg = data.inb[sI:eI]
data.inb = data.inb[eI + len(msgEnd):]
tmpGlobalTD = None
msg = 'tmpGlobalTD = '+msg
exec(msg, locals(), globals())
#Below, we update, how much unit of task has been done by client
nJob = getNextTask(tmpGlobalTD)
print('From : ',sock.getpeername(), 'nJob is: ', nJob)
toSend = msgStart + str(nJob) + msgEnd
toSend = str.encode(toSend) #We need to encode it befre sending
while len(toSend) > 0:
if mask & selectors.EVENT_WRITE:
sent = sock.send(toSend)
toSend = toSend[sent:]
if verbose:
print('Sent: ',sent)
else: #Not needed could be removed.
print('Not ready to send!!!! THIS SHOULD NOT HAPPEN (But who knows!!)')
time.sleep(0.0001)
else: #This will be true when client will close connection. We can even set
print("\n\n closing connection to", data.addr) # a timer here.
sel.unregister(sock)
sock.close()
else:
if verbose:
time.sleep(0.2) #So that output is not flooded
print('Not read ready !!!!', end=" ") #Thre are clients which have registerd
# But no one has sent anything for the server to read.
# In[ ]:
while True:
events = sel.select(timeout=None)
if verbose:
sleepT = 1 - len(events)*0.1
if sleepT <= 0:
sleepT = 0.01
time.sleep(sleepT)
print('NUmber of clinets is: ',len(events))
for key, mask in events:
if key.data is None:
accept_wrapper(key.fileobj)
else:
service_connection(key, mask)
nT = getNextTask(None)
if nT < 0:
if nT == -2:
print('Status: ',underProcess)
time.sleep(1)
continue
print('All done.underProcess ', underProcess)
for key, mask in events:
data = key.data
sock = key.fileobj
print("\n\n closing connection to", data.addr) # a timer here.
sel.unregister(sock)
sock.close()
break
客户端:
#!/usr/bin/env python
# coding: utf-8
# In[ ]:
import socket
import selectors
import types
import time
import random
sel = selectors.DefaultSelector()
msgStart = "crrrazy_msg_START"
msgEnd = "crrrazy_msg_END"
verbose = False
# In[ ]:
def start_connections(host, port):
server_addr = (host, port)
print("\n\nstarting connection to", server_addr)
sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
sock.setblocking(False)
sock.connect_ex(server_addr) #connect_ex() is used instead of connect() since connect() would immediately raise a BlockingIOError exception. connect_ex() initially returns an error indicator, errno.EINPROGRESS, instead of raising an exception while the connection is in progress.
events = selectors.EVENT_READ | selectors.EVENT_WRITE
data = types.SimpleNamespace(
#connid=connid, #In case multiple connections are made from same machine
# we can assign an ID to them.
#Other fields, if other protocols like fixed message lengths etc needs to be
#used then more variables could be added here to keep track of those.
inb="", #Not a binary stream as it will be easier to handle str
)
sel.register(sock, events, data=data)
# In[ ]:
tempTaskID = None
def sendRecv(sel, tS = {}, recv = True, disconnect = False):
'''
This function is written in such a way that one can send a message to server
and server will send a return message and the function will return that.
Since the function may wait endlessly if server does not return it, after some
delay, we are assuming that server didn't listened, so we are resending the message.
This is not an ideal solution, as server might get the message again and falsely
assume that task has been done. This can be fixed by providing unique key to each job,
and IMHO must be done in actual code.
sel: contains the connection to server.
tS: dictionary to send. By default it is empty, which in our case we consider as
blank message.
recv: a flag to indicate, if we are expecting a return message from server.
In our case, this will always be True.
disconnect: When True, this function will close the connection.
'''
global msgStart
global msgEnd
global tempTaskID
global verbose
waitAttempts = 0
while True:
events = sel.select(timeout=1)
if events:
toProcess = True #To ensure that there is only one connection
# If everything goes correctly we don't need this.
# and ideally code involving this shouldnot run.
# Thus all code involving this could be removed,
for key, mask in events: #Idealy we should have just one value in events.
sock = key.fileobj
data = key.data
if not toProcess:
print('Repeated!!!!! ', key.index, ' closing it.')
print('THIS SHOULD NOT HAVE BEEN EXECUTED. But what can we say...')
sel.unregister(sock)
sock.close()
print('Disconnected extra socket.')
continue
toProcess = False
if disconnect: #This time function hsa been called to disconnect the connection.
# Please note, in case we would have been managing multiple
# connections, this would have been added to data when calling
# SimpleNamespace while registering, so that we know which socket
# to close.
sel.unregister(sock)
sock.close()
print('Disconnected')
return {}
msg = msgStart + str(tS) + msgEnd
toSend = str.encode(msg)
if waitAttempts <= 0: #It'll not send again till waitAttempts becomes less
# This is added in case, server does not recieve the message
# for the first time. So after waiting for few times
# ,in our case 10 seconds, as done by sleep in each iteration
# we'll resend the message.
# In practice, this could even happen when message from server
# is lost, thus a better mechanism to detect that should be there
while len(toSend) > 0:
if mask & selectors.EVENT_WRITE:
sent = sock.send(toSend)
if verbose:
print('sent: ',sent, 'From: ',sock.getsockname(),' Cont: ',toSend[:sent])
toSend = toSend[sent:]
else:
print('Not ready to write!!. THIS SHOULD NOT HAVE HAPPENED, but what do I know..')
time.sleep(0.0001)
waitAttempts = 10 #Try again after these many iterations of outer while
else:
waitAttempts -= 1
if not recv: #Return not expected.
return {}
if verbose:
print('wtoR...', end=' ')#For illustration we are SoPing
# wtoR: waiting to READ.
msg = ''
if mask & selectors.EVENT_READ:
recv_data = sock.recv(1024) # Should be ready to read
if recv_data:
data.inb += recv_data.decode()
#print('Now we got:::: ',data.outb,' len: ',len(data.outb), 'and find: ',data.outb.find(msgEnd))
else:
if verbose:
print('NotR..', end=' ')#For illustration we are SoPing
# NotR: Not ready to READ. (Waiting for server to send message)
time.sleep(1) #By giving delay here we will wait for 10 seconds for
# server to reply.
eI = data.inb.find(msgEnd)
if eI >= 0:
sI = data.inb.find(msgStart)
if sI < 0 or sI > eI: #Ideally throw an exception here.
sI = 0
else:
sI += len(msgStart)
msg = data.inb[sI:eI]
if len(msg) <= 0:
print('ERROR, this definitely looks like a protocol failure. Terminating connection, by returning -1')
#We are expecting that server must alway reply an integer eithre
#representing taskID or -1 to terminate.
return -1
data.inb = data.inb[eI + len(msgEnd):] #Simply removing the part
# of the message we have processed. In this case, if everything
# is as planned, this should always set data.inb to empty string.
try:
return int(msg)
except:
print('Integer not returned. Val return is:',msg,'. Protocol break.')
return -1
else:
if verbose:
print('No end on: ', data.inb, end='')
else:
print('No events!!! THIS SHOULD NOT HAPPEN...')
if not sel.get_map():
print('It\'s likely that no socket is opened OR server is not there.')
break
return {}
# In[ ]:
host, port = "<<ENTER YOUR SERVER IP>>", 65432#sys.argv[1:4]
start_connections(host, int(port))
# In[ ]:
d = sendRecv(sel)
while True:
if d <0:
if d == -2:
print('d is: ',d)
time.sleep(0.5)
d = sendRecv(sel)
continue
sendRecv(sel, disconnect=True)
print('Disconnected as server said target achieved',d)
break
tS = {}
tS[d] = random.randint(1,2) #Random number between 1 and 2
print('Sending: ',tS)
d = sendRecv(sel, tS)
# In[ ]:
#sendRecv(sel, disconnect=True)
# In[ ]:
关于python - 在 Python 中通过 TCP 发送/接收多条消息,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/51590277/
我是 ZMQ 的新手。我发现 ZMQ 套接字实现比 winsock 简单得多。但我怀疑 “使用 ZMQ TCP 套接字创建的客户端可以与传统的 TCP 服务器通信吗?” 换句话说我的 ZMQ 客户端可
我想使用 TCP 协议(protocol) 将数据发送到 Logstash。为了发送数据,我正在使用 Node-RED。一个简单的配置如下所示: 在 Logstash 文件夹中,我创建了一个名为 no
当我尝试更改窗口缩放选项时,作为 root,我可以通过在 /proc/sys/net/中执行 net.ipv4.tcp_mem=16777000 来更改值。如果我必须更改这 100 个系统,那将需要大
明天做一些练习题,这道做不出来 TCP 服务器连接 TCP 客户端进行通信所需的最小套接字端口数是多少? 肯定只有两个吧?一个用于服务器,一个用于客户端,但这似乎是显而易见的。我的伙伴们认为 TCP
考虑一个存在一个服务器和多个客户端的场景。每个客户端创建 TCP 连接以与服务器交互。 TCP alive的三种用法: 服务器端保活:服务器发送 TCP 保活以确保客户端处于事件状态。如果客户端死了,
TCP TAHOE 和 TCP RENO 有什么区别。 我想知道的是关于 3-dup-ack 和超时的行为? SST 发生了什么变化? 谢谢! 最佳答案 TCP Tahoe 和 Reno 是处理 TC
大家早上好。我一直在阅读(其中大部分在堆栈溢出中)关于如何进行安全密码身份验证(散列 n 次,使用盐等)但我怀疑我将如何在我的 TCP 客户端中实际实现它-服务器架构。 我已经实现并测试了我需要的方法
在遍历 RFC793 时,我开始知道应该以这种方式选择初始序列号段重叠被阻止。 有人能解释一下如果发生重叠,重复段将如何影响 TCP? 最佳答案 不同的操作系统有不同的行为。参见 http://ins
你能举例说明一下tcp/ip中nagle算法的概念吗? 最佳答案 我认为Wikipedia在开头的段落中做得很好。 Nagle's document, Congestion Control in IP
似乎最大 TCP 接收窗口大小为 1GB(使用缩放时)。因此,仍然可以用一个连接填充 100Gb 管道的最大 RTT 是 40ms(因为 2 * 40E-3 * 100E9/8 = 1GB)。这会将这
考虑在两个 TCP 端点之间建立的 TCP 连接,其中一个调用: 关闭():此处,不允许进一步读取或写入。 关机(fd,SHUT_WR):这会将全双工连接转换为单工连接,其中调用 SHUT_WR 的端
我是在 Lua 中编写解析器的新手,我有两个简短的问题。我有一个包含 TCP 选项的数据包,如 MSS、TCP SACK、时间戳、NOP、窗口比例、未知。我基本上是在尝试剖析 TCP 选项字段中的未知
TCP 是否不负责通过在传输过程中发生丢失等情况时采取任何可能必要的措施来确保通过网络完整地发送流? 它做的不对吗? 为什么更高的应用层协议(protocol)及其应用程序仍然执行校验和? 最佳答案
考虑使用 10 Mbps 链路的单个 TCP (Reno) 连接。假设此链路不缓冲数据并且接收方的接收缓冲区比拥塞窗口大得多。设每个 TCP 段的大小为 1500 字节,发送方和接收方之间连接的双向传
考虑这样一个场景,有client-a和server-b。 server-b 禁用了 TCP keepalive。 server-b 没有任何应用程序逻辑来检查 TCP 连接是否打开。 client-a
我正在尝试用 Rust 编写回显服务器。 use std::net::{TcpStream, TcpListener}; use std::io::prelude::*; fn main() {
听说对于TCP连接,服务器会监听一个端口,并使用另一个端口发送数据。 例如,Web 服务器监听端口 80。每当客户端连接到它时,该服务器将使用另一个端口(比如 9999)向客户端发送数据(Web 内容
我试图了解带有标记 PSH 和标记 URG 的 TCP 段之间的区别。我阅读了 RFC,但仍然无法理解,其中一个在将数据发送到进程之前缓冲数据而另一个没有吗? 最佳答案 它们是两种截然不同的机制。 #
有第三方服务公开 TCP 服务器,我的 Node 服务器(TCP 客户端)应使用 tls Node 模块与其建立 TCP 连接。作为 TCP 客户端, Node 服务器同时也是 HTTP 服务器,它应
我正在发送一些 TCP SYN 数据包以获得 TCP RST 的返回。为了识别每个探测器,我在 TCP 序列字段中包含一个计数器。我注意到以下几点: 当SYN probe中的sequence numb
我是一名优秀的程序员,十分优秀!