如何远程运行python脚本，但在本地使用结果列表？

Question

I have 3 or more nodes, and I want to perform localisation using an estimation of the distances between each based upon the RSSIs between.

To do this I am trying to create a list on each node of the RSSIs it receives then share these lists between all the nodes. I have a python script which captures all the RSSIs along with the receiving and transmitting node into sublists of a list.

The code is as following:

import subprocess

def matching_line(lines, keyword):
    """Returns the first matching line in a list of lines. See match()"""
    for line in lines:
        matching=match(line,keyword)
        if matching!=None:
            return matching
    return None

def match(line,keyword):
    """If the first part of line (modulo blanks) matches keyword,
    returns the end of that line. Otherwise returns None"""
    line=line.lstrip()
    length=len(keyword)
    if line[:length] == keyword:
        return line[length:]
    else:
        return None

neighbour = [] #list of local node and neighbour's addresses
scanned = {}    # dictionary containing iwlist scan results Address: RSSI
single_localisation = [[],[],[]] #list of iwlist scan results at a single node. Node's address, transmitting node's address, RSSI 
all_localisation = [[],[],[]] #list of iwlist scan results from all nodes. Receiving node's address, transmitting node's address, RSSI 

#Save batctl o to file - batctl o shows all the nodes nearby participating in the mesh
Proc=subprocess.Popen("batctl o > bato.txt", shell=true)
Proc.wait()

#Populate neighbour list with addresses of neighbouring nodes
with open("bat.txt") as fd:
    fd.readline() #skip column headings
    for line in fd:
        neighbour.append(line.split()[0])

#Add local node's Address to neighbour list for later comparison
neigbour.append( subprocess.check_output("ip link show wlan0 | grep link | awk '{print $2}'",shell=True).strip())

#Scan wlan2 and save to file
Proc=subprocess.Popen("iwlist wlan2 scan | awk '/ESSID/ {print $1} /level/ {print $3}' > rssi.txt", shell=true)
Proc.wait()

#Populate scanned list with all MAC addresses and RSSIs from file
cells=cells[1:]

with open("rssi.txt") as fd:
    for line in fd:
        cell_line = match(line,"Cell ")
        if cell_line != None:
            cells.append([])
            line = cell_line[-27:]
        cells[-1].append(line.rstrip())


for cell in cells:
    level.append(matching_line(cell,"Quality=").split()[2].split('=')[1])
    address.append(matching_line(cell,"Address: "))

scanned=dict(zip(address, level))

#Test if MAC address in scanned list matches MAC address in neighbour list (and is therefore a meshed node)
for s in scanned:
    if s in neighbour:
    #if it does make an entry in localisation for it
        localisation[0].append( subprocess.check_output("ip link show wlan0 | grep link | awk '{print $2}'",shell=True).strip())
        localisation[1].append(s)
        localisation[2].append(scanned[s])

So

• localisation[0] contains the local node's MAC
• localisation[1] contains the transmitting node's MAC
• localisation[2] contains the signal strength received from [1] by [0]

I want to somehow merge all the localisation lists across all the nodes to create one large list which each node has.

Is there a way I could possibly share the produced list over SSH using paramiko (or alternative)?

Answer 1

Is there a way I could possibly share the produced list over SSH using paramiko (or alternative)?

Yes, you can use paramiko for this. demo_sftp.py from the source distribution should show you how to do it, together with the API docs . Set up your host keys and authorized keys properly on the various hosts, then each remote host just needs to create an SFTPClient and call put on it.

Whether this is the best design depends entirely on how your network is set up and what your logic looks like. The easiest solution, if it's feasible, would be to just store the data on a networked filesystem (NFS, SMB, whatever) that all of the hosts can access. Or maybe you want to set up a socket server or 0MQ or whatever so that each host can upload its data directly instead of storing it in a file in the first place (which also means that the target can be triggered directly on receipt of data, instead of having to set up a filesystem watch or something). Or… really, there are many options.

You should also consider whether you're trying to build a client-server system (where everyone uploads to one "master", which then sends updates to all the "slave" hosts) or a p2p system (where everyone uploads to everyone else). For the latter, you might be better off with a simple DHT design instead of trying to keep everyone in sync.