用XMLRPC开服务进行server/client通信

应用场景:1)需多client访问应用程序给予应答情况——网页服务;  2)数据极大,希望加载一次,后面只用方法调用

解决方案:  开两个服务,一个数据服务,一个网络服务;
  数据服务端加载数据,网络服务端调用数据,并将结果显示在网络服务中;
  外部调用网络服务返回结果;

应用工具:xmlrpc,本文中以python 2.7.3的xmlrpclib为例,其他语言也有相应接口

下面分别说明。

1. 数据端

在本地localhost的8000端口开server服务,load数据,并定义接口查找数据第i个元素(findai).

Server :

from SimpleXMLRPCServer import SimpleXMLRPCServer
global a

def load():
 global a
 a = [1 ,2, 24]
 return a

def findai(i):
 global a
 print a[i]
 return a[i]

server = SimpleXMLRPCServer(("localhost", 8000))
server.register_function(findai,"findai")
load()
server.serve_forever()

Client:

import xmlrpclib
proxy = xmlrpclib.ServerProxy("http://localhost:8000/")

candidate = proxy.findai(1)
print "the %d-th number of a is %d" %(1, candidate)

2. 数据端 + 网络端

Client:

import xmlrpclib
proxy = xmlrpclib.ServerProxy("http://localhost:8000/")

candidate = proxy.findai(1)
print "the %d-th number of a is %d" %(1, candidate)

from bottle import route, run, template
@route('/hello/<name>')
def index(name):
 return template('<b> hello {{name}} </b>', name=candidate)

run(host="localhost", port=8086)

注意事项:

1. 通信数据类型

注意通讯数据类型只能是python的built-in类型(而不能是numpy类型),所以其他类型应转换为str类型(client端用ast.literal_eval从str转回来)或者更方便的用list(直接server端tolist转,client端numpy.array解);

否则会报错:

xmlrpclib.Fault:  <Fault  8002:  "Can't serialize output: cannot marshal <type 'numpy.float64'> objects">

以string为例(其实tolist更简单),

Server:

from SimpleXMLRPCServer import SimpleXMLRPCServer
global a
import ast
from cStringIO import StringIO
from numpy.lib import format
import numpy

class T:
 def to_string(self,arr):
  f = StringIO()
  if type(arr)==numpy.ndarray:
   format.write_array(f,arr)
   s = f.getvalue()
  elif isinstance(arr,str)==False:
   s = str(arr)
  return s

def from_string(self,s):
  if s[0]!="[": # converted from numpy array
   f = StringIO(s)
   arr = format.read_array(f)
  else:
   arr = ast.literal_eval(s)
  return arr

def load(self):
  global a
  a = [1 ,2, 24]
  return a

def ret_a(self):
  global a
  return a
 

server = SimpleXMLRPCServer(("localhost", 8002))
server.register_instance(T())
srv = T()
srv.load()
server.serve_forever()

Client:

import xmlrpclib
proxy = xmlrpclib.ServerProxy("http://localhost:8002/")

candidate = proxy.ret_a()
print "the variable 'a' in server is "+ str((proxy.from_string(candidate)))

2. 通讯字符编码问题

注意通讯字符必须是unicode编码,用中文的时候要小心。

所以中文的case下,在server段执行:

def gbk_to_unicode(s):
    return s.decode('gbk').encode('utf-8').decode('latin1')

client端执行:

def unicode_to_gbk(s):
    return s.encode('latin1').decode('utf-8').encode('gbk')

for example,

Server:

from SimpleXMLRPCServer import SimpleXMLRPCServer
global a
import ast
from cStringIO import StringIO
from numpy.lib import format
import numpy
import sys

def gbk_to_unicode(s):
        return s.decode('gbk').encode('utf-8').decode('latin1')

class T:

def load(self): # load a dictionary with gbk elements
        global a
        a = {"1,1":["小","苹果"],"1,2":[1,2]}

def printf(self,s): # receive unicode, return unicode
        print "received string : "+ s #unicode
        return s

def idx(self,s): # transfer gbk -> unicode to client
        global a
        return [gbk_to_unicode(x) for x in a.get(s,[])]

reload(sys)
sys.setdefaultencoding('gbk')
server = SimpleXMLRPCServer(("localhost", 8002))
server.register_instance(T())
srv = T()
srv.load()
server.serve_forever()

Client:

import xmlrpclib
proxy = xmlrpclib.ServerProxy("http://localhost:8002/")

# method 1. 用unicode编码

s = u"美女"
print "the variable to transfer is "+ s
res_u1 = proxy.printf(s)

# method 2. decode to unicode
s = "美女"
print "the variable to transfer is "+ s
res_u2 = proxy.printf(s.decode('latin1'))

assert res_u1 == res_u2
res_gbk = res_u1.encode('latin1')
print res_gbk

# 再进一步

def unicode_to_gbk(s):
 return s.encode('latin1').decode('utf-8').encode('gbk')

res = proxy.idx("1,1") # receive unicode
a =  [unicode_to_gbk(s) for s in res] # transfer unicode->gbk
print a[0], a[1]

内容版权声明:除非注明,否则皆为本站原创文章。

转载注明出处:https://www.heiqu.com/2015195eb79723ec7c6f573960b61b2e.html