简体   繁体   中英

Benefits of twisted-suds - Async way of using python suds soap lib

I'm using python suds library to make a SOAP client based on a local wsdl file. My goal is to use Twisted as the backend so I query the SOAP servers in a asyncronous way.

I know this topic has been covered different times ( here1 , here2 ), but I still have some questions.


I've seen three different approaches to use twisted with suds:

a) Applying this patch to the suds library.

b) Use twisted-suds , which is a fork of suds.

c) Influenced by this post , I implemented Client_Async suds client using the twisted deferToThread operation, (fully working gist can be found here . I also implemented a Client_Sync suds client as well to do some benchmarks)

# Init approach c) code
from suds.client import Client as SudsClient
from twisted.internet.threads import deferToThread

class MyClient(SudsClient):
   def handleFailure(self, f, key, stats):
        stats.stop_stamp(error=True)
        logging.error("%s. Failure: %s" % (key, str(f)))

    def handleResult(self, result, key, stats):
        stats.stop_stamp(error=False)
        success, text, res = False, None, None
        try:
            success = result.MessageResult.MessageResultCode == 200
            text = result.MessageResult.MessageResultText
            res = result.FooBar
        except Exception, err:
            pass
        logging.debug('%40s : %5s %10s \"%40s\"' % (key, success, text, res))
        logging.debug('%40s : %s' % (key, self.last_sent()))
        logging.debug('%40s : %s' % (key, self.last_received()))

def call(stats, method, service, key, *a, **kw):
    stats.start_stamp()
    logging.debug('%40s : calling!' % (key))
    result = service.__getattr__(method)(*a, **kw)
    return result

class Client_Async(MyClient):
    """ Twisted based async client"""
    def callRemote(self, stats, method, key, *args, **kwargs):
        logging.debug('%s. deferring to thread...' % key)
        d = deferToThread(call, stats, method, self.service, key, *args, **kwargs)
        d.addCallback(self.handleResult, key, stats)
        d.addErrback(self.handleFailure, key, stats)
        return d

class Client_Sync(MyClient):
    def callRemote(self, stats,  method, key, *args, **kwargs):
        result = None
        try:
            result = call(stats, method, self.service, key, *args, **kwargs)
        except Exception, err:
            self.handleFailure(err, key, stats)
        else:
            self.handleResult(result, key, stats)
# End approach c) code

Doing a small benchmark using the c) approach points the benefits of the Async model:

-- Sync model using Client_Sync of approach c).
# python soap_suds_client.py -t 200 --sync
Total requests:800/800. Success:794 Errors:6
Seconds elapsed:482.0
Threads used:1

-- Async model using Client_Async of approach c).
# python soap_suds_client.py -t 200
Total requests:800/800. Success:790 Errors:10
Seconds elapsed:53.0
Threads used:11

I haven't tested approaches a) or b), my question is:

What am I really gaining from them apart from the use of just one thread?

I'm using suds in my projects. I didn't have to do any patches, or use twisted-suds. I'm using the 0.4.1-2 version of python-suds package (on ubuntu) and it comes with very usefull nosend option.

    # This parses the wsdl file. The autoblend option you'd probably skip, 
    # its needed when name spaces are not strictly preserved (case for Echo Sign).
    from suds import client
    self._suds = client.Client('file://' + config.wsdl_path, nosend=True,
                               autoblend=True)

    ....

    # Create a context for the call, example sendDocument() call. This doesn't yet
    # send anything, only creates an object with the request and capable of parsing
    # the response
    context = self._suds.service.sendDocument(apiKey=....)

    # Actually send the request. Use any web client you want. I actually use 
    # something more sophisticated, but below I put the example using 
    # standard twisted web client.
    from twisted.web import client
    d = client.getPage(url=context.client.location(), 
                       postdata=str(context.envelope),
                       method='POST',
                       headers=context.client.headers())

    # The callback() of the above Deferred is fired with the body of the 
    # http response. I parse it using the context object.          
    d.addCallback(context.succeeded)

    # Now in the callback you have the actual python object defined in 
    # your WSDL file. You can print...
    from pprint import pprint
    d.addCallback(pprint)
    # I the response is a failure, your Deferred would be errbacked with 
    # the suds.WebFault exception.

The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.

 
粤ICP备18138465号  © 2020-2024 STACKOOM.COM