How to handle IncompleteRead: in python

You can use requests instead of urllib2. requests is based on urllib3 so it rarely have any problem. Put it in a loop to try it 3 times, and it will be much stronger. You can use it this way:

import requests      

msg = None   
for i in [1,2,3]:        
    try:  
        r = requests.get(self.crawling, timeout=30)
        msg = r.text
        if msg: break
    except Exception as e:
        sys.stderr.write('Got error when requesting URL "' + self.crawling + '": ' + str(e) + '\n')
        if i == 3 :
            sys.stderr.write('{0.filename}@{0.lineno}: Failed requesting from URL "{1}" ==> {2}\n'.                       format(inspect.getframeinfo(inspect.currentframe()), self.crawling, e))
            raise e
        time.sleep(10*(i-1))

Note this answer is Python 2 only (it was published in 2013)

I find out in my case : send HTTP/1.0 request , adding this , fix the problem.

import httplib
httplib.HTTPConnection._http_vsn = 10
httplib.HTTPConnection._http_vsn_str = 'HTTP/1.0'

after I do the request :

req = urllib2.Request(url, post, headers)
filedescriptor = urllib2.urlopen(req)
img = filedescriptor.read()

after I back to http 1.1 with (for connections that support 1.1) :

httplib.HTTPConnection._http_vsn = 11
httplib.HTTPConnection._http_vsn_str = 'HTTP/1.1'

the trick is use http 1.0 instead the default http/1.1 http 1.1 could handle chunks but for some reason webserver don't , so we do the request in http 1.0

for Python3, it will tell you

ModuleNotFoundError: No module named 'httplib'

then try to use http.client Module it will solve the problem

import http.client as http
http.HTTPConnection._http_vsn = 10
http.HTTPConnection._http_vsn_str = 'HTTP/1.0'



What worked for me is catching IncompleteRead as an exception and harvesting the data you managed to read in each iteration by putting this into a loop like below: (Note, I am using Python 3.4.1 and the urllib library has changed between 2.7 and 3.4)

try:
    requestObj = urllib.request.urlopen(url, data)
    responseJSON=""
    while True:
        try:
            responseJSONpart = requestObj.read()
        except http.client.IncompleteRead as icread:
            responseJSON = responseJSON + icread.partial.decode('utf-8')
            continue
        else:
            responseJSON = responseJSON + responseJSONpart.decode('utf-8')
            break

    return json.loads(responseJSON)

except Exception as RESTex:
    print("Exception occurred making REST call: " + RESTex.__str__())

The link you included in your question is simply a wrapper that executes urllib's read() function, which catches any incomplete read exceptions for you. If you don't want to implement this entire patch, you could always just throw in a try/catch loop where you read your links. For example:

try:
    page = urllib2.urlopen(urls).read()
except httplib.IncompleteRead, e:
    page = e.partial

for python3

try:
    page = request.urlopen(urls).read()
except (http.client.IncompleteRead) as e:
    page = e.partial