url_query.py

Go to the documentation of this file.

"""
Joshua Dawes - CERN, CMS, The University of Manchester

File provides a class that handles pycurl requests.

Provides methods for performing/closing the request, as well as getting the request response.
Note: user agent string from current version of cmsDbUpload
"""

import pycurl
from StringIO import StringIO
from urllib import urlencode
import traceback
import sys
import json
from errors import *
from time import sleep

class url_query():

    def __init__(self, url=None, url_data=None, body=None, response_stream=None, timeout=60):
        if not(url):
            return None
        self._url = url
        self._r = pycurl.Curl()

        # set options for the request - of note is the fact that we do not verify the peer or the host - because
        # CERN certificates are self-signed, and we only need the encryption from HTTPS, not the certificate checks.

        self._r.setopt(self._r.CONNECTTIMEOUT, timeout)
        user_agent = "User-Agent: ConditionWebServices/1.0 python/%d.%d.%d PycURL/%s" % (sys.version_info[ :3 ] + (pycurl.version_info()[1],))
        self._r.setopt(self._r.USERAGENT, user_agent)
        # we don't need to verify who signed the certificate or who the host is
        self._r.setopt(self._r.SSL_VERIFYPEER, 0)
        self._r.setopt(self._r.SSL_VERIFYHOST, 0)
        self._response = StringIO()

        if body:
            if isinstance(body, dict):
                body = urlencode(body)
            elif isinstance(body, list):
                body = json.dumps(body)

            self._r.setopt(self._r.POSTFIELDS, body)

        if url_data:
            if isinstance(url_data, dict):
                url_data = urlencode(url_data)
            else:
                exit("URL data '%s' for request to URL '%s' was not valid - should be a dictionary." % (str(url_data), url))

        # set the URL with url parameters if they were given
        self._r.setopt(self._r.URL, url + (("?%s" % url_data) if url_data else ""))

        if response_stream and not isinstance(response_stream, StringIO):
            response_stream = StringIO()
            # copy reference to instance variable
            self._response = response_stream
        elif not(response_stream):
            self._response = StringIO()

        self._r.setopt(self._r.WRITEFUNCTION, self._response.write)

    def send(self):
        failed = True
        max_retries = 5
        attempt = 0
        # retry while we're within the limit for the number of retries
        while failed and attempt < max_retries:
            try:
                self._r.perform()
                failed = False
                self._r.close()
                return self._response.getvalue()
            except Exception as e:
                failed = True
                attempt += 1
                # this catches exceptions that occur with the actual http request
                # not exceptions sent back from server side
                if isinstance(e, pycurl.error) and e[0] in [7, 52]:
                    # wait two seconds to retry
                    print("Request failed - waiting 3 seconds to retry.")
                    sleep(3)
                    # do nothing for now
                    pass
                else:
                    print("Unforesoon error occurred when sending data to server.")
                    traceback.print_exc()
        if attempt == max_retries:
            raise NoMoreRetriesException(max_retries)