Network Transport

Grab can use two libraries to submit network requests: pycurl and urllib3. You may acess transport object with Grab.transport attribute. In most cases you do not need direct access to transport object.

Pycurl transport

The pycurl transport is the default network transport. You can control low-level options of pycurl object by accessing Grab.transport.pycurl object. For example:

from grab import Grab
import pycurl

grab = Grab()
grab.transport.pycurl.setopt(pycurl.LOW_SPEED_LIMIT, 100)

Urllib3 transport

If you want to use Grab in gevent environment then consider to use urllib3 transport. The urllib3 uses native python sockets that could be patched by gevent.monkey.patch_all.

import gevent
import gevent.monkey
from grab import Grab
import time

def worker():
    g = Grab(user_agent='Medved', transport='urllib3')
    # Request the document that is served with 1 second delay
    return g.doc.json['headers']['User-Agent']

started = time.time()
pool = []
for _ in range(10):
for th in pool:
    assert th.value == 'Medved'
# The total time would be less than 2 seconds
# unless you have VERY bad internet connection
assert (time.time() - started) < 2

Use your own transport

You can implement you own transport class and use it. Just pass your transport class to transport option.

Here is the crazy example of wget-powered transport. Note that this is VERY simple transport that understands only one option: the URL.

from grab import Grab
from grab.document import Document
from subprocess import check_output

class WgetTransport(object):
    def __init__(self):
        self.request_head = b''
        self.request_body = b''

    def reset(self): pass

    def process_config(self, grab):
        self._request_url = grab.config['url']

    def request(self):
        out = check_output(['/usr/bin/wget', '-O', '-',
        self._response_body = out

    def prepare_response(self, grab):
        doc = Document()
        doc.body = self._response_body
        return doc

g = Grab(transport=WgetTransport)
assert 'Secure email' in g.doc('//title').text()