User Guide

Installing

urllib3 can be installed with pip

$ python -m pip install urllib3

Making Requests

First things first, import the urllib3 module:

>>> import urllib3

You’ll need a PoolManager instance to make requests. This object handles all of the details of connection pooling and thread safety so that you don’t have to:

>>> http = urllib3.PoolManager()

To make a request use request():

>>> r = http.request('GET', 'http://httpbin.org/robots.txt')
>>> r.data
b'User-agent: *\nDisallow: /deny\n'

request() returns a HTTPResponse object, the Response Content section explains how to handle various responses.

You can use request() to make requests using any HTTP verb:

>>> r = http.request(
...     'POST',
...     'http://httpbin.org/post',
...     fields={'hello': 'world'}
... )

The Request Data section covers sending other kinds of requests data, including JSON, files, and binary data.

Response Content

The HTTPResponse object provides status, data, and headers attributes:

>>> r = http.request('GET', 'http://httpbin.org/ip')
>>> r.status
200
>>> r.data
b'{\n  "origin": "104.232.115.37"\n}\n'
>>> r.headers
HTTPHeaderDict({'Content-Length': '33', ...})

JSON Content

JSON content can be loaded by decoding and deserializing the data attribute of the request:

>>> import json
>>> r = http.request('GET', 'http://httpbin.org/ip')
>>> json.loads(r.data.decode('utf-8'))
{'origin': '127.0.0.1'}

Binary Content

The data attribute of the response is always set to a byte string representing the response content:

>>> r = http.request('GET', 'http://httpbin.org/bytes/8')
>>> r.data
b'\xaa\xa5H?\x95\xe9\x9b\x11'

Note

For larger responses, it’s sometimes better to stream the response.

Using io Wrappers with Response Content

Sometimes you want to use io.TextIOWrapper or similar objects like a CSV reader directly with HTTPResponse data. Making these two interfaces play nice together requires using the auto_close attribute by setting it to False. By default HTTP responses are closed after reading all bytes, this disables that behavior:

>>> import io
>>> r = http.request('GET', 'https://example.com', preload_content=False)
>>> r.auto_close = False
>>> for line in io.TextIOWrapper(r):
>>>     print(line)

Request Data

Headers

You can specify headers as a dictionary in the headers argument in request():

>>> r = http.request(
...     'GET',
...     'http://httpbin.org/headers',
...     headers={
...         'X-Something': 'value'
...     }
... )
>>> json.loads(r.data.decode('utf-8'))['headers']
{'X-Something': 'value', ...}

Query Parameters

For GET, HEAD, and DELETE requests, you can simply pass the arguments as a dictionary in the fields argument to request():

>>> r = http.request(
...     'GET',
...     'http://httpbin.org/get',
...     fields={'arg': 'value'}
... )
>>> json.loads(r.data.decode('utf-8'))['args']
{'arg': 'value'}

For POST and PUT requests, you need to manually encode query parameters in the URL:

>>> from urllib.parse import urlencode
>>> encoded_args = urlencode({'arg': 'value'})
>>> url = 'http://httpbin.org/post?' + encoded_args
>>> r = http.request('POST', url)
>>> json.loads(r.data.decode('utf-8'))['args']
{'arg': 'value'}

Form Data

For PUT and POST requests, urllib3 will automatically form-encode the dictionary in the fields argument provided to request():

>>> r = http.request(
...     'POST',
...     'http://httpbin.org/post',
...     fields={'field': 'value'}
... )
>>> json.loads(r.data.decode('utf-8'))['form']
{'field': 'value'}

JSON

You can send a JSON request by specifying the encoded data as the body argument and setting the Content-Type header when calling request():

>>> import json
>>> data = {'attribute': 'value'}
>>> encoded_data = json.dumps(data).encode('utf-8')
>>> r = http.request(
...     'POST',
...     'http://httpbin.org/post',
...     body=encoded_data,
...     headers={'Content-Type': 'application/json'}
... )
>>> json.loads(r.data.decode('utf-8'))['json']
{'attribute': 'value'}

Files & Binary Data

For uploading files using multipart/form-data encoding you can use the same approach as Form Data and specify the file field as a tuple of (file_name, file_data):

>>> with open('example.txt') as fp:
...     file_data = fp.read()
>>> r = http.request(
...     'POST',
...     'http://httpbin.org/post',
...     fields={
...         'filefield': ('example.txt', file_data),
...     }
... )
>>> json.loads(r.data.decode('utf-8'))['files']
{'filefield': '...'}

While specifying the filename is not strictly required, it’s recommended in order to match browser behavior. You can also pass a third item in the tuple to specify the file’s MIME type explicitly:

>>> r = http.request(
...     'POST',
...     'http://httpbin.org/post',
...     fields={
...         'filefield': ('example.txt', file_data, 'text/plain'),
...     }
... )

For sending raw binary data simply specify the body argument. It’s also recommended to set the Content-Type header:

>>> with open('example.jpg', 'rb') as fp:
...     binary_data = fp.read()
>>> r = http.request(
...     'POST',
...     'http://httpbin.org/post',
...     body=binary_data,
...     headers={'Content-Type': 'image/jpeg'}
... )
>>> json.loads(r.data.decode('utf-8'))['data']
b'...'

Certificate Verification

Note

New in version 1.25:

HTTPS connections are now verified by default (cert_reqs = 'CERT_REQUIRED').

While you can disable certification verification by setting cert_reqs = 'CERT_NONE', it is highly recommend to leave it on.

Unless otherwise specified urllib3 will try to load the default system certificate stores. The most reliable cross-platform method is to use the certifi package which provides Mozilla’s root certificate bundle:

$ python -m pip install certifi

You can also install certifi along with urllib3 by using the secure extra:

$ python -m pip install urllib3[secure]

Once you have certificates, you can create a PoolManager that verifies certificates when making requests:

>>> import certifi
>>> import urllib3
>>> http = urllib3.PoolManager(
...     cert_reqs='CERT_REQUIRED',
...     ca_certs=certifi.where()
... )

The PoolManager will automatically handle certificate verification and will raise SSLError if verification fails:

>>> http.request('GET', 'https://google.com')
(No exception)
>>> http.request('GET', 'https://expired.badssl.com')
urllib3.exceptions.SSLError ...

Note

You can use OS-provided certificates if desired. Just specify the full path to the certificate bundle as the ca_certs argument instead of certifi.where(). For example, most Linux systems store the certificates at /etc/ssl/certs/ca-certificates.crt. Other operating systems can be difficult.

Using Timeouts

Timeouts allow you to control how long (in seconds) requests are allowed to run before being aborted. In simple cases, you can specify a timeout as a float to request():

>>> http.request(
...     'GET', 'http://httpbin.org/delay/3', timeout=4.0
... )
<urllib3.response.HTTPResponse>
>>> http.request(
...     'GET', 'http://httpbin.org/delay/3', timeout=2.5
... )
MaxRetryError caused by ReadTimeoutError

For more granular control you can use a Timeout instance which lets you specify separate connect and read timeouts:

>>> http.request(
...     'GET',
...     'http://httpbin.org/delay/3',
...     timeout=urllib3.Timeout(connect=1.0)
... )
<urllib3.response.HTTPResponse>
>>> http.request(
...     'GET',
...     'http://httpbin.org/delay/3',
...     timeout=urllib3.Timeout(connect=1.0, read=2.0)
... )
MaxRetryError caused by ReadTimeoutError

If you want all requests to be subject to the same timeout, you can specify the timeout at the PoolManager level:

>>> http = urllib3.PoolManager(timeout=3.0)
>>> http = urllib3.PoolManager(
...     timeout=urllib3.Timeout(connect=1.0, read=2.0)
... )

You still override this pool-level timeout by specifying timeout to request().

Retrying Requests

urllib3 can automatically retry idempotent requests. This same mechanism also handles redirects. You can control the retries using the retries parameter to request(). By default, urllib3 will retry requests 3 times and follow up to 3 redirects.

To change the number of retries just specify an integer:

>>> http.requests('GET', 'http://httpbin.org/ip', retries=10)

To disable all retry and redirect logic specify retries=False:

>>> http.request(
...     'GET', 'http://nxdomain.example.com', retries=False
... )
NewConnectionError
>>> r = http.request(
...     'GET', 'http://httpbin.org/redirect/1', retries=False
... )
>>> r.status
302

To disable redirects but keep the retrying logic, specify redirect=False:

>>> r = http.request(
...     'GET', 'http://httpbin.org/redirect/1', redirect=False
... )
>>> r.status
302

For more granular control you can use a Retry instance. This class allows you far greater control of how requests are retried.

For example, to do a total of 3 retries, but limit to only 2 redirects:

>>> http.request(
...     'GET',
...     'http://httpbin.org/redirect/3',
...     retries=urllib3.Retry(3, redirect=2)
... )
MaxRetryError

You can also disable exceptions for too many redirects and just return the 302 response:

>>> r = http.request(
...     'GET',
...     'http://httpbin.org/redirect/3',
...     retries=urllib3.Retry(
...         redirect=2, raise_on_redirect=False)
... )
>>> r.status
302

If you want all requests to be subject to the same retry policy, you can specify the retry at the PoolManager level:

>>> http = urllib3.PoolManager(retries=False)
>>> http = urllib3.PoolManager(
...     retries=urllib3.Retry(5, redirect=2)
... )

You still override this pool-level retry policy by specifying retries to request().

Errors & Exceptions

urllib3 wraps lower-level exceptions, for example:

>>> try:
...     http.request('GET', 'nx.example.com', retries=False)
... except urllib3.exceptions.NewConnectionError:
...     print('Connection failed.')

See exceptions for the full list of all exceptions.

Logging

If you are using the standard library logging module urllib3 will emit several logs. In some cases this can be undesirable. You can use the standard logger interface to change the log level for urllib3’s logger:

>>> logging.getLogger("urllib3").setLevel(logging.WARNING)