Selenium Wire

Selenium Wire extends Selenium's Python bindings to give you access to the underlying requests made by the browser. You author your code in the same way as you do with Selenium, but you get extra APIs for inspecting requests and responses and making changes to them on the fly.

Simple Example

from seleniumwire import webdriver  # Import from seleniumwire

# Create a new instance of the Chrome driver
driver = webdriver.Chrome()

# Go to the Google home page
driver.get('https://www.google.com')

# Access requests via the `requests` attribute
for request in driver.requests:
    if request.response:
        print(
            request.url,
            request.response.status_code,
            request.response.headers['Content-Type']
        )

Prints:

https://www.google.com/ 200 text/html; charset=UTF-8
https://www.google.com/images/branding/googlelogo/2x/googlelogo_color_120x44dp.png 200 image/png
https://consent.google.com/status?continue=https://www.google.com&pc=s&timestamp=1531511954&gl=GB 204 text/html; charset=utf-8
https://www.google.com/images/branding/googlelogo/2x/googlelogo_color_272x92dp.png 200 image/png
https://ssl.gstatic.com/gb/images/i2_2ec824b0.png 200 image/png
https://www.google.com/gen_204?s=webaft&t=aft&atyp=csi&ei=kgRJW7DBONKTlwTK77wQ&rt=wsrt.366,aft.58,prt.58 204 text/html; charset=UTF-8
...

Features

Pure Python, user-friendly API
HTTP and HTTPS requests captured
Intercept requests and responses
Modify headers, parameters, body content on the fly
Capture websocket messages
HAR format supported
Proxy server support

Compatibilty

Python 3.6+
Selenium 3.4.0+
Chrome, Firefox and Remote Webdriver supported

Installation
- Browser Setup
- OpenSSL
Creating the Webdriver
Accessing Requests
Request Objects
Response Objects
Intercepting Requests and Responses
Limiting Request Capture
Proxies
- SOCKS
Bot Detection
Backends
Certificates
All Options
License

Installation

Install using pip:

pip install selenium-wire

If you get an error about not being able to build cryptography, you may be running an old version of pip. Try upgrading pip with python -m pip install --upgrade pip, and then re-run the above command.

Browser Setup

No specific configuration should be necessary except to ensure that you have downloaded the ChromeDriver and GeckoDriver for Chrome and Firefox to be remotely controlled - the same as if you were using Selenium directly. Once downloaded, these executables should be placed somewhere on your PATH.

OpenSSL

Selenium Wire requires OpenSSL for decrypting HTTPS requests. This is normally already installed on most systems, but if it's not you can install it with:

Linux

# For apt based Linux systems
sudo apt install openssl

# For RPM based Linux systems
sudo yum install openssl

# For Linux alpine
sudo apk add openssl

MacOS

brew install openssl

Windows

No installation is required.

Creating the Webdriver

Ensure that you import webdriver from the seleniumwire package:

from seleniumwire import webdriver

For sub-packages of webdriver, you should continue to import these directly from selenium. For example, to import WebDriverWait:

# Sub-packages of webdriver must still be imported from `selenium` itself
from selenium.webdriver.support.ui import WebDriverWait

Chrome and Firefox

For Chrome and Firefox, you don't need to do anything special. Just instantiate the webdriver as you would normally with webdriver.Chrome() or webdriver.Firefox() passing in any desired capabilities and browser specific options for Chrome or Firefox , such as the executable path, headless mode etc. Selenium Wire also has it's own options that can be passed in the seleniumwire_options attribute.

Remote

Selenium Wire has limited support for using the remote webdriver client. When you create an instance of the remote webdriver, you need to specify the hostname or IP address of the machine (or container) running Selenium Wire. This allows the remote instance to communicate back to Selenium Wire with its requests and responses.

options = {
    'addr': 'hostname_or_ip'  # Address of the machine running Selenium Wire
}
driver = webdriver.Remote(
    command_executor='http://www.example.com',
    seleniumwire_options=options
)

If the machine running the browser needs to use a different address to talk to the machine running Selenium Wire, you need to configure the browser manually. This issue goes into more detail.

Accessing Requests

Selenium Wire captures all HTTP/HTTPS traffic made by the browser [1].

driver.requests

The list of captured requests in chronological order.

driver.last_request

Convenience attribute for retrieving the most recently captured request. This is more efficient than using driver.requests[-1].

driver.wait_for_request(pat, timeout=10)

This method will wait until it sees a request matching a pattern. The pat attribute will be matched within the request URL. pat can be a simple sub-string or a regex. Note that driver.wait_for_request() doesn't make a request, it just waits for a previous request made by some other action - and it will return the first request it finds. Also note that since pat can be a regex, you must escape special characters such as question marks with a slash. A TimeoutException is raised if no match is found within the timeout period.

For example, to wait for an AJAX request to return after a button is clicked:

# Click a button that triggers a background request to https://server/api/products/12345/
button_element.click()

# Wait for the request/response to complete
request = driver.wait_for_request('/api/products/12345/')

driver.har

A JSON formatted HAR archive of HTTP transactions that have taken place. HAR capture is turned off by default, and you must set the enable_har option to True before using driver.har.

driver.iter_requests()

Returns an iterator over captured requests. Useful when dealing with a large number of requests.

driver.request_interceptor

Used to set a request interceptor.

driver.response_interceptor

Used to set a response interceptor. See Intercepting Requests and Responses.

Clearing Requests

To clear previously captured requests and HAR entries, use del:

del driver.requests

[1]	Selenium Wire ignores OPTIONS requests by default, as these are typically uninteresting and just add overhead. If you want to capture OPTIONS requests, you need to set the `ignore_http_methods` option to `[]`.

Request Objects

Request objects have the following attributes.

body: The request body as bytes. If the request has no body the value of body will be empty, i.e. b''.
cert: Information about the server SSL certificate in dictionary format. Empty for non-HTTPS requests.
date: The datetime the request was made.
headers: A dictionary-like object of request headers. Headers are case-insensitive and duplicates are permitted. Asking for request.headers['user-agent'] will return the value of the User-Agent header. If you wish to replace a header, make sure you delete the existing header first with del request.headers['header-name'], otherwise you'll create a duplicate.
method: The HTTP method, e.g. GET or POST etc.
params: A dictionary of request parameters. If a parameter with the same name appears more than once in the request, it's value in the dictionary will be a list.
path: The request path, e.g. /some/path/index.html
querystring: The query string, e.g. foo=bar&spam=eggs
response: The response associated with the request. This will be None if the request has no response.
url: The request URL, e.g. https://server/some/path/index.html?foo=bar&spam=eggs
ws_messages: Where the request is a websocket handshake request (normally with a URL starting wss://), then ws_messages will contain a list of any websocket messages sent and received. See WebSocketMessage Objects.

Request objects have the following methods.

abort(error_code=403): Trigger immediate termination of the request with the supplied error code. For use within request interceptors. See Example: Block a request.
create_response(status_code, headers=(), body=b''): Create a response and return it without sending any data to the remote server. For use within request interceptors. See Example: Mock a response.

WebSocketMessage Objects

These objects represent websocket messages sent between the browser and server and vice versa. They are held in a list by request.ws_messages on websocket handshake requests. They have the following attributes.

content: The message content which may be either str or bytes.
date: The datetime of the message.
from_client: True when the message was sent by the client and False when sent by the server.

Response Objects

Response objects have the following attributes.

body: The response body as bytes. If the response has no body the value of body will be empty, i.e. b''.
date: The datetime the response was received.
headers: A dictionary-like object of response headers. Headers are case-insensitive and duplicates are permitted. Asking for response.headers['content-length'] will return the value of the Content-Length header. If you wish to replace a header, make sure you delete the existing header first with del response.headers['header-name'], otherwise you'll create a duplicate.
reason: The reason phrase, e.g. OK or Not Found etc.
status_code: The status code of the response, e.g. 200 or 404 etc.

Intercepting Requests and Responses

Selenium Wire allows you to modify requests and responses on the fly using interceptors. An interceptor is a function that gets invoked with requests and responses as they pass through Selenium Wire. Within an interceptor you can modify the request and response as you see fit.

You set your interceptor functions using the driver.request_interceptor and driver.response_interceptor attributes before you start using the driver. A request interceptor should accept a single argument for the request. A response interceptor should accept two arguments, one for the originating request and one for the response.

Example: Add a request header

def interceptor(request):
    request.headers['New-Header'] = 'Some Value'

driver.request_interceptor = interceptor
driver.get(...)

# All requests will now contain New-Header

How can I check that a header has been set correctly? You can print the headers from captured requests after the page has loaded (using driver.requests), or alternatively point the webdriver at https://httpbin.org/headers which will echo the request headers back to the browser so you can view them.

Example: Replace an existing request header

Duplicate header names are permitted in an HTTP request, so before setting the replacement header you must first delete the existing header using del like in the following example, otherwise two headers with the same name will exist (request.headers is a special dictionary-like object that allows duplicates).

def interceptor(request):
    del request.headers['Referer']  # Remember to delete the header first
    request.headers['Referer'] = 'some_referer'  # Spoof the referer

driver.request_interceptor = interceptor
driver.get(...)

# All requests will now use 'some_referer' for the referer

Example: Add a response header

def interceptor(request, response):  # A response interceptor takes two args
    if request.url == 'https://server.com/some/path':
        response.headers['New-Header'] = 'Some Value'

driver.response_interceptor = interceptor
driver.get(...)

# Responses from https://server.com/some/path will now contain New-Header

Example: Add a request parameter

Request parameters work differently to headers in that they are calculated when they are set on the request. That means that you first have to read them, then update them, and then write them back - like in the following example. Parameters are held in a regular dictionary, so parameters with the same name will be overwritten.

def interceptor(request):
    params = request.params
    params['foo'] = 'bar'
    request.params = params

driver.request_interceptor = interceptor
driver.get(...)

# foo=bar will be added to all requests

Example: Update JSON in a POST request body

import json

def interceptor(request):
    if request.method == 'POST' and request.headers['Content-Type'] == 'application/json':
        # The body is in bytes so convert to a string
        body = request.body.decode('utf-8')
        # Load the JSON
        data = json.loads(body)
        # Add a new property
        data['foo'] = 'bar'
        # Set the JSON back on the request
        request.body = json.dumps(data).encode('utf-8')
        # Update the content length
        del request.headers['Content-Length']
        request.headers['Content-Length'] = str(len(request.body))

driver.request_interceptor = interceptor
driver.get(...)

Example: Block a request

You can use request.abort() to block a request and send an immediate response back to the browser. An optional error code can be supplied. The default is 403 (forbidden).

def interceptor(request):
    # Block PNG, JPEG and GIF images
    if request.path.endswith(('.png', '.jpg', '.gif')):
        request.abort()

driver.request_interceptor = interceptor
driver.get(...)

# Requests for PNG, JPEG and GIF images will result in a 403 Forbidden

Example: Mock a response

You can use request.create_response() to send a custom reply back to the browser. No data will be sent to the remote server.

def interceptor(request):
    if request.url == 'https://server.com/some/path':
        request.create_response(
            status_code=200,
            headers={'Content-Type': 'text/html'},  # Optional headers dictionary
            body='<html>Hello World!</html>'  # Optional body
        )

driver.request_interceptor = interceptor
driver.get(...)

# Requests to https://server.com/some/path will have their responses mocked

Have any other examples you think could be useful? Feel free to submit a PR.

Unset an interceptor

To unset an interceptor, use del:

del driver.request_interceptor
del driver.response_interceptor

Limiting Request Capture

Selenium Wire works by redirecting browser traffic through an internal proxy server it spins up in the background. As requests flow through the proxy they are intercepted and captured. Capturing requests can slow things down a little, but there are a few things you can do to restrict what gets captured.

driver.scopes

This accepts a list of regular expressions that will match the URLs to be captured. It should be set on the driver before making any requests. When empty (the default) all URLs are captured.

driver.scopes = [
    '.*stackoverflow.*',
    '.*github.*'
]

driver.get(...)  # Start making requests

# Only request URLs containing "stackoverflow" or "github" will now be captured

Note that even if a request is out of scope and not captured, it will still travel through Selenium Wire.

seleniumwire_options.disable_capture

Use this option to switch off request interception and capture. Requests will still pass through Selenium Wire and through any upstream proxy you have configured, but they won't be decrypted and nothing will be stored. Useful for boosting performance if all you want is the upstream proxy functionality.

options = {
    'disable_capture': True  # Pass all requests straight through
}
driver = webdriver.Chrome(seleniumwire_options=options)

seleniumwire_options.exclude_hosts

Use this option to bypass Selenium Wire entirely. Any requests made to addresses listed here will go direct from the browser to the server without involving Selenium Wire. Note that if you've configured an upstream proxy then these requests will also bypass that proxy.

options = {
    'exclude_hosts': ['host1.com', 'host2.com']  # Bypass Selenium Wire for these hosts
}
driver = webdriver.Chrome(seleniumwire_options=options)

request.abort()

You can abort a request early by using request.abort() from within a request interceptor. This will send an immediate response back to the client without the request travelling any further. You can use this mechanism to block certain types of requests (e.g. images) to improve page load performance.

def interceptor(request):
    # Block PNG, JPEG and GIF images
    if request.path.endswith(('.png', '.jpg', '.gif')):
        request.abort()

driver.request_interceptor = interceptor

driver.get(...)  # Start making requests

Proxies

If the site you are accessing sits behind a proxy server you can tell Selenium Wire about that proxy server in the options you pass to the webdriver.

The configuration takes the following format:

options = {
    'proxy': {
        'http': 'http://192.168.10.100:8888',
        'https': 'https://192.168.10.100:8888',
        'no_proxy': 'localhost,127.0.0.1'
    }
}
driver = webdriver.Chrome(seleniumwire_options=options)

To use HTTP Basic Auth with your proxy, specify the username and password in the URL:

options = {
    'proxy': {
        'https': 'https://user:pass@192.168.10.100:8888',
    }
}

For authentication other than Basic, you can supply the full value for the Proxy-Authorization header using the custom_authorization option. For example, if your proxy used the Bearer scheme:

options = {
    'proxy': {
        'https': 'https://192.168.10.100:8888',  # No username or password used
        'custom_authorization': 'Bearer mytoken123'  # Custom Proxy-Authorization header value
    }
}

Note that the custom_authorization option is only supported by the default backend. More info on the Proxy-Authorization header can be found here.

The proxy configuration can also be loaded through environment variables called HTTP_PROXY, HTTPS_PROXY and NO_PROXY:

$ export HTTP_PROXY="http://192.168.10.100:8888"
$ export HTTPS_PROXY="https://192.168.10.100:8888"
$ export NO_PROXY="localhost,127.0.0.1"

I just want the proxy functionality, I don't care about request capture

In which case you can disable request capture using the disable_capture option. When this option is set to True Selenium Wire will just pass everything straight through to the upstream proxy. No interception, decryption or capture will take place, which should improve performance.

SOCKS

Using a SOCKS proxy is the same as using an HTTP based one, but with the scheme set to socks5:

options = {
    'proxy': {
        'http': 'socks5://user:pass@192.168.10.100:8888',
        'https': 'socks5://user:pass@192.168.10.100:8888',
        'no_proxy': 'localhost,127.0.0.1'
    }
}
driver = webdriver.Chrome(seleniumwire_options=options)

You can leave out the user and pass if your proxy doesn't require authentication.

As well as socks5, the schemes socks4 and socks5h are supported. Use socks5h when you want DNS resolution to happen on the proxy server rather than on the client.

Using Selenium Wire with Tor

See this example if you want to run Selenium Wire with Tor.

Bot Detection

Selenium Wire will automatically integrate itself with undetected-chromedriver if it finds it in your environment. This library will transparently modify ChromeDriver to prevent it from triggering anti-bot measures on websites.

If you wish to take advantage of this, make sure you have undetected-chromedriver installed:

pip install undetected-chromedriver

Then just use webdriver.Chrome() as you would normally, making sure that you import it from the seleniumwire package. If you use ChromeOptions this should also be imported from the seleniumwire package:

from seleniumwire import webdriver

chrome_options = webdriver.ChromeOptions()
sw_options = {...}

driver = webdriver.Chrome(  # Optimized for bot detection
    options=chrome_options,
    seleniumwire_options=sw_options
)

The first time you run the webdriver it will download and patch the ChromeDriver binary in the background.

You can check that undetected_chromedriver is being used by looking for the log messages it generates. You just need to ensure that you've activated logging at the top of your script or program first - for example:

import logging
logging.basicConfig(level=logging.INFO)
logging.getLogger('undetected_chromedriver').level = logging.INFO

from seleniumwire import webdriver

... code ...

You should see messages similar to:

INFO:undetected_chromedriver:Selenium patched. Safe to import Chrome / ChromeOptions
INFO:undetected_chromedriver:starting undetected_chromedriver.Chrome((), ...

Note that this functionality is currently experimental.

Backends

Selenium Wire allows you to change the backend component that performs request capture. Currently two backends are supported: the backend that ships with Selenium Wire (the default) and the mitmproxy backend.

The default backend is adequate for most purposes. However, in certain cases you may find you get better performance with the mitmproxy backend.

The mitmproxy backend relies upon the powerful open source mitmproxy proxy server being installed in your environment.

To switch to the mitmproxy backend, first install the mitmproxy package:

pip install mitmproxy

Once installed, set the backend option in Selenium Wire's options to mitmproxy:

options = {
    'backend': 'mitmproxy'
}
driver = webdriver.Chrome(seleniumwire_options=options)

You can pass mitmproxy specific options to the mitmproxy backend by prefixing them with mitm_. For example, to change the location of the mitmproxy configuration directory which lives in your home folder by default:

options = {
    'backend': 'mitmproxy',
    'mitm_confdir': '/tmp/.mitmproxy'  # Switch the location to /tmp
}
driver = webdriver.Chrome(seleniumwire_options=options)

Mitmproxy includes options that can help with performance such as mitm_stream_large_bodies. Setting this to a low value (e.g. '1k') has been shown to improve performance, in conjunction with the use of driver.scopes.

Note that the mitmproxy backend won't work with upstream SOCKS proxies.

Certificates

Selenium Wire uses it's own CA certificate to decrypt HTTPS traffic. It is not normally necessary for the browser to trust this certificate because Selenium Wire tells the browser to add it as an exception. This will allow the browser to function normally, but it will display a "Not Secure" message in the address bar. If you wish to get rid of this message you can install the CA certificate manually.

For the default backend, you can download the CA certificate here. Once downloaded, navigate to "Certificates" in your browser settings and import the certificate in the "Authorities" section.

If you are using the mitmproxy backend, you can follow these instructions to install the CA certificate.

All Options

A summary of all options that can be passed to Selenium Wire via the seleniumwire_options webdriver attribute.

addr: The IP address or hostname of the machine running Selenium Wire. This defaults to 127.0.0.1. You may want to change this to the public IP of the machine (or container) if you're using the remote webdriver.

options = {
    'addr': '192.168.0.10'  # Use the public IP of the machine
}
driver = webdriver.Chrome(seleniumwire_options=options)

auto_config: Whether Selenium Wire should auto-configure the browser for request capture. True by default.
backend: The backend component that Selenium Wire will use to capture requests. The currently supported values are default (same as not specifying) or mitmproxy.

options = {
    'backend': 'mitmproxy'  # Use the mitmproxy backend (see limitations above)
}
driver = webdriver.Chrome(seleniumwire_options=options)

disable_capture: Disable request interception and capture. When True all requests are passed straight through and no HTTPS decryption takes place. Useful for boosting performance if all you want is the upstream proxy functionality.

options = {
    'disable_capture': True  # Pass all requests straight through.
}
driver = webdriver.Chrome(seleniumwire_options=options)

disable_encoding: Ask the server to send back un-compressed data. When True this sets the Accept-Encoding header to identity for all outbound requests. Note that it won't always work - sometimes the server may ignore it. The default is False.

options = {
    'disable_encoding': True  # Ask the server not to compress the response
}
driver = webdriver.Chrome(seleniumwire_options=options)

enable_har: When True a HAR archive of HTTP transactions will be kept which can be retrieved with driver.har. False by default.

options = {
    'enable_har': True  # Capture HAR data, retrieve with driver.har
}
driver = webdriver.Chrome(seleniumwire_options=options)

exclude_hosts: A list of addresses for which Selenium Wire should be bypassed entirely. Note that if you have configured an upstream proxy then requests to excluded hosts will also bypass that proxy.

options = {
    'exclude_hosts': ['google-analytics.com']  # Bypass these hosts
}
driver = webdriver.Chrome(seleniumwire_options=options)

ignore_http_methods: A list of HTTP methods (specified as uppercase strings) that should be ignored by Selenium Wire and not captured. The default is ['OPTIONS'] which ignores all OPTIONS requests. To capture all request methods, set ignore_http_methods to an empty list:

options = {
    'ignore_http_methods': []  # Capture all requests, including OPTIONS requests
}
driver = webdriver.Chrome(seleniumwire_options=options)

port: The port number that Selenium Wire's backend listens on. You don't normally need to specify a port as a random port number is chosen automatically.

options = {
    'port': 9999  # Tell the backend to listen on port 9999 (not normally necessary to set this)
}
driver = webdriver.Chrome(seleniumwire_options=options)

proxy: The upstream proxy server configuration (if you're using a proxy).

options = {
    'proxy': {
        'http': 'http://user:pass@192.168.10.100:8888',
        'https': 'https://user:pass@192.168.10.100:8889',
        'no_proxy': 'localhost,127.0.0.1'
    }
}
driver = webdriver.Chrome(seleniumwire_options=options)

request_storage_base_dir: Captured requests and responses are stored in the current user's home folder by default. You might want to change this if you're running in an environment where you don't have access to the user's home folder.

options = {
    'request_storage_base_dir': '/tmp'  # Use /tmp to store captured data
}
driver = webdriver.Chrome(seleniumwire_options=options)

suppress_connection_errors: Whether to suppress connection related tracebacks. The default is True so that harmless errors that commonly occur at browser shutdown do not alarm users. When suppressed, the connection error message is logged at DEBUG level without a traceback. Set to False to allow exception propagation and see full tracebacks. Applies to the default backend only.

options = {
    'suppress_connection_errors': False  # Show full tracebacks for any connection errors
}
driver = webdriver.Chrome(seleniumwire_options=options)

verify_ssl: Whether SSL certificates should be verified. The default is False which prevents errors with self-signed certificates.

options = {
    'verify_ssl': True  # Verify SSL certificates but beware of errors with self-signed certificates
}
driver = webdriver.Chrome(seleniumwire_options=options)

License

MIT

eanon/selenium-wire