splinterext.proxyheader

Abstract

You too love splinter that makes Web functional tests through a real browser so easy.

Unfortunately, splinter has a limited access to the request and response HTTP headers. This is the consequence of the limitation of the browser drivers splinter relies on. So whatever's the talent of the splinter developers, this issue cannot be closed in a clean fashion as long as these drivers don't support HTTP headers.

But there is a solution

This package adds to splinter browser the ability to play with HTTP request and response headers with additional APIs. This is done by instanciating a tests dedicated proxy on localhost on which the tests browser will rely.

This special proxy will give you the ability to control the HTTP request headers as well as readind the response headers through additional splinter browser methods.

Don't worry about starting or stopping that proxy. This is completely transparent: this temporary proxy is built and started when you create your browser through the legacy API, then stopped and deleted when you invoke the quit() method or the test script exits.

Using splinterext.proxyheader

Now, instead of:

from splinter import Browser
...
browser = Browser("whatever", ...)

You will need to write this:

from splinterext.proxyheader import Browser
...
browser = Browser("whatever", ..., use_proxy=3128, ...)
...
# Now you can use ``browser`` as a legacy splinter browser with some adds.

But...

The counterpart of this is that your favorite browser setup must include an HTTP + HTTPS proxy on localhost, on some available port (say 3128 that's the default port for proxies, manageable by non root users)

Firefox profiles manager lets you add named profiles. Read this and create a splinterproxy profile that includes that proxy setup. Then your scripts should look like this

from splinterext.proxyheader import Browser
...
browser = Browser("firefox", ..., profile="spliterproxy", ... use_proxy=3128)

Additional API

The Browser object provided by splinterext.proxyheader provides these additional methods :

• set a header with a value

  browser.set_request_header(name, value)
  

• removes a header (fails silently)

  browser.remove_request_header(name)
  

• set a basic authentication token

  browser.basic_authenticate(login, password)
  

• clear an authentication token

  browser.basic_anonymize()
  

• get a response header (None if header not found)

  browser.get_response_header(name)
  

Known issues

HTTPS support

The (poor) proxy does not support https. Pages published by https server won't show, pages having sub elements (images, ...) published by https servers will be incomplete.

Thread safety

There are some global objects that inject and capture headers. Playing with browsers shared by threads or even threads that control their own browser may drive to weird results.