Contents
Stateful programmatic web browsing in Python
- The browser class mechanize.Browser implements the interface of urllib2.OpenerDirector, so any URL can be opened not just http.
- Easy HTML form filling.
- Convenient link parsing and following.
- Browser history (.back() and .reload() methods).
- The Referer HTTP header is added properly (optional).
- Automatic observance of robots.txt.
- Automatic handling of HTTP-Equiv and Refresh.
To install for normal usage:
sudo pip2 install mechanize
To install for development:
git clone https://github.com/python-mechanize/mechanize.git
cd mechanize
sudo pip2 install -e .
To install manually, simply add the mechanize sub-directory somewhere on your PYTHONPATH.
See https://mechanize.readthedocs.io/en/latest/
python-mechanize was the creation of John J. Lee. Maintenance was taken over by Kovid Goyal in 2017.
Much of the code was originally derived from the work of the following people:
- Gisle Aas -- [libwww-perl]
- Jeremy Hylton (and many others) -- [urllib2]
- Andy Lester -- [WWW::Mechanize]
- Johnny Lee (coincidentally-named) -- MSIE CookieJar Perl code from which mechanize's support for that is derived.
Also:
- Gary Poster and Benji York at Zope Corporation -- contributed significant changes to the HTML forms code
- Ronald Tschalar -- provided help with Netscape cookies
Thanks also to the many people who have contributed bug reports and patches.