A simple, immutable URL class with a clean API for interrogation and manipulation. Supports Pythons 2.7, 3.3, 3.4, 3.5, 3.6, 3.7, 3.8 and pypy.

Also supports template URLs as per RFC 6570

Contents:

http://purl.readthedocs.org/en/latest/

From PyPI (stable):

$ pip install purl

From Github (unstable):

$ pip install git+git://github.com/codeinthehole/purl.git#egg=purl

Construct:

>>> from purl import URL

# String constructor
>>> from_str = URL('https://www.google.com/search?q=testing')

# Keyword constructor
>>> from_kwargs = URL(scheme='https', host='www.google.com', path='/search', query='q=testing')

# Combine
>>> from_combo = URL('https://www.google.com').path('search').query_param('q', 'testing')

URL objects are immutable - all mutator methods return a new instance.

Interrogate:

>>> u = URL('https://www.google.com/search?q=testing')
>>> u.scheme()
'https'
>>> u.host()
'www.google.com'
>>> u.domain()
'www.google.com'
>>> u.username()
>>> u.password()
>>> u.netloc()
'www.google.com'
>>> u.port()
>>> u.path()
'/search'
>>> u.query()
'q=testing'
>>> u.fragment()
''
>>> u.path_segment(0)
'search'
>>> u.path_segments()
('search',)
>>> u.query_param('q')
'testing'
>>> u.query_param('q', as_list=True)
['testing']
>>> u.query_param('lang', default='GB')
'GB'
>>> u.query_params()
{'q': ['testing']}
>>> u.has_query_param('q')
True
>>> u.has_query_params(('q', 'r'))
False
>>> u.subdomains()
['www', 'google', 'com']
>>> u.subdomain(0)
'www'

Note that each accessor method is overloaded to be a mutator method too, similar to the jQuery API. Eg:

>>> u = URL.from_string('https://github.com/codeinthehole')

# Access
>>> u.path_segment(0)
'codeinthehole'

# Mutate (creates a new instance)
>>> new_url = u.path_segment(0, 'tangentlabs')
>>> new_url is u
False
>>> new_url.path_segment(0)
'tangentlabs'

Hence, you can build a URL up in steps:

>>> u = URL().scheme('http').domain('www.example.com').path('/some/path').query_param('q', 'search term')
>>> u.as_string()
'http://www.example.com/some/path?q=search+term'

Along with the above overloaded methods, there is also a add_path_segment method for adding a segment at the end of the current path:

>>> new_url = u.add_path_segment('here')
>>> new_url.as_string()
'http://www.example.com/some/path/here?q=search+term'

Couple of other things:

• Since the URL class is immutable it can be used as a key in a dictionary
• It can be pickled and restored
• It supports equality operations
• It supports equality operations

URL templates can be used either via a Template class:

>>> from purl import Template
>>> tpl = Template("http://example.com{/list*}")
>>> url = tpl.expand({'list': ['red', 'green', 'blue']})
>>> url.as_string()
'http://example.com/red/green/blue'

or the expand function:

>>> from purl import expand
>>> expand(u"{/list*}", {'list': ['red', 'green', 'blue']})
'/red/green/blue'

A wide variety of expansions are possible - refer to the RFC for more details.

• Use pytest insteed of nose.
• Fix warning around regex string.

• Allow @ in passwords.

• Allow usernames and passwords to be removed from URLs.

• Ensure paths always have a leading slash.

• Allow absolute URLs to be converted into relative.

• Support password-less URLs.
• Allow slashes to be passed as path segments.

• Support setting username and password via mutator methods

• Handle some unicode compatibility edge-cases

• Fix template expansion bug with no matching variables being passed in. This ensures purl.Template works correctly with the URLs returned from the Github API.

• Fix bug with special characters in paths not being escaped.

• Slight tidy up. Document support for PyPy and Python 3.4.

• Support for RFC 6570 URI templates

• All internal strings are unicode.
• Support for unicode chars in path, fragment, query, auth added.

• Added append_query_param method
• Added remove_query_param method

• Added support for Python 3.2/3.3 (thanks @pmcnr and @mitchellrj)

• Added API docs
• Added to readthedocs.org

• Modified constructor to accept full URL string as first arg
• Added add_path_segment method

• Fixed bug port number in string when using from_string constructor

• Fixed bug with passing lists to query param setter methods

• Added support for comparison and equality
• Added support for pickling
• Added __slots__ so instances can be used as keys within dictionaries

Clone, create a virtualenv then install purl and the packages required for testing:

$ git clone [email protected]:codeinthehole/purl.git
$ cd purl
$ mkvirtualenv purl  # requires virtualenvwrapper
(purl) $ make

Ensure tests pass using:

(purl) $ pytest

or:

$ tox