def __init__(self, httpResponse):
encoding = httpResponse.getCharset()
if not is_known_encoding( encoding ):
raise ValueError('Unknown encoding: %s' % encoding)
# "setBaseUrl"
url = httpResponse.getURL()
redirURL = httpResponse.getRedirURL()
if redirURL:
url = redirURL
self._baseUrl = url
self._baseDomain = url.getDomain()
self._rootDomain = url.getRootDomain()
self._encoding = httpResponse.getCharset()
# To store results
self._emails = []
self._re_urls = set()
def __init__(self, data, encoding=DEFAULT_ENCODING):
'''
@param data: Either a string representing a URL or a 6-elems tuple
representing the URL components:
<scheme>://<netloc>/<path>;<params>?<query>#<fragment>
Simple generic test, more detailed tests in each method!
>>> u = url_object('http://w3af.com/foo/bar.txt')
>>> u.path
'/foo/bar.txt'
>>> u.scheme
'http'
>>> u.getFileName()
'bar.txt'
>>> u.getExtension()
'txt'
>>>
#
# http is the default protocol, we can provide URLs with no proto
#
>>> u = url_object('w3af.com')
>>> u.getDomain()
'w3af.com'
>>> u.getProtocol()
'http'
#
# But we can't specify a URL without a domain!
#
>>> u = url_object('http://')
Traceback (most recent call last):
File "<stdin>", line 1, in ?
ValueError: Invalid URL "http://"
>>> u = url_object(u'http://w3af.com/foo/bar.txt')
>>> u.path
u'/foo/bar.txt'
>>> u = url_object('http://w3af.org/?foo=http://w3af.com')
>>> u.netloc
'w3af.org'
>>> u = url_object('http://w3af.org/', encoding='x-euc-jp')
Traceback (most recent call last):
File "<stdin>", line 1, in ?
ValueError: Invalid encoding "x-euc-jp" when creating URL.
'''
self._already_calculated_url = None
self._querystr = None
self._changed = True
self._encoding = encoding
if data is None:
raise ValueError('Can not build a url_object from data=None.')
# Verify that the encoding is a valid one. If we don't do it here,
# things might get crazy afterwards.
if not is_known_encoding( encoding ):
raise ValueError('Invalid encoding "%s" when creating URL.' % encoding)
if isinstance(data, tuple):
scheme, netloc, path, params, qs, fragment = data
else:
scheme, netloc, path, params, qs, fragment = \
urlparse.urlparse(data)
#
# This is the case when someone creates a url_object like
# this: url_object('www.w3af.com')
#
if scheme == netloc == '' and path:
# By default we set the protocol to "http"
scheme = 'http'
netloc = path
path = ''
self.scheme = scheme or u''
self.netloc = netloc or u''
self.path = path or u''
self.params = params or u''
self.querystring = qs or u''
self.fragment = fragment or u''
if not self.netloc:
# The URL is invalid, we don't have a netloc!
if isinstance(data, tuple):
invalid_url = urlparse.urlunparse(data)
else:
invalid_url = data
raise ValueError, 'Invalid URL "%s"' % (invalid_url,)