def _pre_parse(self, http_resp):
'''
@parameter http_resp: The HTTP response document that contains the
HTML document inside its body.
'''
SGMLParser._pre_parse(self, http_resp)
assert self._baseUrl, 'The base URL must be set.'
def _pre_parse(self, httpResponse):
'''
@parameter httpResponse: The HTTP response document that contains the WML
document inside its body.
Init,
>>> from core.data.url.httpResponse import httpResponse as httpResponse
>>> u = url_object('http://www.w3af.com/')
Parse a simple form,
>>> form = """
... <go method="post" href="dataReceptor.php">
... <postfield name="clave" value="$(clave)"/>
... <postfield name="cuenta" value="$(cuenta)"/>
... <postfield name="tipdat" value="D"/>
... </go>"""
>>> response = httpResponse( 200, form, {}, u, u )
>>> w = wmlParser(response)
>>> w.getForms()
[Form({'clave': ['$(clave)'], 'cuenta': ['$(cuenta)'], 'tipdat': ['D']})]
Get the simplest link
>>> response = httpResponse( 200, '<a href="/index.aspx">ASP.NET</a>', {}, u, u )
>>> w = wmlParser( response )
>>> re, parsed = w.getReferences()
#
# TODO:
# I don't really understand why I'm getting results @ the "re".
# They should really be inside the "parsed" list.
#
# >>> re
# []
# >>> parsed[0].url_string
# u'http://www.w3af.com/index.aspx'
Get a link by applying regular expressions
>>> response = httpResponse(200, 'header /index.aspx footer', {}, u, u)
>>> w = wmlParser( response )
>>> re, parsed = w.getReferences()
>>> #
>>> # TODO: Shouldn't this be the other way around?!
>>> #
>>> re
[]
>>> parsed[0].url_string
u'http://www.w3af.com/index.aspx'
'''
SGMLParser._pre_parse(self, httpResponse)
assert self._baseUrl is not None, 'The base URL must be set.'
