def testFilter0(self):
testcfg.set('filter.domain.0', '')
testcfg.set('filter.domain.1', '')
testcfg.set('filter.domain.2', '')
testcfg.set('filter.domain.3', '')
testcfg.set('filter.domain.4', '')
domain_filter.g_exdm = None # force reload
self.assertEqual(None, domain_filter.match(''))
self.assertEqual(None, domain_filter.match('http://abc'))
def preparse_filter(first_block, meta):
""" Filter by domain and magic header.
Returns distill result
"""
uri = meta.get('uri','')
dm = domain_filter.match(uri)
if dm:
return EXDOMAIN, dm
guessed = magic.guess_type(first_block)
if guessed and guessed != 'text/html' and guessed != 'text/plain':
return NON_HTML, guessed
return 0
def testFilter1(self):
domain_filter.g_exdm = None # force reload
self.assertEqual(None, domain_filter.match(''))
# exact domain match
self.assertEqual('abc.com', domain_filter.match('http://abc.com/'))
self.assertEqual(None, domain_filter.match('http://www.abc.com/'))
self.assertEqual('abc.com', domain_filter.match('http://abc.com/index.html?a=b#c'))
self.assertEqual('abc.com', domain_filter.match('http://*****:*****@abc.com/index.html?a=b#c'))
self.assertEqual('def', domain_filter.match('http://def/'))
self.assertEqual(None, domain_filter.match('http://www.def.com/'))
# suffix domain match
self.assertEqual(None, domain_filter.match('http://xyz.com/'))
self.assertEqual('.xyz.com', domain_filter.match('http://www.xyz.com/'))
self.assertEqual('.xyz.com', domain_filter.match('http://www.xyz.com/index.html?a=b#c'))
self.assertEqual('.xyz.com', domain_filter.match('http://*****:*****@www.xyz.com/index.html?a=b#c'))
