def test_remove_nasties(self):
nasty_string = "This string! has 9 (&#$# all \n 'sudo rm -rf /*' nasty stuff"
sanitized = [
'this', 'string', 'has', '9', 'all', 'sudo', 'rm', '-rf', 'nasty',
'stuff'
]
self.assertEqual(sanitized_list(nasty_string), sanitized)
self.assertEqual(sanitized_list(nasty_string, '[^a]'), ['a', 'a', 'a'])
def _findText(self, elem, tag_name):
words = []
if elem.nodeType == elem.TEXT_NODE:
words.extend(sanitized_list(elem.data))
elif elem.hasChildNodes():
for node in elem.childNodes:
words.extend(self._findText(node, tag_name))
return words
def _findText(self, elem, tag_name):
words=[]
if elem.nodeType == elem.TEXT_NODE:
words.extend(sanitized_list(elem.data))
elif elem.hasChildNodes():
for node in elem.childNodes:
words.extend(self._findText(node, tag_name))
return words
def test_unicode(self):
s = u'this is a unicode string'
self.assertEqual(sanitized_list(s),
['this', 'is', 'a', 'unicode', 'string'])
def test_unicode(self):
s=u'this is a unicode string'
self.assertEqual(sanitized_list(s), ['this', 'is', 'a', 'unicode', 'string'])
def test_remove_nasties(self):
nasty_string="This string! has 9 (&#$# all \n 'sudo rm -rf /*' nasty stuff"
sanitized=['this', 'string', 'has', '9', 'all', 'sudo', 'rm', '-rf', 'nasty', 'stuff']
self.assertEqual(sanitized_list(nasty_string), sanitized)
self.assertEqual(sanitized_list(nasty_string, '[^a]'), ['a', 'a', 'a'])