2001-10-04 17:05:10 -03:00
|
|
|
import formatter
|
|
|
|
|
import unittest
|
|
|
|
|
|
2002-07-23 16:04:11 -03:00
|
|
|
from test import test_support
|
2008-06-01 18:19:14 -03:00
|
|
|
htmllib = test_support.import_module('htmllib', deprecated=True)
|
2001-10-04 17:05:10 -03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
class AnchorCollector(htmllib.HTMLParser):
|
|
|
|
|
def __init__(self, *args, **kw):
|
|
|
|
|
self.__anchors = []
|
|
|
|
|
htmllib.HTMLParser.__init__(self, *args, **kw)
|
|
|
|
|
|
|
|
|
|
def get_anchor_info(self):
|
|
|
|
|
return self.__anchors
|
|
|
|
|
|
|
|
|
|
def anchor_bgn(self, *args):
|
|
|
|
|
self.__anchors.append(args)
|
|
|
|
|
|
2003-03-30 10:25:40 -04:00
|
|
|
class DeclCollector(htmllib.HTMLParser):
|
|
|
|
|
def __init__(self, *args, **kw):
|
|
|
|
|
self.__decls = []
|
|
|
|
|
htmllib.HTMLParser.__init__(self, *args, **kw)
|
|
|
|
|
|
|
|
|
|
def get_decl_info(self):
|
|
|
|
|
return self.__decls
|
|
|
|
|
|
|
|
|
|
def unknown_decl(self, data):
|
|
|
|
|
self.__decls.append(data)
|
|
|
|
|
|
2001-10-04 17:05:10 -03:00
|
|
|
|
|
|
|
|
class HTMLParserTestCase(unittest.TestCase):
|
|
|
|
|
def test_anchor_collection(self):
|
|
|
|
|
# See SF bug #467059.
|
|
|
|
|
parser = AnchorCollector(formatter.NullFormatter(), verbose=1)
|
|
|
|
|
parser.feed(
|
|
|
|
|
"""<a href='http://foo.org/' name='splat'> </a>
|
|
|
|
|
<a href='http://www.python.org/'> </a>
|
|
|
|
|
<a name='frob'> </a>
|
|
|
|
|
""")
|
|
|
|
|
parser.close()
|
2010-11-21 09:34:58 -04:00
|
|
|
self.assertEqual(parser.get_anchor_info(),
|
|
|
|
|
[('http://foo.org/', 'splat', ''),
|
|
|
|
|
('http://www.python.org/', '', ''),
|
|
|
|
|
('', 'frob', ''),
|
|
|
|
|
])
|
2001-10-04 17:05:10 -03:00
|
|
|
|
2003-03-30 10:25:40 -04:00
|
|
|
def test_decl_collection(self):
|
|
|
|
|
# See SF patch #545300
|
|
|
|
|
parser = DeclCollector(formatter.NullFormatter(), verbose=1)
|
|
|
|
|
parser.feed(
|
|
|
|
|
"""<html>
|
|
|
|
|
<body>
|
|
|
|
|
hallo
|
|
|
|
|
<![if !supportEmptyParas]> <![endif]>
|
|
|
|
|
</body>
|
|
|
|
|
</html>
|
|
|
|
|
""")
|
|
|
|
|
parser.close()
|
2010-11-21 09:34:58 -04:00
|
|
|
self.assertEqual(parser.get_decl_info(),
|
|
|
|
|
["if !supportEmptyParas",
|
|
|
|
|
"endif"
|
|
|
|
|
])
|
2001-10-04 17:05:10 -03:00
|
|
|
|
|
|
|
|
def test_main():
|
|
|
|
|
test_support.run_unittest(HTMLParserTestCase)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
if __name__ == "__main__":
|
|
|
|
|
test_main()
|
