RandomSec/lib/jython-2.5.1/test/test_htmllib.py
Vishal Talwar d0df704d8a added python code part of jython distribution in lib/jython-2.5.1
added python.path vm arg to startup script
fixed infinite loop in unwrap() when displaying sequences of sequences



git-svn-id: http://google-refine.googlecode.com/svn/trunk@509 7d457c2a-affb-35e4-300a-418c747d4874
2010-04-20 18:50:24 +00:00

70 lines
1.9 KiB
Python

import formatter
import htmllib
import unittest
from test import test_support
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)
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)
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()
self.assertEquals(parser.get_anchor_info(),
[('http://foo.org/', 'splat', ''),
('http://www.python.org/', '', ''),
('', 'frob', ''),
])
def test_decl_collection(self):
# See SF patch #545300
parser = DeclCollector(formatter.NullFormatter(), verbose=1)
parser.feed(
"""<html>
<body>
hallo
<![if !supportEmptyParas]>&nbsp;<![endif]>
</body>
</html>
""")
parser.close()
self.assertEquals(parser.get_decl_info(),
["if !supportEmptyParas",
"endif"
])
def test_main():
test_support.run_unittest(HTMLParserTestCase)
if __name__ == "__main__":
test_main()