def test_parse(self):
page = WikiPage(
'Japan', 'en',
"'''Japan''' is a [[Sovereign state|sovereign]] [[island country|island nation]] in [[East Asia]]",
None)
ret = dump_db._parse(page)
eq_('page', ret[0])
eq_(b'Japan', ret[1][0])
paragraph = pickle.loads(zlib.decompress(ret[1][1]))[0][0]
eq_('Japan is a sovereign island nation in East Asia', paragraph[0])
eq_([('Sovereign state', 'sovereign', 11, 20),
('Island country', 'island nation', 21, 34),
('East Asia', 'East Asia', 38, 47)], paragraph[1])
def test_parse_redirect(self):
page = WikiPage('日本', 'en', '#REDIRECT [[Japan]]', 'Japan')
ret = dump_db._parse(page)
eq_('redirect', ret[0])
eq_('日本'.encode('utf-8'), ret[1][0])
eq_(b'Japan', ret[1][1])