def test_migrate_html(self):
# chunks: foo bar baz
# element: ___ ba_ ___
queue = self.reset_queue()
elements = [budou.Element('ba', 'a', '<a href="#">ba</a>', 3)]
expected = ['foo', '<a href="#">ba</a>r', 'baz']
result = self.parser._migrate_html(queue, elements)
self.assertEqual(expected, [chunk.word for chunk in result.chunks])
# chunks: foo bar baz
# element: ___ bar b__
queue = self.reset_queue()
elements = [budou.Element('barb', 'a', '<a href="#">barb</a>', 3)]
expected = ['foo', '<a href="#">barb</a>az']
result = self.parser._migrate_html(queue, elements)
self.assertEqual(expected, [chunk.word for chunk in result.chunks])
def test_get_elements_list(self):
source = u'<a>こちら</a>をクリック'
dom = html.fragment_fromstring(source, create_parent='body')
expected = [budou.Element(u'こちら', 'a', u'<a>こちら</a>', 0)]
result = self.parser._get_elements_list(dom)
self.assertEqual(
result, expected,
'The input DOM should be processed to an element list.')
def test_get_elements_list(self):
dom = html.fragment_fromstring('click <a>this</a>',
create_parent='body')
expected = [budou.Element('this', 'a', '<a>this</a>', 6)]
result = self.parser._get_elements_list(dom)
self.assertEqual(
result, expected,
'The input DOM should be processed to an element list.')