def test_iter_links__yanked_reason(self, anchor_html, expected):
html = (
# Mark this as a unicode string for Python 2 since anchor_html
# can contain non-ascii.
u'<html><head><meta charset="utf-8"><head>'
'<body>{}</body></html>').format(anchor_html)
html_bytes = html.encode('utf-8')
page = HTMLPage(html_bytes, url='https://example.com/simple/')
links = list(page.iter_links())
link, = links
actual = link.yanked_reason
assert actual == expected
def test_base_url(html, url, expected):
assert HTMLPage(html, url).base_url == expected