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 make_fake_html_page(url):
html = dedent(u"""\
<html><head><meta name="api-version" value="2" /></head>
<body>
<a href="/abc-1.0.tar.gz#md5=000000000">abc-1.0.tar.gz</a>
</body></html>
""")
content = html.encode('utf-8')
headers = {}
return HTMLPage(content, url=url, headers=headers)