def test_extract_title_removes_unwanted_characters():
html = """
<meta property="og:title" content="Trump Denies Charitable Donation He Promised If Elizabeth Warren Releases DNA Results And It’s On Video" />
"""
expected = "Trump Denies Charitable Donation He Promised If Elizabeth Warren Releases DNA Results And It’s On Video"
assert extract_title(html) == expected
def test_extract_title_pick_shortest_version_of_equivalent_title():
html = """
<h1 class="entry-title">Pamela Geller in Breitbart News: Dueling Billboards from CAIR, AFDI in Times Square</h1>
<meta property="og:title" content="Pamela Geller in Breitbart News: Dueling Billboards from CAIR, AFDI in Times Square - Geller Report" />
"""
expected = "Pamela Geller in Breitbart News: Dueling Billboards from CAIR, AFDI in Times Square"
assert extract_title(html) == expected
def test_extract_title_prioritises_highest_score_xpath():
html = """
<h2 class="title">Silly title</h2>
<h1 class="entry-title">Example title</h1>
<header><h1>Bad title</h1></header>
<p>Hello world</p>
"""
expected = "Example title"
assert extract_title(html) == expected
def test_extract_title_gets_text_within_hyperlinks():
html = """
<h1 class="entry-title">
<a href="http://addictinginfo.com/2018/10/15/trump-denies-charitable-donation-he-promised-if-elizabeth-warren-releases-dna-results-and-its-on-video/"
title="Permalink to Trump Denies Charitable Donation He Promised If Elizabeth Warren Releases DNA Results And It’s On Video"
rel="bookmark">Trump Denies Charitable Donation He Promised If Elizabeth Warren Releases DNA Results And
It’s On Video</a>
</h1>
"""
expected = "Trump Denies Charitable Donation He Promised If Elizabeth Warren Releases DNA Results And It’s On Video"
assert extract_title(html) == expected
def test_extract_title(html, expected):
assert extract_title(html) == expected