def test_si_sample(self):
"""Using the si sample, load article with only opening body element"""
sample = load_sample('si-game.sample.html')
doc = Document(
sample,
url='http://sportsillustrated.cnn.com/baseball/mlb/gameflash/2012/04/16/40630_preview.html')
res = doc.summary()
self.assertEqual('<html><body><div><div class', res[0:27])
def test_si_sample_html_partial(self):
"""Using the si sample, make sure we can get the article alone."""
sample = load_sample('si-game.sample.html')
doc = Document(sample, url='http://sportsillustrated.cnn.com/baseball/mlb/gameflash/2012/04/16/40630_preview.html')
res = doc.summary(enclose_with_html_tag=False)
self.assertEqual('<div><div class="', res[0:17])
def process_article(article):
sample = load_sample(article)
doc = Document(sample)
res = doc.summary()
failed_msg = "Failed to process the article: " + article
assert '<html><body><div><div class' == res[0:27], failed_msg