def test_link_density(): """Test that we get a link density""" doc = document_fromstring(load_article('ars.001.html')) for node in doc.iter('p', 'td', 'pre'): density = get_link_density(node) # the density must be between 0, 1 assert density >= 0.0 and density <= 1.0
def test_link_density(self): """Test that we get a link density""" doc = document_fromstring(load_article('ars/ars.001.html')) for node in doc.getiterator(): if node.tag in ['p', 'td', 'pre']: density = get_link_density(node) # the density must be between 0, 1 self.assertTrue(density >= 0.0 and density <= 1.0)
def test_small_doc_no_links(self): doc = Article(load_snippet('document_min.html')) assert 0 == get_link_density(doc._readable), "Still no link density"
def test_empty_node(): """An empty node doesn't have much of a link density""" doc = Article("<div></div>") assert get_link_density(doc.readable_dom) == 0.0
def test_small_doc_no_links(): doc = Article(load_snippet('document_min.html')) assert get_link_density(doc.readable_dom) == 0.0
def test_empty_node(self): """An empty node doesn't have much of a link density""" doc = Article("<div></div>") self.assertEqual(get_link_density(doc.readable_dom), 0.0)
def test_several_links(): """This doc has a 3 links with the majority of content.""" doc = Article(load_snippet('document_absolute_url.html')) assert get_link_density(doc.readable_dom) == 22 / 37
def test_several_links(self): """This doc has a 3 links with the majority of content.""" doc = Article(load_snippet('document_absolute_url.html')) self.assertAlmostEqual(get_link_density(doc.readable_dom), 22/37)
def test_empty_node(self): """An empty node doesn't have much of a link density""" empty_div = u"<div></div>" doc = Article(empty_div) assert 0 == get_link_density(doc._readable), "Link density is nadda"
def test_several_links(): """This doc has a 3 links with the majority of content.""" doc = Article(load_snippet('document_absolute_url.html')) assert get_link_density(doc.readable_dom) == 22/37
def test_several_links(self): """This doc has a 3 links with the majority of content.""" doc = Article(load_snippet('document_absolute_url.html')) self.assertAlmostEqual(get_link_density(doc.readable_dom), 22 / 37)
def test_small_doc_no_links(self): doc = Article(load_snippet('document_min.html')) self.assertEqual(get_link_density(doc.readable_dom), 0.0)