def test_calculate_links_length_and_clean_chunk(self):
chunk = parser.Chunk(
'<p><a href="#">asd<a href="#">asd</a>asd</a>asd</p>')
chunk._calculate_length_with_tags()
chunk._calculate_links_length_and_clean_chunk(self.cleaner)
self.assertEqual(chunk.chunk, 'asd' * 4)
self.assertEqual(chunk.length_of_links, 9)
def test_make_calculations(self):
chunk = parser.Chunk(self.chunk)
chunk.make_calculations(self.cleaner, self.punctuation)
text_density = len(self.useful_text) / len(self.chunk)
count_of_punctuation_marks = sum(
(self.chunk.count(mark) for mark in self.punctuation))
weight = (text_density + count_of_punctuation_marks / 100) + 1
weight += (1 - count_of_punctuation_marks / len(self.useful_text))
self.assertEqual(chunk.weight, weight)
def test_calculate_count_of_punctuation_marks(self):
count_of_punctuation_marks = sum(
(self.chunk.count(mark) for mark in self.punctuation))
chunk = parser.Chunk(self.chunk)
self.assertRaises(parser.ChunkProcedureException,
chunk._calculate_count_of_punctuation_marks,
self.punctuation)
self.assertRaises(parser.ChunkProcedureException,
chunk._calculate_links_length_and_clean_chunk,
self.cleaner)
chunk._calculate_length_with_tags()
chunk._calculate_links_length_and_clean_chunk(self.cleaner)
chunk._calculate_count_of_punctuation_marks(self.punctuation)
self.assertEqual(chunk.count_of_punctuation_marks,
count_of_punctuation_marks)
def test_calculate_length_without_tags(self):
chunk = parser.Chunk(self.chunk)
self.assertRaises(parser.ChunkProcedureException,
chunk._calculate_length_without_tags)
chunk._calculate_length_with_tags()
self.assertFalse(chunk._cleaned)
self.assertRaises(parser.ChunkProcedureException,
chunk._calculate_length_without_tags)
chunk._calculate_links_length_and_clean_chunk(self.cleaner)
self.assertEqual(chunk._links_length, 0)
chunk._calculate_length_without_tags()
self.assertEqual(chunk.length_without_tags, len(self.useful_text))
def test_calculate_length_with_tags(self):
chunk = parser.Chunk(self.chunk)
chunk._calculate_length_with_tags()
self.assertEqual(chunk.length_with_tags, len(self.chunk))