def test_term_frequency(self):
self.__ranker.reset(21)
self.__ranker.update("foo", 2, in3120.Posting(21, 4))
self.__ranker.update("bar", 1, in3120.Posting(21, 3))
self.assertEqual(self.__ranker.evaluate(), 11)
self.__ranker.reset(42)
self.__ranker.update("foo", 1, in3120.Posting(42, 1))
self.__ranker.update("baz", 2, in3120.Posting(42, 2))
self.assertEqual(self.__ranker.evaluate(), 5)
def test_static_quality_score(self):
self.__ranker.reset(0)
self.__ranker.update("foo", 1, in3120.Posting(0, 1))
score1 = self.__ranker.evaluate()
self.__ranker.reset(1)
self.__ranker.update("foo", 1, in3120.Posting(1, 1))
score2 = self.__ranker.evaluate()
self.assertGreater(score1, 0.0)
self.assertGreater(score2, 0.0)
self.assertGreater(score1, score2)
def test_term_frequency(self):
self.__ranker.reset(1)
self.__ranker.update("foo", 1, in3120.Posting(1, 1))
score1 = self.__ranker.evaluate()
self.__ranker.reset(2)
self.__ranker.update("foo", 1, in3120.Posting(1, 2))
score2 = self.__ranker.evaluate()
self.assertGreater(score1, 0.0)
self.assertGreater(score2, 0.0)
self.assertGreater(score2, score1)
def test_inverse_document_frequency(self):
self.__ranker.reset(3)
self.__ranker.update("the", 1, in3120.Posting(3, 1))
self.assertAlmostEqual(self.__ranker.evaluate(), 0.0, 8)
self.__ranker.reset(3)
self.__ranker.update("bar", 1, in3120.Posting(3, 1))
score1 = self.__ranker.evaluate()
self.__ranker.reset(5)
self.__ranker.update("baz", 1, in3120.Posting(5, 1))
score2 = self.__ranker.evaluate()
self.assertGreater(score1, 0.0)
self.assertGreater(score2, 0.0)
self.assertGreater(score1, score2)
def test_order_independence(self):
postings1 = [
in3120.Posting(1, 0),
in3120.Posting(2, 0),
in3120.Posting(3, 0)
]
postings2 = [
in3120.Posting(2, 0),
in3120.Posting(3, 0),
in3120.Posting(6, 0)
]
result12 = list(
map(lambda p: p.document_id,
self.__merger.intersection(iter(postings1), iter(postings2))))
result21 = list(
map(lambda p: p.document_id,
self.__merger.intersection(iter(postings2), iter(postings1))))
self.assertListEqual(result12, [2, 3])
self.assertListEqual(result12, result21)
result12 = list(
map(lambda p: p.document_id,
self.__merger.union(iter(postings1), iter(postings2))))
result21 = list(
map(lambda p: p.document_id,
self.__merger.union(iter(postings2), iter(postings1))))
self.assertListEqual(result12, [1, 2, 3, 6])
self.assertListEqual(result12, result21)
def test_empty_lists(self):
posting = in3120.Posting(123, 4)
self.assertListEqual(
list(self.__merger.intersection(iter([]), iter([]))), [])
self.assertListEqual(
list(self.__merger.intersection(iter([]), iter([posting]))), [])
self.assertListEqual(
list(self.__merger.intersection(iter([posting]), iter([]))), [])
self.assertListEqual(list(self.__merger.union(iter([]), iter([]))), [])
self.assertListEqual([
p.document_id
for p in self.__merger.union(iter([]), iter([posting]))
], [posting.document_id])
self.assertListEqual([
p.document_id
for p in self.__merger.union(iter([posting]), iter([]))
], [posting.document_id])
def test_uses_yield(self):
import types
postings1 = [
in3120.Posting(1, 0),
in3120.Posting(2, 0),
in3120.Posting(3, 0)
]
postings2 = [
in3120.Posting(2, 0),
in3120.Posting(3, 0),
in3120.Posting(6, 0)
]
result1 = self.__merger.intersection(iter(postings1), iter(postings2))
result2 = self.__merger.union(iter(postings1), iter(postings2))
self.assertIsInstance(result1, types.GeneratorType,
"Are you using yield?")
self.assertIsInstance(result2, types.GeneratorType,
"Are you using yield?")