| 1 | """Tests for YouTube relevance scoring.""" |
| 2 | |
| 3 | import unittest |
| 4 | |
| 5 | # Add lib to path |
| 6 | |
| 7 | from lib.relevance import token_overlap_relevance as _compute_relevance, tokenize as _tokenize |
| 8 | |
| 9 | |
| 10 | class TestTokenize(unittest.TestCase): |
| 11 | def test_basic(self): |
| 12 | tokens = _tokenize("Claude Code Tutorial") |
| 13 | self.assertIn("claude", tokens) |
| 14 | self.assertIn("code", tokens) |
| 15 | self.assertIn("tutorial", tokens) |
| 16 | |
| 17 | def test_removes_stopwords(self): |
| 18 | tokens = _tokenize("how to use Claude Code for beginners") |
| 19 | self.assertNotIn("how", tokens) |
| 20 | self.assertNotIn("to", tokens) |
| 21 | self.assertNotIn("for", tokens) |
| 22 | self.assertIn("claude", tokens) |
| 23 | self.assertIn("code", tokens) |
| 24 | self.assertIn("beginners", tokens) |
| 25 | |
| 26 | def test_strips_punctuation(self): |
| 27 | tokens = _tokenize("What's new in Claude?") |
| 28 | self.assertIn("claude", tokens) |
| 29 | self.assertNotIn("what's", tokens) |
| 30 | |
| 31 | def test_drops_single_char(self): |
| 32 | tokens = _tokenize("a b c Claude") |
| 33 | self.assertNotIn("a", tokens) |
| 34 | self.assertNotIn("b", tokens) |
| 35 | self.assertIn("claude", tokens) |
| 36 | |
| 37 | def test_empty(self): |
| 38 | tokens = _tokenize("") |
| 39 | self.assertEqual(tokens, set()) |
| 40 | |
| 41 | def test_all_stopwords(self): |
| 42 | tokens = _tokenize("the a an to for how") |
| 43 | self.assertEqual(tokens, set()) |
| 44 | |
| 45 | |
| 46 | class TestComputeRelevance(unittest.TestCase): |
| 47 | def test_exact_match(self): |
| 48 | result = _compute_relevance("Claude Code", "Claude Code") |
| 49 | self.assertEqual(result, 1.0) |
| 50 | |
| 51 | def test_full_match_in_longer_title(self): |
| 52 | result = _compute_relevance("Claude Code", "Claude Code Tutorial") |
| 53 | self.assertEqual(result, 1.0) |
| 54 | |
| 55 | def test_partial_match(self): |
| 56 | result = _compute_relevance("Claude Code Tips", "Claude Tips for Beginners") |
| 57 | self.assertGreater(result, 0.5) |
| 58 | self.assertLess(result, 1.0) |
| 59 | |
| 60 | def test_no_match(self): |
| 61 | result = _compute_relevance("Claude Code", "Python Web Scraping") |
| 62 | self.assertEqual(result, 0.0) |
| 63 | |
| 64 | def test_empty_query_returns_neutral(self): |
| 65 | result = _compute_relevance("", "Some Video Title") |
| 66 | self.assertEqual(result, 0.5) |
| 67 | |
| 68 | def test_stopword_only_query(self): |
| 69 | result = _compute_relevance("how to the", "Some Video Title") |
| 70 | self.assertEqual(result, 0.5) |
| 71 | |
| 72 | def test_empty_title(self): |
| 73 | result = _compute_relevance("Claude Code", "") |
| 74 | self.assertEqual(result, 0.0) |
| 75 | |
| 76 | def test_case_insensitive(self): |
| 77 | result = _compute_relevance("claude code", "CLAUDE CODE Tutorial") |
| 78 | self.assertEqual(result, 1.0) |
| 79 | |
| 80 | def test_stopwords_in_title_dont_inflate(self): |
| 81 | # "Claude Code" query against a title with lots of stopwords |
| 82 | # Should still match well since Claude and Code are present |
| 83 | result = _compute_relevance( |
| 84 | "Claude Code", |
| 85 | "How To Use Claude Code For Complete Beginners" |
| 86 | ) |
| 87 | self.assertEqual(result, 1.0) |
| 88 | |
| 89 | def test_no_match_returns_zero(self): |
| 90 | result = _compute_relevance("quantum computing", "cat videos compilation") |
| 91 | self.assertEqual(result, 0.0) |
| 92 | |
| 93 | def test_cap_at_1_0(self): |
| 94 | result = _compute_relevance("AI", "AI AI AI AI AI") |
| 95 | self.assertLessEqual(result, 1.0) |
| 96 | |
| 97 | def test_single_word_query(self): |
| 98 | result = _compute_relevance("Seedance", "Seedance AI Video Generator Review") |
| 99 | self.assertEqual(result, 1.0) |
| 100 | |
| 101 | def test_single_word_no_match(self): |
| 102 | result = _compute_relevance("Seedance", "Random cooking video") |
| 103 | self.assertEqual(result, 0.0) |
| 104 | |
| 105 | if __name__ == "__main__": |
| 106 | unittest.main() |
| 107 |