| 1 | """Tests for relevance.py — shared relevance scoring. |
| 2 | |
| 3 | Migrated from test_youtube_relevance.py + new hashtag/synonym tests. |
| 4 | """ |
| 5 | |
| 6 | import unittest |
| 7 | |
| 8 | from lib.relevance import STOPWORDS, SYNONYMS, token_overlap_relevance, tokenize |
| 9 | |
| 10 | |
| 11 | class TestTokenize(unittest.TestCase): |
| 12 | """Tests for tokenize().""" |
| 13 | |
| 14 | def test_removes_stopwords(self): |
| 15 | tokens = tokenize("how to use the AI tools") |
| 16 | self.assertNotIn("how", tokens) |
| 17 | self.assertNotIn("to", tokens) |
| 18 | self.assertNotIn("the", tokens) |
| 19 | self.assertIn("ai", tokens) |
| 20 | self.assertIn("tools", tokens) |
| 21 | |
| 22 | def test_lowercases(self): |
| 23 | tokens = tokenize("Python REACT") |
| 24 | self.assertIn("python", tokens) |
| 25 | self.assertIn("react", tokens) |
| 26 | |
| 27 | def test_strips_punctuation(self): |
| 28 | tokens = tokenize("hello, world!") |
| 29 | self.assertIn("hello", tokens) |
| 30 | self.assertIn("world", tokens) |
| 31 | |
| 32 | def test_drops_single_char(self): |
| 33 | tokens = tokenize("a b c python") |
| 34 | self.assertNotIn("a", tokens) |
| 35 | self.assertNotIn("b", tokens) |
| 36 | self.assertNotIn("c", tokens) |
| 37 | self.assertIn("python", tokens) |
| 38 | |
| 39 | def test_expands_synonyms(self): |
| 40 | tokens = tokenize("ai tools") |
| 41 | self.assertIn("artificial", tokens) |
| 42 | self.assertIn("intelligence", tokens) |
| 43 | |
| 44 | def test_expands_js_synonym(self): |
| 45 | tokens = tokenize("js framework") |
| 46 | self.assertIn("javascript", tokens) |
| 47 | |
| 48 | def test_expands_svelte(self): |
| 49 | tokens = tokenize("svelte app") |
| 50 | self.assertIn("sveltejs", tokens) |
| 51 | |
| 52 | def test_expands_vue(self): |
| 53 | tokens = tokenize("vue components") |
| 54 | self.assertIn("vuejs", tokens) |
| 55 | |
| 56 | |
| 57 | class TestTokenOverlapRelevance(unittest.TestCase): |
| 58 | """Tests for token_overlap_relevance().""" |
| 59 | |
| 60 | def test_high_relevance_exact_match(self): |
| 61 | rel = token_overlap_relevance("claude code", "Claude Code tricks and tips") |
| 62 | self.assertGreater(rel, 0.7) |
| 63 | |
| 64 | def test_low_relevance_no_match(self): |
| 65 | rel = token_overlap_relevance("claude code tips", "Best AI tools for coding") |
| 66 | self.assertLess(rel, 0.5) |
| 67 | |
| 68 | def test_empty_query_returns_neutral(self): |
| 69 | rel = token_overlap_relevance("", "Some video title") |
| 70 | self.assertEqual(rel, 0.5) |
| 71 | |
| 72 | def test_floor_at_0_1(self): |
| 73 | rel = token_overlap_relevance("quantum physics", "cat dancing video") |
| 74 | self.assertEqual(rel, 0.0) |
| 75 | |
| 76 | def test_full_match_returns_1(self): |
| 77 | rel = token_overlap_relevance("python tutorial", "Python Tutorial for Beginners") |
| 78 | self.assertEqual(rel, 1.0) |
| 79 | |
| 80 | def test_partial_match(self): |
| 81 | rel = token_overlap_relevance("react native tutorial", "React Native Guide") |
| 82 | self.assertGreater(rel, 0.3) |
| 83 | self.assertLess(rel, 1.0) |
| 84 | |
| 85 | def test_synonym_boosts_relevance(self): |
| 86 | # "js" should match "javascript" via synonym expansion |
| 87 | rel_with_syn = token_overlap_relevance("js framework", "javascript framework comparison") |
| 88 | rel_without = token_overlap_relevance("python framework", "javascript framework comparison") |
| 89 | self.assertGreater(rel_with_syn, rel_without) |
| 90 | |
| 91 | def test_stopword_only_query(self): |
| 92 | rel = token_overlap_relevance("the a is", "some content here") |
| 93 | self.assertEqual(rel, 0.5) |
| 94 | |
| 95 | def test_generic_only_overlap_stays_below_filter_threshold(self): |
| 96 | rel = token_overlap_relevance("anthropic odds", "Republican house odds update") |
| 97 | self.assertLess(rel, 0.3) |
| 98 | |
| 99 | def test_informative_partial_match_stays_above_generic_only(self): |
| 100 | generic_only = token_overlap_relevance("anthropic odds", "Republican house odds update") |
| 101 | informative = token_overlap_relevance("anthropic odds", "Anthropic valuation market") |
| 102 | self.assertGreater(informative, generic_only) |
| 103 | |
| 104 | |
| 105 | class TestHashtagRelevance(unittest.TestCase): |
| 106 | """Tests for hashtag-aware relevance (TikTok/Instagram pattern).""" |
| 107 | |
| 108 | def test_hashtag_boost(self): |
| 109 | rel_no_hash = token_overlap_relevance("claude code", "random video about stuff") |
| 110 | rel_with_hash = token_overlap_relevance( |
| 111 | "claude code", "random video about stuff", ["claudecode", "ai"] |
| 112 | ) |
| 113 | self.assertGreater(rel_with_hash, rel_no_hash) |
| 114 | |
| 115 | def test_concatenated_hashtag_splitting(self): |
| 116 | # "claudecode" should match "claude" from query via substring check |
| 117 | rel = token_overlap_relevance("claude", "video", ["claudecode"]) |
| 118 | self.assertGreater(rel, 0.5) |
| 119 | |
| 120 | def test_none_hashtags_same_as_no_hashtags(self): |
| 121 | rel1 = token_overlap_relevance("test query", "test content", None) |
| 122 | rel2 = token_overlap_relevance("test query", "test content") |
| 123 | self.assertEqual(rel1, rel2) |
| 124 | |
| 125 | if __name__ == "__main__": |
| 126 | unittest.main() |
| 127 |