| 1 | import unittest |
| 2 | |
| 3 | from lib import relevance |
| 4 | |
| 5 | |
| 6 | class RelevanceCoreV3Tests(unittest.TestCase): |
| 7 | def test_tokenize_removes_stopwords_and_expands_synonyms(self): |
| 8 | tokens = relevance.tokenize("How to use JS for hip hop apps") |
| 9 | self.assertIn("js", tokens) |
| 10 | self.assertIn("javascript", tokens) |
| 11 | self.assertIn("hiphop", tokens) |
| 12 | self.assertNotIn("how", tokens) |
| 13 | |
| 14 | def test_token_overlap_relevance_returns_neutral_for_stopword_only_query(self): |
| 15 | self.assertEqual(0.5, relevance.token_overlap_relevance("how to", "anything at all")) |
| 16 | |
| 17 | def test_token_overlap_relevance_returns_zero_for_no_overlap(self): |
| 18 | self.assertEqual(0.0, relevance.token_overlap_relevance("openclaw", "corsair gaming mouse")) |
| 19 | |
| 20 | def test_token_overlap_relevance_rewards_exact_phrase_matches(self): |
| 21 | phrase_score = relevance.token_overlap_relevance( |
| 22 | "openclaw nanoclaw", |
| 23 | "A detailed openclaw nanoclaw comparison for agents.", |
| 24 | ) |
| 25 | partial_score = relevance.token_overlap_relevance( |
| 26 | "openclaw nanoclaw", |
| 27 | "A detailed openclaw comparison for agents.", |
| 28 | ) |
| 29 | self.assertGreater(phrase_score, partial_score) |
| 30 | |
| 31 | def test_token_overlap_relevance_caps_generic_only_matches(self): |
| 32 | score = relevance.token_overlap_relevance( |
| 33 | "anthropic odds", |
| 34 | "Latest odds and prediction updates for markets", |
| 35 | ) |
| 36 | self.assertLessEqual(score, 0.24) |
| 37 | |
| 38 | def test_token_overlap_relevance_splits_concatenated_hashtags(self): |
| 39 | score = relevance.token_overlap_relevance( |
| 40 | "claude code", |
| 41 | "Agent workflow discussion", |
| 42 | hashtags=["ClaudeCode", "BuildInPublic"], |
| 43 | ) |
| 44 | self.assertGreater(score, 0.0) |
| 45 | |
| 46 | if __name__ == "__main__": |
| 47 | unittest.main() |
| 48 |