| 1 | import unittest |
| 2 | |
| 3 | import scripts.sanitize_agent_output as sanitize_agent_output |
| 4 | |
| 5 | |
| 6 | class SanitizeAgentOutputTests(unittest.TestCase): |
| 7 | def test_sanitize_text_removes_leaked_agent_summary_lines(self) -> None: |
| 8 | original = """--- |
| 9 | title: \"Week 21, 2026 Analysis\" |
| 10 | quality_score: 76 |
| 11 | --- |
| 12 | |
| 13 | ## This Week's Trends |
| 14 | |
| 15 | Real article paragraph. |
| 16 | |
| 17 | ✅ Farnsworth is done. data/analyzed/2026-W21-summary.md is written (88 lines). |
| 18 | Editorial thesis: Genuine agentic coding infrastructure is breaking out. |
| 19 | Quality score: 76/100 — publishable. |
| 20 | """ |
| 21 | |
| 22 | sanitized = sanitize_agent_output.sanitize_text(original) |
| 23 | |
| 24 | self.assertNotIn("✅ Farnsworth is done.", sanitized) |
| 25 | self.assertNotIn("Editorial thesis:", sanitized) |
| 26 | self.assertNotIn("Quality score: 76/100", sanitized) |
| 27 | self.assertIn("Real article paragraph.", sanitized) |
| 28 | self.assertIn("quality_score: 76", sanitized) |
| 29 | |
| 30 | def test_sanitize_text_preserves_legitimate_article_content(self) -> None: |
| 31 | original = """--- |
| 32 | title: \"Week 21, 2026 Analysis\" |
| 33 | quality_score: 76 |
| 34 | summary: \"Quality score caveats belong only in frontmatter.\" |
| 35 | --- |
| 36 | |
| 37 | ## Signal & Noise |
| 38 | |
| 39 | The article discusses quality score inflation as a concept without using agent metadata. |
| 40 | """ |
| 41 | |
| 42 | sanitized = sanitize_agent_output.sanitize_text(original) |
| 43 | |
| 44 | self.assertEqual(sanitized, original) |
| 45 | |
| 46 | |
| 47 | if __name__ == "__main__": |
| 48 | unittest.main() |