| 1 | """Qualifier-only GitHub topics must not embed unbounded raw topics. |
| 2 | |
| 3 | Stream fanout calls ``search_github`` once per subquery. Before #954 the |
| 4 | error envelope used ``{topic!r}`` and the strip log used the full core, so |
| 5 | log volume, error-report volume, and doctor-hint size scaled with both |
| 6 | fanout and topic length. |
| 7 | """ |
| 8 | |
| 9 | from unittest.mock import patch |
| 10 | |
| 11 | from lib import github |
| 12 | |
| 13 | _FROM = "2026-07-01" |
| 14 | _TO = "2026-07-31" |
| 15 | # Wrapper text plus a ~120-char topic slice. The pre-fix error embedded |
| 16 | # the entire raw topic (thousands of chars on a pathological planner query). |
| 17 | _ERROR_BOUND = 300 |
| 18 | _STRIP_LOG_BOUND = 400 |
| 19 | |
| 20 | |
| 21 | def _long_qualifier_only_topic() -> str: |
| 22 | return "created:>2025-03-20 " + ("stars:>1 " * 400) |
| 23 | |
| 24 | |
| 25 | def test_qualifier_only_topic_returns_clean_empty_envelope_without_network(): |
| 26 | """#953 supersedes the #954 error bound: a qualifier-only topic is a clean |
| 27 | no-results envelope (no error key), so there is nothing to spam.""" |
| 28 | topic = _long_qualifier_only_topic() |
| 29 | with patch.object(github, "_resolve_token", return_value="t"), patch.object( |
| 30 | github, "_fetch_json" |
| 31 | ) as fetch: |
| 32 | result = github.search_github(topic, _FROM, _TO) |
| 33 | fetch.assert_not_called() |
| 34 | assert result["items"] == [] |
| 35 | assert "error" not in result |
| 36 | |
| 37 | |
| 38 | def test_short_qualifier_only_topic_is_also_clean_no_results(): |
| 39 | topic = "created:>2025-03-20" |
| 40 | with patch.object(github, "_resolve_token", return_value="t"), patch.object( |
| 41 | github, "_fetch_json" |
| 42 | ) as fetch: |
| 43 | result = github.search_github(topic, _FROM, _TO) |
| 44 | fetch.assert_not_called() |
| 45 | assert result["items"] == [] |
| 46 | assert "error" not in result |
| 47 | |
| 48 | |
| 49 | def test_qualifier_only_logs_are_bounded_across_fanout(): |
| 50 | topic = _long_qualifier_only_topic() |
| 51 | logs: list[str] = [] |
| 52 | with patch.object(github, "_resolve_token", return_value="t"), patch.object( |
| 53 | github, "_fetch_json" |
| 54 | ) as fetch, patch.object(github, "_log", side_effect=lambda m: logs.append(m)): |
| 55 | for _ in range(5): |
| 56 | github.search_github(topic, _FROM, _TO) |
| 57 | fetch.assert_not_called() |
| 58 | assert logs |
| 59 | for msg in logs: |
| 60 | assert len(msg) < _ERROR_BOUND |
| 61 | assert topic not in msg |
| 62 | |
| 63 | |
| 64 | def test_mixed_topic_strip_log_is_bounded(): |
| 65 | subject = "open source ai " * 200 |
| 66 | topic = subject + "stars:>1000 created:>2025-03-20" |
| 67 | logs: list[str] = [] |
| 68 | with patch.object(github, "_resolve_token", return_value="t"), patch.object( |
| 69 | github, "_fetch_json", return_value={"items": []} |
| 70 | ), patch.object(github, "_log", side_effect=lambda m: logs.append(m)): |
| 71 | github.search_github(topic, _FROM, _TO) |
| 72 | strip_logs = [m for m in logs if m.startswith("Stripped search qualifiers:")] |
| 73 | assert strip_logs |
| 74 | for msg in strip_logs: |
| 75 | assert len(msg) < _STRIP_LOG_BOUND |
| 76 | assert subject not in msg |
| 77 |