feat(hands,jobs): длящаяся беда не пропадает - суточная сводка и всплеск против своей нормы

tests/test_komodo_logs.py

@@ -8,7 +8,9 @@ from beaver_agent.hands.komodo_logs import (
     StackScan,
     Sweep,
     brief,
+    daily,
     digest,
+    grouped,
     read_stack,
 )
 from beaver_agent.jobs.logs import sweep as sweep_fleet
@@ -467,3 +469,191 @@ def test_a_bare_blob_stays_readable():
     )
     (hit,) = scan.hits
     assert "kJ8sdfLKJ23kjhsdf98234x" in hit.sample
+
+
+def stamps(day, hour, count, start=0):
+    return [
+        f"2026-09-{day:02d}T{hour:02d}:{m:02d}:00.0"
+        for m in range(start, start + count)
+    ]
+
+
+def burst(stack, times, text="ERROR всё та же беда", service="svc"):
+    """Хвост стека, где одна и та же беда повторилась в перечисленные минуты."""
+    return read_stack(stack, log(*(line(service, t, text) for t in times)))
+
+
+def at(day, hour, minute=0):
+    return datetime(2026, 9, day, hour, minute, tzinfo=UTC)
+
+
+def told(state, report):
+    state.mark_told(report)
+    return report
+
+
+def test_a_continuing_trouble_is_told_once_and_then_keeps_quiet(tmp_path):
+    state = seen(tmp_path)
+    first = told(
+        state, digest(state, swept([burst("gitea", stamps(20, 9, 3))], at=at(20, 10)))
+    )
+    assert len(first.new) == 1
+
+    # Хвост перечитан: те же строки плюс свежие - беда идёт, но новостью не стала.
+    later = digest(
+        state,
+        swept([burst("gitea", stamps(20, 9, 3) + stamps(20, 12, 4))], at=at(20, 13)),
+    )
+    assert later.new == () and later.surge == () and later.quiet == ()
+    assert not later.worth_telling
+    assert later.ongoing == ()  # за прошлые сутки её ещё не было
+
+
+def test_the_reread_tail_does_not_inflate_the_count(tmp_path):
+    state = seen(tmp_path)
+    told(state, digest(state, swept([burst("gitea", stamps(20, 9, 3))], at=at(20, 10))))
+    for hour in (13, 16, 19):
+        digest(state, swept([burst("gitea", stamps(20, 9, 3))], at=at(20, hour)))
+    (entry,) = state.entries.values()
+    assert entry.count == 3  # три повтора, а не три на каждый из четырёх проходов
+    assert entry.passes == 1
+
+
+def test_a_continuing_trouble_comes_back_in_the_daily_summary(tmp_path):
+    state = seen(tmp_path)
+    told(state, digest(state, swept([burst("gitea", stamps(20, 9, 5))], at=at(20, 10))))
+
+    # Первый проход следующих суток - 01:40, он же несёт сводку.
+    night = digest(
+        state,
+        swept([burst("gitea", stamps(20, 9, 5) + stamps(21, 1, 2))], at=at(21, 1, 40)),
+    )
+    assert night.summary and len(night.ongoing) == 1
+    assert night.new == () and not night.worth_telling
+    text = daily(night, UTC)
+    assert "длящееся, сводка за 2026-09-21" in text
+    assert "gitea/svc ошибка, с 20.09 09:00, за сутки 5, первые сутки" in text
+    assert "ERROR" not in text  # образцов строк в сводке нет намеренно
+    told(state, night)
+    state.mark_daily(night)
+
+    # Тот же день, следующий проход: сводку уже отдали.
+    again = digest(
+        state,
+        swept([burst("gitea", stamps(20, 9, 5) + stamps(21, 1, 2))], at=at(21, 4, 40)),
+    )
+    assert again.ongoing and not again.summary
+
+
+def test_the_summary_says_which_way_it_is_going(tmp_path):
+    state = seen(tmp_path)
+    told(state, digest(state, swept([burst("gitea", stamps(20, 9, 2))], at=at(20, 10))))
+    day_two = swept(
+        [burst("gitea", stamps(20, 9, 2) + stamps(21, 9, 8))], at=at(21, 10)
+    )
+    told(state, digest(state, day_two))
+    third = digest(
+        state,
+        swept(
+            [burst("gitea", stamps(20, 9, 2) + stamps(21, 9, 8) + stamps(22, 1, 1))],
+            at=at(22, 1, 40),
+        ),
+    )
+    assert "за сутки 8, растёт (было 2)" in daily(third, UTC)
+
+
+def test_a_surge_of_the_familiar_breaks_the_silence(tmp_path):
+    state = seen(tmp_path)
+    state_report = digest(
+        state, swept([burst("gitea", stamps(20, 9, 1))], at=at(20, 10))
+    )
+    told(state, state_report)
+    for hour, minute in ((13, 12), (16, 15)):
+        digest(
+            state,
+            swept([burst("gitea", stamps(20, hour - 1, 1, minute))], at=at(20, hour)),
+        )
+    (entry,) = state.entries.values()
+    assert entry.passes == 3 and entry.norm == 1.0
+
+    spike = digest(state, swept([burst("gitea", stamps(20, 18, 12))], at=at(20, 19)))
+    (item,) = spike.surge
+    assert item.gain == 12 and item.norm == 1.0
+    assert spike.new == () and spike.worth_telling
+    text = brief(spike)
+    assert "Всплеск знакомого (1)" in text
+    assert "×12 за проход при норме ×1.0" in text
+    assert "Нового в этой области нет" not in text
+
+    # Второй такой же проход внутри выдержки молчит: это уже не новость.
+    told(state, spike)
+    more = digest(state, swept([burst("gitea", stamps(20, 21, 12))], at=at(20, 22)))
+    assert more.surge == () and not more.worth_telling
+
+
+def test_a_small_jump_is_not_a_surge(tmp_path):
+    state = seen(tmp_path)
+    told(state, digest(state, swept([burst("gitea", stamps(20, 9, 1))], at=at(20, 10))))
+    for hour in (13, 16):
+        digest(
+            state,
+            swept([burst("gitea", stamps(20, hour - 1, 1, hour))], at=at(20, hour)),
+        )
+    # Вчетверо больше, но всего четыре штуки: пол в SURGE_MIN держит тишину.
+    small = digest(state, swept([burst("gitea", stamps(20, 18, 4))], at=at(20, 19)))
+    assert small.surge == () and not small.worth_telling
+
+
+def test_a_trouble_gone_silent_is_reported_even_while_its_line_lingers(tmp_path):
+    state = seen(tmp_path)
+    tail = burst("gitea", stamps(20, 9, 2))
+    told(state, digest(state, swept([tail], at=at(20, 10))))
+
+    # Сутки та же строка лежит в хвосте, новых событий нет.
+    assert not digest(state, swept([tail], at=at(20, 22))).quiet
+    stopped = digest(state, swept([tail], at=at(21, 12)))
+    (gone,) = stopped.quiet
+    assert gone.last == at(20, 9, 1)
+    assert "последнее в 20.09 09:01" in brief(stopped)
+
+    told(state, stopped)
+    assert digest(state, swept([tail], at=at(21, 16))).quiet == ()
+
+
+def test_a_trouble_that_returns_after_the_all_clear_is_new_again(tmp_path):
+    state = seen(tmp_path)
+    tail = burst("gitea", stamps(20, 9, 2))
+    told(state, digest(state, swept([tail], at=at(20, 10))))
+    told(state, digest(state, swept([tail], at=at(21, 12))))  # стихла
+
+    back = digest(
+        state,
+        swept([burst("gitea", stamps(20, 9, 2) + stamps(21, 18, 3))], at=at(21, 19)),
+    )
+    assert len(back.new) == 1 and back.worth_telling
+
+
+def test_the_background_of_the_baseline_never_claims_to_have_stopped(tmp_path):
+    # Про фон, принятый в базовую линию, мастер не слышал - «стихло» про него врёт.
+    state = seen(tmp_path, known=())
+    state.fresh = False
+    cold = digest(state, swept([burst("lodge", stamps(20, 9, 2))], at=at(20, 10)))
+    assert cold.new == () and cold.baseline == 1
+    gone = digest(state, swept([read_stack("lodge", log())], at=at(20, 13)))
+    assert gone.quiet == () and not gone.worth_telling
+
+
+def test_the_summary_folds_the_variants_of_one_service_into_one_row(tmp_path):
+    state = seen(tmp_path)
+    # Разные по смыслу строки: отличие в одних цифрах - это одна и та же беда.
+    variants = [
+        burst("gitea", stamps(20, 9, 4), text=f"ERROR {what}")
+        for what in ("push отклонён", "миграция не прошла", "ssh-ключ отвергнут")
+    ]
+    told(state, digest(state, swept(variants, at=at(20, 10))))
+    night = digest(state, swept(variants, at=at(21, 1, 40)))
+    (row,) = grouped(night.ongoing)
+    assert row.kinds == 3 and row.day_before == 12
+    text = daily(night, UTC)
+    assert "Идёт до сих пор: 1 рядов (3 разных строк)" in text
+    assert "gitea/svc ошибка, строк 3 видов, с 20.09 09:00, за сутки 12" in text