fix(generator): усилены проверки startup-history

- Зачем:
  - коммит-гейт не запускал корневые контрактные тесты, а часть подтверждённых обходов могла снова смешать разные миры генератора.
- Что:
  - добавлены цели make test, make lint и contract-test с тихим pytest-выводом через Docker.
  - закрыты обходы через generator-reset, неизвестную версию state и fail-open проверку DM-витрин.
  - усилены поведенческие контракты CHECK_LIVE_SEAM, профиля manifest и pause-check etl_pipeline; обновлены документы и issue 19.
- Проверка:
  - make test; make lint; git diff --check.
This commit is contained in:
2026-07-05 22:46:53 +03:00
parent 0faa5cc219
commit 9b0b063fed
18 changed files with 346 additions and 59 deletions
@@ -1,7 +1,110 @@
from pathlib import Path
import os
import subprocess
import textwrap
REPO_ROOT = Path(__file__).resolve().parents[1]
CHECK_SCRIPT = REPO_ROOT / "scripts" / "check_generated_analytics.sh"
def _fake_compose(
tmp_path,
*,
manifest_profile="ci",
live_rows="3",
views_mode="ok",
):
fake = tmp_path / "docker-compose"
fake.write_text(
textwrap.dedent(
f"""\
#!/usr/bin/env bash
set -euo pipefail
args="$*"
query=""
previous=""
for arg in "$@"; do
if [[ "$previous" == "--query" ]]; then
query="$arg"
break
fi
previous="$arg"
done
if [[ "$args" == *"kafka-manifest-summary"* ]]; then
printf '%s\\n' '16054\t2800\t930\t2026-01-01T00:00:00+00:00\t2026-01-01T05:59:00+00:00\t2026-01-01T00:00:00+00:00\t2026-01-01T06:00:00+00:00\t{manifest_profile}'
exit 0
fi
if [[ "$args" == *"clickhouse-client"* ]]; then
if [[ "$query" == *"hex(sipHash128"* ]]; then
printf '%s\\n' '16054\t2800\t930\t2026-01-01 00:00:00.000000\t2026-01-01 05:59:00.000000\t1\t1\t1\tdeadbeef'
elif [[ "$query" == *"returning_users / users"* ]]; then
printf '%s\\n' '930\t120\t0.129\t4'
elif [[ "$query" == *"short_visit_share"* ]]; then
printf '%s\\n' '2800\t0.2\t5\t5.7\t0\t120\t900\t18'
elif [[ "$query" == *"minIf(event_ts"* ]]; then
printf '%s\\n' '2800\t2100\t1500\t900\t650\t1\t0.23'
elif [[ "$query" == *"has_home"* ]]; then
printf '%s\\n' '2800\t2200\t1600\t950\t700\t1\t0.25'
elif [[ "$query" == *"SELECT count()"* && "$query" == *"FROM dds.event"* ]]; then
printf '%s\\n' '{live_rows}'
elif [[ "$query" == *"uniqExact(event_id)"* ]]; then
printf '%s\\n' '16200\t16200\t0'
elif [[ "$query" == *"per_event_homogeneous_visits"* ]]; then
printf '%s\\n' '19\t19\t19'
elif [[ "$query" == *"ods_device_rows"* ]]; then
printf '%s\\n' '19\t19\t19\t0\t0'
elif [[ "$query" == *"SELECT source, rows"* ]]; then
if [[ "{views_mode}" == "fail" ]]; then
exit 42
fi
printf '%s\\n' \
'dm.dq_summary\t1' \
'dm.v_daily_traffic\t1' \
'dm.v_events_enriched\t16054' \
'dm.v_session_overview\t1' \
'dm.v_top_pages_daily\t1' \
'dm.v_utm_effectiveness\t1'
else
echo "unexpected ClickHouse query: $query" >&2
exit 91
fi
exit 0
fi
echo "unexpected compose call: $args" >&2
exit 92
"""
),
encoding="utf-8",
)
fake.chmod(0o755)
return fake
def _run_generated_history_check(tmp_path, *, fake_compose, **env_overrides):
env = os.environ.copy()
env.update(
{
"COMPOSE_BIN": str(fake_compose),
"REQUIRE_SUPERSET": "0",
"WAIT_LIVE_ROWS_SECONDS": "0",
}
)
env.update(env_overrides)
return subprocess.run(
["bash", str(CHECK_SCRIPT)],
cwd=REPO_ROOT,
env=env,
text=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
timeout=30,
check=False,
)
def test_generated_history_check_uses_actual_manifest_boundary_and_profile():
@@ -14,6 +117,7 @@ def test_generated_history_check_uses_actual_manifest_boundary_and_profile():
assert "kafka-manifest-summary" in script
assert "manifest_model_t_end" in script
assert "CH_MODEL_T_END=\"$(clickhouse_datetime_literal \"${manifest_model_t_end}\")\"" in script
assert "PROFILE ?= ci" in makefile
assert "GEN_LAUNCH_PROFILE" in makefile
assert "PROFILE" in makefile
@@ -31,6 +135,51 @@ def test_generated_history_check_does_not_skip_required_live_seam():
assert 'CHECK_LIVE_SEAM="$${CHECK_LIVE_SEAM:-1}"' in makefile
def test_generated_history_check_rejects_manifest_profile_mismatch(tmp_path):
"""Проверка падает, если ожидаемый профиль не совпал с manifest."""
fake_compose = _fake_compose(tmp_path, manifest_profile="ci")
result = _run_generated_history_check(
tmp_path,
fake_compose=fake_compose,
PROFILE="daily-wave",
CHECK_LIVE_SEAM="0",
)
assert result.returncode != 0
assert "профиля daily-wave, но manifest от ci" in result.stderr
def test_generated_history_check_requires_live_rows_when_seam_is_required(tmp_path):
"""CHECK_LIVE_SEAM=1 падает, если live-продолжение не записало строк."""
fake_compose = _fake_compose(tmp_path, live_rows="0")
result = _run_generated_history_check(
tmp_path,
fake_compose=fake_compose,
PROFILE="ci",
CHECK_LIVE_SEAM="1",
)
assert result.returncode != 0
assert "live-продолжение не записало строки" in result.stderr
def test_generated_history_check_fails_when_dm_views_query_fails(tmp_path):
"""Ошибка запроса DM-витрин не превращается в пустой успешный цикл."""
fake_compose = _fake_compose(tmp_path, views_mode="fail")
result = _run_generated_history_check(
tmp_path,
fake_compose=fake_compose,
PROFILE="ci",
CHECK_LIVE_SEAM="0",
)
assert result.returncode != 0
assert "не удалось прочитать основные DM-витрины" in result.stderr
def test_clean_generated_history_run_explicitly_skips_live_seam():
"""Чистый backfill без live-продолжения отключает проверку стыка явно."""
script = (REPO_ROOT / "scripts" / "run_generated_history_analytics.sh").read_text(