ddadminandClaude Fable 5
ac7a973504
- Зачем:
- список DAG'ов должен читаться лесенкой ddl_init → world_init →
world_next_day, а путь менти — проходиться пустыми формами
(issue #4, спека редизайна пути менти, решения 2–3).
- Что:
- generator_control переименован в world_init, дефолт операции —
import; next-day ушёл из выпадашки в отдельный DAG;
- новый беспараметрный world_next_day: расписание */30 * * * *,
создаётся на паузе, catchup=False, max_active_runs=1; общие
задачи вынесены в airflow/dags/utils/startup_history_tasks.py;
- доки и контрактные тесты обновлены синхронно; быстрый старт
README — без make ddl, схему создаёт DAG ddl_init.
- Проверка:
- make test (210 + 31) и make lint зелёные;
- живая приёмка на чистом стенде: world_init пустой формой
импортировал эталонный мир за 217 с (3 дня, 280 437 событий),
world_next_day после снятия с паузы добавляет ровно один день
за прогон, дашборд Superset собирается.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
322 lines
13 KiB
Python
322 lines
13 KiB
Python
"""Контракты DAG world_init и world_next_day без запуска Airflow."""
|
||
|
||
import ast
|
||
from pathlib import Path
|
||
|
||
from clickstream_generator.launch import PROFILES
|
||
|
||
|
||
DAG_PATH = Path(__file__).parents[2] / "airflow" / "dags" / "world_init_dag.py"
|
||
NEXT_DAY_DAG_PATH = Path(__file__).parents[2] / "airflow" / "dags" / "world_next_day_dag.py"
|
||
REPO_ROOT = Path(__file__).parents[2]
|
||
|
||
|
||
def _tree():
|
||
return ast.parse(DAG_PATH.read_text(encoding="utf-8"))
|
||
|
||
|
||
def _declared_param_types() -> dict[str, set[str]]:
|
||
param_types = {}
|
||
for node in ast.walk(_tree()):
|
||
if not isinstance(node, ast.Dict):
|
||
continue
|
||
for key, value in zip(node.keys, node.values, strict=True):
|
||
if not (
|
||
isinstance(key, ast.Constant)
|
||
and isinstance(key.value, str)
|
||
and isinstance(value, ast.Call)
|
||
and isinstance(value.func, ast.Name)
|
||
and value.func.id == "Param"
|
||
):
|
||
continue
|
||
type_keyword = next(
|
||
keyword.value for keyword in value.keywords if keyword.arg == "type"
|
||
)
|
||
if isinstance(type_keyword, ast.List):
|
||
param_types[key.value] = {
|
||
item.value for item in type_keyword.elts if isinstance(item, ast.Constant)
|
||
}
|
||
else:
|
||
param_types[key.value] = {type_keyword.value}
|
||
return param_types
|
||
|
||
|
||
def _declared_param_defaults() -> dict[str, object]:
|
||
param_defaults = {}
|
||
for node in ast.walk(_tree()):
|
||
if not isinstance(node, ast.Dict):
|
||
continue
|
||
for key, value in zip(node.keys, node.values, strict=True):
|
||
if not (
|
||
isinstance(key, ast.Constant)
|
||
and isinstance(key.value, str)
|
||
and isinstance(value, ast.Call)
|
||
and isinstance(value.func, ast.Name)
|
||
and value.func.id == "Param"
|
||
and value.args
|
||
):
|
||
continue
|
||
param_defaults[key.value] = ast.literal_eval(value.args[0])
|
||
return param_defaults
|
||
|
||
|
||
def test_dag_file_exists_and_uses_dynamic_profiles():
|
||
"""DAG берёт варианты профилей из PROFILES, а не из ручного списка."""
|
||
text = DAG_PATH.read_text(encoding="utf-8")
|
||
|
||
assert "dag_id=\"world_init\"" in text
|
||
assert "sorted(PROFILES)" in text
|
||
for profile in PROFILES:
|
||
assert profile not in {"hardcoded-profile"}
|
||
|
||
|
||
def test_trigger_form_has_expected_param_enums():
|
||
"""Форма запуска ограничивает операции и профили."""
|
||
text = DAG_PATH.read_text(encoding="utf-8")
|
||
param_defaults = _declared_param_defaults()
|
||
|
||
assert 'enum=["backfill", "import", "check"]' in text
|
||
assert param_defaults["operation"] == "import"
|
||
assert "enum=sorted(PROFILES)" in text
|
||
assert param_defaults["profile"] == "daily-wave"
|
||
assert '"duration": Param(' in text
|
||
assert '"artifact_path": Param(' in text
|
||
assert '"expected_t_end": Param(' not in text
|
||
|
||
|
||
def test_trigger_form_marks_only_optional_params_as_nullable():
|
||
"""Форма разрешает оставить необязательные поля пустыми."""
|
||
param_types = _declared_param_types()
|
||
|
||
for name in {
|
||
"duration",
|
||
"seed",
|
||
"model_time_speed",
|
||
"artifact_path",
|
||
}:
|
||
assert "null" in param_types[name]
|
||
for name in {"operation", "profile"}:
|
||
assert "null" not in param_types[name]
|
||
|
||
|
||
def test_dag_branches_and_waits_for_etl_completion():
|
||
"""Backfill/import запускают ETL и ждут завершения перед check."""
|
||
text = DAG_PATH.read_text(encoding="utf-8")
|
||
|
||
assert "BranchPythonOperator" in text
|
||
assert "TriggerDagRunOperator" in text
|
||
assert "trigger_dag_id=ETL_DAG_ID" in text
|
||
assert "wait_for_completion=True" in text
|
||
assert 'allowed_states=["success"]' in text
|
||
assert 'failed_states=["failed"]' in text
|
||
|
||
|
||
def test_world_next_day_is_parameterless_paused_half_hour_dag():
|
||
"""DAG следующего дня запускается пустой формой каждые полчаса."""
|
||
text = NEXT_DAY_DAG_PATH.read_text(encoding="utf-8")
|
||
|
||
assert 'dag_id="world_next_day"' in text
|
||
assert 'schedule="*/30 * * * *"' in text
|
||
assert "catchup=False" in text
|
||
assert "is_paused_upon_creation=True" in text
|
||
assert "max_active_runs=1" in text
|
||
assert "params=" not in text
|
||
assert "precheck_next_day" in text
|
||
assert "run_next_day_task" in text
|
||
assert "TriggerDagRunOperator" in text
|
||
assert "wait_for_completion=True" in text
|
||
assert "check_manifest_task" in text
|
||
assert "check_etl_not_paused >> precheck" in text
|
||
|
||
|
||
def test_world_dags_precheck_etl_dag_not_paused_before_waiting():
|
||
"""Пульт проверяет паузу etl_pipeline до долгого ожидания."""
|
||
text = DAG_PATH.read_text(encoding="utf-8")
|
||
shared_tasks = (
|
||
REPO_ROOT / "airflow" / "dags" / "utils" / "startup_history_tasks.py"
|
||
).read_text(encoding="utf-8")
|
||
|
||
assert "assert_target_dag_not_paused" in text
|
||
assert 'ETL_DAG_ID = "etl_pipeline"' in text
|
||
assert 'op_kwargs={"dag_id": ETL_DAG_ID}' in text
|
||
assert "session.query(DagModel)" in shared_tasks
|
||
assert "DagModel.dag_id == dag_id" in shared_tasks
|
||
assert "target_dag_trigger_error" in shared_tasks
|
||
assert "Airflow 2.10.5" in shared_tasks
|
||
assert "fail_when_dag_is_paused" in shared_tasks
|
||
assert 'return "check_etl_not_paused_before_backfill"' in text
|
||
assert 'return "check_etl_not_paused_before_import"' in text
|
||
assert "check_etl_not_paused_before_backfill >> precheck_backfill_task" in text
|
||
assert "check_etl_not_paused_before_import >> precheck_import_task" in text
|
||
assert text.index("check_etl_not_paused_before_backfill >> precheck_backfill_task") < text.index(
|
||
"precheck_backfill_task >> backfill_task"
|
||
)
|
||
assert text.index("check_etl_not_paused_before_import >> precheck_import_task") < text.index(
|
||
"precheck_import_task >> import_task"
|
||
)
|
||
|
||
|
||
def test_make_up_rebuilds_airflow_images_after_repo_update():
|
||
"""make up пересобирает Airflow, но не запускает Superset до DM."""
|
||
text = (REPO_ROOT / "Makefile").read_text(encoding="utf-8")
|
||
|
||
up_block = text.split("\nup:\n", maxsplit=1)[1].split(
|
||
"\n\n# Остановить",
|
||
maxsplit=1,
|
||
)[0]
|
||
assert "$(COMPOSE) up -d --build" in up_block
|
||
assert "airflow-webserver" in up_block
|
||
assert "airflow-scheduler" in up_block
|
||
assert "superset" not in up_block
|
||
assert "superset-init" not in up_block
|
||
|
||
|
||
def test_readme_quick_start_creates_schema_via_ddl_init_before_world_init():
|
||
"""Быстрый старт: терминал — только make up, схему создаёт ddl_init до world_init."""
|
||
text = (REPO_ROOT / "README.md").read_text(encoding="utf-8")
|
||
|
||
assert "uv" in text
|
||
assert "Docker" in text
|
||
assert "docker compose" in text
|
||
assert "make ddl" not in text
|
||
quick_start = text.split("## Быстрый старт", maxsplit=1)[1]
|
||
assert quick_start.index("ddl_init") < quick_start.index("world_init")
|
||
|
||
|
||
def test_course_readme_lists_uv_before_first_command():
|
||
"""Курс называет uv до первой команды."""
|
||
text = (REPO_ROOT / "docs" / "course" / "README.md").read_text(encoding="utf-8")
|
||
|
||
assert "uv" in text
|
||
assert text.index("uv") < text.index("make generated-history-analytics")
|
||
|
||
|
||
def test_startup_history_runbook_warns_about_daily_wave_idle_gap():
|
||
"""Runbook предупреждает о дыре модельного времени при простое daily-wave."""
|
||
text = (REPO_ROOT / "docs" / "runbooks" / "startup-history.md").read_text(
|
||
encoding="utf-8"
|
||
)
|
||
|
||
assert "daily-wave" in text
|
||
assert "прост" in text
|
||
assert "дыр" in text
|
||
assert "make generator-reset" in text
|
||
assert "startup-history-import" in text
|
||
|
||
|
||
def test_no_docker_or_continue_operation_in_dag():
|
||
"""Пульт не управляет Docker и не содержит операцию continue."""
|
||
text = DAG_PATH.read_text(encoding="utf-8")
|
||
|
||
assert "docker" not in text.lower()
|
||
assert '"continue"' not in text
|
||
|
||
|
||
def test_compose_mounts_generator_code_without_socket_and_gates_live_service():
|
||
"""Airflow видит код генератора, но не получает Docker socket."""
|
||
text = (REPO_ROOT / "docker-compose.yml").read_text(encoding="utf-8")
|
||
|
||
assert "PYTHONPATH: /opt/airflow/generator_src" in text
|
||
assert "./generator/src:/opt/airflow/generator_src:ro" in text
|
||
assert "/var/run/docker.sock" not in text
|
||
assert "profiles:\n - live-generator" in text
|
||
|
||
|
||
def test_down_and_clean_include_live_generator_profile():
|
||
"""Остановка стенда удаляет профильный live-генератор."""
|
||
text = (REPO_ROOT / "Makefile").read_text(encoding="utf-8")
|
||
|
||
assert "down:\n\t$(COMPOSE) --profile live-generator down" in text
|
||
assert "clean:\n\t$(COMPOSE) --profile live-generator down -v --remove-orphans" in text
|
||
|
||
|
||
def test_console_generator_paths_precheck_clean_stand_before_writes():
|
||
"""Backfill/import с хоста проверяют чистый стенд до записи."""
|
||
run_generator = (REPO_ROOT / "scripts" / "run_generator.sh").read_text(
|
||
encoding="utf-8"
|
||
)
|
||
import_artifact = (
|
||
REPO_ROOT / "scripts" / "import_startup_history_artifact.sh"
|
||
).read_text(encoding="utf-8")
|
||
|
||
assert "bash \"${SCRIPT_DIR}/assert_stand_clean.sh\" clean" in run_generator
|
||
assert run_generator.index("assert_stand_clean.sh") < run_generator.index(
|
||
"run --rm --no-deps"
|
||
)
|
||
assert "bash \"${SCRIPT_DIR}/assert_stand_clean.sh\" clean" in import_artifact
|
||
assert import_artifact.index("assert_stand_clean.sh") < import_artifact.index(
|
||
"startup_history_artifact import"
|
||
)
|
||
|
||
|
||
def test_reference_artifact_is_import_default_only():
|
||
"""Экспорт не может по умолчанию перезаписать эталонный мир."""
|
||
makefile = (REPO_ROOT / "Makefile").read_text(encoding="utf-8")
|
||
export_script = (
|
||
REPO_ROOT / "scripts" / "export_startup_history_artifact.sh"
|
||
).read_text(encoding="utf-8")
|
||
import_script = (
|
||
REPO_ROOT / "scripts" / "import_startup_history_artifact.sh"
|
||
).read_text(encoding="utf-8")
|
||
|
||
assert "STARTUP_HISTORY_EXPORT_ARTIFACT" in makefile
|
||
assert "STARTUP_HISTORY_IMPORT_ARTIFACT" in makefile
|
||
assert "/tmp/clickstream-startup-history.json" in export_script
|
||
assert "data/startup_history/reference-world.json.xz" in import_script
|
||
|
||
|
||
def test_console_continue_checks_state_or_clean_stand_before_live_start():
|
||
"""Continue не стартует новый live-мир на грязном стенде без state."""
|
||
run_generator = (REPO_ROOT / "scripts" / "run_generator.sh").read_text(
|
||
encoding="utf-8"
|
||
)
|
||
|
||
assert "elif [[ \"${VERB}\" == \"continue\" ]]" in run_generator
|
||
assert "bash \"${SCRIPT_DIR}/assert_stand_clean.sh\" continue" in run_generator
|
||
assert run_generator.index("assert_stand_clean.sh\" continue") < run_generator.index(
|
||
"up -d --build generator"
|
||
)
|
||
|
||
|
||
def test_console_reset_checks_clean_stand_before_live_start():
|
||
"""Reset не пишет новый мир поверх старых данных."""
|
||
run_generator = (REPO_ROOT / "scripts" / "run_generator.sh").read_text(
|
||
encoding="utf-8"
|
||
)
|
||
reset_branch = run_generator.split("else\n", maxsplit=1)[1]
|
||
|
||
assert "bash \"${SCRIPT_DIR}/assert_stand_clean.sh\" clean" in reset_branch
|
||
assert reset_branch.index("assert_stand_clean.sh\" clean") < reset_branch.index(
|
||
"up -d --build generator"
|
||
)
|
||
|
||
|
||
def test_clean_start_paths_include_live_generator_profile():
|
||
"""Все чистые сбросы видят профильный live-генератор."""
|
||
generated_history = (
|
||
REPO_ROOT / "scripts" / "run_generated_history_analytics.sh"
|
||
).read_text(encoding="utf-8")
|
||
|
||
assert "--profile live-generator down -v --remove-orphans" in generated_history
|
||
|
||
|
||
def test_host_clean_precheck_checks_live_before_kafka_and_stg():
|
||
"""Host backfill/import сначала проверяет, что live-генератор не запущен."""
|
||
text = (REPO_ROOT / "scripts" / "assert_stand_clean.sh").read_text(
|
||
encoding="utf-8"
|
||
)
|
||
|
||
assert "--mode live" in text
|
||
assert text.index("--mode live") < text.index("clickhouse-client")
|
||
assert text.index("clickhouse-client") < text.index("--mode \"${CHECK_MODE}\"")
|
||
|
||
|
||
def test_airflow_and_generator_kafka_dependency_versions_match():
|
||
"""Airflow и генератор используют одну версию kafka-python."""
|
||
airflow_req = (REPO_ROOT / "airflow" / "requirements.txt").read_text(encoding="utf-8")
|
||
generator_req = (REPO_ROOT / "generator" / "requirements.txt").read_text(encoding="utf-8")
|
||
|
||
assert "kafka-python==2.0.6" in airflow_req
|
||
assert "kafka-python==2.0.6" in generator_req
|
||
assert "prometheus-client==0.21.1" in airflow_req
|