feat: consolidate legacy workflows into gyxx-flow

2026-07-28 14:51:15 +08:00
commit c23b62a8c8
374 changed files with 132990 additions and 0 deletions
@@ -0,0 +1,86 @@
+from __future__ import annotations
+
+from collections import Counter
+from pathlib import Path
+
+import pytest
+
+from gyxx_flow.catalog import CatalogError, WorkflowCatalog
+
+PROJECT_ROOT = Path(__file__).resolve().parents[1]
+
+
+def test_catalog_maps_exactly_21_current_scheduled_tasks() -> None:
+    catalog = WorkflowCatalog.load(PROJECT_ROOT / "config")
+    scheduled = catalog.scheduled_workflows()
+
+    assert len(scheduled) == 21
+    assert len({item.source_task_name for item in scheduled}) == 21
+    assert Counter(item.module for item in scheduled) == {
+        "content_marketing": 9,
+        "product_commerce": 7,
+        "shop_intelligence": 2,
+        "supply_chain": 3,
+    }
+
+
+def test_catalog_keeps_known_manual_or_unregistered_workflows_unscheduled() -> None:
+    catalog = WorkflowCatalog.load(PROJECT_ROOT / "config")
+
+    assert {
+        "content.mapping.refresh",
+        "content.retry_failed",
+        "product.backfill",
+        "product.market_rank",
+        "product.review_collection",
+        "supply.purchase_order_update",
+    }.issubset({item.workflow_id for item in catalog.manual_workflows()})
+    assert "product.weekly_aggregate.documented_missing" in {
+        item.workflow_id for item in catalog.unavailable_workflows()
+    }
+
+
+def test_every_scheduled_workflow_has_one_valid_asia_shanghai_schedule() -> None:
+    catalog = WorkflowCatalog.load(PROJECT_ROOT / "config")
+
+    assert catalog.timezone == "Asia/Shanghai"
+    assert len(catalog.schedules) == 21
+    for workflow in catalog.scheduled_workflows():
+        schedule = catalog.schedule_for(workflow.workflow_id)
+        assert schedule.workflow_id == workflow.workflow_id
+        assert schedule.kind in {"daily", "weekly", "monthly", "interval_days"}
+        assert schedule.at.count(":") == 1
+
+
+def test_catalog_rejects_duplicate_workflow_ids(tmp_path: Path) -> None:
+    (tmp_path / "workflows.json").write_text(
+        '{"schema_version":2,"workflows":['
+        '{"id":"a.one","module":"content_marketing","trigger":"manual",'
+        '"execution":{"entry":"run.py"}},'
+        '{"id":"a.one","module":"content_marketing","trigger":"manual",'
+        '"execution":{"entry":"other.py"}}]}',
+        encoding="utf-8",
+    )
+    (tmp_path / "schedules.json").write_text(
+        '{"schema_version":1,"timezone":"Asia/Shanghai","schedules":[]}',
+        encoding="utf-8",
+    )
+
+    with pytest.raises(CatalogError, match="duplicate workflow id"):
+        WorkflowCatalog.load(tmp_path)
+
+
+def test_catalog_rejects_absolute_execution_entry(tmp_path: Path) -> None:
+    (tmp_path / "workflows.json").write_text(
+        '{"schema_version":2,"workflows":['
+        '{"id":"a.one","module":"content_marketing","trigger":"manual",'
+        '"execution":{"entry":"D:/old/run.py"}}]}',
+        encoding="utf-8",
+    )
+    (tmp_path / "schedules.json").write_text(
+        '{"schema_version":1,"timezone":"Asia/Shanghai","schedules":[]}',
+        encoding="utf-8",
+    )
+
+    with pytest.raises(CatalogError, match="relative"):
+        WorkflowCatalog.load(tmp_path)