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 +1,144 @@
+from __future__ import annotations
+
+import importlib.util
+import sys
+from pathlib import Path
+
+import pytest
+
+from gyxx_flow.core.layout import DataLayout
+
+PROJECT_ROOT = Path(__file__).resolve().parents[1]
+RUNTIME_ROOT = PROJECT_ROOT / "src" / "gyxx_flow" / "modules"
+
+
+def _load_path_module(module: str, filename: str, monkeypatch, data_root: Path):
+    monkeypatch.setenv("GYXX_DATA_ROOT", str(data_root))
+    runtime_root = RUNTIME_ROOT / module / "runtime"
+    monkeypatch.setenv("GYXX_MODULE_ROOT", str(runtime_root))
+    path = runtime_root / filename
+    name = f"_gyxx_test_paths_{module}"
+    spec = importlib.util.spec_from_file_location(name, path)
+    assert spec and spec.loader
+    loaded = importlib.util.module_from_spec(spec)
+    sys.modules[name] = loaded
+    try:
+        spec.loader.exec_module(loaded)
+    finally:
+        sys.modules.pop(name, None)
+    return loaded
+
+
+@pytest.mark.parametrize(
+    ("module", "filename", "names"),
+    [
+        (
+            "content_marketing",
+            "runtime_paths.py",
+            ("raw_root", "normalized_root", "curated_root", "exports_root"),
+        ),
+        (
+            "product_commerce",
+            "runtime_paths.py",
+            (
+                "RAW_DATA_ROOT",
+                "NORMALIZED_DATA_ROOT",
+                "CURATED_DATA_ROOT",
+                "EXPORTS_DATA_ROOT",
+            ),
+        ),
+        (
+            "shop_intelligence",
+            "paths.py",
+            ("RAW_ROOT", "NORMALIZED_ROOT", "CURATED_ROOT", "EXPORTS_ROOT"),
+        ),
+        (
+            "supply_chain",
+            "paths.py",
+            ("SHARED_ROOT", "NORMALIZED_ROOT", "CURATED_ROOT", "BACKUP_ROOT"),
+        ),
+    ],
+)
+def test_all_module_path_boundaries_use_the_canonical_data_tree(
+    tmp_path: Path,
+    monkeypatch,
+    module: str,
+    filename: str,
+    names: tuple[str, str, str, str],
+) -> None:
+    data_root = tmp_path / "relocatable-data"
+    loaded = _load_path_module(module, filename, monkeypatch, data_root)
+
+    layers = ("raw", "normalized", "curated", "exports")
+    for name, layer in zip(names, layers, strict=True):
+        assert getattr(loaded, name) == data_root / "data" / layer / module
+    assert not data_root.exists(), "path resolution must not mutate the filesystem"
+
+
+def test_module_data_paths_cover_collected_and_exported_file_types(tmp_path: Path) -> None:
+    paths = DataLayout(tmp_path).for_module("content_marketing")
+
+    assert paths.raw_path("api", "response.json") == (
+        tmp_path / "data" / "raw" / "content_marketing" / "api" / "response.json"
+    )
+    assert paths.raw_path("downloads", "source.xlsx") == (
+        tmp_path / "data" / "raw" / "content_marketing" / "downloads" / "source.xlsx"
+    )
+    assert paths.raw_path("downloads", "source.xls") == (
+        tmp_path / "data" / "raw" / "content_marketing" / "downloads" / "source.xls"
+    )
+    assert paths.raw_path("downloads", "source.csv") == (
+        tmp_path / "data" / "raw" / "content_marketing" / "downloads" / "source.csv"
+    )
+    assert paths.export_path("reports", "weekly.md") == (
+        tmp_path / "data" / "exports" / "content_marketing" / "reports" / "weekly.md"
+    )
+    assert list(tmp_path.iterdir()) == []
+
+
+@pytest.mark.parametrize("unsafe", ["", "..", "a/b", "a\\b", "C:\\data"])
+def test_module_data_paths_reject_unsafe_module_names(tmp_path: Path, unsafe: str) -> None:
+    with pytest.raises(ValueError, match="path segment"):
+        DataLayout(tmp_path).for_module(unsafe)
+
+
+@pytest.mark.parametrize("unsafe", [("..", "escape.json"), ("C:\\", "escape.json")])
+def test_module_data_paths_reject_paths_outside_the_layer(
+    tmp_path: Path, unsafe: tuple[str, str]
+) -> None:
+    with pytest.raises(ValueError, match="escapes module data root"):
+        DataLayout(tmp_path).for_module("product_commerce").raw_path(*unsafe)
+
+
+@pytest.mark.parametrize("configured", ["", "   "])
+@pytest.mark.parametrize(
+    ("module", "filename", "root_name"),
+    [
+        ("content_marketing", "runtime_paths.py", "data_root"),
+        ("product_commerce", "runtime_paths.py", "DATA_HOME"),
+        ("shop_intelligence", "paths.py", "DATA_ROOT"),
+        ("supply_chain", "paths.py", "DATA_ROOT"),
+    ],
+)
+def test_blank_data_root_uses_the_project_var_default(
+    monkeypatch,
+    configured: str,
+    module: str,
+    filename: str,
+    root_name: str,
+) -> None:
+    monkeypatch.setenv("GYXX_DATA_ROOT", configured)
+    runtime_root = RUNTIME_ROOT / module / "runtime"
+    monkeypatch.setenv("GYXX_MODULE_ROOT", str(runtime_root))
+    path = runtime_root / filename
+    name = f"_gyxx_test_blank_paths_{module}"
+    spec = importlib.util.spec_from_file_location(name, path)
+    assert spec and spec.loader
+    loaded = importlib.util.module_from_spec(spec)
+    sys.modules[name] = loaded
+    try:
+        spec.loader.exec_module(loaded)
+    finally:
+        sys.modules.pop(name, None)
+
+    assert getattr(loaded, root_name) == PROJECT_ROOT / "var"