migration: establish exact preserved app baseline

2026-07-30 23:51:48 +08:00
parent 41329943c4
commit 4083dceba3
399 changed files with 129967 additions and 23 deletions
@@ -0,0 +1,77 @@
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+from app_config import APP_DIR
+from backend.data.contracts import DataUsage, DatasetContract, ProviderContract
+
+
+class DataPolicyError(RuntimeError):
+    pass
+
+
+class DataSourcePolicy:
+    def __init__(
+        self,
+        providers: dict[str, ProviderContract],
+        datasets: dict[str, DatasetContract],
+    ) -> None:
+        self.providers = dict(providers)
+        self.datasets = dict(datasets)
+
+    @classmethod
+    def load(cls, path: Path | None = None) -> "DataSourcePolicy":
+        config_path = path or APP_DIR / "config" / "data-fields.config.json"
+        payload = json.loads(config_path.read_text(encoding="utf-8"))
+        providers = {
+            provider_id: ProviderContract(
+                id=provider_id,
+                provider_class=str(item["class"]),
+                calculation_allowed=bool(item["calculation_allowed"]),
+            )
+            for provider_id, item in payload["providers"].items()
+        }
+        datasets = {
+            item["id"]: DatasetContract(
+                id=str(item["id"]),
+                entity=str(item["entity"]),
+                frequency=str(item["frequency"]),
+                primary=str(item["primary"]),
+                fallbacks=tuple(str(value) for value in item.get("fallbacks", [])),
+                usage=str(item["usage"]),
+                fields=tuple(str(value) for value in item.get("fields", [])),
+            )
+            for item in payload["datasets"]
+        }
+        return cls(providers, datasets)
+
+    def dataset(self, dataset_id: str) -> DatasetContract:
+        try:
+            return self.datasets[dataset_id]
+        except KeyError as exc:
+            raise DataPolicyError(f"Unregistered dataset: {dataset_id}") from exc
+
+    def assert_allowed(
+        self,
+        dataset_id: str,
+        provider_id: str,
+        usage: DataUsage,
+    ) -> DatasetContract:
+        dataset = self.dataset(dataset_id)
+        if dataset.usage == "blocked":
+            raise DataPolicyError(f"Dataset is blocked: {dataset_id}")
+        if provider_id not in dataset.providers:
+            raise DataPolicyError(
+                f"Provider {provider_id} is not registered for dataset {dataset_id}"
+            )
+        try:
+            provider = self.providers[provider_id]
+        except KeyError as exc:
+            raise DataPolicyError(f"Unregistered provider: {provider_id}") from exc
+        if usage == "calculation":
+            if dataset.usage != "calculation" or not provider.calculation_allowed:
+                raise DataPolicyError(
+                    f"Provider {provider_id} cannot calculate dataset {dataset_id}"
+                )
+        return dataset