feat(core): define durable datasource artifact contract

src/govoplan_core/core/datasources.py

@@ -23,6 +23,7 @@ CAPABILITY_DATASOURCE_CATALOGUE = "datasources.catalogue"
 CAPABILITY_DATASOURCE_LIFECYCLE = "datasources.lifecycle"
 CAPABILITY_DATASOURCE_PUBLICATION = "datasources.publication"
 CAPABILITY_DATASOURCE_ORIGINS = "connectors.datasourceOrigins"
+CAPABILITY_DATASOURCE_ARTIFACT_BACKENDS = "datasources.artifactBackends"
 
 DatasourceMode = Literal["live", "cached", "static"]
 DatasourceKind = Literal[
@@ -37,6 +38,11 @@ DatasourceKind = Literal[
 ]
 DatasourceShape = Literal["tabular", "document", "binary", "directory", "stream"]
 DatasourceConsistency = Literal["current", "live", "frozen"]
+DatasourcePublicationStatus = Literal[
+    "published",
+    "published_with_warnings",
+    "review_required",
+]
 
 
 class DatasourceError(ValueError):
@@ -309,12 +315,73 @@ class DatasourceStageInput:
     governance: DatasourceGovernance | None = None
 
 
+@dataclass(frozen=True, slots=True)
+class DatasourceArtifactReference:
+    """Immutable provider-neutral reference to a durable tabular payload.
+
+    The producer owns creation of the payload. Datasources pins its locator,
+    checksum and declared shape without importing the artifact-owning module;
+    a configured payload backend verifies integrity and provides bounded reads.
+    """
+
+    backend: str
+    locator: str
+    checksum: str
+    row_count: int
+    byte_count: int
+    schema: tuple[DatasourceField, ...]
+    fingerprint: str
+    media_type: str = "application/x-ndjson"
+    checkpoint: Mapping[str, object] = field(default_factory=dict)
+    metadata: Mapping[str, object] = field(default_factory=dict)
+    validation: Mapping[str, object] = field(default_factory=dict)
+
+
+@runtime_checkable
+class DatasourceArtifactBackend(Protocol):
+    """Storage-module boundary for immutable artifact-backed tabular data."""
+
+    backend: str
+
+    def verify(
+        self,
+        session: object,
+        *,
+        tenant_id: str,
+        artifact: DatasourceArtifactReference,
+    ) -> None: ...
+
+    def read_rows(
+        self,
+        session: object,
+        *,
+        tenant_id: str,
+        artifact: DatasourceArtifactReference,
+        offset: int,
+        limit: int,
+    ) -> Sequence[Mapping[str, object]]: ...
+
+    def delete(
+        self,
+        session: object,
+        *,
+        tenant_id: str,
+        artifact: DatasourceArtifactReference,
+    ) -> None: ...
+
+
+@runtime_checkable
+class DatasourceArtifactBackendProvider(Protocol):
+    def artifact_backends(self) -> Sequence[DatasourceArtifactBackend]: ...
+
+
 @dataclass(frozen=True, slots=True)
 class DatasourcePublicationRequest:
     producer_module: str
     producer_run_ref: str
     idempotency_key: str
-    rows: tuple[Mapping[str, object], ...]
+    rows: tuple[Mapping[str, object], ...] | None = None
+    artifact: DatasourceArtifactReference | None = None
     target_datasource_ref: str | None = None
     name: str | None = None
     source_name: str | None = None
@@ -331,7 +398,7 @@ class DatasourcePublicationRequest:
 @dataclass(frozen=True, slots=True)
 class DatasourcePublicationResult:
     ref: str
-    status: str
+    status: DatasourcePublicationStatus
     datasource: DatasourceDescriptor
     materialization: DatasourceMaterialization
     replayed: bool = False
@@ -564,6 +631,17 @@ def datasource_catalogue(registry: object | None) -> DatasourceCatalogueProvider
     return capability if isinstance(capability, DatasourceCatalogueProvider) else None
 
 
+def datasource_artifact_backend_provider(
+    registry: object | None,
+) -> DatasourceArtifactBackendProvider | None:
+    capability = _capability(registry, CAPABILITY_DATASOURCE_ARTIFACT_BACKENDS)
+    return (
+        capability
+        if isinstance(capability, DatasourceArtifactBackendProvider)
+        else None
+    )
+
+
 def datasource_lifecycle(registry: object | None) -> DatasourceLifecycleProvider | None:
     capability = _capability(registry, CAPABILITY_DATASOURCE_LIFECYCLE)
     return capability if isinstance(capability, DatasourceLifecycleProvider) else None
@@ -613,9 +691,14 @@ def _governance_mapping(value: object) -> Mapping[str, object]:
 
 __all__ = [
     "CAPABILITY_DATASOURCE_CATALOGUE",
+    "CAPABILITY_DATASOURCE_ARTIFACT_BACKENDS",
     "CAPABILITY_DATASOURCE_LIFECYCLE",
     "CAPABILITY_DATASOURCE_ORIGINS",
+    "CAPABILITY_DATASOURCE_PUBLICATION",
     "DatasourceAccessError",
+    "DatasourceArtifactReference",
+    "DatasourceArtifactBackend",
+    "DatasourceArtifactBackendProvider",
     "DatasourceCatalogueProvider",
     "DatasourceConsistency",
     "DatasourceDescriptor",
@@ -631,6 +714,10 @@ __all__ = [
     "DatasourceOriginProvider",
     "DatasourceOriginReadRequest",
     "DatasourceOriginReadResult",
+    "DatasourcePublicationProvider",
+    "DatasourcePublicationRequest",
+    "DatasourcePublicationResult",
+    "DatasourcePublicationStatus",
     "DatasourceReadRequest",
     "DatasourceReadResult",
     "DatasourceShape",
@@ -639,6 +726,8 @@ __all__ = [
     "DatasourceUnavailableError",
     "DatasourceValidationError",
     "datasource_catalogue",
+    "datasource_artifact_backend_provider",
     "datasource_lifecycle",
     "datasource_origins",
+    "datasource_publication",
 ]