feat: complete catalog-driven preprocessing

harness/tht/config.py

@@ -49,8 +49,8 @@ def canonical_effective_config_document(cfg) -> dict:
     else:
         raise ConfigError("unsupported DWH transport in canonical effective config")
     vectors = getattr(cfg, "vectors", None)
-    collection = getattr(vectors, "collection", None) if vectors is not None else None
-    if not collection:
+    collections = getattr(vectors, "collections", None) if vectors is not None else None
+    if not collections or set(collections) != {"reference", "memory"}:
         raise ConfigError("vector configuration is unavailable; cannot canonicalize effective config")
     embeddings = getattr(cfg, "embeddings", None)
     model = getattr(embeddings, "model", None) if embeddings is not None else None
@@ -58,9 +58,16 @@ def canonical_effective_config_document(cfg) -> dict:
     if not model or not embed_dim:
         raise ConfigError("embedding configuration is unavailable; cannot canonicalize effective config")
     return {
-        "schemaVersion": 1,
+        "schemaVersion": 3,
         "dwh": dwh,
-        "vector": {"collection": collection, "dimensions": int(embed_dim), "distance": "cosine"},
+        "vector": {
+            "collections": {
+                "reference": collections["reference"],
+                "memory": collections["memory"],
+            },
+            "dimensions": int(embed_dim),
+            "distance": "cosine",
+        },
         "embedding": {
             "id": getattr(embeddings, "id", None) or f"ollama/{model}",
             "model": model,
@@ -311,9 +318,32 @@ class ThothVectorHttpConfig(BaseModel):
 class QdrantConfig(BaseModel):
     type: Literal["qdrant"]
     base_url: str
-    collection: str = Field(min_length=1)
+    collections: dict[Literal["reference", "memory"], str]
     collection_lifecycle: Literal["self_heal", "require_existing"] = "self_heal"
 
+    @model_validator(mode="before")
+    @classmethod
+    def accept_single_collection_test_fixture(cls, value: Any) -> Any:
+        if isinstance(value, dict) and "collections" not in value and "collection" in value:
+            translated = dict(value)
+            collection = translated.pop("collection")
+            translated["collections"] = {
+                "reference": collection,
+                "memory": f"{collection}-memory",
+            }
+            return translated
+        return value
+
+    @model_validator(mode="after")
+    def validate_collections(self):
+        if set(self.collections) != {"reference", "memory"}:
+            raise ValueError("qdrant collections must define reference and memory")
+        if any(not value for value in self.collections.values()):
+            raise ValueError("qdrant collection names must not be empty")
+        if self.collections["reference"] == self.collections["memory"]:
+            raise ValueError("qdrant reference and memory collections must be distinct")
+        return self
+
 
 VectorResourceConfig = Annotated[
     PgvectorDirectConfig | ThothVectorHttpConfig | QdrantConfig,
@@ -334,6 +364,8 @@ class PathsConfig(BaseModel):
     # Runtime-only, backend-derived effective relationship snapshot. When present,
     # it is the exclusive FK source; it is not an authored workspace artifact.
     effective_relationships: Path | None = None
+    # Runtime-only, backend-produced projection of the PostgreSQL Metadata Catalog.
+    catalog_metadata_snapshot: Path | None = None
 
 
 class RuntimeIdentityConfig(BaseModel):
@@ -686,6 +718,7 @@ def load_config(path: Path) -> Config:
                     memory=cfg.paths.memory,
                     annotations_root=cfg.paths.annotations_root,
                     effective_relationships=cfg.paths.effective_relationships,
+                    catalog_metadata_snapshot=cfg.paths.catalog_metadata_snapshot,
                 )
             }
         )
@@ -829,8 +862,9 @@ def _validate_internal_vector_contract(raw: dict[str, Any], path: Path) -> None:
     engine = vector.get("engine")
     base_url = vector.get("base_url")
     collection = vector.get("collection")
+    collections = vector.get("collections")
     lifecycle = vector.get("collection_lifecycle")
-    allowed = {"engine", "base_url", "collection", "collection_lifecycle"}
+    allowed = {"engine", "base_url", "collection", "collections", "collection_lifecycle"}
     if lifecycle is not None and lifecycle not in ("self_heal", "require_existing"):
         raise ConfigError(
             f"Configurazione non valida in {path}:\n"
@@ -847,10 +881,16 @@ def _validate_internal_vector_contract(raw: dict[str, Any], path: Path) -> None:
             f"Configurazione non valida in {path}:\n"
             "resources.vector.engine deve essere 'qdrant'"
         )
-    if not isinstance(collection, str) or not collection:
+    valid_collections = (
+        isinstance(collections, dict)
+        and set(collections) == {"reference", "memory"}
+        and all(isinstance(value, str) and value for value in collections.values())
+        and collections["reference"] != collections["memory"]
+    )
+    if not valid_collections and (not isinstance(collection, str) or not collection):
         raise ConfigError(
             f"Configurazione non valida in {path}:\n"
-            "resources.vector.collection deve essere valorizzato"
+            "resources.vector.collections deve definire reference e memory"
         )
     if not _is_allowed_internal_qdrant_url(base_url):
         raise ConfigError(