feat: complete catalog-driven preprocessing

harness/tht/cli/vector_cmd.py

@@ -6,7 +6,7 @@ import typer
 
 from tht.cli._guards import require_vector_write_allowed
 from tht.cli.config_cmd import CONFIG_OPT
-from tht.cli.schema_cmd import _load_config_or_exit, annotations_path, physical_path
+from tht.cli.schema_cmd import _load_config_or_exit
 from tht.ports.vector import VectorWriteRecord
 from tht.vectorstore.store import SyncStats, content_hash
 
@@ -112,42 +112,14 @@ def index_schema_cmd(
     config: Path = CONFIG_OPT,
     json_output: bool = typer.Option(False, "--json"),
 ) -> None:
-    """Embedda e sincronizza i record schema (tabelle e colonne) nel semantic store."""
-    from tht.adapters.factory import build_vector_store
-    from tht.mschema.models import Annotations, PhysicalSchema
+    """Replace the schema slice from the PostgreSQL Catalog projection."""
     from tht.ports.vector import VectorStoreError
-    from tht.vectorstore.records import schema_records
 
     cfg = _load_config_or_exit(config)
     require_vector_write_allowed(cfg, "vector index-schema")
     require_vector_cfg(cfg)
-    phys_file = physical_path(cfg)
-    if not phys_file.exists():
-        message = f"ERRORE: {phys_file} non trovato. Esegui prima `tht schema introspect`."
-        if json_output:
-            _emit_json({
-                "code": "schema_missing",
-                "error": "physical schema is missing",
-                "operation": "index_schema",
-                "schemaVersion": 1,
-                "status": "failed",
-                "workspaceId": cfg._workspace_id,
-                "workspaceRevision": cfg._workspace_revision,
-            })
-        else:
-            typer.secho(message, fg=typer.colors.RED, err=True)
-        raise typer.Exit(code=1)
-    physical = PhysicalSchema.from_yaml(phys_file)
-    annotations_file = annotations_path(cfg)
-    annotations = Annotations.from_yaml(annotations_file)
-    records = schema_records(physical, annotations)
     try:
-        stats = sync_canonical_records(
-            "schema_records",
-            records,
-            store=build_vector_store(cfg, require_write=True),
-            embedder=make_embedder(cfg.embeddings),
-        )
+        stats, counts, snapshot_path = index_catalog_schema(cfg)
     except VectorStoreError as exc:
         code = str(exc)
         error = "semantic index incompatible" if code == "semantic_index_incompatible" else "schema indexing failed"
@@ -167,17 +139,17 @@ def index_schema_cmd(
     if json_output:
         _emit_json({
             "artifactIdentities": [
-                {"digest": _artifact_digest(annotations_file), "kind": "schema_annotations"},
-                {"digest": _artifact_digest(phys_file), "kind": "physical_schema"},
+                {"digest": _artifact_digest(snapshot_path), "kind": "catalog_metadata_snapshot"},
             ],
             "code": "ok",
-            "collection": cfg.vectors.collection,
+            "collection": cfg.vectors.collections["reference"],
             "counts": {
                 "added": stats.added,
-                "columns": sum(len(table.columns) for table in physical.tables.values()),
+                "columns": counts["columns"],
                 "deleted": stats.deleted,
-                "records": len(records),
-                "tables": len(physical.tables),
+                "records": counts["records"],
+                "relationships": counts["relationships"],
+                "tables": counts["tables"],
                 "unchanged": stats.unchanged,
                 "updated": stats.updated,
             },
@@ -189,3 +161,32 @@ def index_schema_cmd(
         })
         return
     _print_stats(stats)
+
+
+def index_catalog_schema(cfg):
+    from tht.adapters.factory import build_vector_store
+    from tht.mschema.catalog_snapshot import load_catalog_metadata_snapshot
+    from tht.vectorstore.records import catalog_schema_records
+
+    snapshot_path = cfg.paths.catalog_metadata_snapshot
+    if snapshot_path is None:
+        raise ValueError("catalog metadata snapshot is not configured")
+    snapshot = load_catalog_metadata_snapshot(snapshot_path, cfg._workspace_id)
+    records = catalog_schema_records(snapshot)
+    store = build_vector_store(cfg, require_write=True)
+    deleted = store.delete_kinds(
+        "schema_records", ["schema_table", "schema_column", "schema_relationship"]
+    )
+    stats = sync_canonical_records(
+        "schema_records",
+        records,
+        store=store,
+        embedder=make_embedder(cfg.embeddings),
+    )
+    stats.deleted = deleted
+    return stats, {
+        "tables": len(snapshot.tables),
+        "columns": sum(len(table.columns) for table in snapshot.tables),
+        "relationships": len(snapshot.relationships),
+        "records": len(records),
+    }, snapshot_path