fix: preserve Task 2 command loading contracts

harness/tht/cli/vector_cmd.py

@@ -11,7 +11,13 @@ 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,
+    _load_schema_config,
+    annotations_path,
+    physical_path,
+)
+from tht.config import Config, ConfigError
 from tht.ports.vector import VectorWriteRecord
 from tht.vectorstore.store import SyncStats, content_hash
 
@@ -154,7 +160,7 @@ class IndexSchemaResult(TypedDict):
     counts: IndexCounts
 
 
-def _vector_cfg_or_error(cfg) -> None:
+def _vector_cfg_or_error(cfg: Config) -> None:
     missing = []
     if cfg.embeddings is None:
         missing.append("embeddings")
@@ -164,24 +170,42 @@ def _vector_cfg_or_error(cfg) -> None:
         raise _MachineVectorError("vector_configuration_missing")
 
 
-def _vector_write_or_error(cfg) -> None:
+def _vector_write_or_error(cfg: Config) -> None:
     if cfg.profile == "workstation" and not has_vector_write_rest(cfg):
         raise _MachineVectorError("vector_write_not_allowed")
 
 
+def _load_schema_artifacts(cfg: Config):
+    import yaml
+    from pydantic import ValidationError
+
+    from tht.mschema.models import Annotations, PhysicalSchema
+
+    phys_file = physical_path(cfg)
+    if not phys_file.exists():
+        raise _MachineVectorError("physical_schema_missing")
+    try:
+        return (
+            PhysicalSchema.from_yaml(phys_file),
+            Annotations.from_yaml(annotations_path(cfg)),
+        )
+    except (OSError, UnicodeError, TypeError, ValueError, yaml.YAMLError, ValidationError):
+        raise _MachineVectorError("schema_artifacts_invalid") from None
+
 def index_schema_data(
-    config: Path, *, suppress_legacy_warning: bool = False
+    config: Config | Path, *, suppress_legacy_warning: bool = False, physical=None, annotations=None
 ) -> IndexSchemaResult:
     """Synchronize schema records and return a bounded machine result."""
-    from tht.cli.schema_cmd import _load_schema_config
-    from tht.config import ConfigError
     from tht.mschema.models import Annotations, PhysicalSchema
     from tht.vectorstore.records import schema_records
 
-    try:
-        cfg = _load_schema_config(config, suppress_legacy_warning=suppress_legacy_warning)
-    except ConfigError:
-        raise _MachineVectorError("invalid_configuration") from None
+    if isinstance(config, Path):
+        try:
+            cfg = _load_schema_config(config, suppress_legacy_warning=suppress_legacy_warning)
+        except ConfigError:
+            raise _MachineVectorError("invalid_configuration") from None
+    else:
+        cfg = config
     _vector_write_or_error(cfg)
     _vector_cfg_or_error(cfg)
     phys_file = physical_path(cfg)
@@ -191,8 +215,10 @@ def index_schema_data(
     from pydantic import ValidationError
 
     try:
-        physical = PhysicalSchema.from_yaml(phys_file)
-        annotations = Annotations.from_yaml(annotations_path(cfg))
+        if physical is None:
+            physical = PhysicalSchema.from_yaml(phys_file)
+        if annotations is None:
+            annotations = Annotations.from_yaml(annotations_path(cfg))
     except (OSError, UnicodeError, TypeError, ValueError, yaml.YAMLError, ValidationError):
         raise _MachineVectorError("schema_artifacts_invalid") from None
     records = schema_records(physical, annotations)
@@ -226,29 +252,35 @@ def index_schema_cmd(
 
     if json_output:
         try:
-            payload = index_schema_data(config, suppress_legacy_warning=True)
-        except _MachineVectorError as error:
+            cfg = _load_schema_config(config, suppress_legacy_warning=True)
+        except ConfigError:
+            typer.echo(json.dumps({"status": "failed", "code": "invalid_configuration"}, sort_keys=True, separators=(",", ":")))
+            raise typer.Exit(code=1) from None
+    else:
+        cfg = _load_config_or_exit(config)
+    try:
+        physical, annotations = _load_schema_artifacts(cfg)
+        payload = index_schema_data(cfg, physical=physical, annotations=annotations)
+    except _MachineVectorError as error:
+        if json_output:
             typer.echo(json.dumps({"status": "failed", "code": error.code}, sort_keys=True, separators=(",", ":")))
             raise typer.Exit(code=1) from None
-        except Exception:  # noqa: BLE001 - JSON CLI boundary
-            typer.echo(json.dumps({"status": "failed", "code": "schema_index_failed"}, sort_keys=True, separators=(",", ":")))
-            raise typer.Exit(code=1) from None
-        typer.echo(json.dumps(payload, sort_keys=True, separators=(",", ":")))
-        return
-    try:
-        payload = index_schema_data(config, suppress_legacy_warning=False)
-    except _MachineVectorError as error:
-        _render_index_schema_error(error, config)
+        _render_index_schema_error(error, cfg)
     except Exception:
         logger.exception("Schema indexing failed")
+        if json_output:
+            typer.echo(json.dumps({"status": "failed", "code": "schema_index_failed"}, sort_keys=True, separators=(",", ":")))
+            raise typer.Exit(code=1) from None
         typer.secho("ERRORE: impossibile indicizzare lo schema.", fg=typer.colors.RED, err=True)
         raise typer.Exit(code=1) from None
+    if json_output:
+        typer.echo(json.dumps(payload, sort_keys=True, separators=(",", ":")))
+        return
     _print_stats(payload["counts"])
 
 
-def _render_index_schema_error(error: _MachineVectorError, config: Path) -> None:
+def _render_index_schema_error(error: _MachineVectorError, cfg) -> None:
     """Render expected failures without changing the old human CLI messages."""
-    cfg = _load_config_or_exit(config)
     if error.code == "physical_schema_missing":
         typer.secho(
             f"ERRORE: {physical_path(cfg)} non trovato. Esegui prima `tht schema introspect`.",