fix(preprocess): initialize fresh DWH writers safely

harness/tht/cli/lsh_cmd.py

@@ -69,32 +69,30 @@ def build_lsh_artifacts(
 def build_cmd(config: Path = CONFIG_OPT) -> None:
     """Costruisce l'indice LSH dai valori del database e lo salva su pickle."""
     cfg = _load_config_or_exit(config)
-    phys_file = physical_path(cfg)
-    if not phys_file.exists():
-        typer.secho(
-            f"ERRORE: {phys_file} non trovato. Esegui prima `tht schema introspect`.",
-            fg=typer.colors.RED, err=True,
-        )
-        raise typer.Exit(code=1)
-    typer.echo("Estrazione valori (i più frequenti) dalle colonne testuali eligible...")
-    from tht.jobs.dwh_pipeline import resolve_dwh_snapshot
-
-    if resolve_dwh_snapshot(cfg).generation is None:
-        minhashes, skipped, truncated, values = build_lsh_artifacts(cfg, verbose=True)
-        n_values = sum(len(v) for table in values.values() for v in table.values())
-        n_columns = sum(len(table) for table in values.values())
-    else:
-        from tht.cli.preprocess_cmd import run_dwh_from_config
-        from tht.lshindex import load_index
-
-        report = run_dwh_from_config(config, steps=("lsh",))
-        if report.status != "succeeded":
-            typer.secho("ERRORE: DWH preprocessing failed", fg=typer.colors.RED, err=True)
+    dwh_root = cfg.paths.artifacts.parent / ".tht-dwh"
+    initialized = dwh_root.exists() or dwh_root.is_symlink()
+    if initialized:
+        phys_file = physical_path(cfg)
+        if not phys_file.exists():
+            typer.secho(
+                f"ERRORE: {phys_file} non trovato. Esegui prima `tht schema introspect`.",
+                fg=typer.colors.RED, err=True,
+            )
             raise typer.Exit(code=1)
-        _, minhashes, _ = load_index(_lsh_dir(cfg), name=cfg.database.db_schema)
-        skipped, truncated = [], []
-        n_values = len(minhashes)
-        n_columns = len({(entry[1], entry[2]) for entry in minhashes.values()})
+    typer.echo("Estrazione valori (i più frequenti) dalle colonne testuali eligible...")
+    from tht.cli.preprocess_cmd import run_dwh_from_config
+    from tht.lshindex import load_index
+
+    report = run_dwh_from_config(
+        config, steps=("lsh",) if initialized else ("introspect", "lsh")
+    )
+    if report.status != "succeeded":
+        typer.secho("ERRORE: DWH preprocessing failed", fg=typer.colors.RED, err=True)
+        raise typer.Exit(code=1)
+    _, minhashes, _ = load_index(_lsh_dir(cfg), name=cfg.database.db_schema)
+    skipped, truncated = [], []
+    n_values = len(minhashes)
+    n_columns = len({(entry[1], entry[2]) for entry in minhashes.values()})
     typer.echo(f"  {n_values} valori da {n_columns} colonne")
     for s in skipped:
         typer.secho(f"  saltata {s.table}.{s.column}: {s.reason}", fg=typer.colors.YELLOW)