fix(preprocess): restrict DWH root claims to writers

harness/tht/cli/lsh_cmd.py

@@ -77,9 +77,9 @@ def build_cmd(config: Path = CONFIG_OPT) -> None:
         )
         raise typer.Exit(code=1)
     typer.echo("Estrazione valori (i più frequenti) dalle colonne testuali eligible...")
-    from tht.jobs.dwh_pipeline import active_generation_dir
+    from tht.jobs.dwh_pipeline import resolve_dwh_snapshot
 
-    if active_generation_dir(cfg.paths.artifacts.parent) is None:
+    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())

harness/tht/cli/preprocess_cmd.py

@@ -19,15 +19,14 @@ def run_dwh_from_config(
     config: Path, *, steps: tuple[str, ...], resume: str | None = None,
 ):
     from tht.cli.lsh_cmd import build_lsh_artifacts
-    from tht.cli.schema_cmd import _load_config_or_exit, physical_path, refresh_catalog
+    from tht.cli.schema_cmd import _load_config_or_exit, refresh_catalog
     from tht.jobs.dwh_pipeline import (
-        DwhPreprocessPipeline, active_generation_dir, config_dwh_binding,
+        DwhPreprocessPipeline, config_dwh_binding,
     )
 
     cfg = _load_config_or_exit(config)
     binding = config_dwh_binding(cfg)
     workspace_root = cfg.paths.artifacts.parent
-    active = active_generation_dir(workspace_root)
     lsh_names = (
         f"{cfg.database.db_schema}_lsh.pkl",
         f"{cfg.database.db_schema}_minhashes.pkl",
@@ -43,8 +42,8 @@ def run_dwh_from_config(
             cfg, physical_file=physical, output_dir=output
         ),
         lsh_filenames=lsh_names,
-        current_physical=physical_path(cfg),
-        current_lsh_dir=active if active is not None else cfg.paths.indexes / "lsh",
+        current_physical=cfg.paths.artifacts / "mschema" / "physical.yaml",
+        current_lsh_dir=cfg.paths.indexes / "lsh",
     )
     return pipeline.run(steps, resume_run_id=resume)
 

harness/tht/cli/schema_cmd.py

@@ -39,6 +39,8 @@ def _load_config_or_exit(config: Path):
 def physical_path(cfg) -> Path:
     from tht.jobs.dwh_pipeline import resolve_dwh_snapshot
 
+    if not (cfg.paths.artifacts.parent / ".tht-dwh").exists():
+        return cfg.paths.artifacts / "mschema" / "physical.yaml"
     return resolve_dwh_snapshot(cfg).physical
 
 
@@ -92,9 +94,9 @@ def introspect_cmd(
             )
             return
     try:
-        from tht.jobs.dwh_pipeline import active_generation_dir
+        from tht.jobs.dwh_pipeline import resolve_dwh_snapshot
 
-        if active_generation_dir(cfg.paths.artifacts.parent) is None:
+        if resolve_dwh_snapshot(cfg).generation is None:
             phys = refresh_catalog(cfg)
         else:
             from tht.cli.preprocess_cmd import run_dwh_from_config