fix(preprocess): restrict DWH root claims to writers

harness/tests/test_dwh_preprocess_job.py

@@ -6,6 +6,7 @@ from typer.testing import CliRunner
 
 from tht.cli import app
 from tht.jobs.dwh_pipeline import DwhPreprocessPipeline
+from tht.jobs.dwh_pipeline import active_generation_dir, config_dwh_binding
 from tht.jobs.dwh_pipeline import resolve_dwh_snapshot
 from tht.jobs.dwh_pipeline import lease_dwh_snapshot
 from tht.jobs.locking import _lock_name
@@ -30,6 +31,94 @@ def test_dwh_and_evidence_jobs_have_distinct_lock_names():
     assert _lock_name("demo", "dwh") != _lock_name("demo", "evidence")
 
 
+def test_unowned_reads_fail_closed_without_creating_any_files(tmp_path):
+    import pytest
+
+    cfg = snapshot_config(tmp_path)
+    with pytest.raises(Exception, match="not initialized"):
+        resolve_dwh_snapshot(cfg)
+    with pytest.raises(Exception, match="not initialized"):
+        with lease_dwh_snapshot(cfg):
+            pass
+    assert not (tmp_path / ".tht-dwh").exists()
+
+
+def test_writer_claim_allows_only_lock_and_empty_generations(tmp_path):
+    import pytest
+
+    for name, make_entry in (
+        ("unexpected", lambda root: (root / "unexpected").write_text("x")),
+        ("stale-temp", lambda root: (root / ".OWNER.json.stale.tmp").write_text("x")),
+        ("unexpected-dir", lambda root: (root / "other").mkdir()),
+    ):
+        root = tmp_path / name / ".tht-dwh"
+        root.mkdir(parents=True, mode=0o700)
+        make_entry(root)
+        calls = []
+        pipeline = DwhPreprocessPipeline(
+            workspace_id="demo", workspace_root=root.parent,
+            config_fingerprint=FP, input_fingerprint=FP,
+            introspect=lambda output: calls.append("called"),
+            build_lsh=lambda physical, output: None,
+        )
+        with pytest.raises(Exception, match="unbound"):
+            pipeline.run()
+        assert calls == []
+        assert not (root / "OWNER.json").exists()
+
+    allowed = tmp_path / "allowed"
+    (allowed / ".tht-dwh" / "generations").mkdir(parents=True, mode=0o700)
+    report = DwhPreprocessPipeline(
+        workspace_id="demo", workspace_root=allowed,
+        config_fingerprint=FP, input_fingerprint=FP,
+        introspect=lambda output: output.write_text("catalog"),
+        build_lsh=lambda physical, output: _write_lsh([], physical, output),
+    ).run()
+    assert report.status == "succeeded"
+
+
+def test_writer_rejects_unbound_legacy_artifacts_before_building(tmp_path):
+    import pytest
+
+    legacy = tmp_path / "artifacts" / "mschema" / "physical.yaml"
+    legacy.parent.mkdir(parents=True)
+    legacy.write_text("legacy")
+    calls = []
+    pipeline = DwhPreprocessPipeline(
+        workspace_id="demo", workspace_root=tmp_path,
+        config_fingerprint=FP, input_fingerprint=FP,
+        introspect=lambda output: calls.append("called"),
+        build_lsh=lambda physical, output: None,
+        current_physical=legacy,
+    )
+    with pytest.raises(Exception, match="legacy artifacts are unbound"):
+        pipeline.run()
+    assert calls == []
+    assert not (tmp_path / ".tht-dwh" / "OWNER.json").exists()
+
+
+def test_owner_requires_exact_read_only_owner_mode_and_active_requires_binding(tmp_path):
+    import pytest
+
+    pipeline = DwhPreprocessPipeline(
+        workspace_id="demo", workspace_root=tmp_path,
+        config_fingerprint=FP, input_fingerprint=FP,
+        introspect=lambda output: output.write_text("catalog"),
+        build_lsh=lambda physical, output: _write_lsh([], physical, output),
+    )
+    pipeline.run()
+    cfg = snapshot_config(tmp_path)
+    binding = config_dwh_binding(cfg)
+    assert active_generation_dir(tmp_path, binding) is not None
+    with pytest.raises(Exception, match="different workspace configuration"):
+        active_generation_dir(tmp_path, {**binding, "workspace_id": "other"})
+
+    marker = tmp_path / ".tht-dwh" / "OWNER.json"
+    marker.chmod(0o440)
+    with pytest.raises(Exception, match="ownership marker"):
+        resolve_dwh_snapshot(cfg)
+
+
 def test_selected_dwh_stages_run_in_declared_order(tmp_path):
     calls = []
     pipeline = DwhPreprocessPipeline(

harness/tests/test_search_pack.py

@@ -5,6 +5,8 @@ from types import SimpleNamespace
 from typer.testing import CliRunner
 
 from tht.cli import app
+from tht.config import load_config
+from tht.jobs.dwh_pipeline import DwhPreprocessPipeline, config_dwh_binding
 from tht.mschema.models import ColumnPhysical, PhysicalSchema, TablePhysical
 from tht.vectorstore.embeddings import EmbeddingsError
 
@@ -43,11 +45,11 @@ class _FakeSearcher:
 
 
 def _workspace(tmp_path, with_session=None):
-    PhysicalSchema(
+    physical = PhysicalSchema(
         database="d", schema="s", introspected_at=datetime(2026, 1, 1),
         tables={"fact_ablazione": TablePhysical(
             comment="Ablazioni", columns={"cod_paz": ColumnPhysical(type="bigint")})},
-    ).to_yaml(tmp_path / "artifacts" / "mschema" / "physical.yaml")
+    )
     cfg = tmp_path / "workspace.yaml"
     cfg.write_text(
         "database: {database: d, schema: s, user: u, password: p, transport: direct}\n"
@@ -56,6 +58,18 @@ def _workspace(tmp_path, with_session=None):
         f"paths: {{artifacts: {tmp_path/'artifacts'}, indexes: {tmp_path/'i'}, "
         f"sessions: {tmp_path/'sessions'}}}\n"
     )
+    binding = config_dwh_binding(load_config(cfg))
+    DwhPreprocessPipeline(
+        workspace_id=binding["workspace_id"], workspace_root=tmp_path,
+        config_fingerprint=binding["config_fingerprint"],
+        input_fingerprint=binding["input_fingerprint"],
+        introspect=lambda output: physical.to_yaml(output),
+        build_lsh=lambda _physical, output: [
+            (output / name).write_text("index")
+            for name in ("s_lsh.pkl", "s_minhashes.pkl", "s_meta.json")
+        ],
+        lsh_filenames=("s_lsh.pkl", "s_minhashes.pkl", "s_meta.json"),
+    ).run()
     if with_session:
         sdir = tmp_path / "sessions" / with_session
         sdir.mkdir(parents=True)