fix(preprocess): lease DWH generation reads

harness/tests/test_dwh_preprocess_job.py

@@ -7,6 +7,7 @@ from typer.testing import CliRunner
 from tht.cli import app
 from tht.jobs.dwh_pipeline import DwhPreprocessPipeline
 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
 
 
@@ -228,3 +229,67 @@ def test_generation_retention_keeps_active_and_one_rollback(tmp_path):
         run_ids.append(report.run_id)
     remaining = {path.name for path in (tmp_path / ".tht-dwh" / "generations").iterdir()}
     assert remaining == set(run_ids[-2:])
+
+
+def test_reader_lease_blocks_retain_one_publisher_until_file_reads_finish(tmp_path):
+    import threading
+    import time
+    from types import SimpleNamespace
+
+    def make(content, retain=1):
+        return DwhPreprocessPipeline(
+            workspace_id="demo", workspace_root=tmp_path,
+            config_fingerprint=FP, input_fingerprint=FP,
+            introspect=lambda output: output.write_text(content),
+            build_lsh=lambda physical, output: [
+                (output / name).write_text(content)
+                for name in ("demo_lsh.pkl", "demo_minhashes.pkl", "demo_meta.json")
+            ], retain_generations=retain,
+        )
+
+    first = make("old").run()
+    cfg = SimpleNamespace(paths=SimpleNamespace(
+        artifacts=tmp_path / "artifacts", indexes=tmp_path / "indexes"
+    ))
+    completed = threading.Event()
+    with lease_dwh_snapshot(cfg) as snapshot:
+        thread = threading.Thread(target=lambda: (make("new").run(), completed.set()))
+        thread.start()
+        time.sleep(0.05)
+        assert not completed.is_set()
+        assert snapshot.physical.read_text() == "old"
+        assert snapshot.generation == first.run_id
+    thread.join(timeout=2)
+    assert completed.is_set()
+    assert not (tmp_path / ".tht-dwh" / "generations" / first.run_id).exists()
+
+
+def test_cleanup_never_follows_top_level_or_child_symlinks(tmp_path):
+    external = tmp_path / "external"
+    external.mkdir()
+    victim = external / "victim"
+    victim.write_text("safe")
+
+    generations = tmp_path / ".tht-dwh" / "generations"
+    generations.mkdir(parents=True)
+    (generations / ("a" * 32)).symlink_to(external, target_is_directory=True)
+
+    def make(content, retain=1):
+        return DwhPreprocessPipeline(
+            workspace_id="demo", workspace_root=tmp_path,
+            config_fingerprint=FP, input_fingerprint=FP,
+            introspect=lambda output: output.write_text(content),
+            build_lsh=lambda physical, output: [
+                (output / name).write_text(content)
+                for name in ("demo_lsh.pkl", "demo_minhashes.pkl", "demo_meta.json")
+            ], retain_generations=retain,
+        )
+
+    first = make("one", retain=2).run()
+    make("two", retain=2).run()
+    old = generations / first.run_id
+    old.chmod(0o700)
+    (old / "hostile-link").symlink_to(victim)
+    make("three").run()
+    assert victim.read_text() == "safe"
+    assert victim.stat().st_mode & 0o200