feat(preprocess): add resumable DWH jobs

harness/tests/test_dwh_preprocess_job.py

@@ -0,0 +1,83 @@
+import hashlib
+import json
+
+from typer.testing import CliRunner
+
+from tht.cli import app
+from tht.jobs.dwh_pipeline import DwhPreprocessPipeline
+from tht.jobs.locking import _lock_name
+
+
+FP = "sha256:" + hashlib.sha256(b"test").hexdigest()
+
+
+def test_dwh_and_evidence_jobs_have_distinct_lock_names():
+    assert _lock_name("demo", "dwh") != _lock_name("demo", "evidence")
+
+
+def test_selected_dwh_stages_run_in_declared_order(tmp_path):
+    calls = []
+    pipeline = DwhPreprocessPipeline(
+        workspace_id="demo",
+        workspace_root=tmp_path,
+        config_fingerprint=FP,
+        input_fingerprint=FP,
+        introspect=lambda: calls.append("introspect"),
+        build_lsh=lambda: calls.append("lsh"),
+    )
+
+    report = pipeline.run(("introspect", "lsh"))
+
+    assert report.status == "succeeded"
+    assert calls == ["introspect", "lsh"]
+    assert [stage.name for stage in report.stages] == ["introspect", "lsh"]
+
+
+def test_preprocess_dwh_json_is_pristine(monkeypatch, tmp_path):
+    import tht.cli.preprocess_cmd as command
+
+    class Report:
+        status = "succeeded"
+
+        def model_dump(self, mode=None):
+            return {"status": "succeeded", "run_id": "a" * 32, "stages": []}
+
+    seen = {}
+
+    def run(config, *, steps, resume):
+        seen.update(config=config, steps=steps, resume=resume)
+        return Report()
+
+    monkeypatch.setattr(command, "run_dwh_from_config", run)
+    response = CliRunner().invoke(
+        app,
+        [
+            "preprocess", "dwh", "--steps", "introspect,lsh", "--json",
+            "-c", str(tmp_path / "workspace.yaml"),
+        ],
+    )
+
+    assert response.exit_code == 0, response.output
+    assert json.loads(response.output)["run_id"] == "a" * 32
+    assert seen["steps"] == ("introspect", "lsh")
+
+
+def test_preprocess_dwh_rejects_unknown_or_duplicate_steps(monkeypatch, tmp_path):
+    import tht.cli.preprocess_cmd as command
+
+    called = False
+
+    def forbidden(*args, **kwargs):
+        nonlocal called
+        called = True
+
+    monkeypatch.setattr(command, "run_dwh_from_config", forbidden)
+    runner = CliRunner()
+    for value in ("introspect,unknown", "lsh,lsh", ""):
+        response = runner.invoke(
+            app,
+            ["preprocess", "dwh", "--steps", value, "--json", "-c", str(tmp_path / "w.yaml")],
+        )
+        assert response.exit_code == 2
+        assert json.loads(response.output)["status"] == "failed"
+    assert called is False

harness/tests/test_lsh_job_resume.py

@@ -0,0 +1,66 @@
+import hashlib
+
+from tht.jobs.dwh_pipeline import DwhPreprocessPipeline
+
+
+FP = "sha256:" + hashlib.sha256(b"test").hexdigest()
+
+
+def test_lsh_failure_resumes_exact_run_without_repeating_introspection(tmp_path):
+    calls = []
+
+    def introspect():
+        calls.append("introspect")
+
+    def fail_lsh():
+        calls.append("lsh-failed")
+        raise RuntimeError("database detail that must not leak")
+
+    failed = DwhPreprocessPipeline(
+        workspace_id="demo",
+        workspace_root=tmp_path,
+        config_fingerprint=FP,
+        input_fingerprint=FP,
+        introspect=introspect,
+        build_lsh=fail_lsh,
+    ).run(("introspect", "lsh"))
+    assert failed.status == "failed"
+
+    resumed = DwhPreprocessPipeline(
+        workspace_id="demo",
+        workspace_root=tmp_path,
+        config_fingerprint=FP,
+        input_fingerprint=FP,
+        introspect=introspect,
+        build_lsh=lambda: calls.append("lsh-recovered"),
+    ).run(("introspect", "lsh"), resume_run_id=failed.run_id)
+
+    assert resumed.status == "succeeded"
+    assert resumed.resumed_from == failed.run_id
+    assert calls == ["introspect", "lsh-failed", "lsh-recovered"]
+
+
+def test_resume_rejects_a_different_stage_selection(tmp_path):
+    failed = DwhPreprocessPipeline(
+        workspace_id="demo",
+        workspace_root=tmp_path,
+        config_fingerprint=FP,
+        input_fingerprint=FP,
+        introspect=lambda: None,
+        build_lsh=lambda: (_ for _ in ()).throw(RuntimeError()),
+    ).run(("introspect", "lsh"))
+
+    pipeline = DwhPreprocessPipeline(
+        workspace_id="demo",
+        workspace_root=tmp_path,
+        config_fingerprint=FP,
+        input_fingerprint=FP,
+        introspect=lambda: None,
+        build_lsh=lambda: None,
+    )
+    try:
+        pipeline.run(("lsh",), resume_run_id=failed.run_id)
+    except Exception as error:
+        assert "incompatible" in str(error)
+    else:
+        raise AssertionError("resume with different stages must fail")