fix(preprocess): initialize fresh DWH writers safely

This commit is contained in:
2026-07-12 07:07:31 +02:00
parent 22e806a41b
commit 4cf0adbdd2
5 changed files with 190 additions and 70 deletions
+55
View File
@@ -97,6 +97,61 @@ def test_writer_rejects_unbound_legacy_artifacts_before_building(tmp_path):
assert not (tmp_path / ".tht-dwh" / "OWNER.json").exists()
def test_writer_rejects_dangling_legacy_symlinks_before_claim_or_callback(tmp_path):
import pytest
legacy = tmp_path / "artifacts" / "mschema" / "physical.yaml"
legacy.parent.mkdir(parents=True)
legacy.symlink_to(tmp_path / "missing-catalog")
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_publication_remains_on_locked_root_when_path_is_swapped(
monkeypatch, tmp_path,
):
import os
import pytest
import tht.jobs.dwh_pipeline as module
real_replace = module.os.replace
moved = tmp_path / "locked-root"
replacement = tmp_path / ".tht-dwh"
swapped = False
def swapping_replace(source, destination, *args, **kwargs):
nonlocal swapped
if destination == "OWNER.json" and kwargs.get("dst_dir_fd") is not None:
swapped = True
replacement.rename(moved)
replacement.mkdir(mode=0o700)
return real_replace(source, destination, *args, **kwargs)
monkeypatch.setattr(module.os, "replace", swapping_replace)
pipeline = DwhPreprocessPipeline(
workspace_id="demo", workspace_root=tmp_path,
config_fingerprint=FP, input_fingerprint=FP,
introspect=lambda output: (_ for _ in ()).throw(AssertionError("callback called")),
build_lsh=lambda physical, output: None,
)
with pytest.raises(Exception):
pipeline.run()
assert swapped
assert (moved / "OWNER.json").is_file()
assert not (replacement / "OWNER.json").exists()
assert os.stat(moved / "generation.lock").st_ino != os.stat(replacement).st_ino
def test_owner_requires_exact_read_only_owner_mode_and_active_requires_binding(tmp_path):
import pytest