fix(evidence): bind corpus to workspace identity

2026-07-12 05:57:23 +02:00
parent 78ff360882
commit db35ddb041
6 changed files with 109 additions and 10 deletions
@@ -329,6 +329,66 @@ def test_pipeline_result_dump_does_not_deepcopy_frozen_metadata():
    assert payload["manifest"]["metadata"] == {"nested": {"value": ["safe"]}}


+def test_reused_corpus_root_rejects_workspace_rename_before_any_mutation(tmp_path):
+    vectors = Vectors()
+    first = pipeline(tmp_path, Source([(item("one", "a"), "stable")]), vectors=vectors)
+    first.run_as_job(
+        workspace_id="workspace-a", workspace_root=tmp_path,
+        config_fingerprint="sha256:" + "1" * 64,
+        input_fingerprint="sha256:" + "2" * 64,
+    )
+    active = first.store.active_generation()
+    records = list(vectors.records)
+    renamed_source = Source([(item("one", "a"), "stable")])
+    renamed = pipeline(tmp_path, renamed_source, vectors=vectors)
+    with pytest.raises(PipelineError, match="different workspace"):
+        renamed.run_as_job(
+            workspace_id="workspace-b", workspace_root=tmp_path,
+            config_fingerprint="sha256:" + "1" * 64,
+            input_fingerprint="sha256:" + "2" * 64,
+        )
+    assert renamed_source.acquire_calls == []
+    assert renamed.store.active_generation() == active
+    assert vectors.records == records
+
+
+def test_gc_rejects_workspace_mismatch_without_deleting(tmp_path):
+    vectors = Vectors()
+    owner = pipeline(tmp_path, Source([(item("one", "a"), "stable")]), vectors=vectors)
+    owner.run_as_job(
+        workspace_id="workspace-a", workspace_root=tmp_path,
+        config_fingerprint="sha256:" + "1" * 64,
+        input_fingerprint="sha256:" + "2" * 64,
+    )
+    generations = owner.store.list_generations()
+    wrong = pipeline(tmp_path, Source([]), vectors=vectors)
+    wrong.workspace_id = "workspace-b"
+    with pytest.raises(PipelineError, match="different workspace"):
+        wrong.gc(workspace_root=tmp_path)
+    assert wrong.store.list_generations() == generations
+
+
+def test_active_search_rejects_workspace_mismatch_before_delegate(tmp_path):
+    from tht.search.evidence import ActiveEvidenceSearcher, CorpusWorkspaceMismatchError
+
+    vectors = Vectors()
+    owner = pipeline(tmp_path, Source([(item("one", "a"), "stable")]), vectors=vectors)
+    owner.run_as_job(
+        workspace_id="workspace-a", workspace_root=tmp_path,
+        config_fingerprint="sha256:" + "1" * 64,
+        input_fingerprint="sha256:" + "2" * 64,
+    )
+
+    class Delegate:
+        def search(self, *args, **kwargs):
+            raise AssertionError("workspace mismatch reached vector delegate")
+
+    with pytest.raises(CorpusWorkspaceMismatchError, match="different workspace"):
+        ActiveEvidenceSearcher(
+            owner.store, Delegate(), expected_workspace_id="workspace-b",
+        ).search([1.0], kinds=["evidence"])
+
+
 def test_unchanged_documents_skip_acquire_normalize_chunk_and_embed(tmp_path):
    one = item("one", "a")
    first_source = Source([(one, "hello")])