fix(evidence): retain referenced vector generations

2026-07-12 05:34:47 +02:00
parent 3a9808f126
commit 92b1d3ccc3
7 changed files with 127 additions and 28 deletions
@@ -137,3 +137,30 @@ def test_owned_copy_uses_validated_descriptor_bytes_when_source_is_replaced(tmp_
    owned = store.materialize_document(document.document_id, tmp_path / "session" / "evidence.md")
    assert owned.read_text() == content
    assert hashlib.sha256(owned.read_bytes()).hexdigest() == document.content_hash.removeprefix("sha256:")
+
+
+def test_materialized_snapshot_uses_identified_manifest_when_active_changes(tmp_path):
+    from tht.corpus.models import CanonicalDocument
+    import hashlib
+
+    def doc(content, fingerprint):
+        return CanonicalDocument(
+            document_id="doc:" + hashlib.sha256(content.encode()).hexdigest(),
+            source_id="fs:item", source_uri="file:///item",
+            source_fingerprint="sha256:" + fingerprint * 64,
+            content_hash="sha256:" + hashlib.sha256(content.encode()).hexdigest(),
+            content=content, pipeline_version="evidence-v1",
+        )
+
+    store = CorpusStore(tmp_path / "corpus")
+    old = doc("old", "a")
+    old_generation = store.stage(CorpusManifest(documents=(old,)), {old.document_id: old.content})
+    store.publish(old_generation)
+    snapshot = store.active_manifest()
+    new = doc("new", "b")
+    new_generation = store.stage(CorpusManifest(documents=(new,)), {new.document_id: new.content})
+    store.publish(new_generation)
+    path = store.materialize_document(
+        snapshot.documents[0].document_id, tmp_path / "owned.md", generation=snapshot.manifest_id,
+    )
+    assert path.read_text() == "old"