Python: fix(core): resolve unmigrated legacy memory topics in the index

list_topics parses a pre-digest file such as caf.md headed "# café" and returns the current digested slug, but the file on disk still uses the legacy name. before_run then calls get_topic(entry.slug) for the selected entry, where neither the digested path nor a legacy candidate derivable from that slug exists, so memory preparation dies with FileNotFoundError. Consolidation silently skips the topic the same way. A record whose file still sits at the pre-digest name for its topic is now keyed by the on-disk stem, so MEMORY.md pointers and get_topic(entry.slug) resolve to the persisted file before a rewrite migrates it. The heading identity check on the fallback paths is unchanged: a same-stem file owned by a different topic still does not alias. Verified with the full agent-framework-core suite: 8483 tests, 0 failures; ruff and pyright clean on the changed files. Signed-off-by: Yufeng He <40085740+he-yufeng@users.noreply.github.com>
Yufeng He (he-yufeng) · Oct 9, 2026 · Oct 9, 2026 · Oct 9, 2026 · Oct 12, 2026 · Oct 16, 2026
commit 1ba70dbd4afd82d878eb9f1eed652520c8619c6b
@@ -866,6 +866,11 @@ def list_topics(self, session: AgentSession, *, source_id: str) -> list[MemoryTo
                 topic_path.read_text(encoding="utf-8"),
                 fallback_topic=topic_path.stem.replace("-", " "),
             )
+            if topic_path.stem == _legacy_slugify_topic(record.topic) and topic_path.stem != record.slug:
+                # Not yet migrated: the file still sits at its pre-digest name. Key the
+                # record by the stem that exists on disk so index pointers and
+                # get_topic(entry.slug) resolve to it before a rewrite migrates it.
+                record.slug = topic_path.stem
             topics.append(record)
         return sorted(topics, key=lambda record: (record.topic.lower(), record.updated_at))
 

@@ -1131,4 +1131,47 @@ def test_memory_file_store_legacy_fallback_requires_matching_stored_topic(tmp_pa
     assert store.get_topic(session, source_id=DEFAULT_MEMORY_SOURCE_ID, topic="café").memories == ["legacy accent fact"]
 
 
+def test_memory_file_store_lists_pre_digest_topic_under_its_on_disk_stem(tmp_path) -> None:
+    """An unmigrated legacy file must be listed under the stem that exists on disk."""
+    store, session = _topic_store(tmp_path)
+    legacy_path = store._get_topics_directory(session, source_id=DEFAULT_MEMORY_SOURCE_ID) / "caf.md"
+    legacy_path.parent.mkdir(parents=True, exist_ok=True)
+    legacy_path.write_text(f"{_record('café', ['legacy accent fact']).to_markdown()}\n", encoding="utf-8")
+
+    (entry,) = store.list_topics(session, source_id=DEFAULT_MEMORY_SOURCE_ID)
+
+    assert entry.topic == "café"
+    assert entry.slug == "caf"
+    assert store.get_topic(session, source_id=DEFAULT_MEMORY_SOURCE_ID, topic=entry.slug).memories == [
+        "legacy accent fact"
+    ]
+    store.rebuild_index(session, source_id=DEFAULT_MEMORY_SOURCE_ID, line_limit=50, line_length=200)
+    index_text = store._get_index_path(session, source_id=DEFAULT_MEMORY_SOURCE_ID).read_text(encoding="utf-8")
+    assert "topics/caf.md" in index_text
+    # Keying by the on-disk stem must not widen the fallback: a same-stem but
+    # different topic still does not alias this file.
+    with pytest.raises(FileNotFoundError):
+        store.get_topic(session, source_id=DEFAULT_MEMORY_SOURCE_ID, topic="cafè")
+
+
+async def test_memory_context_provider_loads_unmigrated_legacy_topic(tmp_path) -> None:
+    """``before_run`` must resolve an index entry whose file still uses its pre-digest name."""
+    store, session = _topic_store(tmp_path)
+    legacy_path = store._get_topics_directory(session, source_id=DEFAULT_MEMORY_SOURCE_ID) / "caf.md"
+    legacy_path.parent.mkdir(parents=True, exist_ok=True)
+    legacy_path.write_text(f"{_record('café', ['legacy accent fact']).to_markdown()}\n", encoding="utf-8")
+    provider = MemoryContextProvider(store=store, recent_turns=0, max_extractions=0)
+
+    from agent_framework._sessions import SessionContext
+
+    context = SessionContext(
+        session_id=session.session_id,
+        input_messages=[Message(role="user", contents=["tell me the legacy accent fact"])],
+    )
+
+    await provider.before_run(agent=None, session=session, context=context, state={})
+
+    assert any("legacy accent fact" in message.text for message in context.get_messages())
+
+
 # endregion