From 99a200c9d2faf6794d974d7360537c4eeb99f96a Mon Sep 17 00:00:00 2001
From: Chris McDonough <chrism@plope.com>
Date: Mon, 22 Jun 2026 14:05:56 -0400
Subject: [PATCH] Restore None guard, add thread-safety test for _chunk_sync

---
 .../haiku/rag/chunkers/docling_local.py       |  3 ++
 tests/test_chunker.py                         | 28 +++++++++++++++++++
 2 files changed, 31 insertions(+)

diff --git a/haiku_rag_slim/haiku/rag/chunkers/docling_local.py b/haiku_rag_slim/haiku/rag/chunkers/docling_local.py
index 9ab36b88..42bfaffe 100644
--- a/haiku_rag_slim/haiku/rag/chunkers/docling_local.py
+++ b/haiku_rag_slim/haiku/rag/chunkers/docling_local.py
@@ -177,4 +177,7 @@ class DoclingLocalChunker(DocumentChunker):
         Returns:
             List of Chunk containing content and structured metadata.
         """
+        if document is None:
+            return []
+
         return await asyncio.to_thread(self._chunk_sync, document)
diff --git a/tests/test_chunker.py b/tests/test_chunker.py
index e59681a0..3debf339 100644
--- a/tests/test_chunker.py
+++ b/tests/test_chunker.py
@@ -55,6 +55,34 @@ async def test_local_chunker(qa_corpus: list[dict[str, str]]):
     assert abs(total_tokens - original_tokens) <= original_tokens * 0.1
 
 
+@pytest.mark.asyncio
+async def test_local_chunker_runs_off_event_loop_thread():
+    """Chunking is CPU-bound; verify it runs in a worker thread."""
+    import threading
+
+    chunker = DoclingLocalChunker()
+    event_loop_thread = threading.current_thread()
+    called_from: list[threading.Thread] = []
+
+    original = chunker._chunk_sync
+
+    def recording_chunk_sync(document):
+        called_from.append(threading.current_thread())
+        return original(document)
+
+    chunker._chunk_sync = recording_chunk_sync
+
+    converter = get_converter(Config)
+    doc = await converter.convert_text("# Hello\n\nWorld", name="test.md")
+    await chunker.chunk(doc)
+
+    assert called_from, "_chunk_sync was never called"
+    assert called_from[0] is not event_loop_thread, (
+        "_chunk_sync ran on the event-loop thread; "
+        "it must be dispatched via asyncio.to_thread"
+    )
+
+
 @pytest.mark.asyncio
 async def test_local_chunker_custom_config():
     """Test DoclingLocalChunker with custom configuration."""