From 247df097c563f4f5a899b2cd88a108aa4da2a175 Mon Sep 17 00:00:00 2001
From: Dev Chiniwala <dev.chiniwala@gmail.com>
Date: Thu, 17 Sep 2026 16:26:56 +0530
Subject: [PATCH] fix: remove stale chunks for empty reingest

---
 datamind/capabilities/ingest/service.py |  8 +++++++
 datamind/tests/test_ingest_revisions.py | 28 +++++++++++++++++++++++++
 2 files changed, 36 insertions(+)

diff --git a/datamind/capabilities/ingest/service.py b/datamind/capabilities/ingest/service.py
index b04387e..9044ea7 100644
--- a/datamind/capabilities/ingest/service.py
+++ b/datamind/capabilities/ingest/service.py
@@ -725,6 +725,9 @@ async def kb_add_file(self, *, path: str, copy_to_profile: bool = True) -> dict[
             ))
 
         if not chunks:
+            await self._upsert_chunks(
+                chunks, replace_sources={source, str(resolved)},
+            )
             return {"file": str(resolved), "chunks_added": 0, "note": "file was empty"}
 
         parsed_chunks = None
@@ -830,6 +833,11 @@ async def _upsert_chunks(
                 }
                 if recorded_sources & replace_sources and chunk_id not in new_ids:
                     stale_ids.append(chunk_id)
+        if not chunks:
+            if stale_ids:
+                await store.delete(stale_ids)
+            await self._kb.record_incremental_ingest()
+            return
         await store.add(
             ids=[c.id for c in chunks],
             texts=texts,
diff --git a/datamind/tests/test_ingest_revisions.py b/datamind/tests/test_ingest_revisions.py
index bcf4ed7..31adad5 100644
--- a/datamind/tests/test_ingest_revisions.py
+++ b/datamind/tests/test_ingest_revisions.py
@@ -43,6 +43,34 @@ async def generate_text(self, prompt, **kwargs):
         return "[]"
 
 
+@pytest.mark.asyncio
+async def test_reingesting_empty_path_removes_old_kb_chunks(tmp_path: Path):
+    profile = tmp_path / "profile"
+    uploads = profile / "uploads"
+    uploads.mkdir(parents=True)
+    source = uploads / "release_note.txt"
+    kb = _KB()
+    service = IngestService(
+        kb=kb,
+        db=None,
+        graph=None,
+        llm_client=_Model(),
+        llm_model="test",
+        profile_data_dir=profile,
+        chunk_size=512,
+        chunk_overlap=64,
+    )
+
+    source.write_text("old release note", encoding="utf-8")
+    await service.kb_add_file(path=str(source))
+    source.write_text("\n\n", encoding="utf-8")
+
+    result = await service.kb_add_file(path=str(source))
+
+    assert result["chunks_added"] == 0
+    assert kb.vector_store.items == {}
+
+
 @pytest.mark.asyncio
 async def test_reingesting_same_path_replaces_old_kb_chunks(tmp_path: Path):
     profile = tmp_path / "profile"