From b5d35610664b8bc3e68709e0e62a109a0eb21519 Mon Sep 17 00:00:00 2001
From: Minh Vu <vuhoangminh97@gmail.com>
Date: Mon, 2 Nov 2026 12:19:13 +0100
Subject: [PATCH] fix(python): reject empty multivector reranker input

Reject empty multivector result lists with `ValueError` before accessing the first result, matching the existing RRF and MRR behavior. Non-empty lists, zero-row tables, and both deduplication modes retain their existing behavior.

Regression coverage exercises the shared base reranker through both deduplication settings. The exact-head Python suite passed with 3021 passed, 126 skipped, and 70 deselected.
---
 python/python/lancedb/rerankers/base.py |  2 ++
 python/python/tests/test_rerankers.py   | 14 ++++++++++++++
 2 files changed, 16 insertions(+)

diff --git a/python/python/lancedb/rerankers/base.py b/python/python/lancedb/rerankers/base.py
index e1bc7f0dd..966f23eeb 100644
--- a/python/python/lancedb/rerankers/base.py
+++ b/python/python/lancedb/rerankers/base.py
@@ -189,6 +189,8 @@ class Reranker(ABC):
         vector_results = (
             [vector_results] if not isinstance(vector_results, list) else vector_results
         )
+        if not vector_results:
+            raise ValueError("vector_results must not be empty")
 
         # Make sure all elements are of the same type
         if not all(isinstance(v, type(vector_results[0])) for v in vector_results):
diff --git a/python/python/tests/test_rerankers.py b/python/python/tests/test_rerankers.py
index 610f533f5..eb7705f80 100644
--- a/python/python/tests/test_rerankers.py
+++ b/python/python/tests/test_rerankers.py
@@ -28,6 +28,7 @@ from lancedb.rerankers import (
     WatsonxReranker,
 )
 from lancedb.table import LanceTable
+from lancedb.rerankers.base import Reranker
 
 
 def get_test_table(tmp_path):
@@ -358,6 +359,19 @@ def test_mrr_reranker(tmp_path):
     assert len(result_deduped) == len(result)
 
 
+@pytest.mark.parametrize("deduplicate", [False, True])
+def test_base_reranker_multivector_empty_input(deduplicate):
+    class VectorReranker(Reranker):
+        def rerank_hybrid(self, query, vector_results, fts_results):
+            raise AssertionError("should not rerank")
+
+        def rerank_vector(self, query, vector_results):
+            raise AssertionError("should not rerank")
+
+    with pytest.raises(ValueError, match="vector_results must not be empty"):
+        VectorReranker().rerank_multivector([], None, deduplicate=deduplicate)
+
+
 def test_mrr_reranker_empty_input():
     reranker = MRRReranker()
     with pytest.raises(ValueError, match="must not be empty"):