fix: panic if all docs are deleted in a posting list (#3163)

lancedb · Nov 25, 2024 · 196ec06 · 196ec06
1 parent d196ab8
commit 196ec06
Show file tree

Hide file tree

Showing 2 changed files with 8 additions and 1 deletion.
diff --git a/python/python/tests/test_scalar_index.py b/python/python/tests/test_scalar_index.py
@@ -300,6 +300,13 @@ def test_fts_with_other_str_scalar_index(dataset):
     assert dataset.to_table(full_text_query=query).num_rows > 0
 
 
+def test_fts_all_deleted(dataset):
+    dataset.create_scalar_index("doc", index_type="INVERTED", with_position=False)
+    first_row_doc = dataset.take(indices=[0], columns=["doc"]).column(0)[0].as_py()
+    dataset.delete(f"doc = '{first_row_doc}'")
+    dataset.to_table(full_text_query=first_row_doc)
+
+
 def test_bitmap_index(tmp_path: Path):
     """Test create bitmap index"""
     tbl = pa.Table.from_arrays(

diff --git a/rust/lance-index/src/scalar/inverted/wand.rs b/rust/lance-index/src/scalar/inverted/wand.rs
@@ -127,7 +127,7 @@ impl Wand {
             threshold: 0.0,
             cur_doc: None,
             num_docs,
-            postings: postings.collect(),
+            postings: postings.filter(|posting| posting.doc().is_some()).collect(),
             candidates: BinaryHeap::new(),
         }
     }