fix(rust, python): fix anonymous list builder (#6916)

pola-rs · Feb 16, 2023 · c4267fc · c4267fc
1 parent a507e1b
commit c4267fc
Show file tree

Hide file tree

Showing 3 changed files with 27 additions and 5 deletions.
diff --git a/polars/polars-arrow/src/array/list.rs b/polars/polars-arrow/src/array/list.rs
@@ -60,9 +60,7 @@ impl<'a> AnonymousBuilder<'a> {
             self.arrays.push(arr.as_ref());
         }
         self.offsets.push(self.size);
-        if let Some(validity) = &mut self.validity {
-            validity.push(true)
-        }
+        self.update_validity()
     }
 
     pub fn push_null(&mut self) {
@@ -75,6 +73,7 @@ impl<'a> AnonymousBuilder<'a> {
 
     pub fn push_empty(&mut self) {
         self.offsets.push(self.last_offset());
+        self.update_validity()
     }
 
     fn init_validity(&mut self) {
@@ -85,11 +84,15 @@ impl<'a> AnonymousBuilder<'a> {
         validity.set(len - 1, false);
         self.validity = Some(validity)
     }
+    fn update_validity(&mut self) {
+        if let Some(validity) = &mut self.validity {
+            validity.push(true)
+        }
+    }
 
     pub fn finish(self, inner_dtype: Option<&DataType>) -> Result<ListArray<i64>> {
         let inner_dtype = inner_dtype.unwrap_or_else(|| self.arrays[0].data_type());
         let values = concatenate::concatenate(&self.arrays)?;
-
         let dtype = ListArray::<i64>::default_datatype(inner_dtype.clone());
         // Safety:
         // offsets are monotonically increasing

diff --git a/py-polars/tests/unit/test_df.py b/py-polars/tests/unit/test_df.py
@@ -1452,7 +1452,7 @@ def test_unique_unit_rows() -> None:
 
 
 def test_panic() -> None:
-    # may contain some tests that yielded a panic in polars or arrow
+    # may contain some tests that yielded a panic in polars or pl_arrow
     # https://github.com/pola-rs/polars/issues/1110
     a = pl.DataFrame(
         {

diff --git a/py-polars/tests/unit/test_explode.py b/py-polars/tests/unit/test_explode.py
@@ -222,3 +222,22 @@ def test_explode_inner_lists_3985() -> None:
         .agg(pl.col("categories"))
         .with_columns(pl.col("categories").arr.eval(pl.element().arr.explode()))
     ).collect().to_dict(False) == {"id": [1], "categories": [["a", "b", "a", "c"]]}
+
+
+def test_list_struct_explode_6905() -> None:
+    assert pl.DataFrame(
+        {
+            "group": [
+                [],
+                [
+                    {"params": [1]},
+                    {"params": []},
+                ],
+            ]
+        },
+        schema={"group": pl.List(pl.Struct([pl.Field("params", pl.List(pl.Int32))]))},
+    )["group"].arr.explode().to_list() == [
+        {"params": None},
+        {"params": [1]},
+        {"params": []},
+    ]