fix(python): fix from_dict schema_inference=0 (#5948)

pola-rs · Dec 30, 2022 · 24d4209 · 24d4209
1 parent 11a2e14
commit 24d4209
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 4 deletions.
diff --git a/py-polars/polars/datatypes.py b/py-polars/polars/datatypes.py
@@ -11,7 +11,6 @@
     TYPE_CHECKING,
     Any,
     Callable,
-    Dict,
     ForwardRef,
     Mapping,
     Optional,
@@ -86,7 +85,7 @@ def get_args(tp: Any) -> Any:
     Mapping[str, Union[PolarsDataType, PythonDataType]],
     Sequence[Tuple[str, Union[PolarsDataType, PythonDataType, None]]],
 ]
-Schema: TypeAlias = Dict[str, PolarsDataType]
+Schema: TypeAlias = Mapping[str, PolarsDataType]
 
 DTYPE_TEMPORAL_UNITS: frozenset[TimeUnit] = frozenset(["ns", "us", "ms"])
 

diff --git a/py-polars/src/dataframe.rs b/py-polars/src/dataframe.rs
@@ -441,7 +441,8 @@ impl PyDataFrame {
         infer_schema_length: Option<usize>,
         schema_overwrite: Option<Wrap<Schema>>,
     ) -> PyResult<Self> {
-        let (rows, mut names) = dicts_to_rows(dicts, infer_schema_length.unwrap_or(50))?;
+        let infer_schema_length = std::cmp::max(infer_schema_length.unwrap_or(50), 1);
+        let (rows, mut names) = dicts_to_rows(dicts, infer_schema_length)?;
 
         // ensure the new names are used
         if let Some(schema) = &schema_overwrite {
@@ -451,7 +452,7 @@ impl PyDataFrame {
         }
         let mut pydf = Self::finish_from_rows(
             rows,
-            infer_schema_length,
+            Some(infer_schema_length),
             schema_overwrite.map(|wrap| wrap.0),
         )?;
 

diff --git a/py-polars/tests/unit/test_interop.py b/py-polars/tests/unit/test_interop.py
@@ -260,6 +260,12 @@ def test_from_dicts() -> None:
     assert df.schema == {"a": pl.Int64, "b": pl.Int64}
 
 
+def test_from_dict_no_inference() -> None:
+    schema = {"a": pl.Utf8}
+    data = [{"a": "aa"}]
+    pl.from_dicts(data, schema=schema, infer_schema_length=0)
+
+
 def test_from_dicts_struct() -> None:
     data = [{"a": {"b": 1, "c": 2}, "d": 5}, {"a": {"b": 3, "c": 4}, "d": 6}]
     df = pl.from_dicts(data)