python: prefer pyarrow when we can memory map the file (#4182)

py-polars/polars/io.py

@@ -618,6 +618,14 @@ def scan_ipc(
     # Map legacy arguments to current ones and remove them from kwargs.
     n_rows = kwargs.pop("stop_after_n_rows", n_rows)
 
+    if (
+        _PYARROW_AVAILABLE
+        and row_count_name is None
+        and row_count_offset is None
+        and "*" not in file
+    ):
+        read_ipc(file=file, n_rows=n_rows, use_pyarrow=True, cache=False, rechunk=False)
+
     return LazyFrame.scan_ipc(
         file=file,
         n_rows=n_rows,
@@ -729,7 +737,7 @@ def read_ipc(
     file: str | BinaryIO | BytesIO | Path | bytes,
     columns: list[int] | list[str] | None = None,
     n_rows: int | None = None,
-    use_pyarrow: bool = False,
+    use_pyarrow: bool = _PYARROW_AVAILABLE,
     memory_map: bool = True,
     storage_options: dict | None = None,
     row_count_name: str | None = None,

py-polars/tests/io/test_ipc.py

@@ -21,7 +21,7 @@ def test_from_to_buffer(df: pl.DataFrame, compressions: list[str]) -> None:
         buf = io.BytesIO()
         df.write_ipc(buf, compression=compression)  # type: ignore[arg-type]
         buf.seek(0)
-        read_df = pl.read_ipc(buf)
+        read_df = pl.read_ipc(buf, use_pyarrow=False)
         assert_frame_equal_local_categoricals(df, read_df)
 
 
@@ -35,7 +35,7 @@ def test_from_to_file(
         for compression in compressions:
             for f in (str(f_ipc), Path(f_ipc)):
                 df.write_ipc(f, compression=compression)  # type: ignore[arg-type]
-                df_read = pl.read_ipc(f)  # type: ignore[arg-type]
+                df_read = pl.read_ipc(f, use_pyarrow=False)  # type: ignore[arg-type]
                 assert_frame_equal_local_categoricals(df, df_read)
 
 
@@ -134,7 +134,7 @@ def test_glob_ipc(io_test_dir: str) -> None:
     if os.name != "nt":
         path = os.path.join(io_test_dir, "small*.ipc")
         assert pl.scan_ipc(path).collect().shape == (3, 12)
-        assert pl.read_ipc(path).shape == (3, 12)
+        assert pl.read_ipc(path, use_pyarrow=False).shape == (3, 12)
 
 
 def test_from_float16() -> None:
@@ -143,4 +143,4 @@ def test_from_float16() -> None:
     f = io.BytesIO()
     pandas_df.to_feather(f)
     f.seek(0)
-    assert pl.read_ipc(f).dtypes == [pl.Float32]
+    assert pl.read_ipc(f, use_pyarrow=False).dtypes == [pl.Float32]