read_excel accepts PathLike, buffers (#195)

* read_excel accepts PathLike, buffers

pandas.read_excel is documented to accept

> str, bytes, ExcelFile, xlrd.Book, path object, or file-like object

This expands the annotation to include str, bytes, paths,
and file-likes. Bytes are coerced into a BytesIO inside ExcelReader.

* Exercise Excel IO using new types

* blacken

* Review feedback

* Add breadcrumb

* Close file before reopening

* Remove try/except

Author: tdsmith

pandas-stubs/io/excel/_base.pyi

@@ -10,12 +10,14 @@ from pandas.core.frame import DataFrame
 
 from pandas._typing import (
     Dtype,
+    FilePath,
+    ReadBuffer,
     Scalar,
 )
 
 @overload
 def read_excel(
-    filepath: str,
+    filepath: FilePath | ReadBuffer[bytes] | bytes,
     sheet_name: list[int | str] | None,
     header: int | Sequence[int] | None = ...,
     names: list[str] | None = ...,
@@ -42,7 +44,7 @@ def read_excel(
 ) -> dict[int | str, DataFrame]: ...
 @overload
 def read_excel(
-    filepath: str,
+    filepath: FilePath | ReadBuffer[bytes] | bytes,
     sheet_name: int | str = ...,
     header: int | Sequence[int] | None = ...,
     names: list[str] | None = ...,

tests/test_frame.py

@@ -1271,6 +1271,27 @@ def test_read_excel() -> None:
         df17: dict[int | str, pd.DataFrame] = pd.read_excel("foo", sheet_name=None)
 
 
+def test_read_excel_io_types() -> None:
+    # GH 195
+    df = pd.DataFrame([[1, 2], [8, 9]], columns=["A", "B"])
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".xlsx") as file:
+        as_str: str = file.name
+        df.to_excel(file)
+
+    check(assert_type(pd.read_excel(as_str), pd.DataFrame), pd.DataFrame)
+
+    as_path = Path(as_str)
+    check(assert_type(pd.read_excel(as_path), pd.DataFrame), pd.DataFrame)
+
+    with as_path.open("rb") as as_file:
+        check(assert_type(pd.read_excel(as_file), pd.DataFrame), pd.DataFrame)
+
+    as_bytes = as_path.read_bytes()
+    check(assert_type(pd.read_excel(as_bytes), pd.DataFrame), pd.DataFrame)
+
+    Path(as_str).unlink()
+
+
 def test_join() -> None:
     float_frame = pd.DataFrame(getSeriesData())
     # GH 29