fix typing and tests

Author: lithomas1

pandas/io/formats/csvs.py

@@ -12,15 +12,20 @@
 import csv as csvlib
 import os
 from typing import (
+    IO,
     TYPE_CHECKING,
     Any,
+    AnyStr,
     cast,
 )
 
 import numpy as np
 
 from pandas._libs import writers as libwriters
-from pandas.compat import pa_version_under11p0
+from pandas.compat import (
+    pa_version_under8p0,
+    pa_version_under11p0,
+)
 from pandas.compat._optional import import_optional_dependency
 from pandas.util._decorators import cache_readonly
 
@@ -258,9 +263,12 @@ def save(self) -> None:
             is_text=self.engine == "python",
         ) as handles:
             # Note: self.encoding is irrelevant here
-            self._save(handles.handle)
 
-    def _save_pyarrow(self, handle) -> None:
+            # This is a mypy bug?
+            # error: Cannot infer type argument 1 of "_save" of "CSVFormatter"  [misc]
+            self._save(handles.handle)  # type: ignore[misc]
+
+    def _save_pyarrow(self, handle: IO[AnyStr]) -> None:
         pa = import_optional_dependency("pyarrow")
         pa_csv = import_optional_dependency("pyarrow.csv")
         # Convert index to column and rename name to empty string
@@ -289,19 +297,21 @@ def _save_pyarrow(self, handle) -> None:
                 f"Quoting option {self.quoting} is not supported with engine='pyarrow'"
             )
 
-        kwargs = {
+        kwargs: dict[str, Any] = {
             "include_header": self._need_to_save_header,
             "batch_size": self.chunksize,
-            "delimiter": self.sep,
         }
 
+        if not pa_version_under8p0:
+            kwargs["delimiter"] = self.sep
+
         if not pa_version_under11p0:
             kwargs["quoting_style"] = pa_quoting
 
         write_options = pa_csv.WriteOptions(**kwargs)
         pa_csv.write_csv(table, handle, write_options)
 
-    def _save(self, handle) -> None:
+    def _save(self, handle: IO[AnyStr]) -> None:
         if self.engine == "pyarrow":
             self._save_pyarrow(handle)
         else:

pandas/io/formats/format.py

@@ -31,6 +31,7 @@
     Any,
     Callable,
     Final,
+    Union,
     cast,
 )
 from unicodedata import east_asian_width
@@ -1157,11 +1158,14 @@ def to_csv(
         csv_formatter.save()
 
         if created_buffer:
+            path_or_buf = cast(Union[BytesIO, StringIO], path_or_buf)
             content = path_or_buf.getvalue()
-            if isinstance(path_or_buf, BytesIO):
+            if isinstance(content, bytes):
                 # Need to decode into string since the
                 # pyarrow engine only writes binary data
+                # content = cast(bytes, content)
                 content = content.decode("utf-8")
+                # content = cast(str, content)
             path_or_buf.close()
             return content