REGR: codecs.open() is always opened in text mode

doc/source/whatsnew/v1.2.1.rst

@@ -34,6 +34,7 @@ Fixed regressions
 - Fixed regression in repr of float-like strings of an ``object`` dtype having trailing 0's truncated after the decimal (:issue:`38708`)
 - Fixed regression that raised ``AttributeError`` with PyArrow versions [0.16.0, 1.0.0) (:issue:`38801`)
 - Fixed regression in :func:`pandas.testing.assert_frame_equal` raising ``TypeError`` with ``check_like=True`` when :class:`Index` or columns have mixed dtype (:issue:`39168`)
+- Fixed regression in :meth:`~DataFrame.to_csv` opening `codecs.StreamReaderWriter` in binary mode instead of in text mode (:issue:`39247`)
 
 We have reverted a commit that resulted in several plotting related regressions in pandas 1.2.0 (:issue:`38969`, :issue:`38736`, :issue:`38865`, :issue:`38947` and :issue:`39126`).
 As a result, bugs reported as fixed in pandas 1.2.0 related to inconsistent tick labeling in bar plots are again present (:issue:`26186` and :issue:`11465`)

pandas/io/common.py

@@ -2,6 +2,7 @@
 from __future__ import annotations
 
 import bz2
+import codecs
 from collections import abc
 import dataclasses
 import gzip
@@ -857,9 +858,12 @@ def file_exists(filepath_or_buffer: FilePathOrBuffer) -> bool:
 
 def _is_binary_mode(handle: FilePathOrBuffer, mode: str) -> bool:
     """Whether the handle is opened in binary mode"""
+    # classes that expect string but have 'b' in mode
+    text_classes = (codecs.StreamReaderWriter,)
+    if isinstance(handle, text_classes):
+        return False
+
     # classes that expect bytes
-    binary_classes = [BufferedIOBase, RawIOBase]
+    binary_classes = (BufferedIOBase, RawIOBase)
 
-    return isinstance(handle, tuple(binary_classes)) or "b" in getattr(
-        handle, "mode", mode
-    )
+    return isinstance(handle, binary_classes) or "b" in getattr(handle, "mode", mode)

pandas/tests/io/test_common.py

@@ -1,6 +1,7 @@
 """
 Tests for the pandas.io.common functionalities
 """
+import codecs
 from io import StringIO
 import mmap
 import os
@@ -429,3 +430,19 @@ def test_default_errors():
         file = Path(path)
         file.write_bytes(b"\xe4\na\n1")
         tm.assert_frame_equal(pd.read_csv(file, skiprows=[0]), pd.DataFrame({"a": [1]}))
+
+
+@pytest.mark.parametrize("encoding", [None, "utf-8"])
+@pytest.mark.parametrize("format", ["csv", "json"])
+def test_codecs_encoding(encoding, format):
+    # GH39247
+    expected = tm.makeDataFrame()
+    with tm.ensure_clean() as path:
+        with codecs.open(path, mode="w", encoding=encoding) as handle:
+            getattr(expected, f"to_{format}")(handle)
+        with codecs.open(path, mode="r", encoding=encoding) as handle:
+            if format == "csv":
+                df = pd.read_csv(handle, index_col=0)
+            else:
+                df = pd.read_json(handle)
+    tm.assert_frame_equal(expected, df)