Bug report

Bug description:

Since 3.14 (gh-123803), the cpXXX codecs for the Windows code pages that have no module of their own in encodings are provided by encodings._win_cp_codecs. Their stream reader cannot read anything:

import codecs, io

text = 'привет'
reader = codecs.getreader('cp20866')(io.BytesIO(text.encode('cp20866')))
print(reader.read() == text)

Output:

Traceback (most recent call last):
  File "repro.py", line 5, in <module>
    print(reader.read() == text)
          ~~~~~~~~~~~^^
  File ".../Lib/codecs.py", line 512, in read
    newchars, decodedbytes = self.decode(data, self.errors)
                             ~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^
TypeError: create_win32_code_page_codec.<locals>.StreamReader.decode() missing 1 required positional argument: 'final'

Expected output: True.

readline() and codecs.StreamReaderWriter.read() fail in the same way. codecs.iterdecode() and io.TextIOWrapper work, because they use the incremental decoder.

codecs.StreamReader.read() calls self.decode(data, self.errors), but StreamReader.decode() in Lib/encodings/_win_cp_codecs.py is defined as decode(self, input, errors, final), so final is required. It should be optional and default to False (as for encodings.mbcs), so that an incomplete multibyte sequence is kept until the next read.

CPython versions tested on:

CPython main branch (the 3.14 and 3.15 branches contain the same code)

Operating systems tested on:

Windows

Linked PRs

• gh-158980: Fix StreamReader.decode() of the Windows cpXXX codecs #158990