[Py2] UnicodeDecodeError in doctest with Unicode

[inducer]

Save this file as x.rst:

.. doctest::

    >>> print(
    ...    "Hi\n\nByé")
    Hi
    ...
    Byé
    >>> 1/0
    1

Run with

python -m pytest x.rst 

on pytest 3.1.0 with Python 2.7.11, observe lengthy stacktrace ending with a UnicodeDecodeError in _ellipsis_match. Pytest 3.0.4 works fine.

[nicoddemus]

Thanks for the report and the reproducible example, we appreciate it.

[NiklasMM]

I was still able to reproduce this issue in pytest 3.1.2 with the following file:

# -*- encoding: utf-8 -*-
from __future__ import unicode_literals

def fix_bad_unicode(text):
    """
        >>> print fix_bad_unicode('único')
        único
    """
    return "único"

Save as test.py and run python -m pytest --doctest-modules test.py. This results in an equally unpleasant stacktrace.
(Python 2.7.12, pytest 3.1.2)

For what it's worth: I made this file while debugging the same error generated by the file embedded in this blog post.

[nicoddemus]

Indeed I can reproduce the problem using you example, thanks!

[inducer]

I don't have time to come up with another reproducer, but this issue is not entirely fixed: https://gitlab.tiker.net/inducer/loopy/-/jobs/40102.