fix(out): reconfigure stdout to UTF-8 on any non-UTF-8 terminal

commitizen/out.py

@@ -4,9 +4,34 @@
 
 from termcolor import colored
 
-if sys.platform == "win32":
-    if isinstance(sys.stdout, io.TextIOWrapper) and sys.version_info >= (3, 7):
-        sys.stdout.reconfigure(encoding="utf-8")
+
+def _ensure_utf8_stdout(stream: object) -> None:
+    """Reconfigure ``stream`` to UTF-8 if its current encoding can't represent
+    the unicode characters commitizen emits (e.g. ``\U0001f680`` 🚀, the
+    ``\u2019`` typographic apostrophe).
+
+    Without this, ``print`` raises ``UnicodeEncodeError`` mid-output on:
+
+    * Windows ``cmd.exe`` defaulting to ``cp1252`` (the historical case),
+    * Linux/macOS terminals with a non-UTF-8 ``LANG`` such as
+      ``de_CH.ISO8859-1`` (#956).
+
+    ``errors="replace"`` is used as a safety net for terminals that
+    genuinely can't render the bytes, so commitizen falls back to a
+    placeholder character instead of crashing.
+    """
+    if not isinstance(stream, io.TextIOWrapper):
+        return
+    encoding = (stream.encoding or "").lower().replace("-", "").replace("_", "")
+    if encoding == "utf8":
+        return
+    try:
+        stream.reconfigure(encoding="utf-8", errors="replace")
+    except (AttributeError, ValueError):  # pragma: no cover - safety net
+        pass
+
+
+_ensure_utf8_stdout(sys.stdout)
 
 
 def write(value: object, *args: object) -> None:

tests/test_out.py

@@ -0,0 +1,85 @@
+"""Tests for ``commitizen.out``.
+
+Mostly focused on the stdout-encoding helper introduced for #956: the
+function must reconfigure non-UTF-8 streams to UTF-8 with a permissive
+``errors="replace"`` strategy so commitizen output (emoji, typographic
+quotes) doesn't crash with ``UnicodeEncodeError`` on terminals using
+locale-dependent encodings such as ``cp1252`` (Windows) or
+``ISO8859-1`` (Linux/macOS).
+"""
+
+from __future__ import annotations
+
+import io
+from typing import Any
+
+from commitizen.out import _ensure_utf8_stdout
+
+
+class _StubStream(io.TextIOWrapper):
+    """Light-weight ``TextIOWrapper`` that records calls to ``reconfigure``.
+
+    Subclassing ``TextIOWrapper`` keeps the ``isinstance`` check in
+    ``_ensure_utf8_stdout`` happy without monkey-patching ``sys.stdout``.
+    """
+
+    reconfigure_calls: list[dict[str, Any]]
+
+    def __init__(self, encoding: str) -> None:
+        super().__init__(io.BytesIO(), encoding=encoding)
+        self.reconfigure_calls = []
+
+    def reconfigure(self, **kwargs: Any) -> None:
+        self.reconfigure_calls.append(kwargs)
+        super().reconfigure(**kwargs)
+
+
+def test_ensure_utf8_stdout_noop_when_already_utf8():
+    stream = _StubStream(encoding="utf-8")
+    _ensure_utf8_stdout(stream)
+    assert stream.reconfigure_calls == []
+
+
+def test_ensure_utf8_stdout_noop_for_dashless_utf8_alias():
+    stream = _StubStream(encoding="UTF8")
+    _ensure_utf8_stdout(stream)
+    assert stream.reconfigure_calls == []
+
+
+def test_ensure_utf8_stdout_reconfigures_iso8859_1_terminal():
+    """Regression test for #956 (Linux/macOS ``LANG=de_CH.ISO8859-1``)."""
+    stream = _StubStream(encoding="latin-1")
+    _ensure_utf8_stdout(stream)
+    assert stream.reconfigure_calls == [{"encoding": "utf-8", "errors": "replace"}]
+
+
+def test_ensure_utf8_stdout_reconfigures_windows_cp1252():
+    """Regression test for the historical Windows ``cmd.exe`` case."""
+    stream = _StubStream(encoding="cp1252")
+    _ensure_utf8_stdout(stream)
+    assert stream.reconfigure_calls == [{"encoding": "utf-8", "errors": "replace"}]
+
+
+def test_ensure_utf8_stdout_skips_non_textio_streams():
+    class NotATextIO:
+        encoding = "latin-1"
+        reconfigure_calls: list[dict[str, Any]] = []
+
+        def reconfigure(self, **kwargs: Any) -> None:  # pragma: no cover - unused
+            self.reconfigure_calls.append(kwargs)
+
+    stream = NotATextIO()
+    _ensure_utf8_stdout(stream)
+    assert stream.reconfigure_calls == []
+
+
+def test_ensure_utf8_stdout_after_reconfigure_can_emit_emoji():
+    """End-to-end: after reconfiguration, writing an emoji must not raise."""
+    stream = _StubStream(encoding="latin-1")
+    _ensure_utf8_stdout(stream)
+
+    # Should not raise UnicodeEncodeError; ``errors="replace"`` lets
+    # genuinely-unrenderable bytes fall through as ``?`` instead of
+    # crashing the whole command.
+    stream.write("Configuration complete \U0001f680")
+    stream.flush()