René's URL Explorer Experiment

Title: fix(config): preserve unicode characters when writing yaml config by bearomorphism · Pull Request #1966 · commitizen-tools/commitizen · GitHub

Open Graph Title: fix(config): preserve unicode characters when writing yaml config by bearomorphism · Pull Request #1966 · commitizen-tools/commitizen

X Title: fix(config): preserve unicode characters when writing yaml config by bearomorphism · Pull Request #1966 · commitizen-tools/commitizen

Description: Description Closes #1164. Why PyYAML's yaml.dump defaults to ASCII-only output: any non-ASCII codepoint is escaped using Python's \Uxxxx notation. This means a .cz.yaml containing a literal emoji — for example 🚀 in bump_message — is silently mangled every time cz bump rewrites the config, replacing the readable character with \U0001F680. Reported by @syepes on commitizen 3.27.0 / Python 3.12 / macOS (#1164), with before-and-after screenshots showing the escape introduced by cz bump --increment PATCH. A triage note from the open-issues audit (2026-05-09) confirmed the bug still reproduces on master (v4.15.1): after cz bump, the bump_message key reads "\U0001F680 chore(release): …" instead of "🚀 chore(release): …". The root cause is that both yaml.dump call sites in commitizen/config/yaml_config.py — init_empty_config_content (line 33) and set_key (line 66) — omit allow_unicode=True. PyYAML documents that this flag causes non-ASCII codepoints to be written as literal UTF-8 bytes rather than escape sequences; the output remains valid YAML and valid UTF-8. What changed File Change commitizen/config/yaml_config.py Add allow_unicode=True to yaml.dump in init_empty_config_content (line 33) and in set_key (line 66) tests/test_conf.py Add test_set_key_preserves_unicode (regression: emoji survives a set_key round-trip) and test_init_empty_config_content_passes_allow_unicode (spy: keyword argument is forwarded to yaml.dump) How it works set_key (commitizen/config/yaml_config.py:58–68): reads the config with yaml.load, mutates the target key in the parsed dict, then re-serialises with yaml.dump. Without allow_unicode=True, any non-ASCII character already present in the file is escaped on write-back — producing the \U0001F680 symptom. Adding the flag instructs PyYAML to emit those characters as literal UTF-8 bytes. init_empty_config_content (commitizen/config/yaml_config.py:29–33): writes {"commitizen": {}} in append mode during cz init. Its payload is currently ASCII-only, so allow_unicode=True has no observable effect today. It is added for consistency and to guard against future default content that might include non-ASCII characters. Test strategy: test_set_key_preserves_unicode writes a YAML file containing 🚀, calls set_key("version", "0.1.1"), and asserts both that 🚀 survives and that \U0001F680 does not appear — this is the direct regression test for #1164. test_init_empty_config_content_passes_allow_unicode uses mocker.spy(yaml, "dump") to assert the keyword is forwarded rather than testing round-trip behaviour that does not currently occur (see the review note in Additional Context). Why not a custom Dumper? PyYAML's allow_unicode flag is the documented, single-argument way to opt out of ASCII escaping. A custom Dumper subclass would be substantially more complex and carries a maintenance burden with no additional benefit for this use case. Backward compatibility YAML files that already contain \Uxxxx escape sequences are normalised on the next cz bump: yaml.load decodes the escape to the Unicode character on read, and the subsequent yaml.dump re-emits it as a literal codepoint. No data is lost and no manual migration is needed. All 20 existing TestYamlConfig tests in tests/test_conf.py pass without modification. No change to the public API, CLI flags, exit codes, or any config backend other than YAMLConfig. Checklist I have read the contributing guidelines Was generative AI tooling used to co-author this PR? Yes (please specify the tool below) Generated-by: Claude following the guidelines Code Changes Add test cases to all the changes you introduce Run uv run poe all locally to ensure this change passes linter check and tests Manually test the changes (see "Steps to Test" below) Update the documentation for the changes Expected Behavior Scenario Outcome .cz.yaml has bump_message: "🚀 chore: bump …" and cz bump is run Emoji 🚀 is preserved verbatim; \U0001F680 does not appear .cz.yaml already contains an escaped \U0001F680 and cz bump is run PyYAML normalises the escape on read and re-emits the literal 🚀 — file is healed in place cz init creates a new .cz.yaml Output is ---\ncommitizen: {}\n, identical to the pre-fix behaviour Steps to Test This Pull Request git fetch fork fix/1164-yaml-allow-unicode git checkout fork/fix/1164-yaml-allow-unicode # 1. Targeted regression test. uv run pytest tests/test_conf.py::TestYamlConfig::test_set_key_preserves_unicode \ tests/test_conf.py::TestYamlConfig::test_init_empty_config_content_passes_allow_unicode -v # 2. Reproduce the bug, then verify the fix. mkdir /tmp/cz-1164 && cd /tmp/cz-1164 git init -b main git config user.name test && git config user.email test@example.com cat > .cz.yaml << 'EOF' commitizen: name: cz_conventional_commits version: 0.1.0 tag_format: "v$version" bump_message: "🚀 chore(release): bump $current_version to $new_version" EOF git add .cz.yaml && git commit -m "feat: initial" cz bump --increment PATCH --yes grep bump_message .cz.yaml # Expected: bump_message: "🚀 chore(release): ..." # NOT: bump_message: "\U0001F680 chore(release): ..." Additional Context This fix was identified during the open-issues audit tracked in #1964. A triage note (@bearomorphism, 2026-05-09) confirmed the bug reproduces on master (v4.15.1) and pinpointed commitizen/config/yaml_config.py:66 as the root cause, with a symmetric fix needed at line 33 for consistency. Review note — test update: the initial draft included test_init_empty_config_content_uses_allow_unicode, which pre-seeded a YAML file with 🚀 and asserted the emoji survived init_empty_config_content(). This was misleading: because init_empty_config_content opens in append mode and writes only {"commitizen": {}}, the assertion would pass regardless of whether allow_unicode=True was present. The test has been replaced with a mocker.spy-based assertion that directly verifies the keyword argument is forwarded, without pretending to test behaviour that doesn't currently happen (see PR comment).

Open Graph Description: Description Closes #1164. Why PyYAML's yaml.dump defaults to ASCII-only output: any non-ASCII codepoint is escaped using Python's \Uxxxx notation. This means a .cz.yaml containing a literal...

X Description: Description Closes #1164. Why PyYAML&#39;s yaml.dump defaults to ASCII-only output: any non-ASCII codepoint is escaped using Python&#39;s \Uxxxx notation. This means a .cz.yaml containing a...

Opengraph URL: https://github.com/commitizen-tools/commitizen/pull/1966

X: @github

direct link

Domain: github.com

Links:

Viewport: width=device-width