fix(provider): use encoding settings in config

Author: bearomorphism

commitizen/providers/base_provider.py

@@ -49,6 +49,9 @@ class FileProvider(VersionProvider):
     def file(self) -> Path:
         return Path() / self.filename
 
+    def _get_encoding(self) -> str:
+        return self.config.settings["encoding"]
+
 
 class JsonProvider(FileProvider):
     """
@@ -58,11 +61,11 @@ class JsonProvider(FileProvider):
     indent: ClassVar[int] = 2
 
     def get_version(self) -> str:
-        document = json.loads(self.file.read_text())
+        document = json.loads(self.file.read_text(encoding=self._get_encoding()))
         return self.get(document)
 
     def set_version(self, version: str) -> None:
-        document = json.loads(self.file.read_text())
+        document = json.loads(self.file.read_text(encoding=self._get_encoding()))
         self.set(document, version)
         self.file.write_text(json.dumps(document, indent=self.indent) + "\n")
 
@@ -79,11 +82,11 @@ class TomlProvider(FileProvider):
     """
 
     def get_version(self) -> str:
-        document = tomlkit.parse(self.file.read_text())
+        document = tomlkit.parse(self.file.read_text(encoding=self._get_encoding()))
         return self.get(document)
 
     def set_version(self, version: str) -> None:
-        document = tomlkit.parse(self.file.read_text())
+        document = tomlkit.parse(self.file.read_text(encoding=self._get_encoding()))
         self.set(document, version)
         self.file.write_text(tomlkit.dumps(document))
 

tests/data/encoding_test_composer.json

@@ -0,0 +1,6 @@
+{
+  "name": "encoding-test-composer",
+  "description": "Тест описания для проверки кодировки",
+  "version": "0.1.0"
+}
+

tests/data/encoding_test_pyproject.toml

@@ -0,0 +1,36 @@
+[project]
+name = "pythonproject-test"
+version = "0.4.1"
+description = "Add your description here"
+readme = "README.md"
+requires-python = ">=3.12"
+dependencies = []
+
+
+[tool.commitizen]
+name = "cz_customize"
+tag_format = "v$version"
+version_scheme = "pep440"
+version_provider = "uv"
+update_changelog_on_bump = true
+changelog_start_rev = "v1.1.0"
+
+[tool.commitizen.customize]
+message_template = "{{ change_type }}{% if scope != 'none' %}({{ scope }}){% endif %}: {{ message }}"
+commit_parser = '^(?P<change_type>feat|fix|refactor|test|perf|misc):\s(?P<message>.*)'
+schema_pattern = '(feat|fix|refactor|test|perf|misc)(\((api|core)\))?:\s(.{3,})'
+bump_pattern = "^(feat|fix|refactor|test|perf|misc)"
+change_type_map = { "feat" = "Новое", "fix" = "Исправление" }
+
+[[tool.commitizen.customize.questions]]
+name = "change_type"
+type = "list"
+message = "Выберите тип изменений"
+choices = [
+    { value = "feat", name = "feat: Новая функциональность" },
+    { value = "fix", name = "fix: Исправление" },
+    { value = "refactor", name = "refactor: Рефакторинг" },
+    { value = "test", name = "test: Изменение авто-тестов" },
+    { value = "perf", name = "perf: Оптимизации" },
+    { value = "misc", name = "misc: Другое" },
+]

tests/providers/test_base_provider.py

@@ -7,8 +7,13 @@
 from commitizen.exceptions import VersionProviderUnknown
 from commitizen.providers import get_provider
 from commitizen.providers.commitizen_provider import CommitizenProvider
+from commitizen.providers.composer_provider import ComposerProvider
+from commitizen.providers.pep621_provider import Pep621Provider
+from commitizen.providers.uv_provider import UvProvider
 
 if TYPE_CHECKING:
+    from pathlib import Path
+
     from commitizen.config.base_config import BaseConfig
 
 
@@ -22,3 +27,98 @@ def test_raise_for_unknown_provider(config: BaseConfig):
     config.settings["version_provider"] = "unknown"
     with pytest.raises(VersionProviderUnknown):
         get_provider(config)
+
+
+@pytest.mark.parametrize("encoding", ["utf-8", "latin-1"])
+def test_file_provider_get_encoding(config: BaseConfig, encoding: str):
+    """_get_encoding should return the configured encoding."""
+    config.settings["encoding"] = encoding
+    provider = ComposerProvider(config)
+    assert provider._get_encoding() == encoding
+
+
+def test_json_provider_uses_encoding_with_encoding_fixture(
+    config: BaseConfig,
+    chdir: Path,
+    data_dir: Path,
+):
+    """JsonProvider should correctly read a JSON file with non-ASCII content."""
+    source = data_dir / "encoding_test_composer.json"
+    target = chdir / "composer.json"
+    target.write_text(source.read_text(encoding="utf-8"), encoding="utf-8")
+
+    config.settings["encoding"] = "utf-8"
+    config.settings["version_provider"] = "composer"
+
+    provider = get_provider(config)
+    assert isinstance(provider, ComposerProvider)
+    assert provider.get_version() == "0.1.0"
+
+
+def test_toml_provider_uses_encoding_with_encoding_fixture(
+    config: BaseConfig,
+    chdir: Path,
+    data_dir: Path,
+):
+    """TomlProvider should correctly read a TOML file with non-ASCII content."""
+    source = data_dir / "encoding_test_pyproject.toml"
+    target = chdir / "pyproject.toml"
+    target.write_text(source.read_text(encoding="utf-8"), encoding="utf-8")
+
+    config.settings["encoding"] = "utf-8"
+    config.settings["version_provider"] = "uv"
+
+    provider = get_provider(config)
+    assert isinstance(provider, UvProvider)
+    assert provider.get_version() == "0.4.1"
+
+
+def test_json_provider_handles_various_unicode_characters(
+    config: BaseConfig,
+    chdir: Path,
+):
+    """JsonProvider should handle a wide range of Unicode characters."""
+    config.settings["encoding"] = "utf-8"
+    config.settings["version_provider"] = "composer"
+
+    filename = ComposerProvider.filename
+    file = chdir / filename
+    file.write_text(
+        (
+            "{\n"
+            '  "name": "多言語-имя-árbol",\n'
+            '  "description": "Emoji 😀 – 漢字 – العربية",\n'
+            '  "version": "0.1.0"\n'
+            "}\n"
+        ),
+        encoding="utf-8",
+    )
+
+    provider = get_provider(config)
+    assert isinstance(provider, ComposerProvider)
+    assert provider.get_version() == "0.1.0"
+
+
+def test_toml_provider_handles_various_unicode_characters(
+    config: BaseConfig,
+    chdir: Path,
+):
+    """TomlProvider should handle a wide range of Unicode characters."""
+    config.settings["encoding"] = "utf-8"
+    config.settings["version_provider"] = "pep621"
+
+    filename = Pep621Provider.filename
+    file = chdir / filename
+    file.write_text(
+        (
+            "[project]\n"
+            'name = "多言語-имя-árbol"\n'
+            'description = "Emoji 😀 – 漢字 – العربية"\n'
+            'version = "0.1.0"\n'
+        ),
+        encoding="utf-8",
+    )
+
+    provider = get_provider(config)
+    assert isinstance(provider, Pep621Provider)
+    assert provider.get_version() == "0.1.0"