Add set_version encoding tests and fix missing encoding in provider writes

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):
     """
@@ -58,13 +61,16 @@ 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")
+        self.file.write_text(
+            json.dumps(document, indent=self.indent, ensure_ascii=False) + "\n",
+            encoding=self._get_encoding(),
+        )
 
     def get(self, document: Mapping[str, str]) -> str:
         return document["version"]
@@ -79,13 +85,13 @@ 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))
+        self.file.write_text(tomlkit.dumps(document), encoding=self._get_encoding())
 
     def get(self, document: tomlkit.TOMLDocument) -> str:
         return document["project"]["version"]  # type: ignore[index,return-value]

commitizen/providers/cargo_provider.py

@@ -43,8 +43,12 @@ def set_version(self, version: str) -> None:
             self.set_lock_version(version)
 
     def set_lock_version(self, version: str) -> None:
-        cargo_toml_content = parse(self.file.read_text())
-        cargo_lock_content = parse(self.lock_file.read_text())
+        cargo_toml_content = parse(
+            self.file.read_text(encoding=self._get_encoding())
+        )
+        cargo_lock_content = parse(
+            self.lock_file.read_text(encoding=self._get_encoding())
+        )
         packages = cargo_lock_content["package"]
 
         if TYPE_CHECKING:
@@ -75,7 +79,9 @@ def set_lock_version(self, version: str) -> None:
                         continue
 
                     cargo_file = Path(path) / "Cargo.toml"
-                    package_content = parse(cargo_file.read_text()).get("package", {})
+                    package_content = parse(
+                        cargo_file.read_text(encoding=self._get_encoding())
+                    ).get("package", {})
                     if TYPE_CHECKING:
                         assert isinstance(package_content, dict)
                     try:
@@ -92,7 +98,9 @@ def set_lock_version(self, version: str) -> None:
                 if package["name"] in members_inheriting:
                     cargo_lock_content["package"][i]["version"] = version  # type: ignore[index]
 
-        self.lock_file.write_text(dumps(cargo_lock_content))
+        self.lock_file.write_text(
+            dumps(cargo_lock_content), encoding=self._get_encoding()
+        )
 
 
 def _try_get_workspace(document: TOMLDocument) -> dict:

commitizen/providers/npm_provider.py

@@ -20,6 +20,9 @@ class NpmProvider(VersionProvider):
     lock_filename = "package-lock.json"
     shrinkwrap_filename = "npm-shrinkwrap.json"
 
+    def _get_encoding(self) -> str:
+        return self.config.settings["encoding"]
+
     @property
     def package_file(self) -> Path:
         return Path() / self.package_filename
@@ -36,29 +39,48 @@ def get_version(self) -> str:
         """
         Get the current version from package.json
         """
-        package_document = json.loads(self.package_file.read_text())
+        package_document = json.loads(
+            self.package_file.read_text(encoding=self._get_encoding())
+        )
         return self.get_package_version(package_document)
 
     def set_version(self, version: str) -> None:
         package_document = self.set_package_version(
-            json.loads(self.package_file.read_text()), version
+            json.loads(
+                self.package_file.read_text(encoding=self._get_encoding())
+            ),
+            version,
         )
         self.package_file.write_text(
-            json.dumps(package_document, indent=self.indent) + "\n"
+            json.dumps(package_document, indent=self.indent, ensure_ascii=False)
+            + "\n",
+            encoding=self._get_encoding(),
         )
         if self.lock_file.is_file():
             lock_document = self.set_lock_version(
-                json.loads(self.lock_file.read_text()), version
+                json.loads(
+                    self.lock_file.read_text(encoding=self._get_encoding())
+                ),
+                version,
             )
             self.lock_file.write_text(
-                json.dumps(lock_document, indent=self.indent) + "\n"
+                json.dumps(lock_document, indent=self.indent, ensure_ascii=False)
+                + "\n",
+                encoding=self._get_encoding(),
             )
         if self.shrinkwrap_file.is_file():
             shrinkwrap_document = self.set_shrinkwrap_version(
-                json.loads(self.shrinkwrap_file.read_text()), version
+                json.loads(
+                    self.shrinkwrap_file.read_text(encoding=self._get_encoding())
+                ),
+                version,
             )
             self.shrinkwrap_file.write_text(
-                json.dumps(shrinkwrap_document, indent=self.indent) + "\n"
+                json.dumps(
+                    shrinkwrap_document, indent=self.indent, ensure_ascii=False
+                )
+                + "\n",
+                encoding=self._get_encoding(),
             )
 
     def get_package_version(self, document: Mapping[str, str]) -> str:

commitizen/providers/uv_provider.py

@@ -26,15 +26,21 @@ def set_version(self, version: str) -> None:
         self.set_lock_version(version)
 
     def set_lock_version(self, version: str) -> None:
-        pyproject_toml_content = tomlkit.parse(self.file.read_text())
+        pyproject_toml_content = tomlkit.parse(
+            self.file.read_text(encoding=self._get_encoding())
+        )
         project_name = pyproject_toml_content["project"]["name"]  # type: ignore[index]
         normalized_project_name = canonicalize_name(str(project_name))
 
-        document = tomlkit.parse(self.lock_file.read_text())
+        document = tomlkit.parse(
+            self.lock_file.read_text(encoding=self._get_encoding())
+        )
 
         packages: tomlkit.items.AoT = document["package"]  # type: ignore[assignment]
         for i, package in enumerate(packages):
             if package["name"] == normalized_project_name:
                 document["package"][i]["version"] = version  # type: ignore[index]
                 break
-        self.lock_file.write_text(tomlkit.dumps(document))
+        self.lock_file.write_text(
+            tomlkit.dumps(document), encoding=self._get_encoding()
+        )

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 +27,152 @@ 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_set_version_uses_encoding_with_encoding_fixture(
+    config: BaseConfig,
+    chdir: Path,
+    data_dir: Path,
+):
+    """JsonProvider.set_version should correctly write 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)
+    # Update version using set_version
+    provider.set_version("1.0.0")
+
+    # Verify the file can be read back with the same encoding
+    assert provider.get_version() == "1.0.0"
+
+    # Verify the non-ASCII content is preserved
+    content = target.read_text(encoding="utf-8")
+    assert "Тест описания для проверки кодировки" in content
+
+
+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_set_version_uses_encoding_with_encoding_fixture(
+    config: BaseConfig,
+    chdir: Path,
+    data_dir: Path,
+):
+    """TomlProvider.set_version should correctly write 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"] = "pep621"
+
+    provider = get_provider(config)
+    assert isinstance(provider, Pep621Provider)
+    # Update version using set_version
+    provider.set_version("1.0.0")
+
+    # Verify the file can be read back with the same encoding
+    assert provider.get_version() == "1.0.0"
+
+    # Verify the non-ASCII content is preserved
+    content = target.read_text(encoding="utf-8")
+    assert "Новое" in content
+    assert "Исправление" in content
+    assert "Выберите тип изменений" in content
+
+
+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"