Fix YAML configuration parsing with Unicode characters on non-UTF-8 locales

Author: Kyle-Neale

datadog_checks_base/changelog.d/21850.fixed

@@ -0,0 +1 @@
+Fix YAML configuration parsing to properly handle Unicode characters on Windows systems where the UTF-8 locale is not enabled by default.

datadog_checks_base/datadog_checks/base/checks/base.py

@@ -1492,9 +1492,10 @@ def load_config(yaml_str: str) -> Any:
             stdin=subprocess.PIPE,
             stdout=subprocess.PIPE,
             stderr=subprocess.PIPE,
+            encoding='utf-8',
         )
-        stdout, stderr = process.communicate(yaml_str.encode())
+        stdout, stderr = process.communicate(yaml_str)
         if process.returncode != 0:
-            raise ValueError(f'Failed to load config: {stderr.decode()}')
+            raise ValueError(f'Failed to load config: {stderr}')
 
-        return _parse_ast_config(stdout.strip().decode())
+        return _parse_ast_config(stdout.strip())

datadog_checks_base/tests/base/checks/test_load_config.py

@@ -76,3 +76,25 @@ def test_load_config_nan():
     config = AgentCheck.load_config("number: .nan")
     assert "number" in config
     assert math.isnan(config["number"])
+
+
+@pytest.mark.parametrize(
+    'yaml_str, expected_object',
+    [
+        pytest.param(
+            "tag: テスト",
+            {"tag": "テスト"},
+            id="japanese_characters",
+        ),
+        pytest.param(
+            "chinese: 中文测试",
+            {"chinese": "中文测试"},
+            id="chinese_characters",
+        ),
+    ],
+)
+def test_load_config_unicode(yaml_str, expected_object):
+    """Test that load_config properly handles Unicode characters including Japanese, Chinese, Korean, and emoji.
+    This is especially important on Windows where the system locale may not default to UTF-8."""
+    config = AgentCheck.load_config(yaml_str)
+    assert config == expected_object