Fix YamlHelper for thread safety in multi-threaded environments.

soda/core/soda/common/yaml_helper.py

@@ -1,3 +1,5 @@
+import threading
+
 from ruamel.yaml import YAML, StringIO
 
 
@@ -14,16 +16,53 @@ def to_yaml_str(yaml_object) -> str:
 
 
 class YamlHelper:
-    __yaml = create_yaml()
+    """
+    A helper class to serialize and deserialize objects to and from YAML format.
+    This class is thread-safe and ensures that each thread has its own instance of the YAML parser/dumper.
+
+    Usage:
+    YamlHelper.to_yaml(yaml_object)
+    YamlHelper.from_yaml(yaml_string)
+    """
+
+    _thread_local = threading.local()
+
+    @classmethod
+    def _get_yaml(cls):
+        """
+        Returns a thread-local YAML instance for serializing and deserializing YAML data.
+        If no instance exists for the current thread, a new one is created.
+
+        :return: a YAML instance specific to the current thread.
+        """
+        if not hasattr(cls._thread_local, "yaml"):
+            cls._thread_local.yaml = create_yaml()
+        return cls._thread_local.yaml
 
     @classmethod
     def to_yaml(cls, yaml_object) -> str:
+        """
+        Serializes a Python object into a YAML formatted string.
+
+        :param object yaml_object: The Python object to serialize.
+        :return: the YAML formatted string, or an empty string if the input is None.
+        """
         if yaml_object is None:
             return ""
-        stream = StringIO()
-        cls.__yaml.dump(yaml_object, stream)
-        return stream.getvalue()
+        with StringIO() as stream:
+            yaml_instance = cls._get_yaml()
+            yaml_instance.dump(yaml_object, stream)
+            return stream.getvalue()
 
     @classmethod
     def from_yaml(cls, yaml_str) -> object:
-        return cls.__yaml.load(yaml_str)
+        """
+        Deserializes a YAML formatted string into a Python object.
+
+        :param str yaml_str: the YAML formatted string to deserialize.
+        :return: the deserialized Python object, or None if the input is None or empty.
+        """
+        if yaml_str is None or not yaml_str.strip():
+            return None
+        yaml_instance = cls._get_yaml()
+        return yaml_instance.load(yaml_str)