Improve from_dict performance by caching fields

toot/entities.py

@@ -12,7 +12,8 @@
 
 from dataclasses import dataclass, is_dataclass
 from datetime import date, datetime
-from typing import Dict, List, Optional, Type, TypeVar, Union
+from functools import lru_cache
+from typing import Any, Dict, List, Optional, Tuple, Type, TypeVar, Union
 from typing import get_type_hints
 
 from toot.typing_compat import get_args, get_origin
@@ -435,17 +436,27 @@ def from_dict(cls: Type[T], data: Dict) -> T:
         data = prepare(data)
 
     def _fields():
-        hints = get_type_hints(cls)
-        for field in dataclasses.fields(cls):
-            field_type = _prune_optional(hints[field.name])
-            default_value = _get_default_value(field)
-            value = data.get(field.name, default_value)
-            converted = _convert_with_error_handling(cls, field.name, field_type, value)
-            yield field.name, converted
+        for name, type, default in get_fields(cls):
+            value = data.get(name, default)
+            converted = _convert_with_error_handling(cls, name, type, value)
+            yield name, converted
 
     return cls(**dict(_fields()))
 
 
+@lru_cache(maxsize=100)
+def get_fields(cls: Type) -> List[Tuple[str, Type, Any]]:
+    hints = get_type_hints(cls)
+    return [
+        (
+            field.name,
+            _prune_optional(hints[field.name]),
+            _get_default_value(field)
+        )
+        for field in dataclasses.fields(cls)
+    ]
+
+
 def from_dict_list(cls: Type[T], data: List[Dict]) -> List[T]:
     return [from_dict(cls, x) for x in data]
 
@@ -497,7 +508,7 @@ def _convert(field_type, value):
     raise ValueError(f"Not implemented for type '{field_type}'")
 
 
-def _prune_optional(field_type):
+def _prune_optional(field_type: Type) -> Type:
     """For `Optional[<type>]` returns the encapsulated `<type>`."""
     if get_origin(field_type) == Union:
         args = get_args(field_type)