fix(dashscope): preserve cache_control in messages #18165

litellm/llms/dashscope/chat/transformation.py

@@ -2,48 +2,37 @@
 Translates from OpenAI's `/v1/chat/completions` to DashScope's `/v1/chat/completions`
 """
 
-from typing import Any, Coroutine, List, Literal, Optional, Tuple, Union, overload
+from typing import TYPE_CHECKING, List, Optional, Tuple
 
-from litellm.litellm_core_utils.prompt_templates.common_utils import (
-    handle_messages_with_content_list_to_str_conversion,
-)
 from litellm.secret_managers.main import get_secret_str
 from litellm.types.llms.openai import AllMessageValues
 
 from ...openai.chat.gpt_transformation import OpenAIGPTConfig
 
+if TYPE_CHECKING:
+    from litellm.types.llms.openai import ChatCompletionToolParam
+
 
 class DashScopeChatConfig(OpenAIGPTConfig):
-    @overload
-    def _transform_messages(
-        self, messages: List[AllMessageValues], model: str, is_async: Literal[True]
-    ) -> Coroutine[Any, Any, List[AllMessageValues]]:
-        ...
+    """
+    DashScope configuration.
 
-    @overload
-    def _transform_messages(
+    DashScope supports content in list format with cache_control metadata.
+    See: https://github.com/BerriAI/litellm/issues/18165
+    """
+
+    def remove_cache_control_flag_from_messages_and_tools(
         self,
-        messages: List[AllMessageValues],
         model: str,
-        is_async: Literal[False] = False,
-    ) -> List[AllMessageValues]:
-        ...
+        messages: List[AllMessageValues],
+        tools: Optional[List["ChatCompletionToolParam"]] = None,
+    ) -> Tuple[List[AllMessageValues], Optional[List["ChatCompletionToolParam"]]]:
+        """
+        DashScope supports cache_control, so we preserve it instead of removing it.
 
-    def _transform_messages(
-        self, messages: List[AllMessageValues], model: str, is_async: bool = False
-    ) -> Union[List[AllMessageValues], Coroutine[Any, Any, List[AllMessageValues]]]:
+        Override parent behavior that strips cache_control for OpenAI compatibility.
         """
-        DashScope does not support content in list format.
-        """
-        messages = handle_messages_with_content_list_to_str_conversion(messages)
-        if is_async:
-            return super()._transform_messages(
-                messages=messages, model=model, is_async=True
-            )
-        else:
-            return super()._transform_messages(
-                messages=messages, model=model, is_async=False
-            )
+        return messages, tools
 
     def _get_openai_compatible_provider_info(
         self, api_base: Optional[str], api_key: Optional[str]

tests/test_litellm/llms/dashscope/test_dashscope_chat_transformation.py

@@ -111,3 +111,40 @@ class TestDashScopeConfig:
         # Check for specific content in the response
         assert "```python" in response.choices[0].message.content
         assert "Hey from LiteLLM" in response.choices[0].message.content
+
+    def test_dashscope_preserves_cache_control(self):
+        """
+        Test that DashScope preserves cache_control metadata in messages.
+
+        Regression test for: https://github.com/BerriAI/litellm/issues/18165
+        DashScope now supports cache_control, so we should NOT strip it.
+        """
+        config = DashScopeChatConfig()
+
+        messages_with_cache = [
+            {
+                "role": "user",
+                "content": [
+                    {
+                        "type": "text",
+                        "text": "Long document content here...",
+                        "cache_control": {"type": "ephemeral"},
+                    }
+                ],
+            }
+        ]
+
+        # This should preserve cache_control (not strip it like OpenAI does)
+        (
+            transformed_messages,
+            tools,
+        ) = config.remove_cache_control_flag_from_messages_and_tools(
+            model="qwen-turbo",
+            messages=messages_with_cache,
+            tools=None,
+        )
+
+        # Verify cache_control is preserved
+        assert transformed_messages[0]["content"][0]["cache_control"] == {
+            "type": "ephemeral"
+        }