feat(workflow): enhance message stages and image routing

backend/tests/test_user_input.py

@@ -1,5 +1,7 @@
 import unittest
 
+from pipecat.frames.frames import LLMMessagesAppendFrame, UserImageRawFrame
+from services.pipecat.pipeline import _multimodal_user_input_frame
 from services.pipecat.processors import UserInputError, parse_user_input
 
 
@@ -52,6 +54,33 @@ class UserInputParserTests(unittest.TestCase):
                 }
             )
 
+    def test_native_image_uses_the_standard_multimodal_user_turn_path(self):
+        image = UserImageRawFrame(
+            image=bytes([220, 40, 40] * 16 * 16),
+            size=(16, 16),
+            format="RGB",
+        )
+
+        frame = _multimodal_user_input_frame(
+            image,
+            "请根据用户刚提交的图片进行回复。",
+        )
+
+        self.assertIsInstance(frame, LLMMessagesAppendFrame)
+        self.assertTrue(frame.run_llm)
+        self.assertEqual(frame.messages[0]["role"], "user")
+        content = frame.messages[0]["content"]
+        self.assertEqual(
+            content[0],
+            {"type": "text", "text": "请根据用户刚提交的图片进行回复。"},
+        )
+        self.assertEqual(content[1]["type"], "image_url")
+        self.assertTrue(
+            content[1]["image_url"]["url"].startswith(
+                "data:image/jpeg;base64,"
+            )
+        )
+
 
 if __name__ == "__main__":
     unittest.main()