feat(image): 支持多模态模型输入base64格式图片

- 在消息转换中，增加对 `data:image/png;base64,...` 格式图片的支持，允许用户直接在输入中提供base64编码的图片。
- 调整图片处理逻辑，使其能够根据模型名称判断是否启用多模态能力，避免非多模态模型错误处理图片链接。
- 当未配置图床时，模型输出的图片将回退为base64格式，确保图片内容始终可用。
- 优化了相关函数的参数传递和代码格式，提高了代码的可读性和健壮性。

app/service/chat/openai_chat_service.py

@@ -285,7 +285,9 @@ class OpenAIChatService:
         api_key: str,
     ) -> Union[Dict[str, Any], AsyncGenerator[str, None]]:
         """创建聊天完成"""
-        messages, instruction = self.message_converter.convert(request.messages)
+        messages, instruction = self.message_converter.convert(
+            request.messages, request.model
+        )
 
         payload = _build_payload(request, messages, instruction)
 

app/service/image/image_create_service.py

@@ -9,8 +9,8 @@ from app.config.config import settings
 from app.core.constants import VALID_IMAGE_RATIOS
 from app.domain.openai_models import ImageGenerationRequest
 from app.log.logger import get_image_create_logger
-from app.utils.uploader import ImageUploaderFactory
 from app.utils.helpers import is_image_upload_configured
+from app.utils.uploader import ImageUploaderFactory
 
 logger = get_image_create_logger()
 
@@ -99,7 +99,10 @@ class ImageCreateService:
                 image_uploader = None
 
                 # Return base64 if explicitly requested or if no uploader is configured
-                if request.response_format == "b64_json" or not is_image_upload_configured():
+                if (
+                    request.response_format == "b64_json"
+                    or not is_image_upload_configured(settings)
+                ):
                     base64_image = base64.b64encode(image_data).decode("utf-8")
                     images_data.append(
                         {"b64_json": base64_image, "revised_prompt": request.prompt}