From 3860634fd21d85b4fe08d41d507833250c97f41a Mon Sep 17 00:00:00 2001
From: Raven95676 <Raven95676@gmail.com>
Date: Sat, 12 Apr 2025 19:15:39 +0800
Subject: [PATCH] =?UTF-8?q?fix:=20=E4=BF=AE=E5=A4=8D=E4=BA=86=E5=A4=9A?=
 =?UTF-8?q?=E6=A8=A1=E6=80=81=E8=BE=93=E5=87=BA=E6=94=AF=E6=8C=81=E5=88=A4?=
 =?UTF-8?q?=E6=96=AD=E9=97=AE=E9=A2=98=E5=B9=B6=E5=AF=B9=E5=8F=AA=E8=BE=93?=
 =?UTF-8?q?=E5=87=BA=E5=9B=BE=E7=89=87=E7=9A=84=E6=83=85=E5=86=B5=E8=BF=9B?=
 =?UTF-8?q?=E8=A1=8C=E5=A4=84=E7=90=86=E3=80=82?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 .../core/provider/sources/gemini_source.py    | 27 ++++++++++++++++---
 1 file changed, 23 insertions(+), 4 deletions(-)

diff --git a/astrbot/core/provider/sources/gemini_source.py b/astrbot/core/provider/sources/gemini_source.py
index a6de908ce..cc026bb0b 100644
--- a/astrbot/core/provider/sources/gemini_source.py
+++ b/astrbot/core/provider/sources/gemini_source.py
@@ -240,6 +240,13 @@ class ProviderGoogleGenAI(Provider):
 
         chain = []
         part: types.Part
+
+        # 暂时这样Fallback
+        if all(
+            part.inline_data and part.inline_data.mime_type.startswith("image/")
+            for part in result_parts
+        ):
+            chain.append(Comp.Plain("这是图片"))
         for part in result_parts:
             if part.text:
                 chain.append(Comp.Plain(part.text))
@@ -300,7 +307,7 @@ class ProviderGoogleGenAI(Provider):
                     logger.warning(f"{self.get_model()} 不支持函数调用，已自动去除")
                     tools = None
                 elif (
-                    "Multi-modal output is not supported"
+                    "Multi-modal output is not supported" in e.message
                     or "Model does not support the requested response modalities"
                     in e.message
                 ):
@@ -358,9 +365,21 @@ class ProviderGoogleGenAI(Provider):
             if chunk.candidates[0].content.parts and any(
                 part.function_call for part in chunk.candidates[0].content.parts
             ):
-                response = LLMResponse("assistant", is_chunk=False)
-                response.result_chain = self._process_content_parts(chunk, response)
-                yield response
+                llm_response = LLMResponse("assistant", is_chunk=False)
+                llm_response.result_chain = self._process_content_parts(
+                    chunk, llm_response
+                )
+                yield llm_response
+                break
+
+            if chunk.candidates[0].content.parts and any(
+                part.inline_data for part in chunk.candidates[0].content.parts
+            ):
+                llm_response = LLMResponse("assistant", is_chunk=False)
+                llm_response.result_chain = self._process_content_parts(
+                    chunk, llm_response
+                )
+                yield llm_response
                 break
 
             if chunk.text: