Fix: can't upload image in ollama model #10447 (#10717)

### What problem does this PR solve?

Fix: can't upload image in ollama model #10447

### Type of change

- [X] Bug Fix (non-breaking change which fixes an issue)


### Change all `image=[]` to `image = None`

Changing `image=[]` to `images=None` avoids Python’s mutable default
parameter issue.
If you keep `images=[]`, all calls share the same list, so modifying it
(e.g., images.append()) will affect later calls.
Using images=None and creating a new list inside the function ensures
each call is independent.
This change does not affect current behavior — it simply makes the code
safer and more predictable.


把 `images=[]` 改成 `images=None` 是为了避免 Python 默认参数的可变对象问题。
如果保留 `images=[]`，所有调用都会共用同一个列表，一旦修改就会影响后续调用。
改成 None 并在函数内部重新创建列表，可以确保每次调用都是独立的。
这个修改不会影响现有运行结果，只是让代码更安全、更可控。

api/apps/user_app.py

@@ -22,7 +22,7 @@ import secrets
 import time
 from datetime import datetime
 
-from flask import redirect, request, session, Response
+from flask import redirect, request, session, make_response
 from flask_login import current_user, login_required, login_user, logout_user
 from werkzeug.security import check_password_hash, generate_password_hash
 
@@ -866,7 +866,9 @@ def forget_get_captcha():
     from captcha.image import ImageCaptcha
     image = ImageCaptcha(width=300, height=120, font_sizes=[50, 60, 70])
     img_bytes = image.generate(captcha_text).read()
-    return Response(img_bytes, mimetype="image/png")
+    response = make_response(img_bytes)
+    response.headers.set("Content-Type", "image/JPEG")
+    return response
 
 
 @manager.route("/forget/otp", methods=["POST"])  # noqa: F821

api/db/services/llm_service.py

@@ -210,19 +210,18 @@ class LLMBundle(LLM4Tenant):
     def _clean_param(chat_partial, **kwargs):
         func = chat_partial.func
         sig = inspect.signature(func)
-        keyword_args = []
         support_var_args = False
+        allowed_params = set()
+
         for param in sig.parameters.values():
-            if param.kind == inspect.Parameter.VAR_KEYWORD or param.kind == inspect.Parameter.VAR_POSITIONAL:
+            if param.kind == inspect.Parameter.VAR_KEYWORD:
                 support_var_args = True
-            elif param.kind == inspect.Parameter.KEYWORD_ONLY:
+            elif param.kind in (inspect.Parameter.POSITIONAL_OR_KEYWORD, inspect.Parameter.KEYWORD_ONLY):
-                keyword_args.append(param.name)
+                allowed_params.add(param.name)
-
+        if support_var_args:
-        use_kwargs = kwargs
+            return kwargs
-        if not support_var_args:
+        else:
-            use_kwargs = {k: v for k, v in kwargs.items() if k in keyword_args}
+            return {k: v for k, v in kwargs.items() if k in allowed_params}
-        return use_kwargs
-
     def chat(self, system: str, history: list, gen_conf: dict = {}, **kwargs) -> str:
         if self.langfuse:
             generation = self.langfuse.start_generation(trace_context=self.trace_context, name="chat", model=self.llm_name, input={"system": system, "history": history})

rag/llm/cv_model.py

@@ -50,7 +50,7 @@ class Base(ABC):
     def describe_with_prompt(self, image, prompt=None):
         raise NotImplementedError("Please implement encode method!")
 
-    def _form_history(self, system, history, images=[]):
+    def _form_history(self, system, history, images=None):
         hist = []
         if system:
             hist.append({"role": "system", "content": system})
@@ -78,7 +78,7 @@ class Base(ABC):
             })
         return pmpt
 
-    def chat(self, system, history, gen_conf, images=[], **kwargs):
+    def chat(self, system, history, gen_conf, images=None, **kwargs):
         try:
             response = self.client.chat.completions.create(
                 model=self.model_name,
@@ -89,7 +89,7 @@ class Base(ABC):
         except Exception as e:
             return "**ERROR**: " + str(e), 0
 
-    def chat_streamly(self, system, history, gen_conf, images=[], **kwargs):
+    def chat_streamly(self, system, history, gen_conf, images=None, **kwargs):
         ans = ""
         tk_count = 0
         try:
@@ -228,7 +228,7 @@ class QWenCV(GptV4):
             base_url = "https://dashscope.aliyuncs.com/compatible-mode/v1"
         super().__init__(key, model_name, lang=lang, base_url=base_url, **kwargs)
 
-    def chat(self, system, history, gen_conf, images=[], video_bytes=None, filename=""):
+    def chat(self, system, history, gen_conf, images=None, video_bytes=None, filename=""):
         if video_bytes:
             try:
                 summary, summary_num_tokens = self._process_video(video_bytes, filename)
@@ -506,7 +506,7 @@ class OllamaCV(Base):
             options["frequency_penalty"] = gen_conf["frequency_penalty"]
         return options
 
-    def _form_history(self, system, history, images=[]):
+    def _form_history(self, system, history, images=None):
         hist = deepcopy(history)
         if system and hist[0]["role"] == "user":
             hist.insert(0, {"role": "system", "content": system})
@@ -547,7 +547,7 @@ class OllamaCV(Base):
         except Exception as e:
             return "**ERROR**: " + str(e), 0
 
-    def chat(self, system, history, gen_conf, images=[]):
+    def chat(self, system, history, gen_conf, images=None):
         try:
             response = self.client.chat(
                 model=self.model_name,
@@ -561,7 +561,7 @@ class OllamaCV(Base):
         except Exception as e:
             return "**ERROR**: " + str(e), 0
 
-    def chat_streamly(self, system, history, gen_conf, images=[]):
+    def chat_streamly(self, system, history, gen_conf, images=None):
         ans = ""
         try:
             response = self.client.chat(
@@ -596,7 +596,7 @@ class GeminiCV(Base):
         self.lang = lang
         Base.__init__(self, **kwargs)
 
-    def _form_history(self, system, history, images=[]):
+    def _form_history(self, system, history, images=None):
         hist = []
         if system:
             hist.append({"role": "user", "parts": [system, history[0]["content"]]})
@@ -633,7 +633,7 @@ class GeminiCV(Base):
                 return res.text, total_token_count_from_response(res)
 
 
-    def chat(self, system, history, gen_conf, images=[], video_bytes=None, filename=""):
+    def chat(self, system, history, gen_conf, images=None, video_bytes=None, filename=""):
         if video_bytes:
             try:
                 summary, summary_num_tokens = self._process_video(video_bytes, filename)
@@ -651,7 +651,7 @@ class GeminiCV(Base):
         except Exception as e:
             return "**ERROR**: " + str(e), 0
 
-    def chat_streamly(self, system, history, gen_conf, images=[]):
+    def chat_streamly(self, system, history, gen_conf, images=None):
         ans = ""
         response = None
         try:
@@ -782,7 +782,7 @@ class NvidiaCV(Base):
             total_token_count_from_response(response)
         )
 
-    def chat(self, system, history, gen_conf, images=[], **kwargs):
+    def chat(self, system, history, gen_conf, images=None, **kwargs):
         try:
             response = self._request(self._form_history(system, history, images), gen_conf)
             return (
@@ -792,7 +792,7 @@ class NvidiaCV(Base):
         except Exception as e:
             return "**ERROR**: " + str(e), 0
 
-    def chat_streamly(self, system, history, gen_conf, images=[], **kwargs):
+    def chat_streamly(self, system, history, gen_conf, images=None, **kwargs):
         total_tokens = 0
         try:
             response = self._request(self._form_history(system, history, images), gen_conf)
@@ -858,7 +858,7 @@ class AnthropicCV(Base):
             gen_conf["max_tokens"] = self.max_tokens
         return gen_conf
 
-    def chat(self, system, history, gen_conf, images=[]):
+    def chat(self, system, history, gen_conf, images=None):
         gen_conf = self._clean_conf(gen_conf)
         ans = ""
         try:
@@ -879,7 +879,7 @@ class AnthropicCV(Base):
         except Exception as e:
             return ans + "\n**ERROR**: " + str(e), 0
 
-    def chat_streamly(self, system, history, gen_conf, images=[]):
+    def chat_streamly(self, system, history, gen_conf, images=None):
         gen_conf = self._clean_conf(gen_conf)
         total_tokens = 0
         try:
@@ -963,13 +963,13 @@ class GoogleCV(AnthropicCV, GeminiCV):
         else:
             return GeminiCV.describe_with_prompt(self, image, prompt)
 
-    def chat(self, system, history, gen_conf, images=[]):
+    def chat(self, system, history, gen_conf, images=None):
         if "claude" in self.model_name:
             return AnthropicCV.chat(self, system, history, gen_conf, images)
         else:
             return GeminiCV.chat(self, system, history, gen_conf, images)
 
-    def chat_streamly(self, system, history, gen_conf, images=[]):
+    def chat_streamly(self, system, history, gen_conf, images=None):
         if "claude" in self.model_name:
             for ans in AnthropicCV.chat_streamly(self, system, history, gen_conf, images):
                 yield ans