chore: 更新版本号至2.2.0

本次提交引入了一项新功能，允许模型在对话中理解所提供URL的上下文。

主要变更：

- **配置**：新增了 `URL_CONTEXT_ENABLED` 和 `URL_CONTEXT_MODELS` 两个配置项，用于控制此功能的开关和支持的模型列表。
- **后端服务**：在 `gemini_chat_service`、`openai_chat_service` 和 `vertex_express_chat_service` 中，为支持的模型动态添加 `urlContext` 工具。
- **前端界面**：在配置编辑器页面增加了相应的UI控件，方便用户启用/禁用该功能并管理支持的模型列表。
- **文档**：更新了 `.env.example`、`README.md` 和 `README_ZH.md`，包含了新配置项的说明。

.env.example

@@ -20,6 +20,9 @@ THINKING_BUDGET_MAP={"gemini-2.5-flash-preview-04-17": 4000}
 IMAGE_MODELS=["gemini-2.0-flash-exp"]
 SEARCH_MODELS=["gemini-2.0-flash-exp","gemini-2.0-pro-exp"]
 FILTERED_MODELS=["gemini-1.0-pro-vision-latest", "gemini-pro-vision", "chat-bison-001", "text-bison-001", "embedding-gecko-001"]
+# 是否启用网址上下文，默认启用
+URL_CONTEXT_ENABLED=true
+URL_CONTEXT_MODELS=["gemini-2.5-pro","gemini-2.5-flash","gemini-2.5-flash-lite","gemini-2.0-flash","gemini-2.0-flash-live-001"]
 TOOLS_CODE_EXECUTION_ENABLED=false
 SHOW_SEARCH_LINK=true
 SHOW_THINKING_PROCESS=true

README.md

@@ -192,6 +192,8 @@ If you want to run the source code directly locally for development or testing,
 | `THINKING_MODELS`              | Optional, list of models that support thinking functions                    | `[]`                                                                                                                                                                                                                                     |
 | `THINKING_BUDGET_MAP`          | Optional, thinking function budget mapping (model_name:budget_value)        | `{}`                                                                                                                                                                                                                                     |
 | `URL_NORMALIZATION_ENABLED`    | Optional, whether to enable intelligent URL routing mapping                 | `false`                                                                                                                                                                                                                                  |
+| `URL_CONTEXT_ENABLED`          | Optional, whether to enable URL context understanding                       | `false`                                                                                                                                                                                                                                  |
+| `URL_CONTEXT_MODELS`           | Optional, list of models that support URL context understanding             | `[]`                                                                                                                                                                                                                                     |
 | `BASE_URL`                     | Optional, Gemini API base URL, no modification needed by default            | `https://generativelanguage.googleapis.com/v1beta`                                                                                                                                                                                       |
 | `MAX_FAILURES`                 | Optional, number of times a single key is allowed to fail                   | `3`                                                                                                                                                                                                                                      |
 | `MAX_RETRIES`                  | Optional, maximum number of retries for failed API requests                 | `3`                                                                                                                                                                                                                                      |

README_ZH.md

@@ -185,6 +185,8 @@ app/
 | `THINKING_MODELS`            | 可选，支持思考功能的模型列表                                   | `[]`                                                  |
 | `THINKING_BUDGET_MAP`        | 可选，思考功能预算映射 (模型名:预算值)                         | `{}`                                                  |
 | `URL_NORMALIZATION_ENABLED`  | 可选，是否启用智能路由映射功能                                 | `false`                                               |
+| `URL_CONTEXT_ENABLED`        | 可选，是否启用URL上下文理解功能                                | `false`                                               |
+| `URL_CONTEXT_MODELS`         | 可选，支持URL上下文理解功能的模型列表                          | `[]`                                                  |
 | `BASE_URL`                   | 可选，Gemini API 基础 URL，默认无需修改                        | `https://generativelanguage.googleapis.com/v1beta`    |
 | `MAX_FAILURES`               | 可选，允许单个key失败的次数                                    | `3`                                                   |
 | `MAX_RETRIES`                | 可选，API 请求失败时的最大重试次数                             | `3`                                                   |

VERSION

@@ -1 +1 @@
-2.1.10
+2.2.0

app/config/config.py

@@ -51,8 +51,8 @@ class Settings(BaseSettings):
         return v
 
     # API相关配置
-    API_KEYS: List[str]
-    ALLOWED_TOKENS: List[str]
+    API_KEYS: List[str]=[]
+    ALLOWED_TOKENS: List[str]=[]
     BASE_URL: str = f"https://generativelanguage.googleapis.com/{API_VERSION}"
     AUTH_TOKEN: str = ""
     MAX_FAILURES: int = 3
@@ -75,6 +75,9 @@ class Settings(BaseSettings):
     IMAGE_MODELS: List[str] = ["gemini-2.0-flash-exp"]
     FILTERED_MODELS: List[str] = DEFAULT_FILTER_MODELS
     TOOLS_CODE_EXECUTION_ENABLED: bool = False
+    # 是否启用网址上下文
+    URL_CONTEXT_ENABLED: bool = True
+    URL_CONTEXT_MODELS: List[str] = ["gemini-2.5-pro","gemini-2.5-flash","gemini-2.5-flash-lite","gemini-2.0-flash","gemini-2.0-flash-live-001"]
     SHOW_SEARCH_LINK: bool = True
     SHOW_THINKING_PROCESS: bool = True
     THINKING_MODELS: List[str] = []

app/router/gemini_routes.py

@@ -329,7 +329,7 @@ async def verify_key(api_key: str, chat_service: GeminiChatService = Depends(get
                     parts=[{"text": "hi"}],
                 )
             ],
-            generation_config={"temperature": 0.7, "top_p": 1.0, "max_output_tokens": 10}
+            generation_config={"temperature": 0.7, "topP": 1.0, "maxOutputTokens": 10}
         )
         
         response = await chat_service.generate_content(
@@ -339,7 +339,9 @@ async def verify_key(api_key: str, chat_service: GeminiChatService = Depends(get
         )
         
         if response:
-            return JSONResponse({"status": "valid"})        
+            # 如果密钥验证成功，则重置其失败计数
+            await key_manager.reset_key_failure_count(api_key)
+            return JSONResponse({"status": "valid"})
     except Exception as e:
         logger.error(f"Key verification failed: {str(e)}")
         
@@ -374,7 +376,7 @@ async def verify_selected_keys(
         try:
             gemini_request = GeminiRequest(
                 contents=[GeminiContent(role="user", parts=[{"text": "hi"}])],
-                generation_config={"temperature": 0.7, "top_p": 1.0, "max_output_tokens": 10}
+                generation_config={"temperature": 0.7, "topP": 1.0, "maxOutputTokens": 10}
             )
             await chat_service.generate_content(
                 settings.TEST_MODEL,
@@ -382,6 +384,8 @@ async def verify_selected_keys(
                 api_key
             )
             successful_keys.append(api_key)
+            # 如果密钥验证成功，则重置其失败计数
+            await key_manager.reset_key_failure_count(api_key)
             return api_key, "valid", None
         except Exception as e:
             error_message = str(e)

app/service/chat/gemini_chat_service.py

@@ -119,15 +119,32 @@ def _build_tools(model: str, payload: Dict[str, Any]) -> List[Dict[str, Any]]:
         tool["codeExecution"] = {}
     if model.endswith("-search"):
         tool["googleSearch"] = {}
-
+    
+    real_model = _get_real_model(model)
+    if real_model in settings.URL_CONTEXT_MODELS and settings.URL_CONTEXT_ENABLED:
+        tool["urlContext"] = {}
+        
     # 解决 "Tool use with function calling is unsupported" 问题
     if tool.get("functionDeclarations"):
         tool.pop("googleSearch", None)
         tool.pop("codeExecution", None)
+        tool.pop("urlContext", None)
 
     return [tool] if tool else []
 
 
+def _get_real_model(model: str) -> str:
+    if model.endswith("-search"):
+        model = model[:-7]
+    if model.endswith("-image"):
+        model = model[:-6]
+    if model.endswith("-non-thinking"):
+        model = model[:-13]
+    if "-search" in model and "-non-thinking" in model:
+        model = model[:-20]
+    return model
+
+
 def _get_safety_settings(model: str) -> List[Dict[str, str]]:
     """获取安全设置"""
     if model == "gemini-2.0-flash-exp":
@@ -157,11 +174,12 @@ def _filter_empty_parts(contents: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
 
 def _build_payload(model: str, request: GeminiRequest) -> Dict[str, Any]:
     """构建请求payload"""
-    request_dict = request.model_dump(by_alias=True, exclude_none=False)
+    request_dict = request.model_dump(exclude_none=False)
     if request.generationConfig:
         if request.generationConfig.maxOutputTokens is None:
             # 如果未指定最大输出长度，则不传递该字段，解决截断的问题
-            request_dict["generationConfig"].pop("maxOutputTokens")
+            if "maxOutputTokens" in request_dict["generationConfig"]:
+                request_dict["generationConfig"].pop("maxOutputTokens")
 
     # 检查是否为TTS模型
     is_tts_model = "tts" in model.lower()
@@ -205,7 +223,10 @@ def _build_payload(model: str, request: GeminiRequest) -> Dict[str, Any]:
     else:
         # 客户端没有提供思考配置，使用默认配置    
         if model.endswith("-non-thinking"):
-            payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": 0} 
+            if "gemini-2.5-pro" in model:
+                payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": 128}
+            else:
+                payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": 0} 
         elif model in settings.THINKING_BUDGET_MAP:
             if settings.SHOW_THINKING_PROCESS:
                 payload["generationConfig"]["thinkingConfig"] = {

app/service/chat/openai_chat_service.py

@@ -87,6 +87,10 @@ def _build_tools(
 
     if model.endswith("-search"):
         tool["googleSearch"] = {}
+        
+    real_model = _get_real_model(model)
+    if real_model in settings.URL_CONTEXT_MODELS and settings.URL_CONTEXT_ENABLED:
+        tool["urlContext"] = {}
 
     # 将 request 中的 tools 合并到 tools 中
     if request.tools:
@@ -126,10 +130,23 @@ def _build_tools(
     if tool.get("functionDeclarations"):
         tool.pop("googleSearch", None)
         tool.pop("codeExecution", None)
+        tool.pop("urlContext",None)
 
     return [tool] if tool else []
 
 
+def _get_real_model(model: str) -> str:
+    if model.endswith("-search"):
+        model = model[:-7]
+    if model.endswith("-image"):
+        model = model[:-6]
+    if model.endswith("-non-thinking"):
+        model = model[:-13]
+    if "-search" in model and "-non-thinking" in model:
+        model = model[:-20]
+    return model
+
+
 def _get_safety_settings(model: str) -> List[Dict[str, str]]:
     """获取安全设置"""
     # if (
@@ -184,7 +201,10 @@ def _build_payload(
         payload["generationConfig"]["responseModalities"] = ["Text", "Image"]
     
     if request.model.endswith("-non-thinking"):
-        payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": 0}
+        if "gemini-2.5-pro" in request.model:
+            payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": 128}
+        else:
+            payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": 0} 
     
     if request.model in settings.THINKING_BUDGET_MAP:
         if settings.SHOW_THINKING_PROCESS:

app/service/chat/vertex_express_chat_service.py

@@ -97,15 +97,32 @@ def _build_tools(model: str, payload: Dict[str, Any]) -> List[Dict[str, Any]]:
         tool["codeExecution"] = {}
     if model.endswith("-search"):
         tool["googleSearch"] = {}
+        
+    real_model = _get_real_model(model)
+    if real_model in settings.URL_CONTEXT_MODELS and settings.URL_CONTEXT_ENABLED:
+        tool["urlContext"] = {}
 
     # 解决 "Tool use with function calling is unsupported" 问题
     if tool.get("functionDeclarations"):
         tool.pop("googleSearch", None)
         tool.pop("codeExecution", None)
+        tool.pop("urlContext", None)
 
     return [tool] if tool else []
 
 
+def _get_real_model(model: str) -> str:
+    if model.endswith("-search"):
+        model = model[:-7]
+    if model.endswith("-image"):
+        model = model[:-6]
+    if model.endswith("-non-thinking"):
+        model = model[:-13]
+    if "-search" in model and "-non-thinking" in model:
+        model = model[:-20]
+    return model
+
+
 def _get_safety_settings(model: str) -> List[Dict[str, str]]:
     """获取安全设置"""
     if model == "gemini-2.0-flash-exp":
@@ -115,7 +132,7 @@ def _get_safety_settings(model: str) -> List[Dict[str, str]]:
 
 def _build_payload(model: str, request: GeminiRequest) -> Dict[str, Any]:
     """构建请求payload"""
-    request_dict = request.model_dump(by_alias=True, exclude_none=False)
+    request_dict = request.model_dump(exclude_none=False)
     if request.generationConfig:
         if request.generationConfig.maxOutputTokens is None:
             # 如果未指定最大输出长度，则不传递该字段，解决截断的问题
@@ -144,7 +161,10 @@ def _build_payload(model: str, request: GeminiRequest) -> Dict[str, Any]:
     else:
         # 客户端没有提供思考配置，使用默认配置    
         if model.endswith("-non-thinking"):
-            payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": 0} 
+            if "gemini-2.5-pro" in model:
+                payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": 128}
+            else:
+                payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": 0} 
         elif model in settings.THINKING_BUDGET_MAP:
             if settings.SHOW_THINKING_PROCESS:
                 payload["generationConfig"]["thinkingConfig"] = {

app/service/tts/native/tts_chat_service.py

@@ -5,7 +5,7 @@
 
 import time
 import datetime
-from typing import Any, Dict, Optional
+from typing import Any, Dict
 from app.service.chat.gemini_chat_service import GeminiChatService
 from app.service.tts.native.tts_response_handler import TTSResponseHandler
 from app.domain.gemini_models import GeminiRequest
@@ -69,7 +69,7 @@ class TTSGeminiChatService(GeminiChatService):
             # 构建TTS专用的payload - 不包含tools和safetySettings
             from app.service.chat.gemini_chat_service import _filter_empty_parts
 
-            request_dict = request.model_dump(by_alias=True, exclude_none=False)
+            request_dict = request.model_dump(exclude_none=False)
 
             # 构建TTS专用的简化payload
             payload = {
@@ -130,7 +130,7 @@ class TTSGeminiChatService(GeminiChatService):
                 error_type="tts-api-error",
                 error_log=error_msg,
                 error_code=status_code,
-                request_msg=request.model_dump(by_alias=True, exclude_none=False)
+                request_msg=request.model_dump(exclude_none=False)
             )
 
             logger.error(f"TTS API call failed: {error_msg}")

app/static/js/config_editor.js

@@ -13,7 +13,7 @@ const SHOW_CLASS = "show"; // For modals
 const API_KEY_REGEX = /AIzaSy\S{33}/g;
 const PROXY_REGEX =
   /(?:https?|socks5):\/\/(?:[^:@\/]+(?::[^@\/]+)?@)?(?:[^:\/\s]+)(?::\d+)?/g;
-const VERTEX_API_KEY_REGEX = /AQ\.[a-zA-Z0-9_]{50}/g; // 新增 Vertex Express API Key 正则
+const VERTEX_API_KEY_REGEX = /AQ\.[a-zA-Z0-9_\-]{50}/g; // 新增 Vertex Express API Key 正则
 const MASKED_VALUE = "••••••••";
 
 // DOM Elements - Global Scope for frequently accessed elements
@@ -713,7 +713,13 @@ async function initConfig() {
       config.SAFETY_SETTINGS = []; // 默认为空数组
     }
     // --- 结束：处理 SAFETY_SETTINGS 默认值 ---
-
+    if (typeof config.URL_CONTEXT_ENABLED === "undefined") {
+      config.URL_CONTEXT_ENABLED = true;
+    }
+    if (!config.URL_CONTEXT_MODELS || !Array.isArray(config.URL_CONTEXT_MODELS)) {
+      config.URL_CONTEXT_MODELS = [];
+    }
+ 
     // --- 新增：处理自动删除错误日志配置的默认值 ---
     if (typeof config.AUTO_DELETE_ERROR_LOGS_ENABLED === "undefined") {
       config.AUTO_DELETE_ERROR_LOGS_ENABLED = false;
@@ -1462,7 +1468,7 @@ function addArrayItem(key) {
   const modelId = addArrayItemWithValue(key, newItemValue); // This adds the DOM element
 
   if (key === "THINKING_MODELS" && modelId) {
-    createAndAppendBudgetMapItem(newItemValue, 0, modelId); // Default budget 0
+    createAndAppendBudgetMapItem(newItemValue, -1, modelId); // Default budget -1
   }
 }
 
@@ -1572,7 +1578,7 @@ function createAndAppendBudgetMapItem(mapKey, mapValue, modelId) {
   const valueInput = document.createElement("input");
   valueInput.type = "number";
   const intValue = parseInt(mapValue, 10);
-  valueInput.value = isNaN(intValue) ? 0 : intValue;
+  valueInput.value = isNaN(intValue) ? -1 : intValue;
   valueInput.placeholder = "预算 (整数)";
   valueInput.className = `${MAP_VALUE_INPUT_CLASS} w-24 px-3 py-2 border border-gray-300 rounded-md focus:outline-none focus:border-primary-500 focus:ring focus:ring-primary-200 focus:ring-opacity-50`;
   valueInput.min = -1;
@@ -1733,7 +1739,7 @@ function collectFormData() {
         formData["THINKING_BUDGET_MAP"][keyInput.value.trim()] = isNaN(
           budgetValue
         )
-          ? 0
+          ? -1
           : budgetValue;
       }
     });
@@ -2284,7 +2290,7 @@ function handleModelSelection(selectedModelId) {
     );
     if (currentModelHelperTarget.targetKey === "THINKING_MODELS" && modelId) {
       // Automatically add corresponding budget map item with default budget 0
-      createAndAppendBudgetMapItem(selectedModelId, 0, modelId);
+      createAndAppendBudgetMapItem(selectedModelId, -1, modelId);
     }
   }
 

app/templates/config_editor.html

@@ -1421,6 +1421,59 @@ endblock %} {% block head_extra_styles %}
           </div>
         </div>
 
+        <!-- 启用网址上下文 -->
+        <div class="mb-6 flex items-center justify-between">
+          <label for="URL_CONTEXT_ENABLED" class="font-semibold text-gray-700"
+            >启用网址上下文</label
+          >
+          <div
+            class="relative inline-block w-10 mr-2 align-middle select-none transition duration-200 ease-in"
+          >
+            <input
+              type="checkbox"
+              name="URL_CONTEXT_ENABLED"
+              id="URL_CONTEXT_ENABLED"
+              class="toggle-checkbox absolute block w-6 h-6 rounded-full bg-white border-4 appearance-none cursor-pointer"
+            />
+            <label
+              for="URL_CONTEXT_ENABLED"
+              class="toggle-label block overflow-hidden h-6 rounded-full bg-gray-300 cursor-pointer"
+            ></label>
+          </div>
+        </div>
+
+        <!-- 网址上下文模型列表 -->
+        <div class="mb-6">
+          <label
+            for="URL_CONTEXT_MODELS"
+            class="block font-semibold mb-2 text-gray-700"
+            >网址上下文模型列表</label
+          >
+          <div class="array-container" id="URL_CONTEXT_MODELS_container">
+            <!-- 数组项将在这里动态添加 -->
+          </div>
+          <div class="flex justify-end gap-2 mt-2">
+            <button
+              type="button"
+              title="从列表选择模型添加到下方"
+              class="model-helper-trigger-btn p-2 rounded-md text-violet-300 hover:bg-violet-700 transition-colors"
+              data-target-array-key="URL_CONTEXT_MODELS"
+            >
+              <i class="fas fa-list-ul"></i>
+            </button>
+            <button
+              type="button"
+              class="bg-violet-600 hover:bg-violet-700 text-white px-4 py-2 rounded-lg font-medium transition-all duration-200 flex items-center gap-2"
+              onclick="addArrayItem('URL_CONTEXT_MODELS')"
+            >
+              <i class="fas fa-plus"></i> 添加模型
+            </button>
+          </div>
+          <small class="text-gray-500 mt-1 block"
+            >支持网址上下文功能的模型列表</small
+          >
+        </div>
+
         <!-- 显示搜索链接 -->
         <div class="mb-6 flex items-center justify-between">
           <label for="SHOW_SEARCH_LINK" class="font-semibold text-gray-700"
@@ -1523,7 +1576,7 @@ endblock %} {% block head_extra_styles %}
                             </button>
                         </div> -->
           <small class="text-gray-500 mt-1 block"
-            >为每个思考模型设置预算（整数，最大值
+            >为每个思考模型设置预算（-1为auto，最大值
             32767），此项与上方模型列表自动关联。</small
           >
         </div>