chore: 更新版本号至 2.1.9

- 修改了 response_handler.py 中的 _handle_openai_stream_response 和 _handle_openai_normal_response 方法，增加了对推理内容 (reasoning_content) 的支持。
- 更新了 _extract_result 方法的返回值，确保能够提取推理内容。
- 在 gemini_chat_service.py 和 openai_chat_service.py 中，调整了生成配置以包含思考过程的选项。
- 在 vertex_express_chat_service.py 中，增强了对客户端思考配置的处理逻辑，确保优先使用客户端提供的配置。

.env.example

@@ -43,6 +43,7 @@ SMMS_SECRET_TOKEN=XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
 PICGO_API_KEY=xxxx
 CLOUDFLARE_IMGBED_URL=https://xxxxxxx.pages.dev/upload
 CLOUDFLARE_IMGBED_AUTH_CODE=xxxxxxxxx
+CLOUDFLARE_IMGBED_UPLOAD_FOLDER=
 ##########################################################################
 #########################stream_optimizer 相关配置########################
 STREAM_OPTIMIZER_ENABLED=false

.github/workflows/release.yml

@@ -3,7 +3,7 @@ name: Publish Release
 on:
   push:
     tags:
-      - 'v*' # 当推送以 "v" 开头的标签时触发（如 v1.0.0, v2.1.0）
+      - "v*" # 当推送以 "v" 开头的标签时触发（如 v1.0.0, v2.1.0）
 
 jobs:
   update-release-draft:
@@ -15,8 +15,17 @@ jobs:
       # Step 1: 检出代码库
       - name: Checkout code
         uses: actions/checkout@v3
+        with:
+          fetch-depth: 0
 
-      # Step 2: 自动生成 Release
+      # Step 2: 自动生成 Release Notes
+      - name: Generate release notes
+        id: changelog
+        uses: mikepenz/release-changelog-builder-action@v4
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+      # Step 3: 自动生成 Release
       - name: Create Release
         id: create_release
         uses: actions/create-release@v1
@@ -25,15 +34,16 @@ jobs:
         with:
           tag_name: ${{ github.ref_name }}
           release_name: ${{ github.ref_name }}
+          body: ${{ steps.changelog.outputs.changelog }}
           draft: false
           prerelease: false
-      
-      # Step 3: 可选，构建zip文件
+
+      # Step 4: 可选，构建zip文件
       - name: Create ZIP file
         run: |
           zip -r gemini-balance.zip . -x "*.git*" "*.github*" "*.env*" "logs/*" "tests/*"
 
-      # Step 4: 可选，上传构建文件
+      # Step 5: 可选，上传构建文件
       - name: Upload Release Asset
         uses: actions/upload-release-asset@v1
         env:
@@ -41,5 +51,5 @@ jobs:
         with:
           upload_url: ${{ steps.create_release.outputs.upload_url }}
           asset_path: ./gemini-balance.zip # 替换为你的构建文件路径
-          asset_name: gemini-balance.zip  # 替换为你的文件名
+          asset_name: gemini-balance.zip # 替换为你的文件名
           asset_content_type: application/zip

Dockerfile

@@ -15,6 +15,7 @@ ENV TOOLS_CODE_EXECUTION_ENABLED=false
 ENV IMAGE_MODELS='["gemini-2.0-flash-exp"]'
 ENV SEARCH_MODELS='["gemini-2.0-flash-exp","gemini-2.0-pro-exp"]'
 ENV URL_NORMALIZATION_ENABLED=false
+ENV CLOUDFLARE_IMGBED_UPLOAD_FOLDER=""
 
 # Expose port
 EXPOSE 8000

README.md

@@ -2,6 +2,12 @@
 
 # Gemini Balance - Gemini API Proxy and Load Balancer
 
+<p align="center">
+  <a href="https://trendshift.io/repositories/13692" target="_blank">
+    <img src="https://trendshift.io/api/badge/repositories/13692" alt="snailyp%2Fgemini-balance | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/>
+  </a>
+</p>
+
 > ⚠️ This project is licensed under the CC BY-NC 4.0 (Attribution-NonCommercial) license. Any form of commercial resale service is prohibited. See the LICENSE file for details.
 
 > I have never sold this service on any platform. If you encounter someone selling this service, they are definitely a reseller. Please be careful not to be deceived.
@@ -211,6 +217,7 @@ If you want to run the source code directly locally for development or testing,
 | `PICGO_API_KEY`                | Optional, API Key for [PicoGo](https://www.picgo.net/) image hosting        | `your-picogo-apikey`                                                                                                                                                                                                                         |
 | `CLOUDFLARE_IMGBED_URL`        | Optional, [CloudFlare](https://github.com/MarSeventh/CloudFlare-ImgBed) image hosting upload address | `https://xxxxxxx.pages.dev/upload`                                                                                                                                                                                           |
 | `CLOUDFLARE_IMGBED_AUTH_CODE`  | Optional, authentication key for CloudFlare image hosting                   | `your-cloudflare-imgber-auth-code`                                                                                                                                                                                                           |
+| `CLOUDFLARE_IMGBED_UPLOAD_FOLDER` | Optional, upload folder path for CloudFlare image hosting               | `""`                                                                                                                                                                                                                                          |
 | **Stream Optimizer Related**   |                                                                             |                                                                                                                                                                                                                                          |
 | `STREAM_OPTIMIZER_ENABLED`     | Optional, whether to enable stream output optimization                      | `false`                                                                                                                                                                                                                                  |
 | `STREAM_MIN_DELAY`             | Optional, minimum delay for stream output                                   | `0.016`                                                                                                                                                                                                                                  |

README_ZH.md

@@ -1,5 +1,11 @@
 # Gemini Balance - Gemini API 代理和负载均衡器
 
+<p align="center">
+  <a href="https://trendshift.io/repositories/13692" target="_blank">
+    <img src="https://trendshift.io/api/badge/repositories/13692" alt="snailyp%2Fgemini-balance | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/>
+  </a>
+</p>
+
 > ⚠️ 本项目采用 CC BY-NC 4.0（署名-非商业性使用）协议，禁止任何形式的商业倒卖服务，详见 LICENSE 文件。
 
 > 本人从未在各个平台售卖服务，如有遇到售卖此服务者，那一定是倒卖狗，大家切记不要上当受骗。
@@ -204,6 +210,7 @@ app/
 | `PICGO_API_KEY`              | 可选，[PicoGo](https://www.picgo.net/)图床的API Key                                      | `your-picogo-apikey`                                  |
 | `CLOUDFLARE_IMGBED_URL`      | 可选，[CloudFlare](https://github.com/MarSeventh/CloudFlare-ImgBed) 图床上传地址                                  | `https://xxxxxxx.pages.dev/upload`                    |
 | `CLOUDFLARE_IMGBED_AUTH_CODE`| 可选，CloudFlare图床的鉴权key                                  | `your-cloudflare-imgber-auth-code`                    |
+| `CLOUDFLARE_IMGBED_UPLOAD_FOLDER`| 可选，CloudFlare图床的上传文件夹路径                        | `""`                                                  |
 | **流式优化器相关**           |                                                          |                                                       |
 | `STREAM_OPTIMIZER_ENABLED`   | 可选，是否启用流式输出优化                                     | `false`                                               |
 | `STREAM_MIN_DELAY`           | 可选，流式输出最小延迟                                         | `0.016`                                               |

VERSION

@@ -1 +1 @@
-2.1.7
+2.1.9

app/config/config.py

@@ -4,7 +4,7 @@
 
 import datetime
 import json
-from typing import Any, Dict, List, Type
+from typing import Any, Dict, List, Type, get_args, get_origin
 
 from pydantic import ValidationError, ValidationInfo, field_validator
 from pydantic_settings import BaseSettings
@@ -67,6 +67,9 @@ class Settings(BaseSettings):
     # 智能路由配置
     URL_NORMALIZATION_ENABLED: bool = False  # 是否启用智能路由映射功能
 
+    # 自定义 Headers
+    CUSTOM_HEADERS: Dict[str, str] = {}
+
     # 模型相关配置
     SEARCH_MODELS: List[str] = ["gemini-2.0-flash-exp"]
     IMAGE_MODELS: List[str] = ["gemini-2.0-flash-exp"]
@@ -90,6 +93,7 @@ class Settings(BaseSettings):
     PICGO_API_KEY: str = ""
     CLOUDFLARE_IMGBED_URL: str = ""
     CLOUDFLARE_IMGBED_AUTH_CODE: str = ""
+    CLOUDFLARE_IMGBED_UPLOAD_FOLDER: str = ""
 
     # 流式输出优化器配置
     STREAM_OPTIMIZER_ENABLED: bool = False
@@ -137,86 +141,106 @@ def _parse_db_value(key: str, db_value: str, target_type: Type) -> Any:
 
     logger = get_config_logger()
     try:
-        # 处理 List[str]
-        if target_type == List[str]:
-            try:
-                parsed = json.loads(db_value)
-                if isinstance(parsed, list):
-                    return [str(item) for item in parsed]
-            except json.JSONDecodeError:
+        origin_type = get_origin(target_type)
+        args = get_args(target_type)
+
+        # 处理 List 类型
+        if origin_type is list:
+            # 处理 List[str]
+            if args and args[0] == str:
+                try:
+                    parsed = json.loads(db_value)
+                    if isinstance(parsed, list):
+                        return [str(item) for item in parsed]
+                except json.JSONDecodeError:
+                    return [item.strip() for item in db_value.split(",") if item.strip()]
+                logger.warning(
+                    f"Could not parse '{db_value}' as List[str] for key '{key}', falling back to comma split or empty list."
+                )
                 return [item.strip() for item in db_value.split(",") if item.strip()]
-            logger.warning(
-                f"Could not parse '{db_value}' as List[str] for key '{key}', falling back to comma split or empty list."
-            )
-            return [item.strip() for item in db_value.split(",") if item.strip()]
-        # 处理 Dict[str, float]
-        elif target_type == Dict[str, float]:
-            parsed_dict = {}
-            try:
-                parsed = json.loads(db_value)
-                if isinstance(parsed, dict):
-                    parsed_dict = {str(k): float(v) for k, v in parsed.items()}
-                else:
-                    logger.warning(
-                        f"Parsed DB value for key '{key}' is not a dictionary type. Value: {db_value}"
-                    )
-            except (json.JSONDecodeError, ValueError, TypeError) as e1:
-                if isinstance(e1, json.JSONDecodeError) and "'" in db_value:
-                    logger.warning(
-                        f"Failed initial JSON parse for key '{key}'. Attempting to replace single quotes. Error: {e1}"
-                    )
-                    try:
-                        corrected_db_value = db_value.replace("'", '"')
-                        parsed = json.loads(corrected_db_value)
-                        if isinstance(parsed, dict):
-                            parsed_dict = {str(k): float(v) for k, v in parsed.items()}
+            # 处理 List[Dict[str, str]]
+            elif args and get_origin(args[0]) is dict:
+                try:
+                    parsed = json.loads(db_value)
+                    if isinstance(parsed, list):
+                        valid = all(
+                            isinstance(item, dict)
+                            and all(isinstance(k, str) for k in item.keys())
+                            and all(isinstance(v, str) for v in item.values())
+                            for item in parsed
+                        )
+                        if valid:
+                            return parsed
                         else:
                             logger.warning(
-                                f"Parsed DB value (after quote replacement) for key '{key}' is not a dictionary type. Value: {corrected_db_value}"
+                                f"Invalid structure in List[Dict[str, str]] for key '{key}'. Value: {db_value}"
                             )
-                    except (json.JSONDecodeError, ValueError, TypeError) as e2:
-                        logger.error(
-                            f"Could not parse '{db_value}' as Dict[str, float] for key '{key}' even after replacing quotes: {e2}. Returning empty dict."
-                        )
-                else:
-                    logger.error(
-                        f"Could not parse '{db_value}' as Dict[str, float] for key '{key}': {e1}. Returning empty dict."
-                    )
-            return parsed_dict
-        # 处理 List[Dict[str, str]]
-        elif target_type == List[Dict[str, str]]:
-            try:
-                parsed = json.loads(db_value)
-                if isinstance(parsed, list):
-                    # 验证列表中的每个元素是否为字典，并且键和值都是字符串
-                    valid = all(
-                        isinstance(item, dict)
-                        and all(isinstance(k, str) for k in item.keys())
-                        and all(isinstance(v, str) for v in item.values())
-                        for item in parsed
-                    )
-                    if valid:
-                        return parsed
+                            return []
                     else:
                         logger.warning(
-                            f"Invalid structure in List[Dict[str, str]] for key '{key}'. Value: {db_value}"
+                            f"Parsed DB value for key '{key}' is not a list type. Value: {db_value}"
                         )
                         return []
-                else:
-                    logger.warning(
-                        f"Parsed DB value for key '{key}' is not a list type. Value: {db_value}"
+                except json.JSONDecodeError:
+                    logger.error(
+                        f"Could not parse '{db_value}' as JSON for List[Dict[str, str]] for key '{key}'. Returning empty list."
                     )
                     return []
-            except json.JSONDecodeError:
-                logger.error(
-                    f"Could not parse '{db_value}' as JSON for List[Dict[str, str]] for key '{key}'. Returning empty list."
-                )
-                return []
-            except Exception as e:
-                logger.error(
-                    f"Error parsing List[Dict[str, str]] for key '{key}': {e}. Value: {db_value}. Returning empty list."
-                )
-                return []
+                except Exception as e:
+                    logger.error(
+                        f"Error parsing List[Dict[str, str]] for key '{key}': {e}. Value: {db_value}. Returning empty list."
+                    )
+                    return []
+        # 处理 Dict 类型
+        elif origin_type is dict:
+            # 处理 Dict[str, str]
+            if args and args == (str, str):
+                parsed_dict = {}
+                try:
+                    parsed = json.loads(db_value)
+                    if isinstance(parsed, dict):
+                        parsed_dict = {str(k): str(v) for k, v in parsed.items()}
+                    else:
+                        logger.warning(
+                            f"Parsed DB value for key '{key}' is not a dictionary type. Value: {db_value}"
+                        )
+                except json.JSONDecodeError:
+                     logger.error(f"Could not parse '{db_value}' as Dict[str, str] for key '{key}'. Returning empty dict.")
+                return parsed_dict
+            # 处理 Dict[str, float]
+            elif args and args == (str, float):
+                parsed_dict = {}
+                try:
+                    parsed = json.loads(db_value)
+                    if isinstance(parsed, dict):
+                        parsed_dict = {str(k): float(v) for k, v in parsed.items()}
+                    else:
+                        logger.warning(
+                            f"Parsed DB value for key '{key}' is not a dictionary type. Value: {db_value}"
+                        )
+                except (json.JSONDecodeError, ValueError, TypeError) as e1:
+                    if isinstance(e1, json.JSONDecodeError) and "'" in db_value:
+                        logger.warning(
+                            f"Failed initial JSON parse for key '{key}'. Attempting to replace single quotes. Error: {e1}"
+                        )
+                        try:
+                            corrected_db_value = db_value.replace("'", '"')
+                            parsed = json.loads(corrected_db_value)
+                            if isinstance(parsed, dict):
+                                parsed_dict = {str(k): float(v) for k, v in parsed.items()}
+                            else:
+                                logger.warning(
+                                    f"Parsed DB value (after quote replacement) for key '{key}' is not a dictionary type. Value: {corrected_db_value}"
+                                )
+                        except (json.JSONDecodeError, ValueError, TypeError) as e2:
+                            logger.error(
+                                f"Could not parse '{db_value}' as Dict[str, float] for key '{key}' even after replacing quotes: {e2}. Returning empty dict."
+                            )
+                    else:
+                        logger.error(
+                            f"Could not parse '{db_value}' as Dict[str, float] for key '{key}': {e1}. Returning empty dict."
+                        )
+                return parsed_dict
         # 处理 bool
         elif target_type == bool:
             return db_value.lower() in ("true", "1", "yes", "on")
@@ -305,18 +329,12 @@ async def sync_initial_settings():
                         if parsed_db_value != memory_value:
                             # 检查类型是否匹配，以防解析函数返回了不兼容的类型
                             type_match = False
-                            if target_type == List[str] and isinstance(
-                                parsed_db_value, list
-                            ):
-                                type_match = True
-                            elif target_type == Dict[str, float] and isinstance(
-                                parsed_db_value, dict
-                            ):
-                                type_match = True
-                            elif target_type not in (
-                                List[str],
-                                Dict[str, float],
-                            ) and isinstance(parsed_db_value, target_type):
+                            origin_type = get_origin(target_type)
+                            if origin_type:  # It's a generic type
+                                if isinstance(parsed_db_value, origin_type):
+                                    type_match = True
+                            # It's a non-generic type, or a specific generic we want to handle
+                            elif isinstance(parsed_db_value, target_type):
                                 type_match = True
 
                             if type_match:

app/core/constants.py

@@ -76,4 +76,15 @@ DEFAULT_SAFETY_SETTINGS = [
         {"category": "HARM_CATEGORY_SEXUALLY_EXPLICIT", "threshold": "OFF"},
         {"category": "HARM_CATEGORY_DANGEROUS_CONTENT", "threshold": "OFF"},
         {"category": "HARM_CATEGORY_CIVIC_INTEGRITY", "threshold": "BLOCK_NONE"},
-    ]
+    ]
+
+TTS_VOICE_NAMES = [
+    "Zephyr", "Puck", "Charon", "Kore",
+    "Fenrir", "Leda", "Orus", "Aoede",
+    "Callirhoe", "Autonoe", "Enceladus", "Iapetus",
+    "Umbriel", "Algieba", "Despina", "Erinome",
+    "Algenib", "Rasalgethi", "Laomedeia", "Achernar",
+    "Alnilam", "Schedar", "Gacrux", "Pulcherrima",
+    "Achird", "Zubenelgenubi", "Vindemiatrix", "Sadachbia",
+    "Sadaltager", "Sulafat"
+]

app/handler/response_handler.py

@@ -39,13 +39,13 @@ class GeminiResponseHandler(ResponseHandler):
 def _handle_openai_stream_response(
     response: Dict[str, Any], model: str, finish_reason: str, usage_metadata: Optional[Dict[str, Any]]
 ) -> Dict[str, Any]:
-    text, tool_calls, _ = _extract_result(
+    text, reasoning_content, tool_calls, _ = _extract_result(
         response, model, stream=True, gemini_format=False
     )
-    if not text and not tool_calls:
+    if not text and not tool_calls and not reasoning_content:
         delta = {}
     else:
-        delta = {"content": text, "role": "assistant"}
+        delta = {"content": text, "reasoning_content": reasoning_content, "role": "assistant"}
         if tool_calls:
             delta["tool_calls"] = tool_calls
     template_chunk = {
@@ -63,7 +63,7 @@ def _handle_openai_stream_response(
 def _handle_openai_normal_response(
     response: Dict[str, Any], model: str, finish_reason: str, usage_metadata: Optional[Dict[str, Any]]
 ) -> Dict[str, Any]:
-    text, tool_calls, _ = _extract_result(
+    text, reasoning_content, tool_calls, _ = _extract_result(
         response, model, stream=False, gemini_format=False
     )
     return {
@@ -77,6 +77,7 @@ def _handle_openai_normal_response(
                 "message": {
                     "role": "assistant",
                     "content": text,
+                    "reasoning_content": reasoning_content,
                     "tool_calls": tool_calls,
                 },
                 "finish_reason": finish_reason,
@@ -156,19 +157,21 @@ def _extract_result(
     model: str,
     stream: bool = False,
     gemini_format: bool = False,
-) -> tuple[str, List[Dict[str, Any]], Optional[bool]]:
-    text, tool_calls = "", []
-    thought = None
+) -> tuple[str, Optional[str], List[Dict[str, Any]], Optional[bool]]:
+    text, reasoning_content, tool_calls, thought = "", "", [], None
     if stream:
         if response.get("candidates"):
             candidate = response["candidates"][0]
             content = candidate.get("content", {})
             parts = content.get("parts", [])
             if not parts:
-                return "", [], None
+                return "", None, [], None
             if "text" in parts[0]:
                 text = parts[0].get("text")
                 if "thought" in parts[0]:
+                    if not gemini_format and settings.SHOW_THINKING_PROCESS:
+                        reasoning_content = text
+                        text = ""
                     thought = parts[0].get("thought")
             elif "executableCode" in parts[0]:
                 text = _format_code_block(parts[0]["executableCode"])
@@ -187,32 +190,18 @@ def _extract_result(
     else:
         if response.get("candidates"):
             candidate = response["candidates"][0]
-            if "thinking" in model:
-                if settings.SHOW_THINKING_PROCESS:
-                    if len(candidate["content"]["parts"]) == 2:
-                        text = (
-                            "> thinking\n\n"
-                            + candidate["content"]["parts"][0]["text"]
-                            + "\n\n---\n> output\n\n"
-                            + candidate["content"]["parts"][1]["text"]
-                        )
-                    else:
-                        text = candidate["content"]["parts"][0]["text"]
-                else:
-                    if len(candidate["content"]["parts"]) == 2:
-                        text = candidate["content"]["parts"][1]["text"]
-                    else:
-                        text = candidate["content"]["parts"][0]["text"]
-            else:
-                text = ""
-                if "parts" in candidate["content"]:
-                    for part in candidate["content"]["parts"]:
-                        if "text" in part:
+            text, reasoning_content = "", ""
+            if "parts" in candidate["content"]:
+                for part in candidate["content"]["parts"]:
+                    if "text" in part:
+                        if "thought" in part and settings.SHOW_THINKING_PROCESS:
+                            reasoning_content += part["text"]
+                        else:
                             text += part["text"]
-                            if "thought" in part and thought is None:
-                                thought = part.get("thought")
-                        elif "inlineData" in part:
-                            text += _extract_image_data(part)
+                        if "thought" in part and thought is None:
+                            thought = part.get("thought")
+                    elif "inlineData" in part:
+                        text += _extract_image_data(part)
 
             text = _add_search_link_text(model, candidate, text)
             tool_calls = _extract_tool_calls(
@@ -220,7 +209,7 @@ def _extract_result(
             )
         else:
             text = "暂无返回"
-    return text, tool_calls, thought
+    return text, reasoning_content, tool_calls, thought
 
 
 def _extract_image_data(part: dict) -> str:
@@ -238,6 +227,7 @@ def _extract_image_data(part: dict) -> str:
             provider=settings.UPLOAD_PROVIDER,
             base_url=settings.CLOUDFLARE_IMGBED_URL,
             auth_code=settings.CLOUDFLARE_IMGBED_AUTH_CODE,
+            upload_folder=settings.CLOUDFLARE_IMGBED_UPLOAD_FOLDER,
         )
     current_date = time.strftime("%Y/%m/%d")
     filename = f"{current_date}/{uuid.uuid4().hex[:8]}.png"
@@ -293,7 +283,7 @@ def _extract_tool_calls(
 def _handle_gemini_stream_response(
     response: Dict[str, Any], model: str, stream: bool
 ) -> Dict[str, Any]:
-    text, tool_calls, thought = _extract_result(
+    text, reasoning_content, tool_calls, thought = _extract_result(
         response, model, stream=stream, gemini_format=True
     )
     if tool_calls:
@@ -310,16 +300,18 @@ def _handle_gemini_stream_response(
 def _handle_gemini_normal_response(
     response: Dict[str, Any], model: str, stream: bool
 ) -> Dict[str, Any]:
-    text, tool_calls, thought = _extract_result(
+    text, reasoning_content, tool_calls, thought = _extract_result(
         response, model, stream=stream, gemini_format=True
     )
+    parts = []
     if tool_calls:
-        content = {"parts": tool_calls, "role": "model"}
+        parts = tool_calls
     else:
-        part = {"text": text}
         if thought is not None:
-            part["thought"] = thought
-        content = {"parts": [part], "role": "model"}
+            parts.append({"text": reasoning_content,"thought": thought})
+        part = {"text": text}
+        parts.append(part)
+    content = {"parts": parts, "role": "model"}
     response["candidates"][0]["content"] = content
     return response
 

app/router/gemini_routes.py

@@ -151,6 +151,35 @@ async def stream_generate_content(
         return StreamingResponse(response_stream, media_type="text/event-stream")
 
 
+@router.post("/models/{model_name}:countTokens")
+@router_v1beta.post("/models/{model_name}:countTokens")
+@RetryHandler(key_arg="api_key")
+async def count_tokens(
+    model_name: str,
+    request: GeminiRequest,
+    _=Depends(security_service.verify_key_or_goog_api_key),
+    api_key: str = Depends(get_next_working_key),
+    key_manager: KeyManager = Depends(get_key_manager),
+    chat_service: GeminiChatService = Depends(get_chat_service)
+):
+    """处理 Gemini token 计数请求。"""
+    operation_name = "gemini_count_tokens"
+    async with handle_route_errors(logger, operation_name, failure_message="Token counting failed"):
+        logger.info(f"Handling Gemini token count request for model: {model_name}")
+        logger.debug(f"Request: \n{request.model_dump_json(indent=2)}")
+        logger.info(f"Using API key: {api_key}")
+
+        if not await model_service.check_model_support(model_name):
+            raise HTTPException(status_code=400, detail=f"Model {model_name} is not supported")
+
+        response = await chat_service.count_tokens(
+            model=model_name,
+            request=request,
+            api_key=api_key
+        )
+        return response
+
+
 @router.post("/reset-all-fail-counts")
 async def reset_all_key_fail_counts(key_type: str = None, key_manager: KeyManager = Depends(get_key_manager)):
     """批量重置Gemini API密钥的失败计数，可选择性地仅重置有效或无效密钥"""

app/service/chat/gemini_chat_service.py

@@ -28,6 +28,33 @@ def _has_image_parts(contents: List[Dict[str, Any]]) -> bool:
     return False
 
 
+def _clean_json_schema_properties(obj: Any) -> Any:
+    """清理JSON Schema中Gemini API不支持的字段"""
+    if not isinstance(obj, dict):
+        return obj
+    
+    # Gemini API不支持的JSON Schema字段
+    unsupported_fields = {
+        "exclusiveMaximum", "exclusiveMinimum", "const", "examples", 
+        "contentEncoding", "contentMediaType", "if", "then", "else",
+        "allOf", "anyOf", "oneOf", "not", "definitions", "$schema",
+        "$id", "$ref", "$comment", "readOnly", "writeOnly"
+    }
+    
+    cleaned = {}
+    for key, value in obj.items():
+        if key in unsupported_fields:
+            continue
+        if isinstance(value, dict):
+            cleaned[key] = _clean_json_schema_properties(value)
+        elif isinstance(value, list):
+            cleaned[key] = [_clean_json_schema_properties(item) for item in value]
+        else:
+            cleaned[key] = value
+    
+    return cleaned
+
+
 def _build_tools(model: str, payload: Dict[str, Any]) -> List[Dict[str, Any]]:
     """构建工具"""
     
@@ -40,7 +67,15 @@ def _build_tools(model: str, payload: Dict[str, Any]) -> List[Dict[str, Any]]:
             for k, v in item.items():
                 if k == "functionDeclarations" and v and isinstance(v, list):
                     functions = record.get("functionDeclarations", [])
-                    functions.extend(v)
+                    # 清理每个函数声明中的不支持字段
+                    cleaned_functions = []
+                    for func in v:
+                        if isinstance(func, dict):
+                            cleaned_func = _clean_json_schema_properties(func)
+                            cleaned_functions.append(cleaned_func)
+                        else:
+                            cleaned_functions.append(func)
+                    functions.extend(cleaned_functions)
                     record["functionDeclarations"] = functions
                 else:
                     record[k] = v
@@ -78,6 +113,26 @@ def _get_safety_settings(model: str) -> List[Dict[str, str]]:
     return settings.SAFETY_SETTINGS
 
 
+def _filter_empty_parts(contents: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
+    """Filters out contents with empty or invalid parts."""
+    if not contents:
+        return []
+
+    filtered_contents = []
+    for content in contents:
+        if not content or "parts" not in content or not isinstance(content.get("parts"), list):
+            continue
+
+        valid_parts = [part for part in content["parts"] if isinstance(part, dict) and part]
+
+        if valid_parts:
+            new_content = content.copy()
+            new_content["parts"] = valid_parts
+            filtered_contents.append(new_content)
+
+    return filtered_contents
+
+
 def _build_payload(model: str, request: GeminiRequest) -> Dict[str, Any]:
     """构建请求payload"""
     request_dict = request.model_dump()
@@ -87,13 +142,17 @@ def _build_payload(model: str, request: GeminiRequest) -> Dict[str, Any]:
             request_dict["generationConfig"].pop("maxOutputTokens")
     
     payload = {
-        "contents": request_dict.get("contents", []),
+        "contents": _filter_empty_parts(request_dict.get("contents", [])),
         "tools": _build_tools(model, request_dict),
         "safetySettings": _get_safety_settings(model),
         "generationConfig": request_dict.get("generationConfig"),
         "systemInstruction": request_dict.get("systemInstruction"),
     }
 
+    # 确保 generationConfig 不为 None
+    if payload["generationConfig"] is None:
+        payload["generationConfig"] = {}
+
     if model.endswith("-image") or model.endswith("-image-generation"):
         payload.pop("systemInstruction")
         payload["generationConfig"]["responseModalities"] = ["Text", "Image"]
@@ -111,7 +170,13 @@ def _build_payload(model: str, request: GeminiRequest) -> Dict[str, Any]:
         if model.endswith("-non-thinking"):
             payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": 0} 
         elif model in settings.THINKING_BUDGET_MAP:
-            payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": settings.THINKING_BUDGET_MAP.get(model,1000)}
+            if settings.SHOW_THINKING_PROCESS:
+                payload["generationConfig"]["thinkingConfig"] = {
+                    "thinkingBudget": settings.THINKING_BUDGET_MAP.get(model,1000),
+                    "includeThoughts": True
+                }
+            else:
+                payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": settings.THINKING_BUDGET_MAP.get(model,1000)}
 
     return payload
 
@@ -195,6 +260,54 @@ class GeminiChatService:
                 request_time=request_datetime
             )
 
+    async def count_tokens(
+        self, model: str, request: GeminiRequest, api_key: str
+    ) -> Dict[str, Any]:
+        """计算token数量"""
+        # countTokens API只需要contents
+        payload = {"contents": _filter_empty_parts(request.model_dump().get("contents", []))}
+        start_time = time.perf_counter()
+        request_datetime = datetime.datetime.now()
+        is_success = False
+        status_code = None
+        response = None
+
+        try:
+            response = await self.api_client.count_tokens(payload, model, api_key)
+            is_success = True
+            status_code = 200
+            return response
+        except Exception as e:
+            is_success = False
+            error_log_msg = str(e)
+            logger.error(f"Count tokens API call failed with error: {error_log_msg}")
+            match = re.search(r"status code (\d+)", error_log_msg)
+            if match:
+                status_code = int(match.group(1))
+            else:
+                status_code = 500
+
+            await add_error_log(
+                gemini_key=api_key,
+                model_name=model,
+                error_type="gemini-count-tokens",
+                error_log=error_log_msg,
+                error_code=status_code,
+                request_msg=payload
+            )
+            raise e
+        finally:
+            end_time = time.perf_counter()
+            latency_ms = int((end_time - start_time) * 1000)
+            await add_request_log(
+                model_name=model,
+                api_key=api_key,
+                is_success=is_success,
+                status_code=status_code,
+                latency_ms=latency_ms,
+                request_time=request_datetime
+            )
+
     async def stream_generate_content(
         self, model: str, request: GeminiRequest, api_key: str
     ) -> AsyncGenerator[str, None]:

app/service/chat/openai_chat_service.py

@@ -26,16 +26,43 @@ from app.service.key.key_manager import KeyManager
 logger = get_openai_logger()
 
 
-def _has_media_parts(contents: List[Dict[str, Any]]) -> bool:
-    """判断消息是否包含图片、音频或视频部分 (inline_data)"""
-    for content in contents:
-        if content and "parts" in content and isinstance(content["parts"], list):
-            for part in content["parts"]:
-                if isinstance(part, dict) and "inline_data" in part:
+def _has_media_parts(messages: List[Dict[str, Any]]) -> bool:
+    """判断消息是否包含多媒体部分"""
+    for message in messages:
+        if "parts" in message:
+            for part in message["parts"]:
+                if "image_url" in part or "inline_data" in part:
                     return True
     return False
 
 
+def _clean_json_schema_properties(obj: Any) -> Any:
+    """清理JSON Schema中Gemini API不支持的字段"""
+    if not isinstance(obj, dict):
+        return obj
+    
+    # Gemini API不支持的JSON Schema字段
+    unsupported_fields = {
+        "exclusiveMaximum", "exclusiveMinimum", "const", "examples", 
+        "contentEncoding", "contentMediaType", "if", "then", "else",
+        "allOf", "anyOf", "oneOf", "not", "definitions", "$schema",
+        "$id", "$ref", "$comment", "readOnly", "writeOnly"
+    }
+    
+    cleaned = {}
+    for key, value in obj.items():
+        if key in unsupported_fields:
+            continue
+        if isinstance(value, dict):
+            cleaned[key] = _clean_json_schema_properties(value)
+        elif isinstance(value, list):
+            cleaned[key] = [_clean_json_schema_properties(item) for item in value]
+        else:
+            cleaned[key] = value
+    
+    return cleaned
+
+
 def _build_tools(
     request: ChatRequest, messages: List[Dict[str, Any]]
 ) -> List[Dict[str, Any]]:
@@ -76,6 +103,8 @@ def _build_tools(
                 ):
                     function.pop("parameters", None)
 
+                # 清理函数中的不支持字段
+                function = _clean_json_schema_properties(function)
                 function_declarations.append(function)
 
         if function_declarations:
@@ -137,9 +166,13 @@ def _build_payload(
     if request.model.endswith("-non-thinking"):
         payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": 0}
     if request.model in settings.THINKING_BUDGET_MAP:
-        payload["generationConfig"]["thinkingConfig"] = {
-            "thinkingBudget": settings.THINKING_BUDGET_MAP.get(request.model, 1000)
-        }
+        if settings.SHOW_THINKING_PROCESS:
+            payload["generationConfig"]["thinkingConfig"] = {
+                "thinkingBudget": settings.THINKING_BUDGET_MAP.get(request.model, 1000),
+                "includeThoughts": True
+            }
+        else:
+            payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": settings.THINKING_BUDGET_MAP.get(request.model, 1000)}
 
     if (
         instruction

app/service/chat/vertex_express_chat_service.py

@@ -28,6 +28,33 @@ def _has_image_parts(contents: List[Dict[str, Any]]) -> bool:
     return False
 
 
+def _clean_json_schema_properties(obj: Any) -> Any:
+    """清理JSON Schema中Gemini API不支持的字段"""
+    if not isinstance(obj, dict):
+        return obj
+    
+    # Gemini API不支持的JSON Schema字段
+    unsupported_fields = {
+        "exclusiveMaximum", "exclusiveMinimum", "const", "examples", 
+        "contentEncoding", "contentMediaType", "if", "then", "else",
+        "allOf", "anyOf", "oneOf", "not", "definitions", "$schema",
+        "$id", "$ref", "$comment", "readOnly", "writeOnly"
+    }
+    
+    cleaned = {}
+    for key, value in obj.items():
+        if key in unsupported_fields:
+            continue
+        if isinstance(value, dict):
+            cleaned[key] = _clean_json_schema_properties(value)
+        elif isinstance(value, list):
+            cleaned[key] = [_clean_json_schema_properties(item) for item in value]
+        else:
+            cleaned[key] = value
+    
+    return cleaned
+
+
 def _build_tools(model: str, payload: Dict[str, Any]) -> List[Dict[str, Any]]:
     """构建工具"""
     
@@ -40,7 +67,15 @@ def _build_tools(model: str, payload: Dict[str, Any]) -> List[Dict[str, Any]]:
             for k, v in item.items():
                 if k == "functionDeclarations" and v and isinstance(v, list):
                     functions = record.get("functionDeclarations", [])
-                    functions.extend(v)
+                    # 清理每个函数声明中的不支持字段
+                    cleaned_functions = []
+                    for func in v:
+                        if isinstance(func, dict):
+                            cleaned_func = _clean_json_schema_properties(func)
+                            cleaned_functions.append(cleaned_func)
+                        else:
+                            cleaned_functions.append(func)
+                    functions.extend(cleaned_functions)
                     record["functionDeclarations"] = functions
                 else:
                     record[k] = v
@@ -98,10 +133,26 @@ def _build_payload(model: str, request: GeminiRequest) -> Dict[str, Any]:
         payload.pop("systemInstruction")
         payload["generationConfig"]["responseModalities"] = ["Text", "Image"]
         
-    if model.endswith("-non-thinking"):
-        payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": 0} 
-    if model in settings.THINKING_BUDGET_MAP:
-        payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": settings.THINKING_BUDGET_MAP.get(model,1000)}
+    # 处理思考配置：优先使用客户端提供的配置，否则使用默认配置
+    client_thinking_config = None
+    if request.generationConfig and request.generationConfig.thinkingConfig:
+        client_thinking_config = request.generationConfig.thinkingConfig
+    
+    if client_thinking_config is not None:
+        # 客户端提供了思考配置，直接使用
+        payload["generationConfig"]["thinkingConfig"] = client_thinking_config
+    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} 
+        elif model in settings.THINKING_BUDGET_MAP:
+            if settings.SHOW_THINKING_PROCESS:
+                payload["generationConfig"]["thinkingConfig"] = {
+                    "thinkingBudget": settings.THINKING_BUDGET_MAP.get(model,1000),
+                    "includeThoughts": True
+                }
+            else:
+                payload["generationConfig"]["thinkingConfig"] = {"thinkingBudget": settings.THINKING_BUDGET_MAP.get(model,1000)}
 
     return payload
 

app/service/client/api_client.py

@@ -40,6 +40,13 @@ class GeminiApiClient(ApiClient):
             model = model[:-20]
         return model
 
+    def _prepare_headers(self) -> Dict[str, str]:
+        headers = {}
+        if settings.CUSTOM_HEADERS:
+            headers.update(settings.CUSTOM_HEADERS)
+            logger.info(f"Using custom headers: {settings.CUSTOM_HEADERS}")
+        return headers
+
     async def get_models(self, api_key: str) -> Optional[Dict[str, Any]]:
         """获取可用的 Gemini 模型列表"""
         timeout = httpx.Timeout(timeout=5)
@@ -52,10 +59,11 @@ class GeminiApiClient(ApiClient):
                 proxy_to_use = random.choice(settings.PROXIES)
             logger.info(f"Using proxy for getting models: {proxy_to_use}")
 
+        headers = self._prepare_headers()
         async with httpx.AsyncClient(timeout=timeout, proxy=proxy_to_use) as client:
             url = f"{self.base_url}/models?key={api_key}&pageSize=1000"
             try:
-                response = await client.get(url)
+                response = await client.get(url, headers=headers)
                 response.raise_for_status()
                 return response.json()
             except httpx.HTTPStatusError as e:
@@ -78,9 +86,10 @@ class GeminiApiClient(ApiClient):
                 proxy_to_use = random.choice(settings.PROXIES)
             logger.info(f"Using proxy for getting models: {proxy_to_use}")
             
+        headers = self._prepare_headers()
         async with httpx.AsyncClient(timeout=timeout, proxy=proxy_to_use) as client:
             url = f"{self.base_url}/models/{model}:generateContent?key={api_key}"
-            response = await client.post(url, json=payload)
+            response = await client.post(url, json=payload, headers=headers)
             if response.status_code != 200:
                 error_content = response.text
                 raise Exception(f"API call failed with status code {response.status_code}, {error_content}")
@@ -98,9 +107,10 @@ class GeminiApiClient(ApiClient):
                 proxy_to_use = random.choice(settings.PROXIES)
             logger.info(f"Using proxy for getting models: {proxy_to_use}")
 
+        headers = self._prepare_headers()
         async with httpx.AsyncClient(timeout=timeout, proxy=proxy_to_use) as client:
             url = f"{self.base_url}/models/{model}:streamGenerateContent?alt=sse&key={api_key}"
-            async with client.stream(method="POST", url=url, json=payload) as response:
+            async with client.stream(method="POST", url=url, json=payload, headers=headers) as response:
                 if response.status_code != 200:
                     error_content = await response.aread()
                     error_msg = error_content.decode("utf-8")
@@ -108,6 +118,27 @@ class GeminiApiClient(ApiClient):
                 async for line in response.aiter_lines():
                     yield line
 
+    async def count_tokens(self, payload: Dict[str, Any], model: str, api_key: str) -> Dict[str, Any]:
+        timeout = httpx.Timeout(self.timeout, read=self.timeout)
+        model = self._get_real_model(model)
+
+        proxy_to_use = None
+        if settings.PROXIES:
+            if settings.PROXIES_USE_CONSISTENCY_HASH_BY_API_KEY:
+                proxy_to_use = settings.PROXIES[hash(api_key) % len(settings.PROXIES)]
+            else:
+                proxy_to_use = random.choice(settings.PROXIES)
+            logger.info(f"Using proxy for counting tokens: {proxy_to_use}")
+
+        headers = self._prepare_headers()
+        async with httpx.AsyncClient(timeout=timeout, proxy=proxy_to_use) as client:
+            url = f"{self.base_url}/models/{model}:countTokens?key={api_key}"
+            response = await client.post(url, json=payload, headers=headers)
+            if response.status_code != 200:
+                error_content = response.text
+                raise Exception(f"API call failed with status code {response.status_code}, {error_content}")
+            return response.json()
+
 
 class OpenaiApiClient(ApiClient):
     """OpenAI API客户端"""
@@ -116,6 +147,13 @@ class OpenaiApiClient(ApiClient):
         self.base_url = base_url
         self.timeout = timeout
         
+    def _prepare_headers(self, api_key: str) -> Dict[str, str]:
+        headers = {"Authorization": f"Bearer {api_key}"}
+        if settings.CUSTOM_HEADERS:
+            headers.update(settings.CUSTOM_HEADERS)
+            logger.info(f"Using custom headers: {settings.CUSTOM_HEADERS}")
+        return headers
+
     async def get_models(self, api_key: str) -> Dict[str, Any]:
         timeout = httpx.Timeout(self.timeout, read=self.timeout)
 
@@ -127,9 +165,9 @@ class OpenaiApiClient(ApiClient):
                 proxy_to_use = random.choice(settings.PROXIES)
             logger.info(f"Using proxy for getting models: {proxy_to_use}")
 
+        headers = self._prepare_headers(api_key)
         async with httpx.AsyncClient(timeout=timeout, proxy=proxy_to_use) as client:
             url = f"{self.base_url}/openai/models"
-            headers = {"Authorization": f"Bearer {api_key}"}
             response = await client.get(url, headers=headers)
             if response.status_code != 200:
                 error_content = response.text
@@ -147,9 +185,9 @@ class OpenaiApiClient(ApiClient):
                 proxy_to_use = random.choice(settings.PROXIES)
             logger.info(f"Using proxy for getting models: {proxy_to_use}")
 
+        headers = self._prepare_headers(api_key)
         async with httpx.AsyncClient(timeout=timeout, proxy=proxy_to_use) as client:
             url = f"{self.base_url}/openai/chat/completions"
-            headers = {"Authorization": f"Bearer {api_key}"}
             response = await client.post(url, json=payload, headers=headers)
             if response.status_code != 200:
                 error_content = response.text
@@ -166,9 +204,9 @@ class OpenaiApiClient(ApiClient):
                 proxy_to_use = random.choice(settings.PROXIES)
             logger.info(f"Using proxy for getting models: {proxy_to_use}")
 
+        headers = self._prepare_headers(api_key)
         async with httpx.AsyncClient(timeout=timeout, proxy=proxy_to_use) as client:
             url = f"{self.base_url}/openai/chat/completions"
-            headers = {"Authorization": f"Bearer {api_key}"}
             async with client.stream(method="POST", url=url, json=payload, headers=headers) as response:
                 if response.status_code != 200:
                     error_content = await response.aread()
@@ -188,9 +226,9 @@ class OpenaiApiClient(ApiClient):
                 proxy_to_use = random.choice(settings.PROXIES)
             logger.info(f"Using proxy for getting models: {proxy_to_use}")
 
+        headers = self._prepare_headers(api_key)
         async with httpx.AsyncClient(timeout=timeout, proxy=proxy_to_use) as client:
             url = f"{self.base_url}/openai/embeddings"
-            headers = {"Authorization": f"Bearer {api_key}"}
             payload = {
                 "input": input,
                 "model": model,
@@ -212,9 +250,9 @@ class OpenaiApiClient(ApiClient):
                 proxy_to_use = random.choice(settings.PROXIES)
             logger.info(f"Using proxy for getting models: {proxy_to_use}")
 
+        headers = self._prepare_headers(api_key)
         async with httpx.AsyncClient(timeout=timeout, proxy=proxy_to_use) as client:
             url = f"{self.base_url}/openai/images/generations"
-            headers = {"Authorization": f"Bearer {api_key}"}
             response = await client.post(url, json=payload, headers=headers)
             if response.status_code != 200:
                 error_content = response.text

app/service/image/image_create_service.py

@@ -121,6 +121,7 @@ class ImageCreateService:
                             provider=settings.UPLOAD_PROVIDER,
                             base_url=settings.CLOUDFLARE_IMGBED_URL,
                             auth_code=settings.CLOUDFLARE_IMGBED_AUTH_CODE,
+                            upload_folder=settings.CLOUDFLARE_IMGBED_UPLOAD_FOLDER,
                         )
                     else:
                         raise ValueError(

app/service/key/key_manager.py

@@ -34,7 +34,7 @@ class KeyManager:
             return next(self.key_cycle)
 
     async def get_next_vertex_key(self) -> str:
-        """获取下一个 Vertex API key"""
+        """获取下一个 Vertex Express API key"""
         async with self.vertex_key_cycle_lock:
             return next(self.vertex_key_cycle)
 
@@ -98,7 +98,7 @@ class KeyManager:
                 return current_key
 
     async def get_next_working_vertex_key(self) -> str:
-        """获取下一可用的 Vertex API key"""
+        """获取下一可用的 Vertex Express API key"""
         initial_key = await self.get_next_vertex_key()
         current_key = initial_key
 
@@ -124,12 +124,12 @@ class KeyManager:
             return ""
 
     async def handle_vertex_api_failure(self, api_key: str, retries: int) -> str:
-        """处理 Vertex API 调用失败"""
+        """处理 Vertex Express API 调用失败"""
         async with self.vertex_failure_count_lock:
             self.vertex_key_failure_counts[api_key] += 1
             if self.vertex_key_failure_counts[api_key] >= self.MAX_FAILURES:
                 logger.warning(
-                    f"Vertex API key {api_key} has failed {self.MAX_FAILURES} times"
+                    f"Vertex Express API key {api_key} has failed {self.MAX_FAILURES} times"
                 )
 
     def get_fail_count(self, key: str) -> int:
@@ -156,7 +156,7 @@ class KeyManager:
         return {"valid_keys": valid_keys, "invalid_keys": invalid_keys}
 
     async def get_vertex_keys_by_status(self) -> dict:
-        """获取分类后的 Vertex API key 列表，包括失败次数"""
+        """获取分类后的 Vertex Express API key 列表，包括失败次数"""
         valid_keys = {}
         invalid_keys = {}
 
@@ -178,8 +178,7 @@ class KeyManager:
         if self.api_keys:
             return self.api_keys[0]
         if not self.api_keys:
-            logger.warning(
-                "API key list is empty, cannot get first valid key.")
+            logger.warning("API key list is empty, cannot get first valid key.")
             return ""
         return self.api_keys[0]
 
@@ -214,7 +213,7 @@ async def get_key_manager_instance(
                 )
             if vertex_api_keys is None:
                 raise ValueError(
-                    "Vertex API keys are required to initialize or re-initialize the KeyManager instance."
+                    "Vertex Express API keys are required to initialize or re-initialize the KeyManager instance."
                 )
 
             if not api_keys:
@@ -223,12 +222,12 @@ async def get_key_manager_instance(
                 )
             if not vertex_api_keys:
                 logger.warning(
-                    "Initializing KeyManager with an empty list of Vertex API keys."
+                    "Initializing KeyManager with an empty list of Vertex Express API keys."
                 )
 
             _singleton_instance = KeyManager(api_keys, vertex_api_keys)
             logger.info(
-                f"KeyManager instance created/re-created with {len(api_keys)} API keys and {len(vertex_api_keys)} Vertex API keys."
+                f"KeyManager instance created/re-created with {len(api_keys)} API keys and {len(vertex_api_keys)} Vertex Express API keys."
             )
 
             # 1. 恢复失败计数
@@ -253,8 +252,7 @@ async def get_key_manager_instance(
                 _singleton_instance.vertex_key_failure_counts = (
                     current_vertex_failure_counts
                 )
-                logger.info(
-                    "Inherited failure counts for applicable Vertex keys.")
+                logger.info("Inherited failure counts for applicable Vertex keys.")
             _preserved_vertex_failure_counts = None
 
             # 2. 调整 key_cycle 的起始点
@@ -351,7 +349,7 @@ async def get_key_manager_instance(
                             break
                 except ValueError:
                     logger.warning(
-                        f"Preserved next key '{_preserved_vertex_next_key_in_cycle}' not found in preserved old Vertex API keys. "
+                        f"Preserved next key '{_preserved_vertex_next_key_in_cycle}' not found in preserved old Vertex Express API keys. "
                         "New cycle will start from the beginning of the new list."
                     )
                 except Exception as e:
@@ -372,12 +370,12 @@ async def get_key_manager_instance(
                     )
                 except ValueError:
                     logger.warning(
-                        f"Determined start key '{start_key_for_new_vertex_cycle}' not found in new Vertex API keys during cycle advancement. "
+                        f"Determined start key '{start_key_for_new_vertex_cycle}' not found in new Vertex Express API keys during cycle advancement. "
                         "New cycle will start from the beginning."
                     )
                 except StopIteration:
                     logger.error(
-                        "StopIteration while advancing Vertex key cycle, implies empty new Vertex API key list previously missed."
+                        "StopIteration while advancing Vertex key cycle, implies empty new Vertex Express API key list previously missed."
                     )
                 except Exception as e:
                     logger.error(
@@ -386,11 +384,11 @@ async def get_key_manager_instance(
             else:
                 if _singleton_instance.vertex_api_keys:
                     logger.info(
-                        "New Vertex key cycle will start from the beginning of the new Vertex API key list (no specific start key determined or needed)."
+                        "New Vertex key cycle will start from the beginning of the new Vertex Express API key list (no specific start key determined or needed)."
                     )
                 else:
                     logger.info(
-                        "New Vertex key cycle not applicable as the new Vertex API key list is empty."
+                        "New Vertex key cycle not applicable as the new Vertex Express API key list is empty."
                     )
 
             # 清理所有保存的状态
@@ -411,11 +409,15 @@ async def reset_key_manager_instance():
         if _singleton_instance:
             # 1. 保存失败计数
             _preserved_failure_counts = _singleton_instance.key_failure_counts.copy()
-            _preserved_vertex_failure_counts = _singleton_instance.vertex_key_failure_counts.copy()
+            _preserved_vertex_failure_counts = (
+                _singleton_instance.vertex_key_failure_counts.copy()
+            )
 
             # 2. 保存旧的 API keys 列表
             _preserved_old_api_keys_for_reset = _singleton_instance.api_keys.copy()
-            _preserved_vertex_old_api_keys_for_reset = _singleton_instance.vertex_api_keys.copy()
+            _preserved_vertex_old_api_keys_for_reset = (
+                _singleton_instance.vertex_api_keys.copy()
+            )
 
             # 3. 保存 key_cycle 的下一个 key 提示
             try:
@@ -431,8 +433,7 @@ async def reset_key_manager_instance():
                 )
                 _preserved_next_key_in_cycle = None
             except Exception as e:
-                logger.error(
-                    f"Error preserving next key hint during reset: {e}")
+                logger.error(f"Error preserving next key hint during reset: {e}")
                 _preserved_next_key_in_cycle = None
 
             # 4. 保存 vertex_key_cycle 的下一个 key 提示
@@ -449,8 +450,7 @@ async def reset_key_manager_instance():
                 )
                 _preserved_vertex_next_key_in_cycle = None
             except Exception as e:
-                logger.error(
-                    f"Error preserving next key hint during reset: {e}")
+                logger.error(f"Error preserving next key hint during reset: {e}")
                 _preserved_vertex_next_key_in_cycle = None
 
             _singleton_instance = None

app/service/tts/tts_service.py

@@ -8,6 +8,7 @@ from typing import Optional
 from google import genai
 
 from app.config.config import settings
+from app.core.constants import TTS_VOICE_NAMES
 from app.database.services import add_error_log, add_request_log
 from app.domain.openai_models import TTSRequest
 from app.log.logger import get_openai_logger
@@ -47,7 +48,7 @@ class TTSService:
                     "speech_config": {
                         "voice_config": {
                             "prebuilt_voice_config": {
-                                "voice_name": settings.TTS_VOICE_NAME
+                                "voice_name": request.voice if request.voice in TTS_VOICE_NAMES else settings.TTS_VOICE_NAME
                             }
                         }
                     },

app/static/js/config_editor.js

@@ -5,12 +5,15 @@ const ARRAY_INPUT_CLASS = "array-input";
 const MAP_ITEM_CLASS = "map-item";
 const MAP_KEY_INPUT_CLASS = "map-key-input";
 const MAP_VALUE_INPUT_CLASS = "map-value-input";
+const CUSTOM_HEADER_ITEM_CLASS = "custom-header-item";
+const CUSTOM_HEADER_KEY_INPUT_CLASS = "custom-header-key-input";
+const CUSTOM_HEADER_VALUE_INPUT_CLASS = "custom-header-value-input";
 const SAFETY_SETTING_ITEM_CLASS = "safety-setting-item";
 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 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
@@ -32,7 +35,7 @@ const bulkDeleteProxyInput = document.getElementById("bulkDeleteProxyInput");
 const resetConfirmModal = document.getElementById("resetConfirmModal");
 const configForm = document.getElementById("configForm"); // Added for frequent use
 
-// Vertex API Key Modal Elements
+// Vertex Express API Key Modal Elements
 const vertexApiKeyModal = document.getElementById("vertexApiKeyModal");
 const vertexApiKeyBulkInput = document.getElementById("vertexApiKeyBulkInput");
 const bulkDeleteVertexApiKeyModal = document.getElementById(
@@ -383,9 +386,15 @@ document.addEventListener("DOMContentLoaded", function () {
     addSafetySettingBtn.addEventListener("click", () => addSafetySettingItem());
   }
 
+  // Add Custom Header button
+  const addCustomHeaderBtn = document.getElementById("addCustomHeaderBtn");
+  if (addCustomHeaderBtn) {
+    addCustomHeaderBtn.addEventListener("click", () => addCustomHeaderItem());
+  }
+
   initializeSensitiveFields(); // Initialize sensitive field handling
 
-  // Vertex API Key Modal Elements and Events
+  // Vertex Express API Key Modal Elements and Events
   const addVertexApiKeyBtn = document.getElementById("addVertexApiKeyBtn");
   const closeVertexApiKeyModalBtn = document.getElementById(
     "closeVertexApiKeyModalBtn"
@@ -691,6 +700,14 @@ async function initConfig() {
     ) {
       config.THINKING_BUDGET_MAP = {}; // 默认为空对象
     }
+    // --- 新增：处理 CUSTOM_HEADERS 默认值 ---
+    if (
+      !config.CUSTOM_HEADERS ||
+      typeof config.CUSTOM_HEADERS !== "object" ||
+      config.CUSTOM_HEADERS === null
+    ) {
+      config.CUSTOM_HEADERS = {}; // 默认为空对象
+    }
     // --- 新增：处理 SAFETY_SETTINGS 默认值 ---
     if (!config.SAFETY_SETTINGS || !Array.isArray(config.SAFETY_SETTINGS)) {
       config.SAFETY_SETTINGS = []; // 默认为空数组
@@ -756,6 +773,7 @@ async function initConfig() {
       VERTEX_EXPRESS_BASE_URL: "", // 确保默认值存在
       THINKING_MODELS: [],
       THINKING_BUDGET_MAP: {},
+      CUSTOM_HEADERS: {},
       AUTO_DELETE_ERROR_LOGS_ENABLED: false,
       AUTO_DELETE_ERROR_LOGS_DAYS: 7, // 新增默认值
       AUTO_DELETE_REQUEST_LOGS_ENABLED: false, // 新增默认值
@@ -854,6 +872,26 @@ function populateForm(config) {
       '<div class="text-gray-500 text-sm italic">请在上方添加思考模型，预算将自动关联。</div>';
   }
 
+  // Populate CUSTOM_HEADERS
+  const customHeadersContainer = document.getElementById(
+    "CUSTOM_HEADERS_container"
+  );
+  let customHeadersAdded = false;
+  if (
+    customHeadersContainer &&
+    config.CUSTOM_HEADERS &&
+    typeof config.CUSTOM_HEADERS === "object"
+  ) {
+    for (const [key, value] of Object.entries(config.CUSTOM_HEADERS)) {
+      createAndAppendCustomHeaderItem(key, value);
+      customHeadersAdded = true;
+    }
+  }
+  if (!customHeadersAdded && customHeadersContainer) {
+    customHeadersContainer.innerHTML =
+      '<div class="text-gray-500 text-sm italic">添加自定义请求头，例如 X-Api-Key: your-key</div>';
+  }
+
   // 4. Populate other array fields (excluding THINKING_MODELS)
   for (const [key, value] of Object.entries(config)) {
     if (Array.isArray(value) && key !== "THINKING_MODELS") {
@@ -1179,17 +1217,13 @@ function handleBulkDeleteProxies() {
 }
 
 /**
- * Handles the bulk addition of Vertex API keys from the modal input.
+ * Handles the bulk addition of Vertex Express API keys from the modal input.
  */
 function handleBulkAddVertexApiKeys() {
   const vertexApiKeyContainer = document.getElementById(
     "VERTEX_API_KEYS_container"
   );
-  if (
-    !vertexApiKeyBulkInput ||
-    !vertexApiKeyContainer ||
-    !vertexApiKeyModal
-  ) {
+  if (!vertexApiKeyBulkInput || !vertexApiKeyContainer || !vertexApiKeyModal) {
     return;
   }
 
@@ -1239,7 +1273,7 @@ function handleBulkAddVertexApiKeys() {
 }
 
 /**
- * Handles the bulk deletion of Vertex API keys based on input from the modal.
+ * Handles the bulk deletion of Vertex Express API keys based on input from the modal.
  */
 function handleBulkDeleteVertexApiKeys() {
   const vertexApiKeyContainer = document.getElementById(
@@ -1255,7 +1289,7 @@ function handleBulkDeleteVertexApiKeys() {
 
   const bulkText = bulkDeleteVertexApiKeyInput.value;
   if (!bulkText.trim()) {
-    showNotification("请粘贴需要删除的 Vertex API 密钥", "warning");
+    showNotification("请粘贴需要删除的 Vertex Express API 密钥", "warning");
     return;
   }
 
@@ -1263,13 +1297,15 @@ function handleBulkDeleteVertexApiKeys() {
 
   if (keysToDelete.size === 0) {
     showNotification(
-      "未在输入内容中提取到有效的 Vertex API 密钥格式",
+      "未在输入内容中提取到有效的 Vertex Express API 密钥格式",
       "warning"
     );
     return;
   }
 
-  const keyItems = vertexApiKeyContainer.querySelectorAll(`.${ARRAY_ITEM_CLASS}`);
+  const keyItems = vertexApiKeyContainer.querySelectorAll(
+    `.${ARRAY_ITEM_CLASS}`
+  );
   let deleteCount = 0;
 
   keyItems.forEach((item) => {
@@ -1290,7 +1326,10 @@ function handleBulkDeleteVertexApiKeys() {
   closeModal(bulkDeleteVertexApiKeyModal);
 
   if (deleteCount > 0) {
-    showNotification(`成功删除了 ${deleteCount} 个匹配的 Vertex 密钥`, "success");
+    showNotification(
+      `成功删除了 ${deleteCount} 个匹配的 Vertex 密钥`,
+      "success"
+    );
   } else {
     showNotification("列表中未找到您输入的任何 Vertex 密钥进行删除", "info");
   }
@@ -1305,8 +1344,10 @@ function switchTab(tabId) {
   console.log(`Switching to tab: ${tabId}`);
 
   // 定义选中态和未选中态的样式
-  const activeStyle = "background-color: #3b82f6 !important; color: #ffffff !important; border: 2px solid #2563eb !important; box-shadow: 0 4px 12px -2px rgba(59, 130, 246, 0.4), 0 2px 6px -1px rgba(59, 130, 246, 0.2) !important; transform: translateY(-2px) !important; font-weight: 600 !important;";
-  const inactiveStyle = "background-color: #f8fafc !important; color: #64748b !important; border: 2px solid #e2e8f0 !important; box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important; font-weight: 500 !important; transform: none !important;";
+  const activeStyle =
+    "background-color: #3b82f6 !important; color: #ffffff !important; border: 2px solid #2563eb !important; box-shadow: 0 4px 12px -2px rgba(59, 130, 246, 0.4), 0 2px 6px -1px rgba(59, 130, 246, 0.2) !important; transform: translateY(-2px) !important; font-weight: 600 !important;";
+  const inactiveStyle =
+    "background-color: #f8fafc !important; color: #64748b !important; border: 2px solid #e2e8f0 !important; box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important; font-weight: 500 !important; transform: none !important;";
 
   // 更新标签按钮状态
   const tabButtons = document.querySelectorAll(".tab-btn");
@@ -1439,8 +1480,7 @@ function addArrayItemWithValue(key, value) {
   const isThinkingModel = key === "THINKING_MODELS";
   const isAllowedToken = key === "ALLOWED_TOKENS";
   const isVertexApiKey = key === "VERTEX_API_KEYS"; // 新增判断
-  const isSensitive =
-    key === "API_KEYS" || isAllowedToken || isVertexApiKey; // 更新敏感判断
+  const isSensitive = key === "API_KEYS" || isAllowedToken || isVertexApiKey; // 更新敏感判断
   const modelId = isThinkingModel ? generateUUID() : null;
 
   const arrayItem = document.createElement("div");
@@ -1562,6 +1602,67 @@ function createAndAppendBudgetMapItem(mapKey, mapValue, modelId) {
   container.appendChild(mapItem);
 }
 
+/**
+ * Adds a new custom header item to the DOM.
+ */
+function addCustomHeaderItem() {
+  createAndAppendCustomHeaderItem("", "");
+}
+
+/**
+ * Creates and appends a DOM element for a custom header.
+ * @param {string} key - The header key.
+ * @param {string} value - The header value.
+ */
+function createAndAppendCustomHeaderItem(key, value) {
+  const container = document.getElementById("CUSTOM_HEADERS_container");
+  if (!container) {
+    console.error(
+      "Cannot add custom header: CUSTOM_HEADERS_container not found!"
+    );
+    return;
+  }
+
+  const placeholder = container.querySelector(".text-gray-500.italic");
+  if (
+    placeholder &&
+    container.children.length === 1 &&
+    container.firstChild === placeholder
+  ) {
+    container.innerHTML = "";
+  }
+
+  const headerItem = document.createElement("div");
+  headerItem.className = `${CUSTOM_HEADER_ITEM_CLASS} flex items-center mb-2 gap-2`;
+
+  const keyInput = document.createElement("input");
+  keyInput.type = "text";
+  keyInput.value = key;
+  keyInput.placeholder = "Header Name";
+  keyInput.className = `${CUSTOM_HEADER_KEY_INPUT_CLASS} flex-grow px-3 py-2 border border-gray-300 rounded-md focus:outline-none bg-gray-100 text-gray-500`;
+
+  const valueInput = document.createElement("input");
+  valueInput.type = "text";
+  valueInput.value = value;
+  valueInput.placeholder = "Header Value";
+  valueInput.className = `${CUSTOM_HEADER_VALUE_INPUT_CLASS} flex-grow 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`;
+
+  const removeBtn = createRemoveButton();
+  removeBtn.addEventListener("click", () => {
+    headerItem.remove();
+    if (container.children.length === 0) {
+      container.innerHTML =
+        '<div class="text-gray-500 text-sm italic">添加自定义请求头，例如 X-Api-Key: your-key</div>';
+    }
+  });
+
+  headerItem.appendChild(keyInput);
+  headerItem.appendChild(valueInput);
+  headerItem.appendChild(removeBtn);
+
+  container.appendChild(headerItem);
+}
+
 /**
  * Collects all data from the configuration form.
  * @returns {object} An object containing all configuration data.
@@ -1638,6 +1739,26 @@ function collectFormData() {
     });
   }
 
+  const customHeadersContainer = document.getElementById(
+    "CUSTOM_HEADERS_container"
+  );
+  if (customHeadersContainer) {
+    formData["CUSTOM_HEADERS"] = {};
+    const customHeaderItems = customHeadersContainer.querySelectorAll(
+      `.${CUSTOM_HEADER_ITEM_CLASS}`
+    );
+    customHeaderItems.forEach((item) => {
+      const keyInput = item.querySelector(`.${CUSTOM_HEADER_KEY_INPUT_CLASS}`);
+      const valueInput = item.querySelector(
+        `.${CUSTOM_HEADER_VALUE_INPUT_CLASS}`
+      );
+      if (keyInput && valueInput && keyInput.value.trim() !== "") {
+        formData["CUSTOM_HEADERS"][keyInput.value.trim()] =
+          valueInput.value.trim();
+      }
+    });
+  }
+
   if (safetySettingsContainer) {
     formData["SAFETY_SETTINGS"] = [];
     const settingItems = safetySettingsContainer.querySelectorAll(

app/static/js/keys_status.js

@@ -817,58 +817,11 @@ function toggleSection(header, sectionId) {
   }
 }
 
-// 筛选有效密钥（根据失败次数阈值）并更新批量操作状态
+// filterValidKeys 函数已被 filterAndSearchValidKeys 替代，此函数保留为空或可移除
 function filterValidKeys() {
-  const thresholdInput = document.getElementById("failCountThreshold");
-  const validKeysList = document.getElementById("validKeys"); // Get the UL element
-  if (!validKeysList) return; // Exit if the list doesn't exist
-
-  const validKeyItems = validKeysList.querySelectorAll("li[data-key]"); // Select li elements within the list
-  // 读取阈值，如果输入无效或为空，则默认为0（不过滤）
-  const threshold = parseInt(thresholdInput.value, 10);
-  const filterThreshold = isNaN(threshold) || threshold < 0 ? 0 : threshold;
-  let hasVisibleItems = false;
-
-  validKeyItems.forEach((item) => {
-    // 确保只处理包含 data-fail-count 的 li 元素
-    if (item.dataset.failCount !== undefined) {
-      const failCount = parseInt(item.dataset.failCount, 10);
-      // 如果失败次数大于等于阈值，则显示，否则隐藏
-      if (failCount >= filterThreshold) {
-        item.style.display = "flex"; // 使用 flex 因为 li 现在是 flex 容器
-        hasVisibleItems = true;
-      } else {
-        item.style.display = "none"; // 隐藏
-        // 如果隐藏了一个项，取消其选中状态
-        const checkbox = item.querySelector(".key-checkbox");
-        if (checkbox && checkbox.checked) {
-          checkbox.checked = false;
-        }
-      }
-    }
-  });
-
-  // 更新有效密钥的批量操作状态和全选复选框
-  updateBatchActions("valid");
-
-  // 处理"暂无有效密钥"消息
-  const noMatchMsgId = "no-valid-keys-msg";
-  let noMatchMsg = validKeysList.querySelector(`#${noMatchMsgId}`);
-  const initialKeyCount = validKeysList.querySelectorAll("li[data-key]").length; // 获取初始密钥数量
-
-  if (!hasVisibleItems && initialKeyCount > 0) {
-    // 仅当初始有密钥但现在都不可见时显示
-    if (!noMatchMsg) {
-      noMatchMsg = document.createElement("li");
-      noMatchMsg.id = noMatchMsgId;
-      noMatchMsg.className = "text-center text-gray-500 py-4 col-span-full";
-      noMatchMsg.textContent = "没有符合条件的有效密钥";
-      validKeysList.appendChild(noMatchMsg);
-    }
-    noMatchMsg.style.display = "";
-  } else if (noMatchMsg) {
-    noMatchMsg.style.display = "none";
-  }
+  // This function is now handled by filterAndSearchValidKeys
+  // Kept for now to avoid breaking any potential legacy calls, but should be removed later.
+  filterAndSearchValidKeys();
 }
 
 // --- Initialization Helper Functions ---

app/templates/config_editor.html

@@ -96,12 +96,7 @@ endblock %} {% block head_extra_styles %}
 
   /* Theming for select fields - 改进下拉框样式 */
   .form-select-themed {
-    background-color: rgba(
-      255,
-      255,
-      255,
-      0.95
-    ) !important; /* 白色背景 */
+    background-color: rgba(255, 255, 255, 0.95) !important; /* 白色背景 */
     border: 1px solid rgba(0, 0, 0, 0.12) !important; /* 灰色边框 */
     color: #374151 !important; /* gray-700 文字颜色 */
     background-image: url("data:image/svg+xml,%3csvg xmlns='http://www.w3.org/2000/svg' fill='none' viewBox='0 0 20 20'%3e%3cpath stroke='%236b7280' stroke-linecap='round' stroke-linejoin='round' stroke-width='2' d='M6 8l4 4 4-4'/%3e%3c/svg%3e") !important; /* 灰色箭头 */
@@ -124,7 +119,12 @@ endblock %} {% block head_extra_styles %}
   }
 
   .form-select-themed option {
-    background-color: rgba(255, 255, 255, 0.98) !important; /* white background */
+    background-color: rgba(
+      255,
+      255,
+      255,
+      0.98
+    ) !important; /* white background */
     color: #374151 !important; /* gray-700 */
     padding: 8px !important;
   }
@@ -141,22 +141,42 @@ endblock %} {% block head_extra_styles %}
   }
 
   #LOG_LEVEL option[value="INFO"] {
-    background-color: rgba(249, 250, 251, 0.98) !important; /* gray-50 浅灰背景 */
+    background-color: rgba(
+      249,
+      250,
+      251,
+      0.98
+    ) !important; /* gray-50 浅灰背景 */
     color: #374151 !important; /* gray-700 深灰色文字 */
   }
 
   #LOG_LEVEL option[value="WARNING"] {
-    background-color: rgba(243, 244, 246, 0.98) !important; /* gray-100 稍深灰背景 */
+    background-color: rgba(
+      243,
+      244,
+      246,
+      0.98
+    ) !important; /* gray-100 稍深灰背景 */
     color: #374151 !important; /* gray-700 深灰色文字 */
   }
 
   #LOG_LEVEL option[value="ERROR"] {
-    background-color: rgba(229, 231, 235, 0.98) !important; /* gray-200 中灰背景 */
+    background-color: rgba(
+      229,
+      231,
+      235,
+      0.98
+    ) !important; /* gray-200 中灰背景 */
     color: #374151 !important; /* gray-700 深灰色文字 */
   }
 
   #LOG_LEVEL option[value="CRITICAL"] {
-    background-color: rgba(209, 213, 219, 0.98) !important; /* gray-300 深灰背景 */
+    background-color: rgba(
+      209,
+      213,
+      219,
+      0.98
+    ) !important; /* gray-300 深灰背景 */
     color: #374151 !important; /* gray-700 深灰色文字 */
   }
 
@@ -228,7 +248,12 @@ endblock %} {% block head_extra_styles %}
   }
 
   .generate-btn:hover {
-    background-color: rgba(59, 130, 246, 0.2) !important; /* blue-500 more opaque */
+    background-color: rgba(
+      59,
+      130,
+      246,
+      0.2
+    ) !important; /* blue-500 more opaque */
     color: #1d4ed8 !important; /* blue-700 */
     box-shadow: 0 0 8px rgba(59, 130, 246, 0.3) !important;
   }
@@ -302,36 +327,47 @@ endblock %} {% block head_extra_styles %}
   }
 
   /* Override all violet/purple buttons to light blue */
-  .bg-violet-600, button.bg-violet-600 {
+  .bg-violet-600,
+  button.bg-violet-600 {
     background-color: #3b82f6 !important; /* blue-500 - light blue */
   }
 
-  .bg-violet-600:hover, button.bg-violet-600:hover,
+  .bg-violet-600:hover,
+  button.bg-violet-600:hover,
   .hover\\:bg-violet-700:hover {
     background-color: #2563eb !important; /* blue-600 - darker light blue */
   }
 
   /* Override blue buttons to light blue */
-  .bg-blue-600, button.bg-blue-600 {
+  .bg-blue-600,
+  button.bg-blue-600 {
     background-color: #3b82f6 !important; /* blue-500 - light blue */
   }
 
-  .bg-blue-600:hover, button.bg-blue-600:hover,
+  .bg-blue-600:hover,
+  button.bg-blue-600:hover,
   .hover\\:bg-blue-700:hover {
     background-color: #2563eb !important; /* blue-600 - darker light blue */
   }
 
   /* Override red buttons to bright light red */
-  .bg-red-600, button.bg-red-600,
-  .bg-red-700, button.bg-red-700,
-  .bg-red-800, button.bg-red-800 {
+  .bg-red-600,
+  button.bg-red-600,
+  .bg-red-700,
+  button.bg-red-700,
+  .bg-red-800,
+  button.bg-red-800 {
     background-color: #f87171 !important; /* red-400 - bright light red */
   }
 
-  .bg-red-600:hover, button.bg-red-600:hover,
-  .bg-red-700:hover, button.bg-red-700:hover,
-  .bg-red-800:hover, button.bg-red-800:hover,
-  .hover\\:bg-red-700:hover, .hover\\:bg-red-800:hover {
+  .bg-red-600:hover,
+  button.bg-red-600:hover,
+  .bg-red-700:hover,
+  button.bg-red-700:hover,
+  .bg-red-800:hover,
+  button.bg-red-800:hover,
+  .hover\\:bg-red-700:hover,
+  .hover\\:bg-red-800:hover {
     background-color: #ef4444 !important; /* red-500 - darker bright light red */
   }
 
@@ -349,7 +385,8 @@ endblock %} {% block head_extra_styles %}
   button.tab-btn.active {
     background-color: #3b82f6 !important; /* blue-500 */
     color: #ffffff !important; /* 确保白色文字 */
-    box-shadow: 0 4px 12px -2px rgba(59, 130, 246, 0.4), 0 2px 6px -1px rgba(59, 130, 246, 0.2) !important; /* 蓝色阴影 */
+    box-shadow: 0 4px 12px -2px rgba(59, 130, 246, 0.4),
+      0 2px 6px -1px rgba(59, 130, 246, 0.2) !important; /* 蓝色阴影 */
     transform: translateY(-2px) !important; /* 更明显的上移效果 */
     border: 2px solid #2563eb !important; /* blue-600 边框 */
     font-weight: 600 !important; /* 加粗字体 */
@@ -388,7 +425,8 @@ endblock %} {% block head_extra_styles %}
     background-color: #3b82f6 !important;
     color: #ffffff !important;
     border: 2px solid #2563eb !important;
-    box-shadow: 0 4px 12px -2px rgba(59, 130, 246, 0.4), 0 2px 6px -1px rgba(59, 130, 246, 0.2) !important;
+    box-shadow: 0 4px 12px -2px rgba(59, 130, 246, 0.4),
+      0 2px 6px -1px rgba(59, 130, 246, 0.2) !important;
     transform: translateY(-2px) !important;
     font-weight: 600 !important;
   }
@@ -409,7 +447,8 @@ endblock %} {% block head_extra_styles %}
     background-color: rgba(255, 255, 255, 0.98) !important;
     color: #374151 !important; /* gray-700 */
     border-color: rgba(0, 0, 0, 0.08) !important;
-    box-shadow: 0 20px 25px -5px rgba(0, 0, 0, 0.1), 0 10px 10px -5px rgba(0, 0, 0, 0.04) !important;
+    box-shadow: 0 20px 25px -5px rgba(0, 0, 0, 0.1),
+      0 10px 10px -5px rgba(0, 0, 0, 0.04) !important;
   }
 
   /* Fix modal titles */
@@ -428,62 +467,78 @@ endblock %} {% block head_extra_styles %}
   }
 
   /* Fix modal body text */
-  .modal p, .modal label, .modal span {
+  .modal p,
+  .modal label,
+  .modal span {
     color: #374151 !important; /* gray-700 */
   }
 
   /* Fix modal textarea and input styling */
-  .modal textarea, .modal input {
+  .modal textarea,
+  .modal input {
     background-color: rgba(255, 255, 255, 0.95) !important;
     color: #374151 !important; /* gray-700 */
     border: 1px solid rgba(0, 0, 0, 0.12) !important;
   }
 
-  .modal textarea:focus, .modal input:focus {
+  .modal textarea:focus,
+  .modal input:focus {
     border-color: #3b82f6 !important; /* blue-500 */
     box-shadow: 0 0 0 3px rgba(59, 130, 246, 0.1) !important;
   }
 
   /* Fix modal button styling */
-  .modal .bg-violet-600, .modal button.bg-violet-600 {
+  .modal .bg-violet-600,
+  .modal button.bg-violet-600 {
     background-color: #3b82f6 !important; /* blue-500 - light blue */
     color: #ffffff !important;
     border: 1px solid #2563eb !important; /* blue-600 */
   }
 
-  .modal .bg-violet-600:hover, .modal button.bg-violet-600:hover {
+  .modal .bg-violet-600:hover,
+  .modal button.bg-violet-600:hover {
     background-color: #2563eb !important; /* blue-600 - darker light blue */
     transform: translateY(-1px) !important;
     box-shadow: 0 4px 6px rgba(0, 0, 0, 0.1) !important;
   }
 
   /* Fix modal blue button styling */
-  .modal .bg-blue-500, .modal button.bg-blue-500,
-  .modal .bg-blue-600, .modal button.bg-blue-600,
-  .modal .bg-blue-700, .modal button.bg-blue-700 {
+  .modal .bg-blue-500,
+  .modal button.bg-blue-500,
+  .modal .bg-blue-600,
+  .modal button.bg-blue-600,
+  .modal .bg-blue-700,
+  .modal button.bg-blue-700 {
     background-color: #3b82f6 !important; /* blue-500 - light blue */
     color: #ffffff !important;
     border: 1px solid #2563eb !important; /* blue-600 */
   }
 
-  .modal .bg-blue-500:hover, .modal button.bg-blue-500:hover,
-  .modal .bg-blue-600:hover, .modal button.bg-blue-600:hover,
-  .modal .bg-blue-700:hover, .modal button.bg-blue-700:hover {
+  .modal .bg-blue-500:hover,
+  .modal button.bg-blue-500:hover,
+  .modal .bg-blue-600:hover,
+  .modal button.bg-blue-600:hover,
+  .modal .bg-blue-700:hover,
+  .modal button.bg-blue-700:hover {
     background-color: #2563eb !important; /* blue-600 - darker light blue */
     transform: translateY(-1px) !important;
     box-shadow: 0 4px 6px rgba(0, 0, 0, 0.1) !important;
   }
 
   /* Fix modal cancel/secondary buttons */
-  .modal .bg-gray-500, .modal button.bg-gray-500,
-  .modal .bg-gray-600, .modal button.bg-gray-600 {
+  .modal .bg-gray-500,
+  .modal button.bg-gray-500,
+  .modal .bg-gray-600,
+  .modal button.bg-gray-600 {
     background-color: #e5e7eb !important; /* gray-200 - light gray */
     color: #374151 !important; /* gray-700 - dark text for contrast */
     border: 1px solid #d1d5db !important; /* gray-300 */
   }
 
-  .modal .bg-gray-500:hover, .modal button.bg-gray-500:hover,
-  .modal .bg-gray-600:hover, .modal button.bg-gray-600:hover {
+  .modal .bg-gray-500:hover,
+  .modal button.bg-gray-500:hover,
+  .modal .bg-gray-600:hover,
+  .modal button.bg-gray-600:hover {
     background-color: #d1d5db !important; /* gray-300 - darker light gray */
     color: #374151 !important; /* gray-700 - dark text for contrast */
     transform: translateY(-1px) !important;
@@ -491,17 +546,23 @@ endblock %} {% block head_extra_styles %}
   }
 
   /* Fix modal red/danger buttons */
-  .modal .bg-red-500, .modal button.bg-red-500,
-  .modal .bg-red-600, .modal button.bg-red-600,
-  .modal .bg-red-700, .modal button.bg-red-700 {
+  .modal .bg-red-500,
+  .modal button.bg-red-500,
+  .modal .bg-red-600,
+  .modal button.bg-red-600,
+  .modal .bg-red-700,
+  .modal button.bg-red-700 {
     background-color: #f87171 !important; /* red-400 - bright light red */
     color: #ffffff !important;
     border: 1px solid #ef4444 !important; /* red-500 */
   }
 
-  .modal .bg-red-500:hover, .modal button.bg-red-500:hover,
-  .modal .bg-red-600:hover, .modal button.bg-red-600:hover,
-  .modal .bg-red-700:hover, .modal button.bg-red-700:hover {
+  .modal .bg-red-500:hover,
+  .modal button.bg-red-500:hover,
+  .modal .bg-red-600:hover,
+  .modal button.bg-red-600:hover,
+  .modal .bg-red-700:hover,
+  .modal button.bg-red-700:hover {
     background-color: #ef4444 !important; /* red-500 - darker bright light red */
     transform: translateY(-1px) !important;
     box-shadow: 0 4px 6px rgba(0, 0, 0, 0.1) !important;
@@ -542,27 +603,54 @@ endblock %} {% block head_extra_styles %}
   }
 
   /* Comprehensive button text color fixes */
-  .bg-blue-500, .bg-blue-600, .bg-blue-700,
-  .bg-red-500, .bg-red-600, .bg-red-700, .bg-red-800,
-  .bg-green-500, .bg-green-600, .bg-green-700,
-  .bg-sky-500, .bg-sky-600, .bg-sky-700,
-  .bg-purple-500, .bg-purple-600, .bg-purple-700,
-  .bg-violet-500, .bg-violet-600, .bg-violet-700 {
+  .bg-blue-500,
+  .bg-blue-600,
+  .bg-blue-700,
+  .bg-red-500,
+  .bg-red-600,
+  .bg-red-700,
+  .bg-red-800,
+  .bg-green-500,
+  .bg-green-600,
+  .bg-green-700,
+  .bg-sky-500,
+  .bg-sky-600,
+  .bg-sky-700,
+  .bg-purple-500,
+  .bg-purple-600,
+  .bg-purple-700,
+  .bg-violet-500,
+  .bg-violet-600,
+  .bg-violet-700 {
     color: #ffffff !important;
   }
 
   /* Ensure button children inherit white text */
-  .bg-blue-500 *, .bg-blue-600 *, .bg-blue-700 *,
-  .bg-red-500 *, .bg-red-600 *, .bg-red-700 *, .bg-red-800 *,
-  .bg-green-500 *, .bg-green-600 *, .bg-green-700 *,
-  .bg-sky-500 *, .bg-sky-600 *, .bg-sky-700 *,
-  .bg-purple-500 *, .bg-purple-600 *, .bg-purple-700 *,
-  .bg-violet-500 *, .bg-violet-600 *, .bg-violet-700 * {
+  .bg-blue-500 *,
+  .bg-blue-600 *,
+  .bg-blue-700 *,
+  .bg-red-500 *,
+  .bg-red-600 *,
+  .bg-red-700 *,
+  .bg-red-800 *,
+  .bg-green-500 *,
+  .bg-green-600 *,
+  .bg-green-700 *,
+  .bg-sky-500 *,
+  .bg-sky-600 *,
+  .bg-sky-700 *,
+  .bg-purple-500 *,
+  .bg-purple-600 *,
+  .bg-purple-700 *,
+  .bg-violet-500 *,
+  .bg-violet-600 *,
+  .bg-violet-700 * {
     color: inherit !important;
   }
 
   /* Fix page title gradient - comprehensive override */
-  h1.text-transparent, .text-transparent.bg-clip-text,
+  h1.text-transparent,
+  .text-transparent.bg-clip-text,
   .bg-gradient-to-r.from-violet-400.to-pink-400 {
     background: none !important;
     color: #1f2937 !important; /* gray-800 - consistent with other pages */
@@ -579,15 +667,24 @@ endblock %} {% block head_extra_styles %}
   }
 
   /* Ensure all violet/purple colors are converted to blue theme */
-  .text-violet-300, .text-violet-400, .text-violet-100 {
+  .text-violet-300,
+  .text-violet-400,
+  .text-violet-100 {
     color: #3b82f6 !important; /* blue-500 */
   }
 
-  .border-violet-300, .border-violet-400 {
-    border-color: rgba(59, 130, 246, 0.3) !important; /* blue-500 with opacity */
+  .border-violet-300,
+  .border-violet-400 {
+    border-color: rgba(
+      59,
+      130,
+      246,
+      0.3
+    ) !important; /* blue-500 with opacity */
   }
 
-  .hover\\:text-violet-400:hover, .hover\\:text-violet-100:hover {
+  .hover\\:text-violet-400:hover,
+  .hover\\:text-violet-100:hover {
     color: #2563eb !important; /* blue-600 */
   }
 
@@ -613,7 +710,9 @@ endblock %} {% block head_extra_styles %}
   }
 
   /* Even more specific selector targeting the exact auth token wrapper */
-  .mb-6 .flex.items-center div.flex.items-center.flex-grow.border.rounded-md:focus-within {
+  .mb-6
+    .flex.items-center
+    div.flex.items-center.flex-grow.border.rounded-md:focus-within {
     border-color: #3b82f6 !important; /* blue-500 */
   }
 
@@ -625,7 +724,12 @@ endblock %} {% block head_extra_styles %}
 
   .focus\\:ring-primary-200:focus,
   .focus\\:ring-primary-300:focus {
-    --tw-ring-color: rgba(59, 130, 246, 0.2) !important; /* blue-500 with opacity */
+    --tw-ring-color: rgba(
+      59,
+      130,
+      246,
+      0.2
+    ) !important; /* blue-500 with opacity */
   }
 
   /* Fix select element styling */
@@ -645,15 +749,18 @@ endblock %} {% block head_extra_styles %}
   }
 
   /* Override any remaining primary colors */
-  .text-primary-600, .text-primary-500 {
+  .text-primary-600,
+  .text-primary-500 {
     color: #3b82f6 !important; /* blue-500 */
   }
 
-  .bg-primary-600, .bg-primary-500 {
+  .bg-primary-600,
+  .bg-primary-500 {
     background-color: #3b82f6 !important; /* blue-500 */
   }
 
-  .bg-primary-700:hover, .hover\\:bg-primary-700:hover {
+  .bg-primary-700:hover,
+  .hover\\:bg-primary-700:hover {
     background-color: #2563eb !important; /* blue-600 */
   }
 
@@ -693,9 +800,7 @@ endblock %} {% block head_extra_styles %}
       <i class="fas fa-sync-alt"></i>
     </button>
 
-    <h1
-      class="text-3xl font-extrabold text-center text-gray-800 mb-4"
-    >
+    <h1 class="text-3xl font-extrabold text-center text-gray-800 mb-4">
       <img
         src="/static/icons/logo.png"
         alt="Gemini Balance Logo"
@@ -705,11 +810,13 @@ endblock %} {% block head_extra_styles %}
     </h1>
 
     <!-- Navigation Tabs -->
-    <div class="nav-buttons-container flex justify-center mb-8 overflow-x-auto gap-2">
+    <div
+      class="nav-buttons-container flex justify-center mb-8 overflow-x-auto gap-2"
+    >
       <a
         href="/config"
         class="main-nav-btn whitespace-nowrap flex items-center justify-center gap-2 px-6 py-3 font-medium rounded-lg shadow-md hover:shadow-lg transition-all duration-200"
-        style="background-color: #3b82f6 !important; color: #ffffff !important;"
+        style="background-color: #3b82f6 !important; color: #ffffff !important"
       >
         <i class="fas fa-cog"></i> 配置编辑
       </a>
@@ -734,49 +841,93 @@ endblock %} {% block head_extra_styles %}
       <button
         class="tab-btn active px-5 py-2 rounded-full font-medium text-sm transition-all duration-200"
         data-tab="api"
-        style="background-color: #3b82f6 !important; color: #ffffff !important; border: 2px solid #2563eb !important; box-shadow: 0 4px 12px -2px rgba(59, 130, 246, 0.4), 0 2px 6px -1px rgba(59, 130, 246, 0.2) !important; transform: translateY(-2px) !important; font-weight: 600 !important;"
+        style="
+          background-color: #3b82f6 !important;
+          color: #ffffff !important;
+          border: 2px solid #2563eb !important;
+          box-shadow: 0 4px 12px -2px rgba(59, 130, 246, 0.4),
+            0 2px 6px -1px rgba(59, 130, 246, 0.2) !important;
+          transform: translateY(-2px) !important;
+          font-weight: 600 !important;
+        "
       >
         API配置
       </button>
       <button
         class="tab-btn px-5 py-2 rounded-full font-medium text-sm transition-all duration-200"
         data-tab="model"
-        style="background-color: #f8fafc !important; color: #64748b !important; border: 2px solid #e2e8f0 !important; box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important; font-weight: 500 !important;"
+        style="
+          background-color: #f8fafc !important;
+          color: #64748b !important;
+          border: 2px solid #e2e8f0 !important;
+          box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important;
+          font-weight: 500 !important;
+        "
       >
         模型配置
       </button>
       <button
         class="tab-btn px-5 py-2 rounded-full font-medium text-sm transition-all duration-200"
         data-tab="tts"
-        style="background-color: #f8fafc !important; color: #64748b !important; border: 2px solid #e2e8f0 !important; box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important; font-weight: 500 !important;"
+        style="
+          background-color: #f8fafc !important;
+          color: #64748b !important;
+          border: 2px solid #e2e8f0 !important;
+          box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important;
+          font-weight: 500 !important;
+        "
       >
         TTS 配置
       </button>
       <button
         class="tab-btn px-5 py-2 rounded-full font-medium text-sm transition-all duration-200"
         data-tab="image"
-        style="background-color: #f8fafc !important; color: #64748b !important; border: 2px solid #e2e8f0 !important; box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important; font-weight: 500 !important;"
+        style="
+          background-color: #f8fafc !important;
+          color: #64748b !important;
+          border: 2px solid #e2e8f0 !important;
+          box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important;
+          font-weight: 500 !important;
+        "
       >
         图像生成
       </button>
       <button
         class="tab-btn px-5 py-2 rounded-full font-medium text-sm transition-all duration-200"
         data-tab="stream"
-        style="background-color: #f8fafc !important; color: #64748b !important; border: 2px solid #e2e8f0 !important; box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important; font-weight: 500 !important;"
+        style="
+          background-color: #f8fafc !important;
+          color: #64748b !important;
+          border: 2px solid #e2e8f0 !important;
+          box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important;
+          font-weight: 500 !important;
+        "
       >
         流式输出
       </button>
       <button
         class="tab-btn px-5 py-2 rounded-full font-medium text-sm transition-all duration-200"
         data-tab="scheduler"
-        style="background-color: #f8fafc !important; color: #64748b !important; border: 2px solid #e2e8f0 !important; box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important; font-weight: 500 !important;"
+        style="
+          background-color: #f8fafc !important;
+          color: #64748b !important;
+          border: 2px solid #e2e8f0 !important;
+          box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important;
+          font-weight: 500 !important;
+        "
       >
         定时任务
       </button>
       <button
         class="tab-btn px-5 py-2 rounded-full font-medium text-sm transition-all duration-200"
         data-tab="logging"
-        style="background-color: #f8fafc !important; color: #64748b !important; border: 2px solid #e2e8f0 !important; box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important; font-weight: 500 !important;"
+        style="
+          background-color: #f8fafc !important;
+          color: #64748b !important;
+          border: 2px solid #e2e8f0 !important;
+          box-shadow: 0 1px 3px 0 rgba(0, 0, 0, 0.1) !important;
+          font-weight: 500 !important;
+        "
       >
         日志配置
       </button>
@@ -898,11 +1049,48 @@ endblock %} {% block head_extra_styles %}
           />
           <small class="text-gray-500 mt-1 block">Gemini API的基础URL</small>
         </div>
- 
-        <!-- Vertex API密钥列表 -->
+
+        <!-- 自定义Headers -->
         <div class="mb-6">
-          <label for="VERTEX_API_KEYS" class="block font-semibold mb-2 text-gray-700"
-            >Vertex API密钥列表</label
+          <label
+            for="CUSTOM_HEADERS"
+            class="block font-semibold mb-2 text-gray-700"
+            >自定义Headers</label
+          >
+          <div
+            class="bg-white rounded-lg border border-gray-200 p-4 mb-2 space-y-3"
+            id="CUSTOM_HEADERS_container"
+            style="
+              background-color: rgba(255, 255, 255, 0.95);
+              border: 1px solid rgba(0, 0, 0, 0.12);
+              color: #374151;
+            "
+          >
+            <!-- 键值对将在这里动态添加 -->
+            <div class="text-gray-500 text-sm italic">
+              添加自定义请求头，例如 X-Api-Key: your-key
+            </div>
+          </div>
+          <div class="flex justify-end">
+            <button
+              type="button"
+              class="bg-blue-600 hover:bg-blue-700 text-white px-4 py-2 rounded-lg font-medium transition-all duration-200 flex items-center gap-2"
+              id="addCustomHeaderBtn"
+            >
+              <i class="fas fa-plus"></i> 添加Header
+            </button>
+          </div>
+          <small class="text-gray-500 mt-1 block"
+            >在这里添加的键值对将被添加到所有出站API请求的Header中。</small
+          >
+        </div>
+
+        <!-- Vertex Express API密钥列表 -->
+        <div class="mb-6">
+          <label
+            for="VERTEX_API_KEYS"
+            class="block font-semibold mb-2 text-gray-700"
+            >Vertex Express API密钥列表</label
           >
           <div class="array-container" id="VERTEX_API_KEYS_container">
             <!-- 数组项将在这里动态添加 -->
@@ -927,10 +1115,12 @@ endblock %} {% block head_extra_styles %}
             >Vertex AI Platform API密钥列表。点击按钮可批量添加或删除。</small
           >
         </div>
- 
+
         <!-- Vertex Express API基础URL -->
         <div class="mb-6">
-          <label for="VERTEX_EXPRESS_BASE_URL" class="block font-semibold mb-2 text-gray-700"
+          <label
+            for="VERTEX_EXPRESS_BASE_URL"
+            class="block font-semibold mb-2 text-gray-700"
             >Vertex Express API基础URL</label
           >
           <input
@@ -940,30 +1130,32 @@ endblock %} {% block head_extra_styles %}
             placeholder="https://aiplatform.googleapis.com/v1beta1/publishers/google/models"
             class="w-full px-4 py-3 rounded-lg form-input-themed"
           />
-          <small class="text-gray-500 mt-1 block">Vertex Express API的基础URL</small>
+          <small class="text-gray-500 mt-1 block"
+            >Vertex Express API的基础URL</small
+          >
         </div>
         <!-- 智能路由配置 -->
         <div class="mb-6">
           <div class="flex items-center justify-between">
-              <label
-                      for="URL_NORMALIZATION_ENABLED"
-                      class="font-semibold text-gray-700"
+            <label
+              for="URL_NORMALIZATION_ENABLED"
+              class="font-semibold text-gray-700"
               >启用智能路由映射</label
-              >
-              <div
-                      class="relative inline-block w-10 mr-2 align-middle select-none transition duration-200 ease-in"
-              >
-                  <input
-                          type="checkbox"
-                          name="URL_NORMALIZATION_ENABLED"
-                          id="URL_NORMALIZATION_ENABLED"
-                          class="toggle-checkbox absolute block w-6 h-6 rounded-full bg-white border-4 appearance-none cursor-pointer"
-                  />
-                  <label
-                          for="URL_NORMALIZATION_ENABLED"
-                          class="toggle-label block overflow-hidden h-6 rounded-full bg-gray-300 cursor-pointer"
-                  ></label>
-              </div>
+            >
+            <div
+              class="relative inline-block w-10 mr-2 align-middle select-none transition duration-200 ease-in"
+            >
+              <input
+                type="checkbox"
+                name="URL_NORMALIZATION_ENABLED"
+                id="URL_NORMALIZATION_ENABLED"
+                class="toggle-checkbox absolute block w-6 h-6 rounded-full bg-white border-4 appearance-none cursor-pointer"
+              />
+              <label
+                for="URL_NORMALIZATION_ENABLED"
+                class="toggle-label block overflow-hidden h-6 rounded-full bg-gray-300 cursor-pointer"
+              ></label>
+            </div>
           </div>
           <small class="text-gray-500 mt-1 block">
             自动客户端请求的url拼接为正确格式（仅保证正常聊天，出现问题请关闭）
@@ -1057,28 +1249,28 @@ endblock %} {% block head_extra_styles %}
         <!-- 代理使用策略 -->
         <div class="mb-6">
           <div class="flex items-center justify-between">
-              <label
-                      for="PROXIES_USE_CONSISTENCY_HASH_BY_API_KEY"
-                      class="font-semibold text-gray-700"
+            <label
+              for="PROXIES_USE_CONSISTENCY_HASH_BY_API_KEY"
+              class="font-semibold text-gray-700"
               >是否开启固定代理策略</label
-              >
-              <div
-                      class="relative inline-block w-10 mr-2 align-middle select-none transition duration-200 ease-in"
-              >
-                  <input
-                          type="checkbox"
-                          name="PROXIES_USE_CONSISTENCY_HASH_BY_API_KEY"
-                          id="PROXIES_USE_CONSISTENCY_HASH_BY_API_KEY"
-                          class="toggle-checkbox absolute block w-6 h-6 rounded-full bg-white border-4 appearance-none cursor-pointer"
-                  />
-                  <label
-                          for="PROXIES_USE_CONSISTENCY_HASH_BY_API_KEY"
-                          class="toggle-label block overflow-hidden h-6 rounded-full bg-gray-300 cursor-pointer"
-                  ></label>
-              </div>
+            >
+            <div
+              class="relative inline-block w-10 mr-2 align-middle select-none transition duration-200 ease-in"
+            >
+              <input
+                type="checkbox"
+                name="PROXIES_USE_CONSISTENCY_HASH_BY_API_KEY"
+                id="PROXIES_USE_CONSISTENCY_HASH_BY_API_KEY"
+                class="toggle-checkbox absolute block w-6 h-6 rounded-full bg-white border-4 appearance-none cursor-pointer"
+              />
+              <label
+                for="PROXIES_USE_CONSISTENCY_HASH_BY_API_KEY"
+                class="toggle-label block overflow-hidden h-6 rounded-full bg-gray-300 cursor-pointer"
+              ></label>
+            </div>
           </div>
           <small class="text-gray-500 mt-1 block"
-          >开启后，对于每一个API_KEY将根据算法从代理列表中选取同一个代理IP，防止一个API_KEY同时被多个IP访问，也同时防止了一个IP访问了过多的API_KEY。</small
+            >开启后，对于每一个API_KEY将根据算法从代理列表中选取同一个代理IP，防止一个API_KEY同时被多个IP访问，也同时防止了一个IP访问了过多的API_KEY。</small
           >
         </div>
       </div>
@@ -1377,8 +1569,8 @@ endblock %} {% block head_extra_styles %}
           </div>
         </div>
       </div>
- 
-       <!-- TTS配置 -->
+
+      <!-- TTS配置 -->
       <div class="config-section" id="tts-section">
         <h2
           class="text-xl font-bold mb-6 pb-3 border-b flex items-center gap-2 text-gray-800 border-violet-300 border-opacity-30"
@@ -1396,15 +1588,21 @@ endblock %} {% block head_extra_styles %}
             name="TTS_MODEL"
             class="w-full px-4 py-3 rounded-lg form-select-themed"
           >
-            <option value="gemini-2.5-flash-preview-tts">gemini-2.5-flash-preview-tts</option>
-            <option value="gemini-2.5-pro-preview-tts">gemini-2.5-pro-preview-tts</option>
+            <option value="gemini-2.5-flash-preview-tts">
+              gemini-2.5-flash-preview-tts
+            </option>
+            <option value="gemini-2.5-pro-preview-tts">
+              gemini-2.5-pro-preview-tts
+            </option>
           </select>
           <small class="text-gray-500 mt-1 block">用于TTS的模型</small>
         </div>
 
         <!-- TTS 语音名称 -->
         <div class="mb-6">
-          <label for="TTS_VOICE_NAME" class="block font-semibold mb-2 text-gray-700"
+          <label
+            for="TTS_VOICE_NAME"
+            class="block font-semibold mb-2 text-gray-700"
             >TTS 语音名称</label
           >
           <select
@@ -1443,7 +1641,9 @@ endblock %} {% block head_extra_styles %}
             <option value="Sadaltager">Sadaltager (博学)</option>
             <option value="Sulafat">Sulafat (温暖)</option>
           </select>
-          <small class="text-gray-500 mt-1 block">TTS 的语音名称，控制风格、语调、口音和节奏</small>
+          <small class="text-gray-500 mt-1 block"
+            >TTS 的语音名称，控制风格、语调、口音和节奏</small
+          >
         </div>
 
         <!-- TTS 语速 -->
@@ -1464,9 +1664,9 @@ endblock %} {% block head_extra_styles %}
         </div>
       </div>
 
-       <!-- 图像生成相关配置 -->
-       <div class="config-section" id="image-section">
-         <h2
+      <!-- 图像生成相关配置 -->
+      <div class="config-section" id="image-section">
+        <h2
           class="text-xl font-bold mb-6 pb-3 border-b flex items-center gap-2 text-gray-800 border-violet-300 border-opacity-30"
         >
           <i class="fas fa-image text-violet-400"></i> 图像生成配置
@@ -1602,6 +1802,25 @@ endblock %} {% block head_extra_styles %}
           />
           <small class="text-gray-500 mt-1 block">Cloudflare图床的认证码</small>
         </div>
+
+        <!-- Cloudflare上传文件夹 -->
+        <div class="mb-6 provider-config" data-provider="cloudflare_imgbed">
+          <label
+            for="CLOUDFLARE_IMGBED_UPLOAD_FOLDER"
+            class="block font-semibold mb-2 text-gray-700"
+            >Cloudflare上传文件夹</label
+          >
+          <input
+            type="text"
+            id="CLOUDFLARE_IMGBED_UPLOAD_FOLDER"
+            name="CLOUDFLARE_IMGBED_UPLOAD_FOLDER"
+            placeholder=""
+            class="w-full px-4 py-3 rounded-lg form-input-themed"
+          />
+          <small class="text-gray-500 mt-1 block"
+            >Cloudflare图床的上传文件夹路径（可选）</small
+          >
+        </div>
       </div>
 
       <!-- 流式输出优化配置 -->
@@ -1956,12 +2175,17 @@ endblock %} {% block head_extra_styles %}
       </div>
 
       <!-- Action Buttons -->
-      <div class="flex flex-col md:flex-row justify-center gap-4 mt-8 pt-4 pb-2">
+      <div
+        class="flex flex-col md:flex-row justify-center gap-4 mt-8 pt-4 pb-2"
+      >
         <button
           type="button"
           id="saveBtn"
           class="action-btn text-white px-8 py-3 rounded-xl font-semibold transition-all duration-300 hover:shadow-lg flex items-center justify-center gap-2"
-          style="background-color: #3b82f6 !important; color: #ffffff !important;"
+          style="
+            background-color: #3b82f6 !important;
+            color: #ffffff !important;
+          "
         >
           <i class="fas fa-save"></i> 保存配置
         </button>
@@ -1969,7 +2193,10 @@ endblock %} {% block head_extra_styles %}
           type="button"
           id="resetBtn"
           class="action-btn bg-gradient-to-r from-gray-600 to-gray-700 text-white px-8 py-3 rounded-xl font-semibold transition-all duration-300 hover:shadow-lg flex items-center justify-center gap-2"
-          style="background-color: #6b7280 !important; color: #ffffff !important;"
+          style="
+            background-color: #6b7280 !important;
+            color: #ffffff !important;
+          "
         >
           <i class="fas fa-undo"></i> 重置配置
         </button>
@@ -2230,8 +2457,8 @@ endblock %} {% block head_extra_styles %}
     </div>
   </div>
 </div>
- 
-<!-- Vertex API Key Add Modal -->
+
+<!-- Vertex Express API Key Add Modal -->
 <div id="vertexApiKeyModal" class="modal">
   <div
     class="w-full max-w-lg mx-auto rounded-2xl shadow-2xl overflow-hidden animate-fade-in"
@@ -2243,7 +2470,9 @@ endblock %} {% block head_extra_styles %}
   >
     <div class="p-6">
       <div class="flex justify-between items-center mb-4">
-        <h2 class="text-xl font-bold text-gray-800">批量添加 Vertex API 密钥</h2>
+        <h2 class="text-xl font-bold text-gray-800">
+          批量添加 Vertex Express API 密钥
+        </h2>
         <button
           id="closeVertexApiKeyModalBtn"
           class="text-gray-300 hover:text-gray-800 text-xl"
@@ -2252,12 +2481,13 @@ endblock %} {% block head_extra_styles %}
         </button>
       </div>
       <p class="text-gray-300 mb-4">
-        每行粘贴一个或多个密钥，将自动提取有效密钥 (格式: AQ.开头，共53位) 并去重。
+        每行粘贴一个或多个密钥，将自动提取有效密钥 (格式: AQ.开头，共53位)
+        并去重。
       </p>
       <textarea
         id="vertexApiKeyBulkInput"
         rows="10"
-        placeholder="在此处粘贴 Vertex API 密钥..."
+        placeholder="在此处粘贴 Vertex Express API 密钥..."
         class="w-full px-4 py-3 rounded-lg font-mono text-sm form-input-themed"
       ></textarea>
       <div class="flex justify-end gap-3 mt-6">
@@ -2279,8 +2509,8 @@ endblock %} {% block head_extra_styles %}
     </div>
   </div>
 </div>
- 
-<!-- Bulk Delete Vertex API Key Modal -->
+
+<!-- Bulk Delete Vertex Express API Key Modal -->
 <div id="bulkDeleteVertexApiKeyModal" class="modal">
   <div
     class="w-full max-w-lg mx-auto rounded-2xl shadow-2xl overflow-hidden animate-fade-in"
@@ -2292,7 +2522,9 @@ endblock %} {% block head_extra_styles %}
   >
     <div class="p-6">
       <div class="flex justify-between items-center mb-4">
-        <h2 class="text-xl font-bold text-gray-800">批量删除 Vertex API 密钥</h2>
+        <h2 class="text-xl font-bold text-gray-800">
+          批量删除 Vertex Express API 密钥
+        </h2>
         <button
           id="closeBulkDeleteVertexModalBtn"
           class="text-gray-300 hover:text-gray-800 text-xl"
@@ -2306,7 +2538,7 @@ endblock %} {% block head_extra_styles %}
       <textarea
         id="bulkDeleteVertexApiKeyInput"
         rows="10"
-        placeholder="在此处粘贴要删除的 Vertex API 密钥..."
+        placeholder="在此处粘贴要删除的 Vertex Express API 密钥..."
         class="w-full px-4 py-3 rounded-lg font-mono text-sm form-input-themed"
       ></textarea>
       <div class="flex justify-end gap-3 mt-6">
@@ -2328,7 +2560,7 @@ endblock %} {% block head_extra_styles %}
     </div>
   </div>
 </div>
- 
+
 <!-- Model Helper Modal -->
 <div id="modelHelperModal" class="modal">
   <div
@@ -2456,8 +2688,10 @@ endblock %} {% block head_extra_styles %}
       input.addEventListener("focus", function () {
         const parentItem = this.closest(".map-item");
         if (parentItem) {
-          parentItem.style.backgroundColor = "rgba(243, 244, 246, 1)"; /* gray-100 */
-          parentItem.style.borderColor = "rgba(59, 130, 246, 0.5)"; /* blue-500 */
+          parentItem.style.backgroundColor =
+            "rgba(243, 244, 246, 1)"; /* gray-100 */
+          parentItem.style.borderColor =
+            "rgba(59, 130, 246, 0.5)"; /* blue-500 */
         }
       });
 

app/templates/keys_status.html

@@ -1245,6 +1245,7 @@ endblock %} {% block head_extra_styles %}
               <option value="20">20</option>
               <option value="50">50</option>
               <option value="100">100</option>
+              <option value="500">500</option>
             </select>
             <span class="text-sm select-none font-semibold" style="color: #1f2937 !important;">项</span>
           </div>

app/utils/uploader.py

@@ -261,18 +261,20 @@ class PicGoUploader(ImageUploader):
 
 class CloudFlareImgBedUploader(ImageUploader):
     """CloudFlare图床上传器"""
-    
-    def __init__(self, auth_code: str, api_url: str):
+
+    def __init__(self, auth_code: str, api_url: str, upload_folder: str = ""):
         """
         初始化CloudFlare图床上传器
         
         Args:
             auth_code: 认证码
             api_url: 上传API地址
+            upload_folder: 上传文件夹路径（可选）
         """
         self.auth_code = auth_code
         self.api_url = api_url
-        
+        self.upload_folder = upload_folder
+
     def upload(self, file: bytes, filename: str) -> UploadResponse:
         """
         上传图片到CloudFlare图床
@@ -288,12 +290,16 @@ class CloudFlareImgBedUploader(ImageUploader):
             UploadError: 上传失败时抛出异常
         """
         try:
-            # 准备请求URL（添加认证码参数，如果存在）
+            # 准备请求URL参数
+            params = []
+            if self.upload_folder:
+                params.append(f"uploadFolder={self.upload_folder}")
             if self.auth_code:
-                request_url = f"{self.api_url}?authCode={self.auth_code}&uploadNameType=origin"
-            else:
-                request_url = f"{self.api_url}?uploadNameType=origin"
-            
+                params.append(f"authCode={self.auth_code}")
+            params.append("uploadNameType=origin")
+
+            request_url = f"{self.api_url}?{'&'.join(params)}"
+
             # 准备文件数据
             files = {
                 "file": (filename, file)
@@ -388,6 +394,7 @@ class ImageUploaderFactory:
         elif provider == "cloudflare_imgbed":
             return CloudFlareImgBedUploader(
                 credentials["auth_code"],
-                credentials["base_url"]
+                credentials["base_url"],
+                credentials.get("upload_folder", ""),
             )
         raise ValueError(f"Unknown provider: {provider}")