向量化使用了来自客户端的 OpenAI API Key 调用其它厂商的向量化模型

[ShellWen]

📦 部署环境

Docker

📦 部署模式

服务端模式(lobe-chat-database 镜像)

📌 软件版本

1.79.8

💻 系统环境

Other Linux

🌐 浏览器

Firefox

🐛 问题描述

在客户端（浏览器）的用户已设置 OpenAI API Key 的情况下，不管环境变量中自定义的向量化模型是哪家的，都会在请求的时候使用来自浏览器中 OpenAI API Key 去请求 向量化模型厂商 的 API，导致 API Key 因为无法匹配报错

📷 复现步骤

1. 在 .env 中设置

SILICONCLOUD_API_KEY="sk-😭"
DEFAULT_FILES_CONFIG=embedding_model=siliconcloud/BAAI/bge-m3

以启用来自硅基流动的向量化模型
2. 用户浏览器启用 OpenAI，并设置 API Key
3. 新建知识库，上传文件，在向量化阶段会因为使用 OpenAI 的 API Key 去请求硅基流动的 API 导致报错

🚦 期望结果

No response

📝 补充信息

No response

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

📦 Deployment environment

Docker

📦 Deployment mode

Server-side mode (lobe-chat-database mirror)

📌 Software version

1.79.8

💻 System environment

Other Linux

🌐 Browser

Firefox

🐛 Question description

When the user of the client (browser) has set the OpenAI API Key, no matter which vectorized model is customized in the environment variable, it will use the OpenAI API Key in the browser to request the API from the vectorized model manufacturer when requesting, resulting in an error in the API Key because it cannot match.

📷 Reproduction steps

1. Set it in .env

SILICONCLOUD_API_KEY="sk-😭"
DEFAULT_FILES_CONFIG=embedding_model=siliconcloud/BAAI/bge-m3

To enable vectorized models from silicon-based flow
2. Enable OpenAI in the user's browser and set up API Key
3. Create a new knowledge base and upload a file. During the vectorization stage, an error will be reported because of using OpenAI's API Key to request a silicon-based flow API.

🚦 Expected results

No response

📝 Supplementary information

No response

[lobehubbot]

👀 @ShellWen

Thank you for raising an issue. We will investigate into the matter and get back to you as soon as possible.
Please make sure you have given us as much context as possible.
非常感谢您提交 issue。我们会尽快调查此事，并尽快回复您。 请确保您已经提供了尽可能多的背景信息。

[dosubot]

Hello @ShellWen! I'm here to help you with any bugs or questions you might have while you wait for a human maintainer. Feel free to ask me anything!

To address the issue where the vectorization process uses the OpenAI API key from the browser instead of the custom vectorization model's API key set in the environment variables, you should ensure that the API key is correctly configured in your Docker environment variables or .env file. This setup should override any API key set in the browser and prevent errors when requesting the vectorization model provider's API. Additionally, make sure to remove any conflicting API key and URL configurations, such as those for OpenAI, from your environment settings to avoid conflicts ^[1].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}