Coco-AI 接入自定义数据源

https://appstore.lazycat.cloud/#/shop/detail/xu.deploy.coco-ai


之前使用 Hugo Connector 接入和 hexo 和任意 Markdown，后来官方也支持了对于任意数据源的支持，主要是开放了这个接口：

![image-20250417194847718](https://raw.githubusercontent.com/cloudsmithy/picgo-imh/master/image-20250417194847718.png)

具体操作如下：



设置 - conntor - 新增，让输入名称和描述等信息，新建出来 conntor

![image-20250417194910829](https://raw.githubusercontent.com/cloudsmithy/picgo-imh/master/image-20250417194910829.png)

然后我们就能在数据源上的页面上看到刚刚添加的了 Customize Connector 了

![image-20250417195015291](https://raw.githubusercontent.com/cloudsmithy/picgo-imh/master/image-20250417195015291.png)

点开提示，给了一个 API

![image-20250417194847718](https://raw.githubusercontent.com/cloudsmithy/picgo-imh/master/image-20250417194847718.png)

然后我们去创建 token，如图

![image-20250417195602722](https://raw.githubusercontent.com/cloudsmithy/picgo-imh/master/image-20250417195602722.png)

我这边使用 Postman 进行设置

![](https://raw.githubusercontent.com/cloudsmithy/picgo-imh/master/image-20250417195729078.png)

如果你的请求没有带 token，就是这样的。

![image-20250417195729078](https://raw.githubusercontent.com/cloudsmithy/picgo-imh/master/image-20250417195729078.png)

转成代码的是这样的，当然也可以开发自己的 agent。

```python
import requests
import json

url = "http://localhost:9000/datasource/d00eeolvf2xxx/_doc"

payload = json.dumps({
  "title": "I am just a Coco doc that you can search",
  "summary": "Nothing but great start",
  "content": "Coco is a unified private search engien that you can trust.",
  "url": "http://coco.rs/",
  "icon": "default"
})
headers = {
  'Content-Type': 'application/json'
}

response = requests.request("POST", url, headers=headers, data=payload)

print(response.text)

```