# 文件自动导入知识库

**URL:** <https://forum.jit.pro/t/topic/219>\
**Category:** About AI\
**Created:** [September 1, 2025, 3:27am UTC](https://forum.jit.pro/t/topic/219 "2025-09-01T03:27:51Z")\
**Posts on this page:** 7\
**Page:** 1

<div class="post-metadata">

**Author:** ![return0](https://avatars.discourse-cdn.com/v4/letter/r/a9a28c/32.png) [@return0](https://forum.jit.pro/u/return0)\
**Post date:** [September 1, 2025, 3:27am UTC](https://forum.jit.pro/t/topic/219/1 "2025-09-01T03:27:51Z")

</div>

现在有这样一个需求。

公司在使用好队友搭建的信通小精灵过程中会上传各种文档附件，类似于这样的：

 ![image](https://us1.discourse-cdn.com/flex002/uploads/jitai/original/1X/879356df8f15bf3cc94ba4b32a6bd5f49c737176.png)

有没有相关的方法实现自动导入文档到jit的AI知识库进行分块，在这边每有一个文档，jit的AI知识库就可以同步获得。

---

<div class="post-metadata">

**Author:** ![zjfjiayou](https://sea2.discourse-cdn.com/flex002/user_avatar/forum.jit.pro/zjfjiayou/32/10_2.png) [@zjfjiayou](https://forum.jit.pro/u/zjfjiayou)\
**Post date:** [September 2, 2025, 2:05am UTC](https://forum.jit.pro/t/topic/219/2 "2025-09-02T02:05:36Z")

</div>

主要是好队友那边没有什么机会可以调用jit的接口，jit这边倒是可以通过服务函数写入文件到知识库

 ![image](https://us1.discourse-cdn.com/flex002/uploads/jitai/original/1X/26008e69313ce0833618d4e4cd024ee2070006da.png)

---

<div class="post-metadata">

**Author:** ![return0](https://avatars.discourse-cdn.com/v4/letter/r/a9a28c/32.png) [@return0](https://forum.jit.pro/u/return0)\
**Post date:** [September 2, 2025, 3:02am UTC](https://forum.jit.pro/t/topic/219/3 "2025-09-02T03:02:46Z")

</div>

行，我先试试看

---

<div class="post-metadata">

**Author:** ![return0](https://avatars.discourse-cdn.com/v4/letter/r/a9a28c/32.png) [@return0](https://forum.jit.pro/u/return0)\
**Post date:** [September 3, 2025, 6:35am UTC](https://forum.jit.pro/t/topic/219/4 "2025-09-03T06:35:14Z")

</div>

关于这个的文档添加配置要怎么用呢？

 ![image](https://us1.discourse-cdn.com/flex002/uploads/jitai/original/1X/1fa171286bd1d61c6a7a674d6a4a9d2d75d58977.png)

---

<div class="post-metadata">

**Author:** ![zjfjiayou](https://sea2.discourse-cdn.com/flex002/user_avatar/forum.jit.pro/zjfjiayou/32/10_2.png) [@zjfjiayou](https://forum.jit.pro/u/zjfjiayou)\
**Post date:** [September 3, 2025, 9:53am UTC](https://forum.jit.pro/t/topic/219/5 "2025-09-03T09:53:11Z")

</div>

示例代码如下:

```python
def add(self, documentul):
        app.rags.information.addDocumentByBusinessId("TEST001", [{ "fileName": "WSL + Claude Code + Cursor 完整安装指南.pdf", "size": 143898, "url": documentul.value, "type": "application/pdf"}], {"chunkSeparator": ["\n"], "chunkSize": 1024, "chunkOverlap": 100}, {"chunkCleaning": True})

```

其中注意

1. url使用的是参数值
2. 其他文件名称，大小，类型你都可以用变量或者你自己写个函数去url中读取
3. app.rags.information.addDocumentByBusinessId 中的rags.information换成你自己知识库

的fullname入下图

 ![image](https://us1.discourse-cdn.com/flex002/uploads/jitai/original/1X/c3b9a614023c385698f64bb4842717a31d1c5ec4.png)

---

<div class="post-metadata">

**Author:** ![lee](https://avatars.discourse-cdn.com/v4/letter/l/bc8723/32.png) [@lee](https://forum.jit.pro/u/lee)\
**Post date:** [September 4, 2025, 1:03am UTC](https://forum.jit.pro/t/topic/219/6 "2025-09-04T01:03:06Z")

</div>

这个方法是添加文档的时候使用的固定值吧。我们想知道这个生成附件的使用方法以及在使用变量值的时候有没有方法实现从外部添加文档（比如通过写一个新的服务）。

 ![image](https://us1.discourse-cdn.com/flex002/uploads/jitai/original/1X/bfe331ad6cb540ff962bbe4a9e658613f69be579.png)

---

<div class="post-metadata">

**Author:** ![zjfjiayou](https://sea2.discourse-cdn.com/flex002/user_avatar/forum.jit.pro/zjfjiayou/32/10_2.png) [@zjfjiayou](https://forum.jit.pro/u/zjfjiayou)\
**Post date:** [September 4, 2025, 2:18am UTC](https://forum.jit.pro/t/topic/219/7 "2025-09-04T02:18:44Z")

</div>

我给的例子里面附件的url不是固定值，你看下代码里面的url是被传参覆盖了的