fix: 去掉.doc分段支持

apps/common/handle/impl/text_split_handle.py

@@ -29,7 +29,8 @@ class TextSplitHandle(BaseSplitHandle):
         if file_name.endswith(".md") or file_name.endswith('.txt'):
             return True
         result = detect(buffer)
-        if result['encoding'] != 'ascii' and result['confidence'] > 0.5:
+        if result['encoding'] is not None and result['confidence'] is not None and result['encoding'] != 'ascii' and \
+                result['confidence'] > 0.5:
             return True
         return False
 

ui/src/utils/utils.ts

@@ -31,18 +31,21 @@ export const randomId = function () {
 */
 export function fileType(name: string) {
   const suffix = name.split('.')
-
-  return suffix[suffix.length - 1] === 'docx' ? 'doc' : suffix[suffix.length - 1]
+  return suffix[suffix.length - 1]
 }
 
 /*
   获得文件对应图片
 */
 export function getImgUrl(name: string) {
-  const typeList = ['txt', 'pdf', 'doc', 'csv', 'md']
-  const type = typeList.includes(fileType(name)) ? fileType(name) : 'unknow'
+  const type = isRightType(name) ? fileType(name) : 'unknow'
   return new URL(`../assets/${type}-icon.svg`, import.meta.url).href
 }
+// 是否是白名单后缀
+export function isRightType(name: string) {
+  const typeList = ['txt', 'pdf', 'docx', 'csv', 'md']
+  return typeList.includes(fileType(name))
+}
 
 /*
   从指定数组中过滤出对应的对象

ui/src/views/dataset/component/UploadComponent.vue

@@ -16,7 +16,7 @@
         action="#"
         :auto-upload="false"
         :show-file-list="false"
-        accept=".txt, .md, .csv, .log, .doc, .docx, .pdf"
+        accept=".txt, .md, .csv, .log, .docx, .pdf"
         :limit="50"
         :on-exceed="onExceed"
         :on-change="filehandleChange"
@@ -29,7 +29,7 @@
           </p>
           <div class="upload__decoration">
             <p>
-              支持格式：TXT、Markdown、PDF、DOC、DOCX，每次最多上传50个文件，每个文件不超过 100MB
+              支持格式：TXT、Markdown、PDF、DOCX，每次最多上传50个文件，每个文件不超过 100MB
             </p>
             <p>若使用【高级分段】建议上传前规范文件的分段标识</p>
           </div>
@@ -61,7 +61,7 @@
 <script setup lang="ts">
 import { ref, reactive, onUnmounted, onMounted, computed, watch } from 'vue'
 import type { UploadFile, UploadFiles } from 'element-plus'
-import { filesize, getImgUrl } from '@/utils/utils'
+import { filesize, getImgUrl, isRightType } from '@/utils/utils'
 import { MsgError } from '@/utils/message'
 import useStore from '@/stores'
 const { dataset } = useStore()
@@ -91,6 +91,11 @@ const filehandleChange = (file: any, fileList: UploadFiles) => {
     fileList.splice(-1, 1) //移除当前超出大小的文件
     return false
   }
+  if (!isRightType(file?.name)) {
+    MsgError('文件格式不支持')
+    fileList.splice(-1, 1)
+    return false
+  }
 }
 
 const onExceed = () => {