fix: upload file (#2992)

* fix: upload file

* chore: remove wasm, support html image parse

* chore: adjust

* chore: move base64match function into htmlstr2md

packages/service/worker/readFile/extension/docx.ts

@@ -1,20 +1,39 @@
-import mammoth from 'mammoth';
-import { ReadRawTextByBuffer, ReadFileResponse } from '../type';
+import mammoth, { images } from 'mammoth';
+import { ReadRawTextByBuffer, ReadFileResponse, ImageType } from '../type';
 import { html2md } from '../../htmlStr2Md/utils';
 
 /**
  * read docx to markdown
  */
 export const readDocsFile = async ({ buffer }: ReadRawTextByBuffer): Promise<ReadFileResponse> => {
+  const imageList: ImageType[] = [];
   try {
-    const { value: html } = await mammoth.convertToHtml({
-      buffer
-    });
+    const { value: html } = await mammoth.convertToHtml(
+      {
+        buffer
+      },
+      {
+        convertImage: images.imgElement(async (image) => {
+          const imageBase64 = await image.readAsBase64String();
+          const uuid = crypto.randomUUID();
+          const mime = image.contentType;
+          imageList.push({
+            uuid,
+            base64: imageBase64,
+            mime
+          });
+          return {
+            src: uuid
+          };
+        })
+      }
+    );
 
-    const rawText = html2md(html);
+    const { rawText } = html2md(html);
 
     return {
-      rawText
+      rawText,
+      imageList
     };
   } catch (error) {
     console.log('error doc read:', error);

packages/service/worker/readFile/extension/html.ts

@@ -5,9 +5,10 @@ import { html2md } from '../../htmlStr2Md/utils';
 export const readHtmlRawText = async (params: ReadRawTextByBuffer): Promise<ReadFileResponse> => {
   const { rawText: html } = readFileRawText(params);
 
-  const rawText = html2md(html);
+  const { rawText, imageList } = html2md(html);
 
   return {
-    rawText
+    rawText,
+    imageList
   };
 };