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/htmlStr2Md/utils.ts

@@ -1,7 +1,14 @@
 import TurndownService from 'turndown';
+import { ImageType } from '../readFile/type';
+// @ts-ignore
 const turndownPluginGfm = require('joplin-turndown-plugin-gfm');
 
-export const html2md = (html: string): string => {
+export const html2md = (
+  html: string
+): {
+  rawText: string;
+  imageList: ImageType[];
+} => {
   const turndownService = new TurndownService({
     headingStyle: 'atx',
     bulletListMarker: '-',
@@ -15,12 +22,32 @@ export const html2md = (html: string): string => {
 
   try {
     turndownService.remove(['i', 'script', 'iframe', 'style']);
-
     turndownService.use(turndownPluginGfm.gfm);
 
-    return turndownService.turndown(html);
+    const base64Regex = /"(data:image\/[^;]+;base64[^"]+)"/g;
+    const imageList: ImageType[] = [];
+    const images = Array.from(html.match(base64Regex) || []);
+    for (const image of images) {
+      const uuid = crypto.randomUUID();
+      const mime = image.split(';')[0].split(':')[1];
+      const base64 = image.split(',')[1];
+      html = html.replace(image, uuid);
+      imageList.push({
+        uuid,
+        base64,
+        mime
+      });
+    }
+
+    return {
+      rawText: turndownService.turndown(html),
+      imageList
+    };
   } catch (error) {
     console.log('html 2 markdown error', error);
-    return '';
+    return {
+      rawText: '',
+      imageList: []
+    };
   }
 };

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
   };
 };

packages/service/worker/readFile/type.d.ts

@@ -8,7 +8,14 @@ export type ReadRawTextProps<T> = {
 
 export type ReadRawTextByBuffer = ReadRawTextProps<Buffer>;
 
+export type ImageType = {
+  uuid: string;
+  base64: string;
+  mime: string;
+};
+
 export type ReadFileResponse = {
   rawText: string;
   formatText?: string;
+  imageList?: ImageType[];
 };