fix: 知识库无法向量化 MD 文件 #569

.yarn/patches/@llm-tools-embedjs-loader-markdown-npm-0.1.25-d1d536d640.patch

@@ -0,0 +1,19 @@
+diff --git a/src/markdown-loader.js b/src/markdown-loader.js
+index 8a17cb7f5a68d90d2be21682db6e95ce22a3e71c..9ee868ef9d4ff3dc914b3abc3c8006deb1e9c6c6 100644
+--- a/src/markdown-loader.js
++++ b/src/markdown-loader.js
+@@ -1,5 +1,4 @@
+ import { micromark } from 'micromark';
+-import { mdxJsx } from 'micromark-extension-mdx-jsx';
+ import { gfmHtml, gfm } from 'micromark-extension-gfm';
+ import createDebugMessages from 'debug';
+ import fs from 'node:fs';
+@@ -21,7 +20,7 @@ export class MarkdownLoader extends BaseLoader {
+             ? (await getSafe(this.filePathOrUrl, { format: 'buffer' })).body
+             : await stream2buffer(fs.createReadStream(this.filePathOrUrl));
+         this.debug('MarkdownLoader stream created');
+-        const result = micromark(buffer, { extensions: [gfm(), mdxJsx()], htmlExtensions: [gfmHtml()] });
++        const result = micromark(buffer, { extensions: [gfm()], htmlExtensions: [gfmHtml()] });
+         this.debug('Markdown parsed...');
+         const webLoader = new WebLoader({
+             urlOrContent: result,

package.json

@@ -53,7 +53,7 @@
     "@llm-tools/embedjs": "patch:@llm-tools/embedjs@npm%3A0.1.25#~/.yarn/patches/@llm-tools-embedjs-npm-0.1.25-ec5645cf36.patch",
     "@llm-tools/embedjs-libsql": "patch:@llm-tools/embedjs-libsql@npm%3A0.1.25#~/.yarn/patches/@llm-tools-embedjs-libsql-npm-0.1.25-fad000d74c.patch",
     "@llm-tools/embedjs-loader-csv": "^0.1.25",
-    "@llm-tools/embedjs-loader-markdown": "^0.1.25",
+    "@llm-tools/embedjs-loader-markdown": "patch:@llm-tools/embedjs-loader-markdown@npm%3A0.1.25#~/.yarn/patches/@llm-tools-embedjs-loader-markdown-npm-0.1.25-d1d536d640.patch",
     "@llm-tools/embedjs-loader-msoffice": "^0.1.25",
     "@llm-tools/embedjs-loader-pdf": "^0.1.25",
     "@llm-tools/embedjs-loader-sitemap": "^0.1.25",

src/main/services/KnowledgeService.ts

@@ -122,7 +122,7 @@ class KnowledgeService {
         return await ragApplication.addLoader(new ExcelLoader({ filePathOrUrl: file.path }) as any, forceReload)
       }
 
-      if (['.md', '.mdx'].includes(file.ext)) {
+      if (['.md'].includes(file.ext)) {
         return await ragApplication.addLoader(new MarkdownLoader({ filePathOrUrl: file.path }) as any, forceReload)
       }
 

src/renderer/src/pages/knowledge/KnowledgeContent.tsx

@@ -29,7 +29,7 @@ interface KnowledgeContentProps {
   selectedBase: KnowledgeBase
 }
 
-const fileTypes = ['.pdf', '.docx', '.pptx', '.xlsx', '.txt', '.md', '.mdx']
+const fileTypes = ['.pdf', '.docx', '.pptx', '.xlsx', '.txt', '.md']
 
 const FlexColumn = styled.div`
   display: flex;