From 36a573fce9696a8feac9574d03ab8eb3e987ceeb Mon Sep 17 00:00:00 2001
From: Philipinho <16838612+Philipinho@users.noreply.github.com>
Date: Mon, 26 May 2025 15:37:35 -0700
Subject: [PATCH] import embeds

---
 .../integrations/import/file-task.service.ts  | 10 ++---
 .../integrations/import/import-formatter.ts   | 43 +++++++++++++++++++
 2 files changed, 46 insertions(+), 7 deletions(-)

diff --git a/apps/server/src/integrations/import/file-task.service.ts b/apps/server/src/integrations/import/file-task.service.ts
index 2ed96d6e..3ad92d3c 100644
--- a/apps/server/src/integrations/import/file-task.service.ts
+++ b/apps/server/src/integrations/import/file-task.service.ts
@@ -28,8 +28,7 @@ import { markdownToHtml } from '@docmost/editor-ext';
 import { getAttachmentFolderPath } from '../../core/attachment/attachment.utils';
 import { AttachmentType } from '../../core/attachment/attachment.constants';
 import { getProsemirrorContent } from '../../common/helpers/prosemirror/utils';
-import { not } from 'rxjs/internal/util/not';
-import { notionFormatter } from './import-formatter';
+import { formatImportHtml, notionFormatter } from './import-formatter';
 
 @Injectable()
 export class FileTaskService {
@@ -68,9 +67,8 @@ export class FileTaskService {
     await pipeline(fileStream, createWriteStream(tmpZipPath));
 
     await extractZip(tmpZipPath, tmpExtractDir);
-    console.log('extract here');
 
-    // TODO: internal link mentions, backlinks, attachments
+    // TODO: backlinks
     try {
       await this.updateTaskStatus(fileTaskId, FileTaskStatus.Processing);
       // if type == generic
@@ -127,8 +125,6 @@ export class FileTaskService {
           content = await markdownToHtml(content);
         }
 
-        //content = this.stripAllStyles(content)
-
         content = await this.rewriteLocalFilesInHtml({
           html: content,
           pageRelativePath: relPath,
@@ -209,7 +205,7 @@ export class FileTaskService {
         );
 
         const pmState = getProsemirrorContent(
-          await this.importService.processHTML(notionFormatter(htmlContent)),
+          await this.importService.processHTML(formatImportHtml(htmlContent)),
         );
 
         const { title, prosemirrorJson } =
diff --git a/apps/server/src/integrations/import/import-formatter.ts b/apps/server/src/integrations/import/import-formatter.ts
index ef37a7e4..a9413d45 100644
--- a/apps/server/src/integrations/import/import-formatter.ts
+++ b/apps/server/src/integrations/import/import-formatter.ts
@@ -1,10 +1,53 @@
 import { Window } from 'happy-dom';
+import { cleanUrlString } from './file.utils';
+import { getEmbedUrlAndProvider } from '@docmost/editor-ext';
+
+export function formatImportHtml(html: string) {
+  const pmHtml = notionFormatter(html);
+  return defaultHtmlFormatter(pmHtml);
+}
+
+export function defaultHtmlFormatter(html: string): string {
+  const window = new Window();
+  const doc = window.document;
+  doc.body.innerHTML = html;
+
+  // embed providers
+  const anchors = Array.from(doc.getElementsByTagName('a'));
+  for (const a of anchors) {
+    const href = cleanUrlString(a.getAttribute('href')) ?? '';
+    if (!href) continue;
+
+    const embedProvider = getEmbedUrlAndProvider(href);
+
+    if (embedProvider) {
+      const embed = doc.createElement('div');
+      embed.setAttribute('data-type', 'embed');
+      embed.setAttribute('data-src', href);
+      embed.setAttribute('data-provider', embedProvider.provider);
+      embed.setAttribute('data-align', 'center');
+      embed.setAttribute('data-width', '640');
+      embed.setAttribute('data-height', '480');
+
+      a.replaceWith(embed);
+    }
+  }
+
+  return doc.body.innerHTML;
+}
 
 export function notionFormatter(html: string): string {
   const window = new Window();
   const doc = window.document;
   doc.body.innerHTML = html;
 
+  // remove empty description paragraph
+  doc.querySelectorAll('p.page-description').forEach((p) => {
+    if (p.textContent?.trim() === '') {
+      p.remove();
+    }
+  });
+
   // Block math
   for (const fig of Array.from(doc.querySelectorAll('figure.equation'))) {
     // get TeX source from the MathML <annotation>