From 68309247b5362871b7b0ae7b9730810f8ac7ce69 Mon Sep 17 00:00:00 2001
From: Philipinho <16838612+Philipinho@users.noreply.github.com>
Date: Wed, 22 Apr 2026 01:36:14 +0100
Subject: [PATCH] share html normalization between zip and single-file imports

---
 .../import/services/import.service.ts         |  6 +++++-
 .../import/utils/import-formatter.ts          | 21 ++++++++++++++++---
 2 files changed, 23 insertions(+), 4 deletions(-)

diff --git a/apps/server/src/integrations/import/services/import.service.ts b/apps/server/src/integrations/import/services/import.service.ts
index 231a6c89a..2b02c16b6 100644
--- a/apps/server/src/integrations/import/services/import.service.ts
+++ b/apps/server/src/integrations/import/services/import.service.ts
@@ -30,6 +30,8 @@ import { InjectQueue } from '@nestjs/bullmq';
 import { Queue } from 'bullmq';
 import { QueueJob, QueueName } from '../../queue/constants';
 import { ModuleRef } from '@nestjs/core';
+import { load } from 'cheerio';
+import { normalizeImportHtml } from '../utils/import-formatter';
 
 @Injectable()
 export class ImportService {
@@ -137,7 +139,9 @@ export class ImportService {
 
   async processHTML(htmlInput: string): Promise<any> {
     try {
-      return htmlToJson(htmlInput);
+      const $ = load(htmlInput);
+      normalizeImportHtml($, $.root());
+      return htmlToJson($.html() || '');
     } catch (err) {
       throw err;
     }
diff --git a/apps/server/src/integrations/import/utils/import-formatter.ts b/apps/server/src/integrations/import/utils/import-formatter.ts
index 12e617c3c..0333b75e7 100644
--- a/apps/server/src/integrations/import/utils/import-formatter.ts
+++ b/apps/server/src/integrations/import/utils/import-formatter.ts
@@ -52,9 +52,7 @@ export async function formatImportHtml(opts: {
     }
   }
 
-  notionFormatter($, $root);
-  xwikiFormatter($, $root);
-  defaultHtmlFormatter($, $root);
+  normalizeImportHtml($, $root);
 
   const backlinks = await rewriteInternalLinksToMentionHtml(
     $,
@@ -74,6 +72,23 @@ export async function formatImportHtml(opts: {
   };
 }
 
+/**
+ * Contextless HTML cleanup shared by every import path.
+ * - notionFormatter: no-op on non-Notion HTML (class-selector-based).
+ * - xwikiFormatter: no-op on non-XWiki HTML (looks for #xwikicontent).
+ * - defaultHtmlFormatter: table column widths + provider auto-embeds.
+ *
+ * Does NOT run rewriteInternalLinksToMentionHtml — that requires zip context.
+ */
+export function normalizeImportHtml(
+  $: CheerioAPI,
+  $root: Cheerio<any>,
+): void {
+  notionFormatter($, $root);
+  xwikiFormatter($, $root);
+  defaultHtmlFormatter($, $root);
+}
+
 export function xwikiFormatter($: CheerioAPI, $root: Cheerio<any>) {
   const $content = $root.find('#xwikicontent');
   if ($content.length) {