From 8d8d6d91aa8cdec49e7e6ef87cbc5622d0bc16ed Mon Sep 17 00:00:00 2001
From: MaheshtheDev <38828053+MaheshtheDev@users.noreply.github.com>
Date: Mon, 3 Nov 2025 00:41:58 +0000
Subject: feat(browser-extension): webpages capture with markdown conversion
 (#548)

Improved browser extension memory saving with better content handling and added markdown conversion.

### What changed?

- Enhanced memory content handling in the background script to prioritize different content types (explicit content, highlighted text, markdown, HTML, or URL)
- Added HTML to markdown conversion using TurndownService when saving entire pages
- Improved HTML handling by removing script tags before processing
- Updated the web app to display the saved URL from metadata when available
- Added turndown library and its type definitions as dependencies
---
 apps/web/components/masonry-memory-list.tsx         | 21 +++++++++++++++++----
 apps/web/components/memories-utils/index.tsx        |  3 +++
 .../web/components/memories-utils/memory-detail.tsx |  2 +-
 3 files changed, 21 insertions(+), 5 deletions(-)

(limited to 'apps/web')

diff --git a/apps/web/components/masonry-memory-list.tsx b/apps/web/components/masonry-memory-list.tsx
index 16cd65c9..1a29093f 100644
--- a/apps/web/components/masonry-memory-list.tsx
+++ b/apps/web/components/masonry-memory-list.tsx
@@ -83,12 +83,25 @@ const DocumentCard = memo(
 			)
 		}
 
-		if (document.url?.includes("https://")) {
+		// Check if this is a website document saved from the Chrome extension
+		const websiteUrl =
+			(document.metadata?.website_url as string | undefined) ||
+			(document.url?.includes("https://") ? document.url : undefined)
+
+		if (websiteUrl) {
 			return (
 				<WebsiteCard
-					url={document.url}
-					title={document.title || "Untitled Document"}
-					image={document.ogImage}
+					url={websiteUrl}
+					title={
+						(document.metadata?.website_title as string | undefined) ||
+						document.title ||
+						"Untitled Document"
+					}
+					image={
+						(document.metadata?.website_og_image as string | undefined) ||
+						document.ogImage
+					}
+					description={document.content && typeof document.content === "string" ? document.content : undefined}
 					onOpenDetails={() => onOpenDetails(document)}
 					onDelete={() => onDelete(document)}
 				/>
diff --git a/apps/web/components/memories-utils/index.tsx b/apps/web/components/memories-utils/index.tsx
index 3052f3c7..fca4d367 100644
--- a/apps/web/components/memories-utils/index.tsx
+++ b/apps/web/components/memories-utils/index.tsx
@@ -48,6 +48,9 @@ export const getSourceUrl = (document: DocumentWithMemories) => {
 	if (document.type === "google_slide" && document.customId) {
 		return `https://docs.google.com/presentation/d/${document.customId}`
 	}
+	if(document.metadata?.website_url) {
+		return document.metadata?.website_url as string
+	}
 	// Fallback to existing URL for all other document types
 	return document.url
 }
diff --git a/apps/web/components/memories-utils/memory-detail.tsx b/apps/web/components/memories-utils/memory-detail.tsx
index 378d3f79..dbbc51cb 100644
--- a/apps/web/components/memories-utils/memory-detail.tsx
+++ b/apps/web/components/memories-utils/memory-detail.tsx
@@ -161,7 +161,7 @@ export const MemoryDetail = memo(
 							<span>{formatDate(document.createdAt)}</span>
 						</div>
 					</div>
-					{document.url && (
+					{(document.url || document.metadata?.website_url) && (
 						<div className="flex items-end">
 							<Button
 								onClick={() => {
-- 
cgit v1.2.3