From 6fe20639702e3eb81bd262075094fb5d1f7033b9 Mon Sep 17 00:00:00 2001
From: Mohamed Bassem <me@mbassem.com>
Date: Sat, 3 Jan 2026 10:39:45 +0000
Subject: fix: Eliminate the O(n2) parsing of the netscape import parsing
 (#2338)

* fix: Eliminate the O(n2) parsing of the netscape import parsing

* remove unneeded tests
---
 packages/shared/import-export/parsers.ts | 84 +++++++++++++++++++-------------
 1 file changed, 50 insertions(+), 34 deletions(-)

(limited to 'packages/shared/import-export/parsers.ts')

diff --git a/packages/shared/import-export/parsers.ts b/packages/shared/import-export/parsers.ts
index a56cbb98..df3d2c45 100644
--- a/packages/shared/import-export/parsers.ts
+++ b/packages/shared/import-export/parsers.ts
@@ -1,5 +1,6 @@
 // Copied from https://gist.github.com/devster31/4e8c6548fd16ffb75c02e6f24e27f9b9
 
+import type { AnyNode } from "domhandler";
 import * as cheerio from "cheerio";
 import { parse } from "csv-parse/sync";
 import { z } from "zod";
@@ -35,43 +36,58 @@ function parseNetscapeBookmarkFile(textContent: string): ParsedBookmark[] {
   }
 
   const $ = cheerio.load(textContent);
-
-  return $("a")
-    .map(function (_index, a) {
-      const $a = $(a);
-      const addDate = $a.attr("add_date");
-      let tags: string[] = [];
-
-      const tagsStr = $a.attr("tags");
-      try {
-        tags = tagsStr && tagsStr.length > 0 ? tagsStr.split(",") : [];
-      } catch {
-        /* empty */
-      }
-      const url = $a.attr("href");
-
-      // Build folder path by traversing up the hierarchy
-      const path: string[] = [];
-      let current = $a.parent();
-      while (current && current.length > 0) {
-        const h3 = current.find("> h3").first();
-        if (h3.length > 0) {
-          const folderName = h3.text().trim();
-          // Use "Unnamed" for empty folder names
-          path.unshift(folderName || "Unnamed");
+  const bookmarks: ParsedBookmark[] = [];
+
+  // Recursively traverse the bookmark hierarchy top-down
+  function traverseFolder(
+    element: cheerio.Cheerio<AnyNode>,
+    currentPath: string[],
+  ) {
+    element.children().each((_index, child) => {
+      const $child = $(child);
+
+      // Check if this is a folder (DT with H3)
+      const h3 = $child.children("h3").first();
+      if (h3.length > 0) {
+        const folderName = h3.text().trim() || "Unnamed";
+        const newPath = [...currentPath, folderName];
+
+        // Find the DL that follows this folder and recurse into it
+        const dl = $child.children("dl").first();
+        if (dl.length > 0) {
+          traverseFolder(dl, newPath);
+        }
+      } else {
+        // Check if this is a bookmark (DT with A)
+        const anchor = $child.children("a").first();
+        if (anchor.length > 0) {
+          const addDate = anchor.attr("add_date");
+          const tagsStr = anchor.attr("tags");
+          const tags = tagsStr && tagsStr.length > 0 ? tagsStr.split(",") : [];
+          const url = anchor.attr("href");
+
+          bookmarks.push({
+            title: anchor.text(),
+            content: url
+              ? { type: BookmarkTypes.LINK as const, url }
+              : undefined,
+            tags,
+            addDate:
+              typeof addDate === "undefined" ? undefined : parseInt(addDate),
+            paths: [currentPath],
+          });
         }
-        current = current.parent();
       }
+    });
+  }
 
-      return {
-        title: $a.text(),
-        content: url ? { type: BookmarkTypes.LINK as const, url } : undefined,
-        tags,
-        addDate: typeof addDate === "undefined" ? undefined : parseInt(addDate),
-        paths: [path],
-      };
-    })
-    .get();
+  // Start traversal from the root DL element
+  const rootDl = $("dl").first();
+  if (rootDl.length > 0) {
+    traverseFolder(rootDl, []);
+  }
+
+  return bookmarks;
 }
 
 function parsePocketBookmarkFile(textContent: string): ParsedBookmark[] {
-- 
cgit v1.3-1-g0d28