Find near-duplicate bookmarks in linkding, and tag them for bulk delete

linkding keeps one bookmark per address (adding an address you already have updates that bookmark), but it treats http://www.example.com/page, https://example.com/page and https://example.com/page?utm_source=feed as three different addresses. A library imported from several browsers keeps all three. This page finds them, then writes a file for linkding's own import that adds a tag to the extra copies, so you can delete them with linkding's bulk edit. This page deletes nothing itself.

  1. In linkding, open Settings, then General, and under "Export" click "Download (.html)".
  2. Choose that file below and check the groups the page finds. It is read in your browser and nothing in this page's code sends it anywhere; the page's security policy also blocks its scripts from fetch and similar requests.
  3. Download the import file.
  4. In linkding, open Settings, then General, and under "Import" choose the downloaded file and click "Upload". Leave "Import public bookmarks as shared" unticked.
  5. Search for #dupe-to-delete (or your own tag) and check the results. Click the "Bulk edit" button (an icon in the bar above the list; hover it to see its name; not shown on small screens), tick the box at the start of the bar, then the "All ... bookmarks" box if it appears, choose "Delete" and click "Execute". Do the same on the Archived page, where archived copies are listed.
Treat as the same address


Before you import. Keep the export you chose above: it is your backup. Importing the file changes your library and cannot be undone by importing again, because linkding only adds tags and never removes them. The extra copies get the tag. The kept bookmark in each group gets the other copies' tags, the group's earliest date added, and their notes after its own. The page refuses a tag that is already on one of your bookmarks, so that deleting everything with the tag only removes extra copies; still look at the tagged bookmarks before you delete. No warranty.

What the import file does, as read in linkding's source code: each line finds the bookmark you already have with that exact address. On the extra copies nothing changes but the tag: their dates and read state are written back as they were, and sharing is not touched as long as "Import public bookmarks as shared" stays unticked. Titles and descriptions are left as they are. Tags from an archived copy are carried over, its archived state is not. The kept bookmark is chosen by date only, so it can be the http copy or the one with tracking parameters.

Possible later tool: find duplicates hidden behind redirects (3 USD, one time, not built).

It would follow each address (moved pages, short links) and group bookmarks that end up on the same page, which a web page cannot do from your browser. Nothing is charged now and no date is promised. If you want it, say so in a public comment on the issue below; there is no payment of any kind before I answer there.

Tell me you want it (free GitHub account needed)

Reads the HTML export of linkding. The import file has been read in the tests by linkding's own import parser and address normalization, taken from its source code, but it has not been imported into a running linkding, the page has not been tried in a real browser, and no real export has been converted. If it does not work for you, an issue on the repo helps; please do not attach your real export, only the linkding version and the error shown. Not affiliated with the linkding project; its name is used here only to say which files this reads and writes. Built with AI assistance, tested with automated tests, not reviewed line by line. No cookies, no browser storage, no analytics, no third-party code. MIT licensed, source on GitHub. Younes Z.