[BUG] Hardlink cross-seeds using categories still creates two different tracker categories

[OscarsWorldTech]

Description

When using hardlink cross-seeds and then tracker name as the category, qui will still create two different tracker names as tracker name and tracker name (api). See screenshot below:

Steps to reproduce

1. Go to cross-seed
2. Go to rules
3. Set mode to hardlink
4. Set categories to "use indexer name as category"

Environment

- qui: 1.26.0
- qbittorrent: v5.2.3
- Ubuntu 24.04 LTS
- Binary

Relevant logs

No response

Checklist

• I have searched existing discussions and issues
• I have checked the documentation

[github-actions[bot]]

This was generated by AI during triage.

Triage Notes

What we have established so far:

• With "Use indexer name as category" enabled, the cross-seed add sets the qBittorrent category to the indexer name verbatim. Nothing trims a trailing (API), (Prowlarr), or (Jackett) suffix, so an indexer whose name carries one produces a category that carries it too. The documented behavior of this option is "category = indexer name".
• qui does strip those suffixes elsewhere, but only when matching an indexer name against a tracker domain. Category naming does not use that normalization.
• The link mode is not the deciding factor here: hardlink, reflink, and regular adds all take the category from the same resolver. In hardlink mode with the "by tracker" directory preset, the save path is built from the tracker display name instead (tracker customization, then indexer name, then domain), so the folder name and the category name can disagree.
• Two cross-seed paths feed that resolver different spellings of the same indexer name. RSS automation resolves the name from the indexer record stored in qui. Seeded torrent search passes the name the indexer feed reported (its Torznab channel title). For Prowlarr-backed indexers qui overwrites the feed name with the stored one, so both paths agree. For Jackett and native Torznab indexers the feed name wins whenever it is non-empty, so those two paths can create two category spellings for one tracker.

What we still need from you (@OscarsWorldTech):

• On the Indexers page, is the tracker from your screenshot configured once or twice? Please give the exact name or names qui shows for it, and the backend of each (Prowlarr, Jackett, or native Torznab).
• Do the two categories differ only by a trailing " (api)", or is there another difference as well (letter case, spacing, punctuation)?
• Which cross-seed source added the torrents in each of the two categories: RSS automation, seeded torrent search (manual or scheduled), webhook, or Dir Scan? The cross-seed run and search history pages show which run added a given torrent.
• Does this still happen on v1.29.0? Your report is from v1.26.0.

[OscarsWorldTech]

1. From the indexers page:

2. It is mostly the trailing "(api)" however, for tracker LST it deviates by going fully lowercase so normal is "LST" and deviation is "lst"

3. This is the log entry after I uploaded a torrent to a tracker via upbrr and skipping client injection: { "level": "info", "time": "2026-09-17T18:17:20-04:00", "message": "[CROSSSEED] Hardlink mode: created hardlink tree", "instanceID": 1, "torrentName": "The.Adventures.Of.Tintin.2011.BluRay.1080p.DTS-HD.MA.7.1.AVC.REMUX-FONDU3.mkv", "destDir": "/data/torrents/cross-seed/Seedpool/The.Adventures.Of.Tintin.2011.BluRay.1080p.DTS-HD.MA.7.1.AVC.REMUX-FONDU3.mkv--83258ed2", "fileCount": 1 }

4. Yes, I can confirm I am on v1.29.0 and it is still happening

[OscarsWorldTech]

For added information, here is what the cross-seed rules are set to:

And here is what the directory looks like in my NFS share:

[github-actions[bot]]

This was generated by AI during triage.

Closed: no reply in 30 days. Reply here with the requested details and a maintainer will reopen it.