mirror of
https://github.com/calibrain/shelfmark.git
synced 2026-09-24 19:40:26 +01:00
CamGoed
e93fbd2a9b
## Problem
Shelfmark can lose track of hybrid v1/v2 torrents after qBittorrent
completes their metadata download.
Shelfmark initially identifies the torrent by its v1 infohash. Once
metadata resolves, qBittorrent may switch the torrent’s primary `hash`
to the truncated v2 hash, causing lookups using the original v1 hash to
return nothing.
For example:
- v1: `edf46c7f938a3c678081734d7bff8b9c652ba5e5`
- qBittorrent `hash`: `0bed5f40753b342cb143e83c2b21924cc8474731`
- full v2:
`0bed5f40753b342cb143e83c2b21924cc847473134e44d1bd300bdc58c13010f`
At that point, querying `/api/v2/torrents/info` with the original v1
hash returns no records. Querying with the new primary hash works, and
the returned record still contains the original hash in `infohash_v1`.
`find_existing()` also ignored its provided category, so audiobook
torrents fall back to the default ebook category instead. This means the
fallback method for a mismatched download ID never occurs for
audiobooks, leading to a "failed" download that is actually successful
in qBittorrent. As a result, the downloaded files are not automatically
transferred/hardlinked to the output directory.
## Fix
- Match torrents against `hash`, `infohash_v1`, and `infohash_v2`.
- Wait for magnet metadata to finish downloading before returning the
torrent ID.
- Return qBittorrent’s current primary `hash`.
- Search the provided category first, then the configured default, and
finally the full torrent list.
Logs showing the issue:
```
2026-08-03 22:42:30,262 - shelfmark.release_sources.audiobookbay.scraper - DEBUG - scraper.py:450 - Generated Magnet Link: magnet:?xt=urn:btih:EDF46C7F938A3C678081734D7BFF8B9C652BA5E5&tr=...
2026-08-03 22:42:30,376 - shelfmark.download.clients.qbittorrent - DEBUG - qbittorrent.py:521 - qBittorrent add result: TorrentsAddedMetadata({'added_torrent_ids': ['edf46c7f938a3c678081734d7bff8b9c652ba5e5'], 'failure_count': 0, 'pending_count': 0, 'success_count': 1})
2026-08-03 22:42:30,427 - shelfmark.download.clients.qbittorrent - INFO - qbittorrent.py:545 - Added torrent: edf46c7f938a3c678081734d7bff8b9c652ba5e5
2026-08-03 22:42:30,427 - shelfmark.download.clients.base_handler - INFO - base_handler.py:868 - Added to qbittorrent: edf46c7f938a3c678081734d7bff8b9c652ba5e5 for 'Pathogenesis: A History of the World in Eight Plagues'
2026-08-03 22:42:30,427 - shelfmark.download.clients.base_handler - DEBUG - base_handler.py:906 - Starting poll for edf46c7f938a3c678081734d7bff8b9c652ba5e5 (content_type=audiobook)
2026-08-03 22:42:32,590 - shelfmark.download.clients.base_handler - DEBUG - base_handler.py:958 - Download edf46c7f938a3c678081734d7bff8b9c652ba5e5 not yet visible in client (attempt 1/15)
2026-08-03 22:43:02,345 - shelfmark.download.clients.base_handler - ERROR - base_handler.py:969 - Download edf46c7f938a3c678081734d7bff8b9c652ba5e5 not found after 15 attempts
2026-08-03 22:43:02,345 - shelfmark.download.clients.base_handler - INFO - base_handler.py:426 - Skipping download client cleanup for protocol=torrent after download error (client=qbittorrent id=edf46c7f938a3c678081734d7bff8b9c652ba5e5)
```
<br>
The successful torrent:
<br>
<img width="968" height="159" alt="image"
src="https://github.com/user-attachments/assets/d63c4444-6a4d-4214-97a0-732acc338970"
/>
<br>
v1 vs v2 hash:
<br>
<img width="749" height="212" alt="image"
src="https://github.com/user-attachments/assets/d985b78a-84fe-4a26-9697-126c787e7303"
/>
I ran some python queries from the shelfmark container that show the
mismatch:
```
qBittorrent URL: http://gluetun-mam:8081
Tracked hash: edf46c7f938a3c678081734d7bff8b9c652ba5e5
=== PROPERTIES LOOKUP USING SHELFMARK HASH ===
HTTP status: 404
Not Found
=== EXACT /torrents/info HASH LOOKUP ===
Returned torrents: 0
=== FIND VISIBLE PATHOGENESIS TORRENT ===
Matching visible torrents: 1
Name: Pathogenesis: A History of the World in Eight Plagues
Primary hash: 0bed5f40753b342cb143e83c2b21924cc8474731
Category: audiobooks
State: stalledUP
Progress: 1
Properties HTTP status: 200
Infohash v1: edf46c7f938a3c678081734d7bff8b9c652ba5e5
Infohash v2: 0bed5f40753b342cb143e83c2b21924cc847473134e44d1bd300bdc58c13010f
```