Auto import incorrectly calculates custom format score

[thovin]

Is there an existing issue for this?

• I have searched the existing open and closed issues

Current Behavior

The activity screen correctly calculates custom format score, but the element that determines whether to import sees a different number and so chooses not to import. I suspect it is missing certain custom formats entirely, rather than being a math problem.

Here we can see that the activity screen shows a custom format score of +2051, but when you hover the warning icon it shows a score of +51 on the new file. This means my custom format with a +2000 is being missed. Additionally, we see that one custom format is factored into that +51 score but the third photo shows 3 custom formats matching.

Here we can see that the manual import tool also reflects a score of +51. It correctly shows multi-language, but fails to identify that my dual audio format matches.

Expected Behavior

Custom format matching should be consistent between activity screen and import tool

Steps To Reproduce

No response

Environment

- OS: Windows 10 V.22H2
- Sonarr: 4.0.15.2940
- Docker Install: Running as service
- Using Reverse Proxy: N/A
- Browser: Chrome
- Database: N/A

What branch are you running?

Develop

Trace Logs?

NOTE: I got logs from a different episode than those in the screenshots. This episode is actually worse - it misses the HEVC and dual audio formats and only sees the v2. As this episode was from a batch download, I suspect the import tool is only using the filename to calculate custom formats. This means elements like HEVC in the batch name or the fact of multiple languages being present are missed entirely.

2025-07-26 11:45:31.9|Debug|Parser|Parsing string '[Judas] Kimi to Boku no Saigo no Senjou, Aruiwa Sekai ga Hajimaru Seisen (Our Last Crusade or the Rise of a New World) (Season 02) [1080p][HEVC x265 10bit][Dual-Audio][Multi-Subs] (Batch)'
2025-07-26 11:45:31.9|Trace|Parser|^[(?.+?)][-. ]?(?<title>.+?)[-. ]+(?(?:[-. ]?#?(?\d{2,3}(.\d{1,2})?(?!\d+|-[a-z]+)))+)?(?:[-. ]+(?special|ova|ovd))?.*?(?[([]\w{8}[)]])?(?:$|.mkv)
2025-07-26 11:45:31.9|Debug|Parser|Episode Parsed. Kimi to Boku no Saigo no Senjou, Aruiwa Sekai ga Hajimaru Seisen (Our Last Crusade or the Rise of a New World) (Season - 002
2025-07-26 11:45:31.9|Debug|Parser|Languages parsed: Unknown
2025-07-26 11:45:31.9|Debug|QualityParser|Trying to parse quality for '[Judas] Kimi to Boku no Saigo no Senjou, Aruiwa Sekai ga Hajimaru Seisen (Our Last Crusade or the Rise of a New World) (Season 02) [1080p][HEVC x265 10bit][Dual-Audio][Multi-Subs] (Batch)'
2025-07-26 11:45:31.9|Debug|Parser|Quality parsed: HDTV-1080p v1
2025-07-26 11:45:31.9|Debug|Parser|Release Group parsed: Judas

Anything else?

No response

[PearsonFlyer]

CF scores are matched/calculated twice. Once on grab from your indexer, using only the release name. Anything there matches because of the release name only. The second calculation is On Import, which is calculated from the Filename AND ffprobe results. This can mean that if something is in a release name that is not ALSO in the filename, the score changes. Or if it's in the release name, but the ffprobe or filename differs.

This is done, entirely separately, to ensure that what was downloaded was the same thing that was grabbed. It's on purpose, and is causing your issue.