Matching has been worse/buggy with latest version

[candre23]

For the last couple months, since upgrading to 5.1.7, movie matching has been somewhat janky. I'm not sure what version I was running before the recent update. The previous matching algorithm was certainly not perfect either, but it tended to err on the side of caution. The new algorithm makes some *wild* matches, without asking for verification. For example, it just matched "ROSA.Republic.Of.South.Ah.Sh.t.2023.720p.WEBRip.x264.AAC-[YTS.MX].mp4" as "The Rose (1979)". It doesn't get every match this wrong. In fact, it's still probably 95% accurate in matching. But previously, whenever there was doubt, it would ask for verification. Now, it is just confidently wrong whenever it's wrong - even when it should be extremely obvious that it's a bad match.

Furthermore, the verification process seems to now be broken. When it does correctly recognize that it's not sure about a match and brings up the window to manually match a movie, it will only work once. Previously, if there were multiple ambiguous matches that needed manual verification, it would present them one after another until they had all been dealt with. Now, the manual matching window closes after the first match has been made. Even if there were half a dozen files that still needed manual intervention, it simply quits trying after the first one and you have to go back through the file list and fix the rest one at a time by hand.

Requisite sysinfo:

Code: Select all

FileBot 5.1.7 (r10514)
JNA Native: 7.0.0
MediaInfo: 24.12
7-Zip-JBinding: 16.02
Tools: fpcalc/1.5.0
Extended Attributes: OK
Unicode Filesystem: OK
Script Bundle: 2025-03-06 (r1003)
Groovy: 4.0.21
JRE: OpenJDK Runtime Environment 21.0.3
JVM: OpenJDK 64-Bit Server VM
CPU/MEM: 16 Core / 8 GB Max Memory / 58 MB Used Memory
OS: Windows 10 (amd64)
STORAGE: NTFS [(C:)] @ 744 GB | NTFS [2TB SSD] @ 1.1 TB | NTFS [1TB SSD] @ 835 GB | NTFS [DrivePool] @ 37 TB
DATA: C:\Users\chip\AppData\Roaming\FileBot
Package: MSI
Activate License [P61853015] on [Wed Apr 30 08:48:45 EDT 2025]
License: FileBot License P61853015 (Valid-Until: 2025-05-28)

[rednoah]

1.

For example, it just matched "ROSA.Republic.Of.South.Ah.Sh.t.2023.720p.WEBRip.x264.AAC-[YTS.MX].mp4" as "The Rose (1979)"

API search for "ROSA Republic Of South Ah Sh t" doesn't yield any results, FileBot therefore does not know that ROSA Republic of south ah sh**t exists, and so the "best result" is going to be completely wrong:

Console Output: Select all

$ filebot -list --q "ROSA Republic Of South Ah Sh t" --db TheMovieDB
No search results

ROSA Republic of south ah sh**t (????) notably does not seem to be very popular so nobody has bothered to enter a Release Year yet, so FileBot cannot come up with a movie match, which requires Name (Year) information at the very least:

Console Output: Select all

$ filebot -list --q "ROSA Republic Of South Ah Sh**t" --db TheMovieDB
Missing data: release_date: {"adult":false,"backdrop_path":"/kVErgZBXkllfFieSTuFxt7owXRZ.jpg","genre_ids":[],"id":1400528,"original_language":"zu","original_title":"ROSA Republic of south ah sh**t","overview":"","popularity":0.383,"poster_path":"/rxhH0qCnzLyoabsrfieyq7iHh34.jpg","release_date":"","title":"ROSA Republic of south ah sh**t","video":true,"vote_average":0.0,"vote_count":0}
No search results

You can use Match Mode: Strict to disallow obvious mismatches like The Rose (1979). The trade-off may be worth it depending on the files at hand. Match Mode: Opportunistic will always give you a match, but unfortunately, in this case, the best possible match is no match at all.


Please open ROSA Republic of south ah sh**t on TheMovieDB and then click EDIT PAGE and then contribute the missing information. Please make sure to fill in all the basic information: (1) release date, (2) movie name for en-US locale, (3) "ROSA Republic Of South Ah Sh t" as Alternative Title to make search & lookup work better for everyone in the future.





2.

When it does correctly recognize that it's not sure about a match and brings up the window to manually match a movie, it will only work once. Previously, if there were multiple ambiguous matches that needed manual verification, it would present them one after another until they had all been dealt with.

Sounds like you may have selected Automatic Selection which reduces prompts by always auto-selecting the first most likely option. Please toggle back to Manual Confirmation if you want to be prompted more often:




Note that you can always use Edit Match to fix a movie match after the fact. It is often easier to just let FileBot run and guess everything without interruption first, while you do something else, so you can then go through the list item by item afterwards and confirm or Edit Match as necessary.

[candre23]

The manual confirmation window had somehow changed to auto mode. Honestly I had never even noticed that icon in the corner.

The matching mode was set to opportunistic, and presumably always has been since I've never changed it. I gave strict a try, and I do like that it is picker about matches. However, strict matching mode doesn't bring up the manual match window to sort out the files it can't match automatically. Ideally, it would still give me the option to manually match the files which it couldn't.

[rednoah]

I gave strict a try, and I do like that it is picker about matches. However, strict matching mode doesn't bring up the manual match window to sort out the files it can't match automatically. Ideally, it would still give me the option to manually match the files which it couldn't.

I recommend using Match Mode: Opportunistic with Automatic Selection enabled (to disable prompts just like in Strict Mode) and then use Edit Match if you need to bring up the manual match window for the file / movie match at hand. That'd effectively be the workflow described above, except that you yourself decide when to bring up the manual match window rather than FileBot - correctly or incorrectly - itself.