Episodes matching out of order using AniDB

[sticke4]

I am having a problem matching an anime. I am using the AniDB database matching the following files:

Code: Select all

[HorribleSubs] Shingeki no Kyojin S3 - 50 [1080p].mkv
[HorribleSubs] Shingeki no Kyojin S3 - 51 [1080p].mkv
[HorribleSubs] Shingeki no Kyojin S3 - 53 [1080p].mkv
[HorribleSubs] Shingeki no Kyojin S3 - 52 [1080p].mkv

Regardless of my matching scheme I get the following match

Code: Select all

Attack on Titan Season 3 (2019) - 3
Attack on Titan Season 3 (2019) - 1
Attack on Titan Season 3 (2019) - 4
Attack on Titan Season 3 (2019) - 2

which is wrong since episode 50 is the first episode of the season. That is the correct ordering should be the following:

Code: Select all

Attack on Titan Season 3 (2019) - 1
Attack on Titan Season 3 (2019) - 2
Attack on Titan Season 3 (2019) - 3
Attack on Titan Season 3 (2019) - 4

The results don't meaningfully change if I switch the preferred episode order scheme between airdate order and absolute order, nor do they change when I follow [Anime] Convert Absolute to SxE numbers. I can't tell if this is an anidb problem or not as searching on the show's anidb page, I can't seem to find where they store the absolute order metadata.

Of note, if I switch to absolute airdate order I get the following

Code: Select all

Attack on Titan Season 3 (2019) - E20190429
Attack on Titan Season 3 (2019) - E20190513
Attack on Titan Season 3 (2019) - E20190520
Attack on Titan Season 3 (2019) - E20190506

which would correspond to the following episodes

Code: Select all

Attack on Titan Season 3 (2019) - 1
Attack on Titan Season 3 (2019) - 3
Attack on Titan Season 3 (2019) - 4
Attack on Titan Season 3 (2019) - 2

which is a different matching scheme to the other two.

If I instead switch to TheTVDB then the episodes match appropriately in order. I think that the cause is either a) I've configured something wrong, b) the AniDB database has the absolute orders coded wrong, or c) there is a bug in filebot causing it to match the episodes out of order. Any guidance would be greatly appreciated.

Brief details about my system:

Code: Select all

OS: Windows 10
Filebot Install: MSI Package
Filebot version: 4.8.5

Expanded Logs (the F7 output and the F5 output): https://pastebin.com/wUrGBvB7

[rednoah]

Since the episode numbering doesn't match AniDB at all, there's no way to process these files with AniDB, reliably in an automated manner at least. Hence the results are effectively random.

If you must use AniDB, then you can do that via manual matching. See Getting Started, Page 2.

AniDB supports Absolute Order only, and doesn't support seasons, hence there's a multiple distinct Anime entries for what would be different seasons for the same TV show entry on TheTVDB. This makes translating between different databases particularly tricky. In this case, your original episode numbering seems to match the TheTVDB Absolute Numbering, and so TheTVDB will work best, since only TheTVDB knows what "Episode #53" might be.



PS: An excellent bug report. Includes all the information one could hope for on the first try. Very much appreciated.

[sticke4]

Thanks for the very quick reply. Unfortunately, I'm using Filebot primarily throught the AMC so was hoping to get something more automated than what manual matching would allow. I've played around with --filter and fine-tuning but it seems that as you said this is a database problem so can't be solved that easily through existing command line or expression tweaks.

Searching the internet, I see that this affects a few users at the least.

• Anime Renamed to Wrong Season by AniDB
• Force amc script to use TheTVDB with anime
• Anime Renamed to Wrong Season by AniDB

In the latter link you state the following:

RFE:

This thread has lead me to consider switching the amc script to use TheTVDB with Absolute Order instead of AniDB in upcoming releases.

I think switching the default database for anime may cause more problems problems than it solves. I ran some tests on my own library to see what the episodes would have been renamed to if I had used theTVDB with absolute ordering and many series were not able to match at all - many of which were matched using AniDB without a problem.

My suggestion would be to specify a feature that allowed a user to specify a list of anime series and their preferred database. Something akin to what was asked in the following two links:

• Force amc script to use TheTVDB with anime
• Forcing anime to use tvdb

I think an implementation that maintains ease of use in the defaults but flexibility for power users could be similar to the series overrides and mappings like in [SNIPPET] Override / Hardcode Series Names and Exclude Blacklist & Series-Mappings. The user could specify in the expressions or through a command line input file a matching scheme that, when evaluated as true, would change the database that is used for the match. This way when AniDB is not working well for a fansub group's particular anime release, in my case HorribleSub's Shingeki no Kyojin, I can set the AMC script to use theTVDB when matching this series in the future.

[sticke4]

A bit more information: In my particular situtation I am using the AMC script to watch a directory rather than using it through a torrent client after each download.

I had considered that one solution for those using it with a torrent client would be to create a shell/python script that handrolled an "Anime exclude list". I imagine that the script could compare the exclude list (provided by a file ideally) against the filename. On a match, the script would pass the

Code: Select all

--def ut_label=TV

option to the AMC script to force the use of theTVDB. That way, if someone notices, to use an example from another post, that One Punch Man S2 is not matching well for them using AniDB, they could put it in their exclude list so that it uses theTVDB instead.

Unfortunately, this solution wouldn't work as well for a directory watching script. This is mainly what motivated my suggestion for a more flexible approach that is more like the series overrides functionality built into filebot itself.

[rednoah]

If you're using the amc script, then you are likely processing files shortly after the airdate, so the age filter approach should work quite well. No? Because while episode numbers don't match, the airdate should roughly match your files last modified date.


EDIT:

If you could post additional logs (with full debug logging enabled) so i can see how things match up, that would be appreciated.