Open atagen opened 1 month ago
Goldenfreddy0703
commented
1 month ago Heyyyy so from what your talking about, we do have prioritization settings and we also have a setting where you can exclude any keywords so that way, you can find the best sources. Tomorrow we can talk about this more and see on how we can improve scrapping an all. Thank you
atagen
commented
1 month ago we do have prioritization settings and we also have a setting where you can exclude any keywords
As far as I've seen these are global though, right? I don't, f.e., necessarily want to exclude everything with a "II" in it, I just want to be able to tell Otaku that "Goblin Slayer" scrapes explicitly should exclude "II". In general, prioritisation is one thing, but what I'm thinking of here is more dynamic - scoped by series, and responsive to the previous scrape's results.
Look forward to hearing your conclusions. If you need any clarification or back n forth on these ideas I'm happy to help here too.
Hello, and thanks for the excellent addon.
I occasionally find that the scraper is mislead by common words - scraping for Monster (the 00s series) yields episodes of Monster Girl Doctor, Goblin Slayer and Goblin Slayer II seem very tough for it to tell between, and so on.
I can think of two possible solutions, one simpler and reactive, and one proactive, more precise and convenient for users:
allow blacklisting of keywords per series, so for example I can blacklist "Girl", "Doctor", "Musume" for Monster
allow users to lock a show to a pack and/or filename schema, for as long as that pack/schema can still provide episodes
To explain the latter in more detail: if I watch S1E1 and Otaku scrapes a pack that contains all of S1, Otaku should source E2 from the same pack and save ourselves the scrape. This could be done transparently, and would also provide a smoother transition between episodes.
Moreover, even if rescrape is required, Otaku could prioritise for files named in a similar fashion, f.e. prefixed with "[BigJimsRaws]", having similar sequence of words/alphanumeric sequences, low levenshtein distance, etc.
Please let me know if these are feasible/desired features - if need be, I can take a hack at a PR.