[iwara] Add support #7785
[iwara] Add support #7785
Conversation
|
Thank you, but there are so many usage and coding patterns that have proven quite useful throughout gallery-dl's development which are present in many places in its code base, that your code mostly ignores. If you don't mind, I'd like to overhaul your code a bit and deal with all the issues myself. Please don't take it the wrong way. Having a working solution and improving upon it is always a lot nicer and easier than having to start from scratch. |
|
Thanks for the thoughtful feedback — I completely understand and agree that consistency with established patterns is important. I would've liked to use said patterns and whatnot, but I'm sure — as you would agree to as well — the documentation regarding developing extractors for new sites is not great. I tried to go by example (mostly from I'll try to learn from your adjustments so future contributions can better align with gallery-dl's structure. |
|
Well... I wanted to work on this this evening but
Guess I'll try my luck tomorrow. edit: Iwara is working again. |
|
It happened to me once in a while as well. I'll also say this, search and tag extractors take significantly more time (regardless of how much data is in each search/tag) to resolve requests and sometimes they fail with no explanation other than timeout. Since both of them mostly use code from other extractors (video/image), I'd advise you to focus on anything other than those two and iron them out at the end. That's just my two cents.
All good. I like working on gallery-dl. In case of Iwara, I just wanted to have a working extractor for my use case. |
- define endpoints inside methods - implement and use _call() and _pagination() - cache auth tokens
TODO: - update test results - simplify code
and move user data extraction into 'yield_video'
and move user info extraction into 'yield_image()'
perform API calls to get full 'files' list inside the function
remove urllib.parse usage
* get("key", {}) -> get("key") or {}
* split("…", 1) -> partition("…")
* use f-strings for all patterns
include 'date' in filenames to order them chronologically
- use less generators - make processing individual media items non-fatal
Fixes #2652 and #5840.
Weird hash hack taken from here and
yt-dlp.