[hitomi] fix image path

[ombe1229]

No description provided.

[Hrxn]

Again?

[God-damnit-all]

It looks like they might have adopted something that's effectively a rotating cipher (it's not literally that in this case, but it's an apt enough comparison).

I've been seeing more and more manga sites adopting systems like this to prevent scraping. If that's what's happening here, a system to keep rotating it until it no longer 404s and then caching that offset might be needed.

Any 404 error could check the next offset before throwing the error, just in case it changes in the middle of extraction.

[Hrxn]

Or just extract the correct values from https://ltn.hitomi.la/common.js programmatically?

[God-damnit-all]

Or just extract the correct values from https://ltn.hitomi.la/common.js programmatically?

I didn't realize it was exposed there so that probably a better option. Either way though, it would need to be cached (if only for the duration of the extraction) and potential 404 errors would need to double check common.js.

[mikf]

It looks like they might have adopted something that's effectively a rotating cipher

I don't think so, especially when the method to get the correct subdomain now is evem simpler than it was a few months ago.

Or just extract the correct values from https://ltn.hitomi.la/common.js programmatically?

Too unreliable since it would have to parse arbitrary Javascript code. If it's just simple number change like in this PR, then sure, but they usually change a bit more than that.