so an update on the kanken extraction thing, i've given up figuring out the binary file structure of the extracted files and instead run the game in an emulator and extract as the text and answers are streamed into memory, still required some tricks but now it works flawlessly

the only disadvantage is that it takes a really long time, my script skips in the GUI until we're reasonably sure all questions were seen and even with emulator speedup this takes a fair bit of time to be statistically sure we've seen all questions, it might be faster if i was able to figure out a way to make an "always correct" cheat but im not sure how to find the function that checks for equality, the answer is only written after clicking the reveal button and what im seeing there is probably gui streamed text idk where the comparison happens and im a reverse engineering noob, just too hard (i'll probably learn more about this stuff later) i've experimented with pointer maps and aob injection but neither really helped me

another issue im facing and something im kind of disappointed by is that the amount of questions is actually pretty low... like even with all kakitori questions extracted we don't have full kanji coverage on any level, in fact this pisses me off the most i mean what the fuck

i suppose i can manually supplement sentences to achieve full coverage but still, what's their problem? they say the game contains like 20 kanken books and they can't even include 人 as a writing question in level 10 and all levels only have like 60-80% level coverage for the kanji in the level