format notes.txt

romfs/
- allpac{,hd}ml <- MRG archive containing chapter title images (and some MZP archives with unknown content)
- allscr.mrg <- MZP archive containing game script
	- 0 <- names of the script files (32-byte chunks, zero padded
	- 1 <- MZP archive containing label offsets for scripts (index is script index)
	- 2 <- MZP archive with unknown (empty?) content
	- 3... <- MXZ compressed script text. 3 corresponds with script index 0 in names and label offsets
		- commands:
			- _RTM_() returns to previous
			- _IF__(var, op, val, label) conditionally jumps to a label
			- _ZYxxxxx(label) marks a label
			- _SCH2(?,img,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?) displays an images
			- _END_() marks the end of the script
			- _JUMP(number) jumps to a different script
				- for script ID 1, this seems to be encoded as 201?
			- _WKST(var,op,val) performs an operation on a variable
				- op = "eq" stores val to the var
			- _ZZxxxxx() marks the start of the script?
			- _ZZxxxxx(name) marks the start label of the script?
			- _ZMxxxxx(textspec) displays text (textspec has odd grammar, but $xxxxxx refers to the index of the text in script_text)
- allui <- MRG archive containing UI graphics
- jallpac{,hd} <- MRG archive containing graphics for the game
	- NXZ files are in fact BNTX images
- parts <- MRG archive containing startup messages (TM logo, JP warning text. This might not actually be used in the Switch release.)
- script_text.mrg <- MZP archive containing script text
	- the sections come in pairs here
		- the first section in each pair is a list of 4-byte begin and end offsets for each line of text
		- the second section contains the actual text
	- the languages are in the following order:
		- JP
		- EN
		- ZH
		- ZH2
		- KO
			- the null language just has U+3000 as each line of text
- witch.bfsar <- BFSAR archive containing audio (https://mk8.tockdom.com/wiki/BFSAR_(File_Format))

----