shithub: mc

ref: 7439e43b38569064395880551b1541db79bcfed5
dir: /lib/std/striter.myr/

View raw version
use "die"
use "types"
use "utf"
use "strfind"
use "option"
use "chartype"

pkg std =
	type chariter = struct
		rest	: byte[:]
	;;

	type charoffiter = struct
		str	: byte[:]
		idx	: size
	;;

	type splititer = struct
		rest	: byte[:]
		split	: byte[:]
	;;

	type tokiter = struct
		str	: byte[:]
		idx	: size
	;;

	impl iterable chariter	-> char
	impl iterable charoffiter	-> (char, size)
	impl iterable splititer -> byte[:]
	impl iterable tokiter -> byte[:]

	const bychar	: (str : byte[:] -> chariter)
	const bycharoff	: (str : byte[:] -> charoffiter)
	const bysplit	: (str : byte[:], split : byte[:] -> splititer)
	const bytok	: (str : byte[:] -> tokiter)
;;

impl iterable chariter -> char =
	__iternext__ = {ci, c
		if ci.rest.len == 0
			-> false
		;;
		(c#, ci.rest) = charstep(ci.rest)
		-> true
	}

	__iterfin__ = {ci, c
	}
;;

const bychar = {str
	-> [.rest = str]
}


impl iterable charoffiter -> (char, size) = 
	__iternext__ = {ci, cv
		var c

		if ci.idx == ci.str.len
			-> false
		;;
		c = std.decode(ci.str[ci.idx:])
		ci.idx += std.charlen(c)
		cv# = (c, ci.idx)
		-> true
	}

	__iterfin__ = {ci, c
	}
;;

const bycharoff = {s
	-> [.str=s, .idx=0]
}

impl iterable splititer -> byte[:] =
	__iternext__ = {si, sp
		match std.strfind(si.rest, si.split)
		| `Some off:
			sp# = si.rest[:off]
			si.rest = si.rest[off + si.split.len:]
			-> true
		| `None:
			if si.rest.len > 0
				sp# = si.rest
				si.rest = ""
				-> true
			;;
		;;
		-> false
	}

	__iterfin__ = {ci, c
	}
;;

const bysplit = {str, split
	-> [.rest = str, .split = split]
}

impl iterable tokiter -> byte[:] =
	__iternext__ = {it, sp
		var s, lo, hi, c
		
		s = it.str
		lo = it.idx
		while lo < s.len
			c = std.decode(s[lo:])
			if !isspace(c)
				break
			;;
			lo += charlen(c)
		;;
		hi = lo
		while hi < s.len
			c = std.decode(s[hi:])
			if isspace(c)
				break
			;;
			hi += charlen(c)
		;;
		it.idx = hi
		sp# = s[lo:hi]
		-> hi > lo
	}

	__iterfin__ = {ci, c
	}
;;

const bytok = {str
	-> [.str = str, .idx = 0]
}