-- Copyright 2023 Sam Nystrom <sam@samnystrom.dev>
-- Hare LPeg lexer

local l = require('lexer')
local C, Cmt, P, R, S = lpeg.C, lpeg.Cmt, lpeg.P, lpeg.R, lpeg.S

local lex = l.new('hare')

-- Whitespace
lex:add_rule('whitespace', l.token(l.WHITESPACE, l.space^1))

-- Comments
lex:add_rule('comment', l.token(l.COMMENT, l.to_eol('//', true)))

-- Keywords
lex:add_rule('keyword', l.token(l.KEYWORD, l.word_match{
	'as', 'is',
	'if', 'else', 'match', 'switch',
	'break', 'continue', 'defer', 'return', 'yield',
	'const', 'def', 'let',
	'fn',
	'case',
	'for',
	'export', 'static',
	'enum', 'struct', 'union',
	'type',
}))

-- Builtins
lex:add_rule('function', l.token(l.FUNCTION, l.word_match{
	'abort', 'assert',
	'align', 'len', 'offset',
	'alloc', 'free',
	'append', 'insert', 'delete',
	-- C ABI
	'vastart', 'vaarg', 'vaend',
}))

-- Types
lex:add_rule('type', l.token(l.TYPE, l.word_match{
	'bool',
	'f32', 'f64',
	'i8', 'i16', 'i32', 'i64', 'u8', 'u16', 'u32', 'u64',
	'size', 'int', 'uint', 'uintptr',
	'never',
	'nullable',
	'opaque',
	'rune', 'str',
	'void',
	-- C ABI
	'valist',
}))

-- Constants
lex:add_rule('constant', l.token(l.CONSTANT, l.word_match{'true', 'false', 'null'}))

-- Numbers
--[[local identifier = P('r#')^-1 * l.word
local function opt_cap(patt) return C(patt^-1) end
local float = l.digit *
	(Cmt(opt_cap('.' * l.digit) * opt_cap(S('eE') * S('+-')^-1 * l.digit) *
		opt_cap(P('f32') + 'f64'), function(input, index, decimals, exponent, type)
		return decimals ~= "" or exponent ~= "" or type ~= ""
	end) + '.' * -(S('._') + identifier))
local bin = P('0b') * S('01')
local oct = P('0o') * lpeg.R('07')
local hex = P('0x') * l.xdigit
local integer = (bin + oct + hex + decimal_literal) *
	(S('iu') * (P('8') + '16' + '32' + '64' + '128' + 'size'))^-1
--]]
lex:add_rule('number', l.token(l.NUMBER, l.float + l.integer))

-- Strings
lex:add_rule('string', l.token(l.STRING, l.range('"') + l.range('\'')))

-- Operators
lex:add_rule('operator', l.token(l.OPERATOR, S('+-/*<>!=@~&|^?:;,.()[]{}')))

lex:add_fold_point(l.OPERATOR, '{', '}')
lex:add_fold_point(l.COMMENT, l.fold_consecutive_lines('//'))

return lex