WHITESPACE = _{ " " | "\t" | "\n" | "\r" | "\x0B" | "\x0C" | "\xA0" | SPACE_SEPARATOR }
COMMENT = _{ ";" ~ (!"\n" ~ ANY)* }

assembly = {
    SOI ~ (origin | include_bin | data | constant | label | instruction)* ~ EOI
}

origin = {
    origin_padding    |
    origin_no_padding
}
origin_padding = { "org.pad" ~ operand_value }
origin_no_padding = { "org" ~ operand_value }

include_bin = { "#include_bin" ~ immediate_str }

data = {
    data_byte |
    data_half |
    data_word |
    data_str  |
    data_strz |
    data_fill
}
data_byte = { "data.8" ~ operand_value }
data_half = { "data.16" ~ operand_value }
data_word = { "data.32" ~ operand_value }
data_str  = { "data.str" ~ immediate_str }
data_strz = { "data.strz" ~ immediate_str }
data_fill = { "data.fill" ~ operand_value ~ "," ~ operand_value }

constant = { "const" ~ constant_name ~ operand_value }
constant_name = ${label_name ~ ":"}

label = { label_kind? ~ label_name ~ ":" }
label_kind = {
    label_external |
    label_global
}
label_name = @{ label_name_chars+ }
label_name_chars = @{ ASCII_ALPHANUMERIC | "_" }
label_external = { "extern" }
label_global   = { "global" }

operand = {
    "[" ~ operand_value_ptr ~ "]" |
    operand_value
}
operand_value_ptr = {
    operand_value
}
operand_value = {
    register       |
    immediate_bin  |
    immediate_hex  |
    immediate_dec  |
    immediate_char |
    label_name
}

size = @{
    ".8"  |
    ".16" |
    ".32"
}

condition = @{
    "ifz"    |
    "ifnz"   |
    "ifc"    |
    "ifnc"   |
    "ifgteq" |
    "ifgt"   |
    "iflteq" |
    "iflt"
}

instruction = {
    condition? ~ instruction_conditional
}

instruction_conditional = {
    instruction_zero          |
    instruction_one ~ size? ~ operand |
    instruction_two ~ size? ~ operand ~ "," ~ operand
}

instruction_zero = @{
    "nop"  |
    "halt" |
    "brk"  |
    "reti" |
    "ret"  |
    "ise"  |
    "icl"  |
    "mse"  |
    "mcl"
}

instruction_one = @{
    "inc"   |
    "dec"   |
    "not"   |
    "jmp"   |
    "call"  |
    "loop"  |
    "rjmp"  |
    "rcall" |
    "rloop" |
    "push"  |
    "pop"   |
    "int"   |
    "tlb"   |
    "flp"
}

instruction_two = @{
    "add"  |
    "sub"  |
    "mul"  |
    "imul" |
    "div"  |
    "idiv" |
    "rem"  |
    "irem" |
    "and"  |
    "or"   |
    "xor"  |
    "sla"  |
    "sra"  |
    "srl"  |
    "rol"  |
    "ror"  |
    "bse"  |
    "bcl"  |
    "bts"  |
    "cmp"  |
    "movz" |
    "mov"  |
    "rta"  |
    "in"   |
    "out"
}

immediate_bin  = ${ "0b" ~ body_bin }
body_bin       = @{ (ASCII_BIN_DIGIT | "_")+ }

immediate_hex  = ${ "0x" ~ body_hex }
body_hex       = @{ (ASCII_HEX_DIGIT | "_")+ }

immediate_dec  = ${ body_dec }
body_dec       = @{ (ASCII_DIGIT | "_")+ }

immediate_char = ${ "'" ~ body_char ~ "'" }
body_char      = @{ '\x00'..'\x7F' }

immediate_str  = ${ "\"" ~ body_str ~ "\"" }
body_str       = @{ body_str_chars* }
body_str_chars = {
    !("\"" | "\\") ~ ANY
    | "\\" ~ body_str_chars
    | "\\" ~ ("u" ~ ASCII_HEX_DIGIT{4})
}

register = ${ "r" ~ register_num }
register_num = @{ ASCII_DIGIT+ | "sp" | "esp" | "fp" }