compile_hex.nula

// bootstrap/compile_hex.nula — Compile expressions to hex-encoded VM bytecode.
// Outputs u32 instruction words as 8-char hex, one per line.
// Run: echo "1 + 2" | nulang bootstrap/compile_hex.nula

fn is_digit(c: Int) -> Bool { c >= 48 and c <= 57 }
fn is_alpha(c: Int) -> Bool { (c >= 65 and c <= 90) or (c >= 97 and c <= 122) }
fn is_alphanum(c: Int) -> Bool { is_alpha(c) or is_digit(c) or c == 95 }
fn is_space(c: Int) -> Bool { c == 32 or c == 10 or c == 13 or c == 9 }

fn skip_ws(src: String, pos: Int, len: Int) -> Int {
    if pos >= len then pos
    else if is_space(perform String.charAt(src, pos)) then skip_ws(src, pos + 1, len)
    else pos
}

fn read_int(src: String, pos: Int, len: Int, acc: Int) -> Int {
    if pos >= len then (acc << 16) + pos
    else {
        let c = perform String.charAt(src, pos);
        if is_digit(c) then read_int(src, pos + 1, len, acc * 10 + (c - 48))
        else (acc << 16) + pos
    }
}

fn read_ident(src: String, pos: Int, len: Int, h: Int) -> Int {
    if pos >= len then (h << 16) + pos
    else {
        let c = perform String.charAt(src, pos);
        if is_alphanum(c) then {
            let nh = h * 5 + c;
            read_ident(src, pos + 1, len, nh)
        } else (h << 16) + pos
    }
}

fn env_lookup(e0: Int, e1: Int, e2: Int, e3: Int, elen: Int, h: Int) -> Int {
    if elen <= 0 then 0
    else {
        let slot = if elen >= 4 then e3
        else if elen >= 3 then e2
        else if elen >= 2 then e1
        else e0;
        let sh = slot >> 32;
        if sh == h then slot & 0xFFFFFFFF
        else env_lookup(e0, e1, e2, 0, elen - 1, h)
    }
}

fn prec_of(c: Int) -> Int {
    if c == 43 then 2 else if c == 45 then 2
    else if c == 42 then 3 else if c == 47 then 3
    else 0
}

fn parse_cmp(src: String, p: Int, len: Int) -> Int {
    if p >= len then 0 else {
        let c = perform String.charAt(src, p);
        if c == 61 and p + 1 < len then {
            let c2 = perform String.charAt(src, p + 1);
            if c2 == 61 then (1 << 16) + (0x40 << 8) + 2 else 0
        } else if c == 33 and p + 1 < len then {
            let c2 = perform String.charAt(src, p + 1);
            if c2 == 61 then (1 << 16) + (0xFF << 8) + 2 else 0
        } else if c == 60 then {
            if p + 1 < len then { let c2 = perform String.charAt(src, p + 1);
                if c2 == 61 then (1 << 16) + (0x43 << 8) + 2 else (1 << 16) + (0x41 << 8) + 1
            } else (1 << 16) + (0x41 << 8) + 1
        } else if c == 62 then {
            if p + 1 < len then { let c2 = perform String.charAt(src, p + 1);
                if c2 == 61 then (1 << 16) + (0x44 << 8) + 2 else (1 << 16) + (0x42 << 8) + 1
            } else (1 << 16) + (0x42 << 8) + 1
        } else 0
    }
}

// Convert nybble (0-15) to hex char
fn hex_digit(n: Int) -> String {
    if n < 10 then perform Int.to_string(n)
    else if n == 10 then "a" else if n == 11 then "b"
    else if n == 12 then "c" else if n == 13 then "d"
    else if n == 14 then "e" else "f"
}

// Print a u32 as 8-char hex. Start with "" to work around string-var concat bug.
fn remark(msg: String) {
    perform IO.print("; " + msg)
}

fn emit_hex(w: Int) {
    perform IO.print("" +
        hex_digit((w >> 28) & 0xF) +
        hex_digit((w >> 24) & 0xF) +
        hex_digit((w >> 20) & 0xF) +
        hex_digit((w >> 16) & 0xF) +
        hex_digit((w >> 12) & 0xF) +
        hex_digit((w >> 8) & 0xF) +
        hex_digit((w >> 4) & 0xF) +
        hex_digit(w & 0xF)
    )
}

// Encode instruction: opcode<<24 | op1<<16 | op2<<8 | op3
fn instr(opcode: Int, op1: Int, op2: Int, op3: Int) -> Int {
    (opcode << 24) + (op1 << 16) + (op2 << 8) + op3
}

// Returns (result_reg << 16) | pos.
fn comp(src: String, pos: Int, len: Int, left: Int, min_prec: Int, nr: Int, e0: Int, e1: Int, e2: Int, e3: Int, elen: Int) -> Int {
    let no_left = 1 << 40;
    let pair = if left == no_left then {
        let p = skip_ws(src, pos, len);
        if p >= len then (nr << 18) + p
        else {
            let c = perform String.charAt(src, p);
            if is_digit(c) then {
                let packed = read_int(src, p, len, 0);
                let num = packed >> 16;
                let q = packed & 0xFFFF;
                if num == 0 then emit_hex(instr(0x03, nr, 0, 0))
                else if num == 1 then emit_hex(instr(0x04, nr, 0, 0))
                else if num == 2 then emit_hex(instr(0x05, nr, 0, 0))
                else if num == -1 then emit_hex(instr(0x06, nr, 0, 0))
                else {
                    remark("const " + perform Int.to_string(num));
                    emit_hex(instr(0x07, 0, 0, nr))
                };
                (nr << 18) + q
            } else if c == 40 then {
                let inner = comp(src, p + 1, len, no_left, -1, nr, e0, e1, e2, e3, elen);
                let rv = inner >> 18;
                let rp = inner & 0x1FFFF;
                let q2 = skip_ws(src, rp, len);
                if q2 < len then {
                    let c2 = perform String.charAt(src, q2);
                    if c2 == 41 then (rv << 18) + (q2 + 1)
                    else (rv << 18) + q2
                } else (rv << 18) + q2
            } else if is_alpha(c) then {
                let ident = read_ident(src, p, len, 0);
                let h = ident >> 16;
                let q = ident & 0xFFFF;
                if h == 3321 then {
                    let q2 = skip_ws(src, q, len);
                    let var_id = read_ident(src, q2, len, 0);
                    let vh = var_id >> 16;
                    let q3 = var_id & 0xFFFF;
                    let q4 = skip_ws(src, q3, len);
                    if q4 < len then {
                        let c3 = perform String.charAt(src, q4);
                        if c3 == 61 then {
                            let bound = comp(src, q4 + 1, len, no_left, -1, nr, e0, e1, e2, e3, elen);
                            let vr = bound >> 18;
                            let vp = bound & 0x1FFFF;
                            let vp2 = skip_ws(src, vp, len);
                            if vp2 + 1 < len then {
                                let ci = perform String.charAt(src, vp2);
                                let cn = perform String.charAt(src, vp2 + 1);
                                if ci == 105 and cn == 110 then {
                                    let entry = (vh << 32) + vr;
                                    let nxt = nr + 1;
                                    if elen == 0 then comp(src, vp2 + 2, len, no_left, -1, nxt, entry, 0, 0, 0, 1)
                                    else if elen == 1 then comp(src, vp2 + 2, len, no_left, -1, nxt, e0, entry, 0, 0, 2)
                                    else if elen == 2 then comp(src, vp2 + 2, len, no_left, -1, nxt, e0, e1, entry, 0, 3)
                                    else comp(src, vp2 + 2, len, no_left, -1, nxt, e0, e1, e2, entry, 4)
                                } else (vr << 18) + vp2
                            } else (vr << 18) + vp2
                        } else (nr << 18) + p
                    } else (nr << 18) + p
                } else if h == 627 then {
                    let cond = comp(src, q, len, no_left, -1, nr, e0, e1, e2, e3, elen);
                    let cr = cond >> 18;
                    let cp = cond & 0x1FFFF;
                    let cnext = nr + 1;
                    let cp2 = skip_ws(src, cp, len);
                    let then_id = read_ident(src, cp2, len, 0);
                    let th = then_id >> 16;
                    let tp = then_id & 0xFFFF;
                    if th == 17715 then {
                        let lb = (cond >> 17) & 1;
                        remark("if");
                        if lb == 0 then {
                            emit_hex(instr(0x03, 7, 0, 0));
                            let tmp = cnext + 1;
                            emit_hex(instr(0x40, cr, 7, tmp));
                            emit_hex(instr(0x49, tmp, tmp, 0));
                            remark("JmpF -> else");
                            emit_hex(instr(0x52, tmp, 0, 0))
                        } else {
                            remark("JmpF -> else");
                            emit_hex(instr(0x52, cr, 0, 0))
                        };
                        remark("then:");
                        let then_body = comp(src, tp, len, no_left, -1, cnext, e0, e1, e2, e3, elen);
                        let tr = then_body >> 18;
                        let tbp = then_body & 0x1FFFF;
                        remark("Jmp -> end");
                        emit_hex(instr(0x50, 0, 0, 0));
                        remark("else:");
                        let tbp2 = skip_ws(src, tbp, len);
                        let else_id = read_ident(src, tbp2, len, 0);
                        let eh = else_id >> 16;
                        let ep = else_id & 0xFFFF;
                        if eh == 16001 then {
                            let else_body = comp(src, ep, len, no_left, -1, cnext, e0, e1, e2, e3, elen);
                            let er = else_body >> 18;
                            let ebp = else_body & 0x1FFFF;
                            remark("end:");
                            (er << 18) + ebp
                        } else (nr << 18) + p
                    } else (nr << 18) + p
                } else if h == 18036 then {
                    emit_hex(instr(0x04, nr, 0, 0));
                    (nr << 18) + q
                } else if h == 79251 then {
                    emit_hex(instr(0x03, nr, 0, 0));
                    (nr << 18) + q
                } else if h == 3421 then {
                    let inner = comp(src, q, len, no_left, -1, nr, e0, e1, e2, e3, elen);
                    let iv = inner >> 18;
                    let ip = inner & 0x1FFFF;
                    let dr = nr + 1;
                    emit_hex(instr(0x03, dr, 0, 0));
                    let dr2 = nr + 2;
                    emit_hex(instr(0x40, iv, dr, dr2));
                    (dr2 << 18) + (1 << 17) + ip
                } else {
                    let reg = env_lookup(e0, e1, e2, e3, elen, h);
                    if reg == 0 then (nr << 18) + p
                    else {
                        emit_hex(instr(0x12, reg, nr, 0));
                        (nr << 18) + q
                    }
                }
            } else { (nr << 18) + p }
        }
    } else {
        let is_bool_bit = 1 << 39;
        let lb = if left >= is_bool_bit then 1 else 0;
        let rl = if left >= is_bool_bit then left - is_bool_bit else left;
        (rl << 18) + (lb << 17) + pos
    };

    let lr = pair >> 18;
    let lp = pair & 0x1FFFF;
    let p = skip_ws(src, lp, len);
    if p >= len then pair
    else {
        let c = perform String.charAt(src, p);
        if c == 61 or c == 33 or c == 60 or c == 62 then {
            let cmp = parse_cmp(src, p, len);
            if cmp == 0 then pair else {
                let cprec = cmp >> 16;
                let cop = (cmp >> 8) & 0xFF;
                let clen = cmp & 0xFF;
                if cprec < min_prec then pair else {
                    let rhs = comp(src, p + clen, len, no_left, cprec, nr + 1, e0, e1, e2, e3, elen);
                    let rr = rhs >> 18;
                    let rp = rhs & 0x1FFFF;
                    if cop == 0xFF then {
                        emit_hex(instr(0x40, lr, rr, dr));
                        emit_hex(instr(0x49, dr, dr, 0))
                    } else {
                        emit_hex(instr(cop, lr, rr, dr))
                    };
                    comp(src, rp, len, dr + (1 << 39), min_prec, dr + 1, e0, e1, e2, e3, elen)
                }
            }
        } else if is_alpha(c) then {
            let ident = read_ident(src, p, len, 0);
            let kw = ident >> 16;
            let q = ident & 0xFFFF;
            if kw == 3075 then {
                if 0 < min_prec then pair else {
                    let lb = (pair >> 17) & 1;
                    remark("and");
                    if lb == 0 then {
                        emit_hex(instr(0x03, 7, 0, 0));
                        let tmp = nr + 1;
                        emit_hex(instr(0x40, lr, 7, tmp));
                        emit_hex(instr(0x49, tmp, tmp, 0));
                        remark("JmpF -> and_end");
                        emit_hex(instr(0x52, tmp, 0, 0))
                    } else {
                        remark("JmpF -> and_end");
                        emit_hex(instr(0x52, lr, 0, 0))
                    };
                    let rhs = comp(src, q, len, no_left, 0, nr + 1, e0, e1, e2, e3, elen);
                    let rr = rhs >> 18;
                    let rp = rhs & 0x1FFFF;
                    remark("and_end:");
                    comp(src, rp, len, rr, min_prec, rr + 1, e0, e1, e2, e3, elen)
                }
            } else if kw == 669 then {
                if 0 < min_prec then pair else {
                    let lb = (pair >> 17) & 1;
                    remark("or");
                    if lb == 0 then {
                        emit_hex(instr(0x03, 7, 0, 0));
                        let tmp = nr + 1;
                        emit_hex(instr(0x40, lr, 7, tmp));
                        emit_hex(instr(0x49, tmp, tmp, 0));
                        remark("JmpF -> or_right");
                        emit_hex(instr(0x52, tmp, 0, 0))
                    } else {
                        remark("JmpF -> or_right");
                        emit_hex(instr(0x52, lr, 0, 0))
                    };
                    let dr = nr + 2;
                    emit_hex(instr(0x04, dr, 0, 0));
                    remark("Jmp -> or_end");
                    emit_hex(instr(0x50, 0, 0, 0));
                    remark("or_right:");
                    let rhs = comp(src, q, len, no_left, 0, dr + 1, e0, e1, e2, e3, elen);
                    let rr = rhs >> 18;
                    let rp = rhs & 0x1FFFF;
                    remark("or_end:");
                    comp(src, rp, len, rr, min_prec, rr + 1, e0, e1, e2, e3, elen)
                }
            } else pair
        } else {
            let prec = prec_of(c);
            if prec == 0 or prec < min_prec then pair else {
                let opcode = if c == 43 then 0x20 else if c == 45 then 0x21
                else if c == 42 then 0x22 else 0x23;
                let rhs = comp(src, p + 1, len, no_left, prec, nr + 1, e0, e1, e2, e3, elen);
                let rr = rhs >> 18;
                let rp = rhs & 0x1FFFF;
                let dr = nr + 2;
                emit_hex(instr(opcode, lr, rr, dr));
                comp(src, rp, len, dr, min_prec, dr + 1, e0, e1, e2, e3, elen)
            }
        }
    }
}

fn compile(src: String) {
    let len = perform String.length(src);
    perform IO.print("; " + src);
    let result = comp(src, 0, len, 1 << 40, -1, 8, 0, 0, 0, 0, 0);
    let reg = result >> 18;
    emit_hex(instr(0x12, reg, 0, 0));
    emit_hex(instr(0x01, 0, 0, 0));
    perform IO.print("; result in r" + perform Int.to_string(result >> 18))
}

fn main() {
    let input = perform IO.read();
    if input == "" then compile("1 < 2 and 2 < 3") else compile(input)
}