semi-libc/tools/compile-iconv.rs

#[derive(Debug, Clone)]
struct Octo {
    tag: Option<String>,
    content: [u8; 8],
    reloc: Vec<(usize, String)>,
}

fn main() {
    let mut args = std::env::args().skip(1);
    let src = args.next().expect("please specify source file");
    let dst = args.next().expect("please specify output file");

    let mut octos = Vec::new();
    let src_content = std::fs::read_to_string(&src)
        .expect("cannot read source file");
    let mut current_tag = None;
    for (nline, line) in src_content.lines().enumerate() {
        let line = line.trim();
        if line.is_empty() || line.starts_with("//") {
            continue;
        }
        if let Some(tag) = line.strip_suffix(":") {
            current_tag = Some(tag);
            continue;
        }
        let Some((op, arglist)) = line.split_once(' ') else {
            error(nline, "operation without arglist is currently disallowed");
        };
        let mut content = [0u8; 8];
        content[0] = do_op(op).unwrap_or_else(|| error(nline, "unrecognized op"));
        let mut cursor = 1;
        if content[0] == 0 {
            cursor = 0;
        }
        let reloc = do_arglist(arglist, &mut content, &mut cursor);
        octos.push(Octo {
            tag: current_tag.take().map(ToString::to_string),
            content,
            reloc,
        });
    }

    let mut tags = std::collections::HashMap::new();
    for (nocto, octo) in octos.iter().enumerate() {
        if let Some(tag) = &octo.tag {
            tags.insert(tag.to_string(), nocto * 8);
        }
    }

    let mut content = Vec::new();
    for octo in octos {
        let mut current = octo.content.to_vec();
        for (offset, tag) in octo.reloc {
            let tag_addr = *tags.get(&tag).expect(&format!("tag {tag} undefined")) as u32;
            current[offset..offset + 4].copy_from_slice(&tag_addr.to_be_bytes()[..]);
        }
        content.append(&mut current);
    }

    std::fs::write(dst, &content).expect("failed to write dst file");
}

fn error(line: usize, reason: &str) -> ! {
    eprintln!("error: {reason} (source:{})", line + 1);
    std::process::exit(1);
}

fn do_op(s: &str) -> Option<u8> {
    match s {
        "place64" => Some(0),
        "input8" => Some(1),
        "input16" => Some(2),
        "input32" => Some(3),
        "input64" => Some(4),
        "output8" => Some(5),
        "output16" => Some(6),
        "output32" => Some(7),
        "output64" => Some(8),
        "xor" => Some(9),
        "and" => Some(10),
        "or" => Some(11),
        "add" => Some(13),
        "sub" => Some(14),
        "smul" => Some(15),
        "umul" => Some(17),
        "eq" => Some(19),
        "mt" => Some(20),
        "lt" => Some(21),
        "shl" => Some(29),
        "shr" => Some(30),
        "not" => Some(12),
        "jnz" => Some(22),
        "jmp" => Some(23),
        "load8" => Some(24),
        "load16" => Some(25),
        "load32" => Some(26),
        "load64" => Some(27),
        "store64" => Some(28),
        "ret" => Some(31),
        "ptrload8" => Some(33),
        "ptrload16" => Some(34),
        "ptrload32" => Some(35),
        "ptrload64" => Some(36),
        _ => None,
    }
}

fn do_arglist(s: &str, content: &mut [u8; 8], cursor: &mut usize) -> Vec<(usize, String)> {
    let mut reloc = Vec::new();
    for arg in s.split(',') {
        let arg = arg.trim();
        if let Some(x) = do_arg(arg, content, cursor) {
            reloc.push(x);
        }
    }
    reloc
}

fn do_arg(s: &str, content: &mut [u8; 8], cursor: &mut usize) -> Option<(usize, String)> {
    if let Ok(n) = s.parse::<u64>() {
        *cursor = 8;
        *content = n.to_be_bytes();
        None
    } else if let Some(digit) = s.strip_prefix('0') {
        let radix = match digit.as_bytes()[0] {
            b'x' => 16,
            b'o' => 8,
            b'b' => 2,
            _ => panic!("unknown radix"),
        };
        *cursor = 8;
        *content = u64::from_str_radix(&digit[1..], radix).unwrap().to_be_bytes();
        None
    } else if let Some(reg_id) = s.strip_prefix("&r") {
        content[*cursor] = reg_id.parse::<u8>().unwrap();
        *cursor += 1;
        None
    } else if s == "@sicv_magic_v1" {
        *content = [b'S', b'I', b'C', b'V', 0, 0, 0, 1];
        *cursor = 8;
        None
    } else {
        *cursor += 4;
        Some((*cursor - 4, s.into()))
    }
}