deimos-lang/src/vm/module.rs

use crate::get_32_le;
use std::collections::HashMap;
use std::io::Read;

pub const DEIMOS_MAGIC_NUMBER: u64 = 0x00_00_64_65_69_6d_6f_73; // ascii 'deimos'
const DEIMOS_MAGIC_STRING: [u8; 6] = [0x64, 0x65, 0x69, 0x6d, 0x6f, 0x73]; // ascii 'deimos'

pub const VERSION_STRING: &str = "0.1.0";

pub struct DmModule {
    name: String,
    version: String,
    constants: HashMap<String, DmConstant>,
    functions: HashMap<String, DmFunction>,
}

pub enum DmConstant {
    Int(i32),
    Long(i64),
    Double(f64),
    String(String),
}

pub struct DmFunction {
    pub name: String,
    pub byte_code: Vec<u8>,
    pub num_registers: usize,
}

const CONST_SYMBOL: u8 = 0x01;
const FUNCTION_SYMBOL: u8 = 0x02;

enum SymbolType {
    Constant,
    Function,
}

struct DmSymbol {
    name: String,
    symbol_type: SymbolType,
    address: u32,
}

macro_rules! push_byte_array {
    ( $dest: expr, $arr: expr ) => {
        for b in $arr {
            $dest.push(b);
        }
    };
}

macro_rules! push_number_le {
    ( $dest: expr, $num: expr ) => {
        for b in $num.to_le_bytes() {
            $dest.push(b);
        }
    };
}

macro_rules! push_string {
    ( $dest: expr, $s: expr ) => {
        for b in $s.len().to_le_bytes() {
            $dest.push(b);
        }
        for b in $s.bytes() {
            $dest.push(b);
        }
    };
}

pub fn write_module(module: DmModule) -> Vec<u8> {
    // Push magic number
    let mut result: Vec<u8> = Vec::new();
    push_byte_array!(result, DEIMOS_MAGIC_STRING);

    // Push version string
    push_string!(result, module.version);

    // Push module name length, little endian
    push_string!(result, module.name);

    result
}

pub fn load_module(bytes: &[u8]) -> Result<DmModule, String> {
    let mut ip: usize = 0;
    // Check for magic number at bytes 0..5
    if !check_deimos(&bytes) {
        return Err(String::from("Not a valid Deimos module."));
    }
    ip = 6;

    // Get version string length from bytes 6..9
    let version_string_length = get_32_le!(bytes, 6, 0, usize);
    ip = 10;

    // Bytes 10..(10 + version_string_length) are the version string, in utf8
    let mut version_string_raw: Vec<u8> = Vec::with_capacity(version_string_length);
    let version_string_end = ip + version_string_length;
    while ip < version_string_end {
        version_string_raw.push(bytes[ip]);
        ip += 1;
    }
    let version_string = String::from_utf8(version_string_raw).unwrap();

    // Check version string. We'll use this in the future to not load modules compiled later than
    // current version.
    if version_string != "0.1.0" {
        return Err(String::from("Invalid Deimos module version."));
    }

    // TODO: extract module name

    // Holder for Symbols we will extract from the symbol table bytes
    let mut symbols: Vec<DmSymbol> = Vec::new();

    // Get the symbol table length and calculate how far we need to read
    let symbol_table_length = get_32_le!(bytes, 10 + version_string_length, 0, usize);
    let symbol_table_end = ip + symbol_table_length;

    // For each "row" in the symbol table,
    // 1. Get the type
    // 2. Obtain the name's length and then get the name in utf8
    // 3. Grab the address to the actual symbol in the module bytes
    while ip < symbol_table_end {
        let type_byte = bytes[ip];
        ip += 1;
        let name_string_length = get_32_le!(bytes, ip, 0, usize);
        ip += 4;
        let name_string_raw = bytes[ip..ip + name_string_length].to_vec();
        let name = String::from_utf8(name_string_raw).unwrap();
        ip += name_string_length;
        let address = get_32_le!(bytes, ip, 0, u32);
        ip += 4;
        let symbol_type = match type_byte {
            CONST_SYMBOL => SymbolType::Constant,
            FUNCTION_SYMBOL => SymbolType::Function,
            _ => return Err(String::from("Invalid Deimos symbol type.")),
        };
        symbols.push(DmSymbol {
            name,
            address,
            symbol_type,
        });
    }

    todo!()
}

fn check_deimos(bytes: &[u8]) -> bool {
    if bytes.len() < 6 {
        return false;
    }
    let first_six = read_as_u64(&bytes[0..6]);
    first_six == DEIMOS_MAGIC_NUMBER
}

fn read_as_u64(bytes: &[u8]) -> u64 {
    let mut result = 0u64;
    let max_shift = (bytes.len() - 1) * 8;
    for i in 0..bytes.len() {
        result |= (bytes[i] as u64) << (max_shift - i * 8);
    }
    result
}

#[cfg(test)]
mod read_as_u64_tests {
    use crate::vm::module::{read_as_u64, DEIMOS_MAGIC_NUMBER};

    #[test]
    fn read_6_bytes() {
        let bytes = vec![0x64u8, 0x65u8, 0x69u8, 0x6du8, 0x6fu8, 0x73u8]; // ascii 'deimos'
        let result = read_as_u64(&bytes);
        assert_eq!(DEIMOS_MAGIC_NUMBER, result);
    }
}