varint_bitcoin.rs

varint_bitcoin.rs

/// A verbose example of Bitcoin's two variable integer types.
fn main() {
    let value: u64 = 515;

    // 1. CompactSize (Little-Endian) - Used in P2P/Transactions
    let compact = encode_compact_size(value);
    println!("Value: {}", value);
    println!("CompactSize (LE-based) hex: {:02x?}", compact);

    // 2. VarInt (Base-128 Big-Endian) - Used in LevelDB/Disk
    let varint = encode_varint(value);
    println!("Internal VarInt (BE-based) hex: {:02x?}", varint);
}

/// CompactSize: The standard "P2P" encoding (Little-Endian bytes after prefix)
fn encode_compact_size(v: u64) -> Vec<u8> {
    if v < 253 {
        vec![v as u8]
    } else if v <= 0xFFFF {
        let mut b = vec![0xfd];
        b.extend_from_slice(&(v as u16).to_le_bytes());
        b
    } else if v <= 0xFFFFFFFF {
        let mut b = vec![0xfe];
        b.extend_from_slice(&(v as u32).to_le_bytes());
        b
    } else {
        let mut b = vec![0xff];
        b.extend_from_slice(&v.to_le_bytes());
        b
    }
}

/// VarInt: Internal Base-128 encoding (Big-Endian)
/// Logic from Bitcoin Core's `serialize.h` / `WriteVarInt`
fn encode_varint(mut n: u64) -> Vec<u8> {
    let mut res = Vec::new();
    let mut tmp = Vec::new();
    
    loop {
        // The last byte has MSB 0, others have MSB 1
        let mask = if tmp.is_empty() { 0x00 } else { 0x80 };
        tmp.push((n & 0x7F) as u8 | mask);
        if n <= 0x7F { break; }
        n = (n >> 7) - 1; // The "minus one" prevents redundant zero-padding
    }
    
    // Reverse because we calculated from least-significant to most-significant
    // but the format is Big-Endian.
    res.extend(tmp.into_iter().rev());
    res
}

https://play.rust-lang.org/?version=stable&mode=debug&edition=2024&gist=87a47ecc26e26be1c44699d0309f1091