abi-decode

Decode static ABI-encoded arguments from calldata or return data, with no dependencies.

Paste a transaction input from an explorer and you get the arguments back as typed values. This covers the static types — the ones that sit on a 32-byte boundary — which is the majority of real calldata.

Usage

npx github:kakmano723-ctrl/abi-decode --hex 0xa9059cbb<...> address uint256

address    0xd8da6bf26964af9d7eed9e03e53415d37aa96045
uint256    1000000000000000000

As a library:

import { decode } from "abi-decode";
decode("0x" + "00".repeat(31) + "01", ["uint256"]); // [1n]

Rules it enforces

• Word alignment. A payload that is not a multiple of 32 bytes is rejected rather than silently truncated.
• Canonical bool. Only 0 and 1 decode; 2 is an error, because a non-canonical bool is a real source of divergent behaviour between clients.
• Clean bytesN padding. The bytes after N must be zero.
• Two's complement for int<M>. int256(-1) decodes to -1n, not a huge positive number.

What it does not do

• Dynamic types. string, bytes and arrays are offset-based, and resolving them needs the entire parameter list. Use a full ABI decoder for those.
• Signature lookup. It decodes arguments; it does not tell you which function was called. evm-calldata does that.
• No type inference. You state the types. Guessing from the bytes is how you get plausible-but-wrong output.

Development

npm test

License

MIT