Ethers.js: How to check if a contract is deployed or not?

[jinserk]

Hi,

I want to check a contract was deployed or not on a specific network. Is there any way to do this? I've tried Contract.deployed() but don't know it works well.

[ricmoo]

The deployed is more for contracts deployed using a ContractFactory. If you did use a contract factory, then deployed is very reliable, since it knows the transaction hash that deployed it, and checks the transaction receipt for success.

If you have the transaction hash of a contract deployment, you can get the receipt. If the receipt status is success, the deployment was good.

[jinserk]

Thanks, @ricmoo! I want to check if the contract address exists on the network, since the contract could be deployed already on the network using other Truffle Framework or something else. So I am just using not ContractFactory but Contract. I have no transaction hash for the deployment. Can I check the address in this situation?

[ricmoo]

If you use deployed on a Contract, it simply uses getCode then, which will return true if there is code at the address. This can be unreliable in a few cases:

• if the contract was deployed recently, in which case it may not be mined yet
• the contract called the self-destruct op-code (which means deployment was successful at some point, but that the contract is no longer there)
• the contract has 0 bytes of code (which is a fairly rare use case)

For most cases though, it should be reliable to use the .deployed, or just use getCode directly on the address.

[jinserk]

Hmm. do you mean that we don't have any effective way to confirm that a contract exists or not on a network? I guess that we can't check a contract is usable or not in the second and the third cases even though we use deployed() with proper transaction hash.

[ricmoo]

I have never experimented with what really happens when you deploy a contract with 0 bytes of init code (I'll be experimenting with that this weekend, as part of ETHCapeTown).

I do not believe there is any way to use the JSON-RPC API to determine for certain whether an address is a deployed contract or not, but if it is a contract you deployed, you likely are not deploying zero-byte contracts, so you should be safe to use the provider.getCode(address) method.

You can detect the second case in some cases, if you know a block between the deployment and the self-destruct blocks, since deployed takes in a n optional parameter of what blockTag to use; however, this also is not reliable on many backends. For example, Etherscan ignores blockTag for getCode, so the EtherscanProvider will detect this and throw an error.

[jinserk]

Thank you for the detailed answer! It's very informative!

[jtakalai]

For clarification, using Contract.deployed() seems to throw in case there's nothing there (4.0.47), and getCode is a method in Provider, not Contract. I think contract's bytecode can be very interesting, and if you're holding a Contract, first place you'd look is Contract's methods, something like bytecode, getByteCode(), or just code or getCode(). I can open a new feature request if that makes more sense :)

So instead of

if (await contract.provider.getCode(contract.address) === "0x") { ... }

it would be cool to be able to do

if (await contract.getCode() === "0x") { ... }