Testing Smart Contracts
ℹ️ Info To use the library functions designed for testing smart contracts, you need to add
snforge_std
package as a dependency in yourScarb.toml
using appropriate release tag.[dev-dependencies] snforge_std = { git = "https://github.com/foundry-rs/starknet-foundry.git", tag = "v0.12.0" }
Using unit testing as much as possible is a good practice, as it makes your test suites run faster. However, when writing smart contracts, you often want to test their interactions with the blockchain state and with other contracts.
The Test Contract
In this tutorial we will be using this Starknet contract
#[starknet::interface]
trait IHelloStarknet<TContractState> {
fn increase_balance(ref self: TContractState, amount: felt252);
fn get_balance(self: @TContractState) -> felt252;
}
#[starknet::contract]
mod HelloStarknet {
#[storage]
struct Storage {
balance: felt252,
}
#[abi(embed_v0)]
impl HelloStarknetImpl of super::IHelloStarknet<ContractState> {
// Increases the balance by the given amount.
fn increase_balance(ref self: ContractState, amount: felt252) {
self.balance.write(self.balance.read() + amount);
}
// Gets the balance.
fn get_balance(self: @ContractState) -> felt252 {
self.balance.read()
}
}
}
Note that the name after mod
will be used as the contract name for testing purposes.
Writing Tests
Let's write a test that will deploy the HelloStarknet
contract and call some functions.
use snforge_std::{ declare, ContractClassTrait };
#[test]
fn call_and_invoke() {
// First declare and deploy a contract
let contract = declare("HelloStarknet").unwrap();
// Alternatively we could use `deploy_syscall` here
let (contract_address, _) = contract.deploy(@array![]).unwrap();
// Create a Dispatcher object that will allow interacting with the deployed contract
let dispatcher = IHelloStarknetDispatcher { contract_address };
// Call a view function of the contract
let balance = dispatcher.get_balance();
assert(balance == 0, 'balance == 0');
// Call a function of the contract
// Here we mutate the state of the storage
dispatcher.increase_balance(100);
// Check that transaction took effect
let balance = dispatcher.get_balance();
assert(balance == 100, 'balance == 100');
}
📝 Note
Notice that the arguments to the contract's constructor (the
deploy
'scalldata
argument) need to be serialized withSerde
.
HelloStarknet
contract has no constructor, so the calldata remains empty in the example above.
$ snforge test
Collected 1 test(s) from using_dispatchers package
Running 0 test(s) from src/
Running 1 test(s) from tests/
[PASS] tests::call_and_invoke
Tests: 1 passed, 0 failed, 0 skipped, 0 ignored, 0 filtered out
Handling Errors
Sometimes we want to test contracts functions that can panic, like testing that function that verifies caller address
panics on invalid address. For that purpose Starknet also provides a SafeDispatcher
, that returns a Result
instead of
panicking.
First, let's add a new, panicking function to our contract.
#[starknet::interface]
trait IHelloStarknet<TContractState> {
// ...
fn do_a_panic(self: @TContractState);
fn do_a_string_panic(self: @TContractState);
}
#[starknet::contract]
mod HelloStarknet {
use array::ArrayTrait;
// ...
#[abi(embed_v0)]
impl HelloStarknetImpl of super::IHelloStarknet<ContractState> {
// ...
// Panics
fn do_a_panic(self: @ContractState) {
let mut arr = ArrayTrait::new();
arr.append('PANIC');
arr.append('DAYTAH');
panic(arr);
}
fn do_a_string_panic(self: @ContractState) {
assert!(false, "This is panicking with a string, which can be longer than 31 characters");
}
}
}
If we called this function in a test, it would result in a failure.
#[test]
#[feature("safe_dispatcher")]
fn failing() {
// ...
let (contract_address, _) = contract.deploy(@calldata).unwrap();
let dispatcher = IHelloStarknetDispatcher { contract_address };
dispatcher.do_a_panic();
}
$ snforge test
Collected 1 test(s) from package_name package
Running 0 test(s) from src/
Running 1 test(s) from tests/
[FAIL] tests::failing
Failure data:
(0x50414e4943 ('PANIC'), 0x444159544148 ('DAYTAH'))
Tests: 0 passed, 1 failed, 0 skipped, 0 ignored, 0 filtered out
Failures:
tests::failing
SafeDispatcher
Using SafeDispatcher
we can test that the function in fact panics with an expected message.
#[test]
#[feature("safe_dispatcher")]
fn handling_errors() {
// ...
let (contract_address, _) = contract.deploy(@calldata).unwrap();
let safe_dispatcher = IHelloStarknetSafeDispatcher { contract_address };
match safe_dispatcher.do_a_panic() {
Result::Ok(_) => panic_with_felt252('shouldve panicked'),
Result::Err(panic_data) => {
assert(*panic_data.at(0) == 'PANIC', *panic_data.at(0));
assert(*panic_data.at(1) == 'DAYTAH', *panic_data.at(1));
}
};
}
Now the test passes as expected.
$ snforge test
Collected 1 test(s) from package_name package
Running 0 test(s) from src/
Running 1 test(s) from tests/
[PASS] tests::handling_errors
Tests: 1 passed, 0 failed, 0 skipped, 0 ignored, 0 filtered out
Similarly, you can handle the panics which use string as an argument (like an assert!
macro)
// Necessary struct and trait imports for string errors mapping
use snforge_std::errors::{ SyscallResultStringErrorTrait, PanicDataOrString };
// ...
#[test]
#[feature("safe_dispatcher")]
fn handling_string_errors() {
// ...
let (contract_address, _) = contract.deploy(@calldata).unwrap();
let safe_dispatcher = IHelloStarknetSafeDispatcher { contract_address };
// Notice the `map_error_to_string` helper usage here, and different error type
match safe_dispatcher.do_a_string_panic().map_error_to_string() {
Result::Ok(_) => panic_with_felt252('shouldve panicked'),
Result::Err(panic_data) => {
match x {
PanicDataOrString::PanicData(_) => panic_with_felt252('wrong format'),
PanicDataOrString::String(str) => {
assert(
str == "This a panicking with a string, which can be longer",
'wrong string received'
);
}
}
}
};
}
You also could skip the de-serialization of the panic_data
, and not use map_error_to_string
, but this way you can actually use assertions on the ByteArray
that was used to panic.
📝 Note
To operate with
SafeDispatcher
it's required to annotage its usage with#[feature("safe_dispatcher")]
.There are 3 options:
- module-level declaration
#[feature("safe_dispatcher")] mod my_module;
- function-level declaration
#[feature("safe_dispatcher")] fn my_function() { ... }
- directly before the usage
#[feature("safe_dispatcher")] let result = safe_dispatcher.some_function();
Expecting Test Failure
Sometimes the test code failing can be a desired behavior.
Instead of manually handling it, you can simply mark your test as #[should_panic(...)]
.
See here for more details.