Fix panic in ExitToNear

[guidovranken]

Description

If input is empty a panic occurs. This PR sets flag to the default value in that case.

Performance / NEAR gas cost considerations

Testing

Fuzzing.

How should this be reviewed

Ensure that flag assuming the default value if input is empty is the intended behavior.

Additional information

[birchmd]

Good catch, thanks!

If the input is empty it will fail the validate_input_size check in parse_input immediately after the assignment of flag, so the default value will never be used, which I think is a good thing. Instead of assigning a default value, we could potentially move flag assignment input parse_input after the size is already validated. But maybe that's more complexity than necessary.

[aleksuss]

Yes. But the only problem is that the parse_input returns modified input. I would propose modifying the parse_input function so it will return the flag as well:

#[cfg(feature = "error_refund")]
let (flag, refund_address, mut input) = parse_input(input)?;
#[cfg(not(feature = "error_refund"))]
let (flag, mut input) = parse_input(input)?;

[birchmd]

@aleksuss Yes, that is what I suggested above. My worry about complexity is because of the conditional compilation you would need to change both versions of the parse_input function (duplicating the flag assignment logic). It's not a big deal to duplicate that one line, and I agree it's nicer to encapsulate all the input handling into one function (even if there are two different versions of that function). So I think I agree with you that it's the better solution overall.

[aleksuss]

On the other hand, even if we get the default value for the flag, we will fail later because we don't pass the validation. So, I guess it's absolutely fine and could be merged.