Structs

Practice using structs.

The struct type allows you to organize related data of different types.

---

Objectives

By the end of this lesson you should be able to:

• Construct a struct (user-defined type) that contains several different data types
• Declare members of the struct to maximize storage efficiency
• Describe constraints related to the assignment of structs depending on the types they contain

---

Creating a Struct

In the last exercise, we used a mapping to create a relationship between an address and a uint. But what if your users have favorite colors too? Or favorite cars? You could create a mapping for each of these, but it would quickly get awkward. Instead, a struct can be used to create a custom type that can store all of a user’s favorites within one data type.

Create a new contract called Structs.

Setting up the Struct

Instantiate a struct with the keyword, followed by a name for the type, curly brackets, and the variables that make up the type. Add a stub for Favorites:

struct Favorites {

}

After consulting with the designers, we need to store the following for each address’s favorites:

• Favorite number
• Birth Day of Month
• Favorite color
• Lucky Lottery numbers

Let’s pause for a moment and do some technical design around how to save our favorites.

The product team has confirmed for us that we can safely expect that no users have a favorite number greater than 65,536, and of course, everyone is born on a day of the month between 1-31.

Variable packing also works inside structs, so we could potentially save on storage by using smaller uints for those variables. However, people don’t change their favorite number very often, and the day of the month that they were born on never changes.

Therefore, it’s probably more gas-efficient and less cumbersome to write other parts of the code, if we just use uint for both variables.

Favorite color can be a string.

For Lucky Lottery Numbers, we need a collection. We could use a dynamic array, since this will be in storage, but we already know that the lottery has 5 numbers.

Try to use this information to build the struct on your own. You should end up with something similar to:

Reveal code

struct Favorites {
    uint favoriteNumber;
    uint birthDay;
    string favoriteColor;
    uint[5] lotteryNumbers;
}

Instantiating a Struct with Its Name

There are two ways to instantiate a struct using its name. The first is similar to instantiating a new object in JavaScript:

Favorites memory myFavorites = Favorites({
    favoriteNumber: 29,
    birthDay: 14,
    favoriteColor: "red",
    lotteryNumbers: [uint(1), 2, 3, 4, 5]
});

You can also use a shorthand method where you skip the member names and just list a value for each one. Note that the curly brackets are not included in this format:

Favorites memory myFavorites = Favorites(
    29,
    14,
    "red",
    [uint(1), 2, 3, 4, 5]
);

There’s no difference in gas costs with either of these methods. Use the one that makes the most sense for the given situation.

Saving Multiple Instances to Storage

Next, we need to figure out the best way to organize the Favorites in storage. There are a few options, as always, each with tradeoffs. You could match the pattern you used for favorite numbers and utilize a mapping to match addresses to Favorites.

Another popular method is to use an array, which takes advantage of .push returning a reference to the newly added element, and the fact that the concept of undefined does not exist in Solidity.

First, instantiate an array of Favorites:

Favorites[] public userFavorites;

Next, add a public function to add submitted favorites to the list. It should take each of the members as an argument. Then, assign each argument to the new element via the reference returned by push().

Reveal code

function addFavorite(
    uint _favoriteNumber,
    uint _birthDay,
    string calldata _favoriteColor,
    uint[5] calldata _lotteryNumbers
) public {
    // .push() returns a reference to the new element
    Favorites storage newFavorite = userFavorites.push();
    newFavorite.favoriteNumber = _favoriteNumber;
    newFavorite.birthDay = _birthDay;
    newFavorite.favoriteColor = _favoriteColor;
    newFavorite.lotteryNumbers = _lotteryNumbers;
}

Alternatively, you can create an instance in memory, then push it to storage.

Reveal code

function addFavorite(
    uint _favoriteNumber,
    uint _birthDay,
    string calldata _favoriteColor,
    uint[5] calldata _lotteryNumbers
) public {
    Favorites memory myFavorites = Favorites(
        29,
        14,
        "red",
        [uint(1), 2, 3, 4, 5]
    );

    userFavorites.push(myFavorites);
}

The gas cost is similar for each of these methods.

---

Unexpected Behavior in Structs

Structs in Solidity exhibit some properties that are unexpected, or even frustrating. Working with them often includes untangling a set of mutually-exclusive properties and needs.

Dynamic Storage Arrays in Structs

The product team has contacted you to let you know that the beta testers are complaining about the lotteryNumbers. As it turns out, not every locality has lotteries where 5 numbers are drawn. Some have 3, 4, or even 6!.

You might think this is an easy enough change. After all, you can just remove the size from the array declaration inside Favorites. Go ahead and try it:

struct Favorites {
    uint favoriteNumber;
    uint birthDay;
    string favoriteColor;
    uint[] lotteryNumbers; // Removed the '5'
}

You’ll get an error if you’re using the memory method shown above.

from solidity:
TypeError: Invalid type for argument in function call. Invalid implicit conversion from uint256[5] memory to uint256[] memory requested.
  --> contracts/mappings_exercise.sol:70:13:
   |
70 |             [uint(1), 2, 3, 4, 5]
   |             ^^^^^^^^^^^^^^^^^^^^^

The simplest resolution here is to switch back to using push() to create an empty instance of Favorites, then assigning the values.

The reason this works is a little obtuse. In the failing example, an unsized uint array is the expected type for the argument, but a sized uint array is provided. Solidity cannot perform implicit conversions like this most of the time and you’ll get a compiler error if you provide the wrong type for an argument, even if it is convertible.

One exception to this rule is that Solidity can perform an implicit conversion during assignment if the variable on the right side “fits” into the variable on the left side.

uint[5] fits in uint[], so Solidity will allow it to sit 🐈.

But what happens if you use the getter for userFavorites to retrieve your entry?

{
	"0": "uint256: favoriteNumber 29",
	"1": "uint256: birthDay 14",
	"2": "string: favoriteColor red"
}

What happened to the array? It’s not there, and it turns out that this is on purpose.

Mappings Inside of Structs

You may add mappings inside of structs, subject to a few quirks and restrictions. Add mapping (uint => uint) numberPairs; to Favorites.

In addFavorites, assign newFavorite.numberPairs[33] = 66;

Deploy and test. So far, so good!

Déjà vu ahead: But what happens if you use the getter for userFavorites to retrieve your entry?

{
	"0": "uint256: favoriteNumber 29",
	"1": "uint256: birthDay 14",
	"2": "string: favoriteColor red"
}

It’s not there, and it turns out that this is on purpose.

Another issue emerges if you try to return the struct from a public function. What if you wanted your addFavorite function to return a reference to the new favorite?

// Bad code example, will not work
function addFavorite(
    uint _favoriteNumber,
    uint _birthDay,
    string calldata _favoriteColor,
    uint[] calldata _lotteryNumbers
) public returns (newFavorite memory) {
    // .push() returns a reference to the new element
    Favorites storage newFavorite = userFavorites.push();
    newFavorite.favoriteNumber = _favoriteNumber;
    newFavorite.birthDay = _birthDay;
    newFavorite.favoriteColor = _favoriteColor;
    newFavorite.lotteryNumbers = _lotteryNumbers;
    newFavorite.numberPairs[33] = 66;

    return newFavorite;
}

You’ll get an error. The mapping type cannot be returned by a public or external function, so neither can a struct that contains one.

from solidity:
TypeError: Types containing (nested) mappings can only be parameters or return variables of internal or library functions.
  --> contracts/mappings_exercise.sol:64:23:
   |
64 |     ) public returns (Favorites memory) {
   |                       ^^^^^^^^^^^^^^^^

Finally, what happens if you try to assign newFavorite to a memory variable? Again, an error occurs because mappings can only be in storage.

// Bad code example, will not work
Favorites memory secondFavorite = newFavorite;

from solidity:
TypeError: Type struct Structs.Favorites memory is only valid in storage because it contains a (nested) mapping.
  --> contracts/mappings_exercise.sol:82:9:
   |
82 |         Favorites memory secondFavorite = newFavorite;
   |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Automatic Getters for Public Structs

As with other types, if you put a public struct in storage at the contract level, the compiler will generate a getter automatically. However, these don’t work quite the way you might expect. For example, imagine:

struct MyStruct {
    uint first;
    uint second;
    uint third;
}

MyStruct myStruct;

The automatic getter for myStruct will not be:

// Approximate example, not real code
function myStruct() public view returns (MyStruct memory) {
    return myStruct;
}

Instead, it returns the members individually:

// Approximate example, not real code
function myStruct() public view returns (uint, uint, uint) {
    return (myStruct.first, myStruct.second, myStruct.third);
}

Create your own getter to return the data as a tuple, which will be interpreted as the appropriate type if it’s called from another contract via an interface.

function getMyStruct() public view returns (MyStruct memory) {
    return myStruct;
}

---

Conclusion

In this lesson, you’ve learned how to use the struct keyword to create a custom type that stores related data. You’ve also learned three methods of instantiating them and common patterns for storing structs in storage. Finally, you’ve explored some of the constraints that emerge when working with more complex data types within a struct.