My Profile Photo

Daniel Perez


software engineer interested in programming languages and functional programming


Go serialization

I have been using Golang to build some REST API recently, and I was having some trouble to serialize my data properly to JSON.

Almost all tutorial around there tell about how we should return JSON, using some struct and tags.

type User struct {
    ID        int `json:"id"`
    Email     string `json:"email"`
    HideEmail bool `json:"hide_email"`
    FirstName string `json:"first_name"`
    LastName  string `json:"last_name"`
    CreatedAt time.Time `json:"created_at"`
    UpdatedAt time.Time `json:"updated_at"`
}

user := User{
    ID: 1, Email: "x@example.com", FirstName: "Foo", LastName:  "Bar",HideEmail: true
}

data, err := json.Marshal(user)

This is indeed useful, but there are many cases when we want to dynamically choose what to serialize depending some privacy settings, whether the user is logged in or not, or a lot of other factors.

For these situations, I found it a bit verbose and repetitive to have to enter each field in a map to be able to have the wanted serialization output, so I decided to create a little library to help me do this. Its goal is basically to help converting a struct to a map with the most flexibility as possible.

With the above struct, given I want to hide the Email field when HideEmail is true, I can write:

userSerializer := structomap.New().
                             UseSnakeCase().
                             Pick("ID", "Email", "FirstName", "LastName").
                             Omit("HideEmail").
                             OmitIf(func(u interface{}) bool {
                                return u.(User).HideEmail
                            }, "Email")

and then I just need to call userSerializer.Transform on any user to get the result I want as a map[string]interface{}. What is nice about this is that it also works with arrays: userSerializer.TransformArray will transform an array of User in an array of map[string]interface{} which can be directly serialized to JSON.

You can find more information on the project page.

comments powered by Disqus