Rust에서 매크로와 함수: 언제 어떤 것을 사용해야 하나?

When developing in Rust, we often face a dilemma: when should we use macros to simplify our code, and when should we rely on functions instead?

This article will analyze scenarios for using macros, helping you understand when macros are appropriate. Let's start with a conclusion:

Macros and functions are not interchangeable but complementary. Each has its own strengths, and only by using them properly can we write excellent Rust code.

Now, let's explore the use cases for macros.

Categories of Macros

Macros in Rust are categorized into:

• Declarative Macros (macro_rules!)
• Procedural Macros

Procedural macros can be further divided into:

• Custom Derive Macros
• Attribute Macros
• Function-like Macros

In Rust, both functions and macros serve as essential tools for code reuse and abstraction. Functions encapsulate logic, handle a fixed number of parameters with known types, and provide type safety and readability. Macros, on the other hand, generate code at compile time, enabling capabilities that functions cannot achieve, such as handling a variable number and type of parameters, code generation, and metaprogramming.

Specific Use Cases

Declarative Macros (macro_rules!)

Scenario: Handling a Variable Number and Type of Parameters

Problem Description:

• Functions must specify the number and types of parameters at definition and cannot directly accept a variable number or type of parameters.
• A mechanism is needed to handle functionalities like println!, which accepts an arbitrary number and type of arguments.

Macro Solution:

• Declarative macros use pattern matching to accept arbitrary numbers and types of parameters.
• Repetition patterns ($()*) and metavariables ($var) are used to capture parameter lists.

Example Code:

// Define a macro that accepts variable arguments
macro_rules! my_println {
    ($($arg:tt)*) => {
        println!($($arg)*);
    };
}

fn main() {
    my_println!("Hello, world!");
    my_println!("Number: {}", 42);
    my_println!("Multiple values: {}, {}, {}", 1, 2, 3);
}

Limitations of Functions:

• Functions cannot define signatures that accept arbitrary numbers and types of parameters.
• Even using variadic parameters, Rust does not directly support them without special constructs like format_args!.

Coordination Between Macros and Functions:

• Macros collect and expand parameters, then call underlying functions (e.g., println! ultimately calls std::io::stdout().write_fmt()).
• Functions handle the core execution logic, while macros parse parameters and generate code.

Scenario: Simplifying Repetitive Code Patterns

Problem Description:

• When there are many repetitive code patterns, such as test cases or field accessors.
• Writing such code manually is error-prone and has high maintenance costs.

Macro Solution:

• Declarative macros match patterns to generate repetitive code structures automatically.
• Using macros reduces the manual effort of writing repetitive code.

Example Code:

// Define a macro to generate getter methods for a struct
macro_rules! generate_getters {
    ($struct_name:ident, $($field:ident),*) => {
        impl $struct_name {
            $(
                pub fn $field(&self) -> &str {
                    &self.$field
                }
            )*
        }
    };
}

struct Person {
    name: String,
    email: String,
}

generate_getters!(Person, name, email);

fn main() {
    let person = Person {
        name: "Alice".to_string(),
        email: "alice@example.com".to_string(),
    };
    println!("Name: {}", person.name());
    println!("Email: {}", person.email());
}

Limitations of Functions:

• Functions cannot generate multiple functions based on input at definition time, requiring manual writing of each getter method.
• Functions lack compile-time code generation and metaprogramming capabilities.

Coordination Between Macros and Functions:

• Macros generate code and create function implementations.
• Functions serve as the final callable entities generated by macros.

Scenario: Implementing Small Embedded DSLs

Problem Description:

• The need for more natural, domain-specific syntax to enhance readability and expressiveness.
• The desire to embed syntax structures similar to other languages, such as HTML or SQL, directly in code.

Macro Solution:

• Declarative macros can match specific syntax patterns and generate corresponding Rust code.
• Recursive pattern matching allows building embedded DSLs (Domain-Specific Languages).

Example Code:

// A simple HTML DSL macro
macro_rules! html {
    // Match a tag with inner content
    ($tag:ident { $($inner:tt)* }) => {
        format!("<{tag}>{content}</{tag}>", tag=stringify!($tag), content=html!($($inner)*))
    };
    // Match a text node
    ($text:expr) => {
        $text.to_string()
    };
    // Match multiple child nodes
    ($($inner:tt)*) => {
        vec![$(html!($inner)),*].join("")
    };
}

fn main() {
    let page = html! {
        html {
            head {
                title { "My Page" }
            }
            body {
                h1 { "Welcome!" }
                p { "This is a simple HTML page." }
            }
        }
    };
    println!("{}", page);
}

Limitations of Functions:

• Functions cannot accept or parse custom syntax structures; parameters must be valid Rust expressions.
• Functions cannot provide nested syntax in an intuitive way, leading to verbose and less readable code.

Coordination Between Macros and Functions:

• Macros parse custom syntax structures and convert them into Rust code.
• Functions execute the core logic, such as format! or string concatenation.

Procedural Macros

Procedural macros are a more powerful type of macro that can manipulate Rust’s Abstract Syntax Tree (AST) for complex code generation and transformation. They are mainly categorized into:

• Custom Derive Macros
• Attribute Macros
• Function-like Macros

Custom Derive Macros

Scenario: Automatically Implementing Traits for Types

Problem Description:

• Need to automatically implement a trait (e.g., Debug, Clone, Serialize, etc.) for multiple types to avoid writing repetitive code.
• Need to generate implementation code dynamically based on type attributes.

Macro Solution:

• Custom derive macros analyze type definitions at compile time and generate trait implementations accordingly.
• Common use cases include auto-deriving traits such as serde’s serialization/deserialization or the built-in Debug and Clone traits.

Example Code:

// Import necessary macro support
use serde::{Serialize, Deserialize};

// Use a custom derive macro to automatically implement Serialize and Deserialize
#[derive(Serialize, Deserialize)]
struct Person {
    name: String,
    age: u8,
}

fn main() {
    let person = Person {
        name: "Alice".to_string(),
        age: 30,
    };

    // Serialize to JSON string
    let json = serde_json::to_string(&person).unwrap();
    println!("Serialized: {}", json);

    // Deserialize back into a struct
    let deserialized: Person = serde_json::from_str(&json).unwrap();
    println!("Deserialized: {} is {} years old.", deserialized.name, deserialized.age);
}

Limitations of Functions:

• Functions cannot automatically generate trait implementations based on type definitions.
• Functions cannot inspect struct fields or attributes at compile time to generate relevant code.

Coordination Between Macros and Functions:

• Custom derive macros generate the required trait implementation code.
• Functions provide the logic for each trait’s behavior.

Attribute Macros

Scenario: Modifying Function or Type Behavior

Problem Description:

• Need to modify function or type behavior at compile time, such as automatically adding logging, performance profiling, or injecting additional logic.
• Prefer using annotations instead of manually modifying every function.

Macro Solution:

• Attribute macros can be attached to functions, types, or modules to modify or generate new code at compile time.
• These macros provide a flexible way to enhance code behavior without directly modifying function definitions.

Example Code:

// Define a simple attribute macro that prints logs before and after a function executes
use proc_macro::TokenStream;

#[proc_macro_attribute]
pub fn log_execution(_attr: TokenStream, item: TokenStream) -> TokenStream {
    let input = syn::parse_macro_input!(item as syn::ItemFn);
    let fn_name = &input.sig.ident;
    let block = &input.block;

    let expanded = quote::quote! {
        fn #fn_name() {
            println!("Entering function {}", stringify!(#fn_name));
            #block
            println!("Exiting function {}", stringify!(#fn_name));
        }
    };

    TokenStream::from(expanded)
}

// Use the attribute macro
#[log_execution]
fn my_function() {
    println!("Function body");
}

fn main() {
    my_function();
}

Limitations of Functions:

• Functions cannot modify their own execution behavior externally. They must manually include logging or profiling code.
• Functions do not have a built-in mechanism to inject behavior dynamically at compile time.

Coordination Between Macros and Functions:

• Attribute macros modify the function’s definition at compile time by injecting additional logic.
• Functions remain focused on their core business logic.

Function-like Macros

Scenario: Creating Custom Syntax or Code Generation

Problem Description:

• Need to accept specific input formats and generate corresponding Rust code, such as initializing configurations or generating routing tables.
• Want to use a function-like syntax (my_macro!(...)) to define custom logic.

Macro Solution:

• Function-like macros take TokenStream input, process it, and generate new Rust code.
• They are suitable for scenarios requiring complex parsing and code generation.

Example Code:

// Define a function macro that converts a string to uppercase at compile time
use proc_macro::TokenStream;

#[proc_macro]
pub fn make_uppercase(input: TokenStream) -> TokenStream {
    let s = input.to_string();
    let uppercased = s.to_uppercase();
    let output = quote::quote! {
        #uppercased
    };
    TokenStream::from(output)
}

// Use the function macro
fn main() {
    let s = make_uppercase!("hello, world!");
    println!("{}", s); // Output: HELLO, WORLD!
}

Limitations of Functions:

• Functions cannot modify string literals at compile time; all transformations happen at runtime.
• Runtime transformations have additional performance overhead compared to compile-time transformations.

Coordination Between Macros and Functions:

• Function-like macros generate required code or data at compile time.
• Functions operate on the generated code during runtime.

How to Choose Between Macros and Functions?

In practical development, the choice between macros and functions should be based on specific needs:

Prefer Functions When Possible

Whenever a problem can be solved with a function, functions should be the first choice due to their:

• Readability
• Maintainability
• Type safety
• Debugging and testing ease

Use Macros When Functions Are Insufficient

Use macros in scenarios where functions are not sufficient, such as:

• Handling a variable number and type of parameters (e.g., println!).
• Generating repetitive code at compile time to avoid boilerplate (e.g., auto-implementing getters).
• Creating embedded DSLs for domain-specific syntax (e.g., html!).
• Automatically implementing traits (e.g., #[derive(Serialize, Deserialize)]).
• Modifying code structure or behavior at compile time (e.g., #[log_execution]).

Situations Where Functions Are Preferable to Macros

• Handling complex business logic → Functions are better suited for implementing intricate logic and algorithms.
• Ensuring type safety and error checking → Functions have explicit type signatures, allowing Rust’s compiler to check for errors.
• Code readability and maintainability → Functions are structured and easier to understand than macros, which expand into complex code.
• Ease of debugging and testing → Functions can be unit-tested and debugged more easily than macros, which often produce obscure error messages.

Final Thoughts

By following these guidelines, you can make an informed decision on whether to use macros or functions in your Rust projects. Combining both effectively will help you write more efficient, maintainable, and scalable Rust code.

---

We are Leapcell, your top choice for hosting Rust projects.

Leapcell is the Next-Gen Serverless Platform for Web Hosting, Async Tasks, and Redis:

Multi-Language Support

• Develop with Node.js, Python, Go, or Rust.

Deploy unlimited projects for free

• pay only for usage — no requests, no charges.

Unbeatable Cost Efficiency

• Pay-as-you-go with no idle charges.
• Example: $25 supports 6.94M requests at a 60ms average response time.

Streamlined Developer Experience

• Intuitive UI for effortless setup.
• Fully automated CI/CD pipelines and GitOps integration.
• Real-time metrics and logging for actionable insights.

Effortless Scalability and High Performance

• Auto-scaling to handle high concurrency with ease.
• Zero operational overhead — just focus on building.

Explore more in the Documentation!

Follow us on X: @LeapcellHQ