23 KiB
[TOC]
Modules
When you write a program in Rust, your code might start off living solely in
the main
function. As your code grows, you eventually move functionality out
into functions, both for re-use and for nicer organization. By splitting your
code up into smaller chunks, each chunk is easier to understand on its own. So
what happens when you start having too many functions? Rust has a module system
that tackles both the problem of wanting to be able to re-use code and the
problem of keeping your code organized.
In the same way that you extract lines of code into a function, you can extract functions (and other code like structs and enums too) into different modules. A module is a namespace that contains definitions of functions or types, and those definitions can be visible outside their module or not. Here's an overview of how the bits fit together:
mod
declares a new module.- Everything starts off as private, but the
pub
keyword makes it public. - The
use
keyword allows you to bring modules, or definitions inside of them, into scope so that it's easier to refer to them.
We'll take a look at each of these parts and see how they fit into the whole.
mod
and the Filesystem
Every module in Rust starts with the mod
keyword. In this next example, we'll
start again by making a new project with Cargo. This time, instead of a binary,
we're going to make a library: a project that other people would pull into their
projects as a dependency. We saw this with the rand
crate in Chapter 2.
Imagine that we're creating a library to provide some general networking
functionality, and we decide to call our library communicator
. To create this
library, we won't use the --bin
option like we have before. This is because
by default cargo will create a library:
$ cargo new communicator
$ cd communicator
Notice that Cargo generated src/lib.rs
instead of src/main.rs
for us, and
inside it we'll find this:
Filename: src/lib.rs
#[cfg(test)]
mod tests {
#[test]
fn it_works() {
}
}
This is an empty test to help us get our library started, instead of the binary
that says "Hello, world!" that we get with the --bin
option. Let's ignore the
#[]
stuff and mod tests
for a little bit, but we'll make sure to leave it
in src/lib.rs
for later.
We're going to look at different ways we could choose to organize our library's code, any of which could make sense depending on exactly what we were trying to do. To start, add this code at the beginning of the file:
Filename: src/lib.rs
mod network {
fn connect() {
}
}
This is our first module declaration. We use the mod
keyword, followed by the
name of the module, and then a block of code in curly braces. Everything inside
this block is inside the namespace network
. In this case, we have a single
function, connect
. If we wanted to try and call this function from outside
the network
module, we would say network::connect()
rather than connect()
.
We could have multiple modules, side-by-side. For example, if we wanted a
client
module:
Filename: src/lib.rs
mod network {
fn connect() {
}
}
mod client {
fn connect() {
}
}
Now we have a network::connect
function and a client::connect
function.
And we can put modules inside of modules. If we wanted to have client
be
within network
:
Filename: src/lib.rs
mod network {
fn connect() {
}
mod client {
fn connect() {
}
}
}
This gives us network::connect
and network::client::connect
.
In this way, modules form a tree. The contents of src/lib.rs
are at the root
of the project's tree, and the submodules form the leaves. Here's what our
first example looks like when thought of this way:
communicator
├── network
└── client
And here's the second:
communicator
└── network
└── client
More complicated projects can have a lot of modules.
Putting Modules in Another File
Modules form a hierarchical, tree-like structure. So does another thing: file systems! The module system is the way that we split larger Rust projects up into multiple files. Let's imagine we have a module layout like this:
File: src/lib.rs
mod client {
fn connect() {
}
}
mod network {
fn connect() {
}
mod server {
fn connect() {
}
}
}
Let's extract the client
module into another file. First, we need to change
our code in src/lib.rs
:
File: src/lib.rs
mod client;
mod network {
fn connect() {
}
mod server {
fn connect() {
}
}
}
We still say mod client
, but instead of curly braces, we have a semicolon.
This lets Rust know that we have a module, but it's in another file with that
module's name. Open up src/client.rs
and put this in it:
File: src/client.rs
fn connect() {
}
Note that we don't need a mod
declaration in this file. mod
is for
declaring a new module, and we've already declared this module in src/lib.rs
.
This file provides the contents of the client
module. If we put a mod client
here, we'd be giving the client
module its own submodule named
client
!
Now, everything should compile successfully, but with a few warnings:
$ cargo build
Compiling communicator v0.1.0 (file:///projects/communicator)
warning: function is never used: `connect`, #[warn(dead_code)] on by default
--> src/client.rs:1:1
|
1 | fn connect() {
| ^
warning: function is never used: `connect`, #[warn(dead_code)] on by default
--> src/lib.rs:4:5
|
4 | fn connect() {
| ^
warning: function is never used: `connect`, #[warn(dead_code)] on by default
--> src/lib.rs:8:9
|
8 | fn connect() {
| ^
Don't worry about those warnings for now; we'll clear them up in a future section. They're just warnings, we've built things successfully!
Let's extract the network
module into its own file next, using the same
pattern. Change src/lib.rs
to look like this:
Filename: src/lib.rs
mod client;
mod network;
And then put this in src/network.rs
Filename: src/network.rs
fn connect() {
}
mod server {
fn connect() {
}
}
And then run cargo build
again. Success! We have one more module to extract:
server
. Unfortunately, our current tactic of extracting a module into a file
named after that module won't work. Let's try it anyway. Modify
src/network.rs
to look like this:
Filename: src/network.rs
fn connect() {
}
mod server;
Put this in src/server.rs
Filename: src/server.rs
fn connect() {
}
When we try to cargo build
, we'll get an error:
$ cargo build
Compiling communicator v0.1.0 (file:///projects/communicator)
error: cannot declare a new module at this location
--> src/network.rs:4:5
|
4 | mod server;
| ^^^^^^
|
note: maybe move this module `network` to its own directory via `network/mod.rs`
--> src/network.rs:4:5
|
4 | mod server;
| ^^^^^^
note: ... or maybe `use` the module `server` instead of possibly redeclaring it
--> src/network.rs:4:5
|
4 | mod server;
| ^^^^^^
This error is actually pretty helpful. It points out something we didn't know that we could do yet:
note: maybe move this module
network
to its own directory vianetwork/mod.rs
Here's the problem: in our case, we have different names for our modules:
client
and network::server
. But what if we had client
and
network::client
, or server
and network::server
? Having two modules at
different places in the module hierarchy have the same name is completely
valid, but then which module would the files src/client.rs
and
src/server.rs
, respectively, be for?
Instead of continuing to follow the same file naming pattern we used
previously, we can do what the error suggests. We'll make a new directory,
move src/server.rs
into it, and change src/network.rs
to
src/network/mod.rs
. Then, when we try to build:
$ mkdir src/network
$ mv src/server.rs src/network
$ mv src/network.rs src/network/mod.rs
$ cargo build
Compiling communicator v0.1.0 (file:///projects/communicator)
<warnings>
$
It works! So now our module layout looks like this:
communicator
├── client
└── network
└── server
And the corresponding file layout looks like this:
├── src
│ ├── client.rs
│ ├── lib.rs
│ └── network
│ ├── mod.rs
│ └── server.rs
In summary, these are the rules of modules with regards to files:
-
If a module named
foo
has no submodules, you should put the declarations in thefoo
module in a file namedfoo.rs
. -
If a module named
foo
does have submodules, you should put the declarations forfoo
in a file namedfoo/mod.rs
. -
The first two rules apply recursively, so that if a module named
foo
has a submodule namedbar
andbar
does not have submodules, you should have the following files in yoursrc
directory:├── foo │ ├── bar.rs (contains the declarations in `foo::bar`) │ └── mod.rs (contains the declarations in `foo`, including `mod bar`)
-
The modules themselves should be declared in their parent module's file using the
mod
keyword.
Next, we'll talk about the pub
keyword, and get rid of those warnings!
Controlling Visibility with pub
At the end of the last section, we had a project, communicator
, and when we compiled it, we got some strange warnings:
Compiling communicator v0.1.0 (file:///projects/communicator)
warning: function is never used: `connect`, #[warn(dead_code)] on by default
--> src/client.rs:1:1
|
1 | fn connect() {
| ^
warning: function is never used: `connect`, #[warn(dead_code)] on by default
--> src/network/mod.rs:1:1
|
1 | fn connect() {
| ^
warning: function is never used: `connect`, #[warn(dead_code)] on by default
--> src/network/server.rs:1:1
|
1 | fn connect() {
| ^
Why does this happen? After all, we're building a library. What if these three
functions are the public interface that we want our users to use? We won't
necessarily be using them within our own project, but the point of creating them
is that they will be used by another project. Let's try using them as if we
were another project using our library to see what happens and understand why
we're getting these unused function warnings. Create a src/main.rs
file with
this code:
Filename: src/main.rs
extern crate communicator;
fn main() {
communicator::client::connect();
}
We need the extern crate
line to bring our communicator
library crate into
scope, because our package actually now contains two crates. Cargo treats
src/main.rs as the crate root of a binary crate, and we also have our existing
library crate. This pattern is quite common for executable crates: most
functionality is in a library crate, and the executable crate uses that
library. This way, other programs can also use the library crate, and it’s a
nice separation of concerns.
Our binary crate right now just calls our library's connect
function from
the client
module; we picked that one since it's the first warning in our
build output above. Invoking cargo build
will now give us an error after the
warnings:
$ cargo build
Compiling communicator v0.1.0 (file:///projects/communicator)
error: module `client` is private
--> src/main.rs:4:5
|
4 | communicator::client::connect();
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Ah ha! The client
module is private. This is the first time we've run into
the concepts of 'public' and 'private' in the context of Rust. There's no
keyword to make something private; that's the default state. In this default
state, no one else could possibly use it, so if we don't use it within our
library crate, Rust will warn us that it's unused. Once we tell Rust something
is public, Rust knows that we intend for code external to our crate to use it,
and Rust considers theoretical external usage that is now possible to count as
being used. Thus, when something is marked as public, Rust will stop warning us
that it is unused.
To tell Rust we want to make something public, we add the pub
keyword. This
keyword goes before the declaration of the item we want to make public. Let's
modify src/lib.rs
to make the client
module public to fix the error we got:
Filename: src/lib.rs
pub mod client;
mod network;
The pub
goes right before mod
. Let's try building again:
$ cargo build
Compiling communicator v0.1.0 (file:///projects/communicator)
<warnings>
error: function `connect` is private
--> src/main.rs:4:5
|
4 | communicator::client::connect();
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Hooray! We have a different error! Yes, different error messages are a cause
for celebration. The new error says "function connect
is private", so let's
edit src/client.rs
to make client::connect
public:
Filename: src/client.rs
pub fn connect() {
}
And run cargo build
again:
cargo build
Compiling communicator v0.1.0 (file:///projects/communicator)
warning: function is never used: `connect`, #[warn(dead_code)] on by default
--> src/network/mod.rs:1:1
|
1 | fn connect() {
| ^
warning: function is never used: `connect`, #[warn(dead_code)] on by default
--> src/network/server.rs:1:1
|
1 | fn connect() {
| ^
It compiled! And the warning about client::connect
not being used is gone!
Making functions public isn't the only way to fix unused code warnings: if we didn't want these functions to be part of our public API and we got these warnings, the warnings could be alerting us to code we no longer needed and could safely delete. They could also be alerting us to a bug, if we had just accidentally removed all places within our library where we called this function.
However, we do want the other two functions to be part of our crate's public
API, so let's mark them as pub
as well to get rid of the remaining warnings.
Modify src/network/mod.rs
to be:
Filename: src/network/mod.rs
pub fn connect() {
}
mod server;
And compile:
$ cargo build
Compiling communicator v0.1.0 (file:///projects/communicator)
warning: function is never used: `connect`, #[warn(dead_code)] on by default
--> src/network/mod.rs:1:1
|
1 | pub fn connect() {
| ^
warning: function is never used: `connect`, #[warn(dead_code)] on by default
--> src/network/server.rs:1:1
|
1 | fn connect() {
| ^
Hmmm, it says this is still dead, even though it's pub
. While the function is
public within the module, the network
module it's in is not public. We're
working from the interior of the library out this time, as opposed to with
client
where we worked from the outside in. Let's change src/lib.rs
to add
the same fix though, by making network
public like client
is:
Filename: src/lib.rs
pub mod client;
pub mod network;
Now if we compile, that warning is gone:
$ cargo build
Compiling communicator v0.1.0 (file:///projects/communicator)
warning: function is never used: `connect`, #[warn(dead_code)] on by default
--> src/network/server.rs:1:1
|
1 | fn connect() {
| ^
Only one last warning! Try to fix this one on your own!
Privacy Rules
Overall, these are the rules for item visibility:
- If an item is public, then it can be accessed through any of its parent modules.
- If an item is private, it may be accessed by the current module and its child modules.
Let's look at a few more examples to get some practice. What if we had this
code in a new project's src/lib.rs
:
Filename: src/lib.rs
mod outermost {
pub fn middle_function() {}
fn middle_secret_function() {}
mod inside {
pub fn inner_function() {}
fn secret_function() {}
}
}
fn try_me() {
outermost::middle_function();
outermost::middle_secret_function();
outermost::inside::inner_function();
outermost::inside::secret_function();
}
Before you try to compile this code, make a guess about which lines in
try_me
will have errors.
Ready? Let's talk through them!
The try_me
function is in the root module of our project. The module named
outermost
is private, but the second rule says we're allowed to access it
since outermost
is in our current, root module.
The function call outermost::middle_function()
will work. middle_function
is public, and we are accessing it through its parent module, outermost
,
which we just determined we can access in the previous paragraph.
outermost::middle_secret_function()
will cause a compilation error.
middle_secret_function
is private, so the second rule applies. Our current
root module is neither the current module of middle_secret_function
(outermost
is), nor is it a child module of the current module of
middle_secret_function
.
The module named inside
is private and has no child modules, so it can only
be accessed by its current module, outermost
. That means the try_me
function is not allowed to call outermost::inside::inner_function()
or
outermost::inside::secret_function()
.
Here are some changes to try making with this code. Try each one, make a guess about what will be allowed or not, compile to see if you're right, and use the rules to understand why.
- What if the
inside
module was public? - What if
outside
was public andinside
was private? - What if, in the body of
inner_function
, we called::outermost::middle_secret_function()
? (The two colons at the beginning mean that we want to refer to the namespaces starting from the root namespace.)
Feel free to design more experiments and try them out!
Next, let's talk about bringing items into a scope with the use
keyword.
Importing Names with use
We've seen how we can call functions defined within a module by using the module name as part of the call, like this:
Filename: src/main.rs
pub mod a {
pub mod series {
pub mod of {
pub fn namespaces() {}
}
}
}
fn main() {
a::series::of::namespaces();
}
However, referring to the fully qualified name can get quite lengthy, as we see
in that example. To solve this issue, Rust has a keyword, use
. It works like
this:
Filename: src/main.rs
pub mod a {
pub mod series {
pub mod of {
pub fn namespaces() {}
}
}
}
use a::series::of;
fn main() {
of::namespaces();
}
We can use
a module, and that will bring its name into scope. This allows us
to shorten our function call, only requiring us to type the final module name,
not the entire chain of them. use
is quite powerful and can bring all kinds
of things into scope. For example, we could use
the function itself:
pub mod a {
pub mod series {
pub mod of {
pub fn namespaces() {}
}
}
}
use a::series::of::namespaces;
fn main() {
namespaces();
}
Enums also form this kind of namespace; we can import an enum's variants with
use
as well. For any kind of use
statement, if you are importing multiple
items from one namespace, you can list them using curly braces and commas in
the last position, like so:
enum TrafficLight {
Red,
Yellow,
Green,
}
use TrafficLight::{Red, Yellow};
fn main() {
let red = Red;
let yellow = Yellow;
let green = TrafficLight::Green; // because we didn't use TrafficLight::Green
}
Glob Imports with *
If you'd like to import all the items in a namespace at once, you can use *
:
enum TrafficLight {
Red,
Yellow,
Green,
}
use TrafficLight::*;
fn main() {
let red = Red;
let yellow = Yellow;
let green = Green;
}
The *
is called a 'glob', and it will import everything that's visible inside
of the namespace. Globs should be used sparingly: they are convenient, but you
might also pull in more things than you expected and cause naming conflicts.
Using super
to Access a Parent Module
Remember when we created our crate that Cargo made a tests
module for us?
Let's talk about that now. It was in src/lib.rs
:
Filename: src/lib.rs
pub mod client;
pub mod network;
#[cfg(test)]
mod tests {
#[test]
fn it_works() {
}
}
We'll explain more about testing in Chapter XX, but parts of this should make
sense now: we have a module named tests
that lives next to our other modules
and contains one function named it_works
. Even though there are special
annotations, the tests
module is just another module!
Since tests are for exercising the code within our library, let's try to call
our client::connect
function from this it_works
function, even though
we're not going to be checking any functionality right now:
Filename: src/lib.rs
#[cfg(test)]
mod tests {
#[test]
fn it_works() {
client::connect();
}
}
Run the tests by invoking the cargo test
command:
$ cargo test
Compiling communicator v0.1.0 (file:///projects/communicator)
error[E0433]: failed to resolve. Use of undeclared type or module `client`
--> src/lib.rs:9:9
|
9 | client::connect();
| ^^^^^^^^^^^^^^^ Use of undeclared type or module `client`
warning: function is never used: `connect`, #[warn(dead_code)] on by default
--> src/network/server.rs:1:1
|
1 | fn connect() {
| ^
Why doesn't this compile? It's not because we don't have communicator::
in
front of the function like we had in src/main.rs
: we are definitely within
the communicator
library crate here. The reason is that paths anywhere except
in a use
statement are relative to the current module (In a use
statement,
they're relative to the crate root by default). Our tests
module doesn't have
a client
module in its scope!
So how do we get back up one module? We can either use leading colons to say that we want to start from the root and list the whole path:
::client::connect();
Or we can use super
to move up one module in the hierarchy:
super::client::connect();
If we were deep in the module hierarchy, starting from the root every time would get long. Plus, if we rearrange our modules by moving a subtree to another place, there might be a lot of places the path would need to be updated if we always used the path from the root.
It would also be annoying to have to type super::
all the time in each test,
but we now have a tool for that solution: use
! super::
is special and
changes the path we give to use
so that it is relative to the parent module
instead of to the root module.
For these reasons, in the tests
module especially, use super::something
is
usually the way to go. So now our test looks like this:
Filename: src/lib.rs
#[cfg(test)]
mod tests {
use super::client;
#[test]
fn it_works() {
client::connect();
}
}
If we run cargo test
again, the test will pass and the first part of the test
result output will be:
$ cargo test
Compiling communicator v0.1.0 (file:///projects/communicator)
Running target/debug/communicator-92007ddb5330fa5a
running 1 test
test tests::it_works ... ok
test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured
Now you know techniques for organizing your code! Use these to group related functionality together, keep files from getting too long, and present a tidy public API to users of your library.