This is a library for accessing Unix users and groups. It supports getting the system users and groups, storing them in a cache, and creating your own mock tables.
This crate works with Cargo. Add the following to your Cargo.toml
dependencies section:
[dependencies]
users = "0.11"
The earliest version of Rust that this crate is tested against is Rust v1.31.0.
In Unix, each user has an individual user ID, and each process has an effective user ID that says which user’s permissions it is using.
Furthermore, users can be the members of groups, which also have names and IDs.
This functionality is exposed in libc, the C standard library, but as an unsafe Rust interface.
This wrapper library provides a safe interface, using User
and Group
types and functions such as get_user_by_id
instead of low-level pointers and strings.
It also offers basic caching functionality.
It does not (yet) offer editing functionality; the values returned are read-only.
The function get_current_uid
returns a uid_t
value representing the user currently running the program, and the get_user_by_uid
function scans the users database and returns a User
with the user’s information.
This function returns None
when there is no user for that ID.
A User
has the following accessors:
- uid: The user’s ID
- name: The user’s name
- primary_group: The ID of this user’s primary group
Here is a complete example that prints out the current user’s name:
use users::{get_user_by_uid, get_current_uid};
let user = get_user_by_uid(get_current_uid()).unwrap();
println!("Hello, {}!", user.name());
This code assumes (with unwrap()
) that the user hasn’t been deleted after the program has started running.
For arbitrary user IDs, this is not a safe assumption: it’s possible to delete a user while it’s running a program, or is the owner of files, or for that user to have never existed.
So always check the return values!
There is also a get_current_username
function, as it’s such a common operation that it deserves special treatment.
Despite the above warning, the users and groups database rarely changes. While a short program may only need to get user information once, a long-running one may need to re-query the database many times, and a medium-length one may get away with caching the values to save on redundant system calls.
For this reason, this crate offers a caching interface to the database, which offers the same functionality while holding on to every result, caching the information so it can be re-used.
To introduce a cache, create a new UsersCache
and call the same methods on it.
For example:
use users::{Users, Groups, UsersCache};
let mut cache = UsersCache::new();
let uid = cache.get_current_uid();
let user = cache.get_user_by_uid(uid).unwrap();
println!("Hello again, {}!", user.name());
This cache is only additive: it’s not possible to drop it, or erase selected entries, as when the database may have been modified, it’s best to start entirely afresh.
So to accomplish this, just start using a new UsersCache
.
Finally, it’s possible to get groups in a similar manner.
A Group
has the following accessors:
- gid: The group’s ID
- name: The group’s name
And again, a complete example:
use users::{Users, Groups, UsersCache};
let mut cache = UsersCache::new();
let group = cache.get_group_by_name("admin").expect("No such group 'admin'!");
println!("The '{}' group has the ID {}", group.name(), group.gid());
The logging
feature, which is on by default, uses the log
crate to record all interactions with the operating system at Trace log level.
You should be prepared for the users and groups tables to be completely broken: IDs shouldn’t be assumed to map to actual users and groups, and usernames and group names aren’t guaranteed to map either!
When you’re testing your code, you don’t want to actually rely on the system actually having various users and groups present - it’s much better to have a custom set of users that are guaranteed to be there, so you can test against them.
The mock
module allows you to create these custom users and groups definitions, then access them using the same Users
trait as in the main library, with few changes to your code.
The only thing a mock users table needs to know in advance is the UID of the current user.
Aside from that, you can add users and groups with add_user
and add_group
to the table:
use std::sync::Arc;
use users::mock::{MockUsers, User, Group};
use users::os::unix::{UserExt, GroupExt};
let mut users = MockUsers::with_current_uid(1000);
let bobbins = User::new(1000, "Bobbins", 1000).with_home_dir("/home/bobbins");
users.add_user(bobbins);
users.add_group(Group::new(100, "funkyppl"));
The exports get re-exported into the mock module, for simpler use
lines.
To set your program up to use either type of Users
table, make your functions and structs accept a generic parameter that implements the Users
trait.
Then, you can pass in a value of either OS or Mock type.
Here’s a complete example:
use std::sync::Arc;
use users::{Users, UsersCache, User};
use users::os::unix::UserExt;
use users::mock::MockUsers;
fn print_current_username<U: Users>(users: &mut U) {
println!("Current user: {:?}", users.get_current_username());
}
let mut users = MockUsers::with_current_uid(1001);
users.add_user(User::new(1001, "fred", 101));
print_current_username(&mut users);
let mut actual_users = UsersCache::new();
print_current_username(&mut actual_users);