rstest
uses procedural macros to help you on writing
fixtures and table-based tests. To use it, add the
following lines to your Cargo.toml
file:
[dev-dependencies]
rstest = "0.5"
The core idea is that you can inject your test dependencies
by passing them as test arguments. In the following example
a fixture
is defined and then used in two tests,
simply indicating it as argument:
```rust use rstest::*;
pub fn fixture() -> u32 { 42 }
fn shouldsuccess(fixture: u32) { asserteq!(fixture, 42); }
fn shouldfail(fixture: u32) { assertne!(fixture, 42); } ```
You can also inject values in some other ways. For instance, you can
create a set of tests by simply indicating the injected values for each
case: rstest
will generate an independent test for each case.
```rust use rstest::rstest;
case(0, 0),
case(1, 1),
case(2, 1),
case(3, 2),
case(4, 3)
)] fn fibonaccitest(input: u32, expected: u32) { asserteq!(expected, fibonacci(input)) } ```
Running cargo test
in this case executes five tests:
```bash running 5 tests test fibonaccitest::case1 ... ok test fibonaccitest::case2 ... ok test fibonaccitest::case3 ... ok test fibonaccitest::case4 ... ok test fibonaccitest::case5 ... ok
test result: ok. 5 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out ```
If you need to just indicate a bunch of values for which you
need to run your test you can use var => [list, of, values]
syntax:
```rust use rstest::rstest;
value => [None, Some(""), Some(" ")]
)] fn shouldbeinvalid(value: Option<&str>) { assert!(!valid(value)) } ```
Or create a matrix test by using list of values for some variables that will generate the cartesian product of all the values.
All these features can be used together with mix fixture variables, fixed cases and bunch of values. For instance you need two tests that given your repository in cases of both logged in or guest user should return an invalid query error.
```rust use rstest::*;
fn repository() -> InMemoryRepository { let mut r = InMemoryRepository::default(); // fill repository by some data r }
fn alice() -> User { User::logged("Alice", "2001-10-04", "London", "UK") }
case::logged_user(alice()), // We can use `fixture` also as standard function
case::guest(User::Guest), // We can give a name to every case : `guest` in this case
query => [" ", "^%$#@!", "...." ]
)]
fn shouldbeinvalidqueryerror(repository: impl Repository, user: User, query: &str) { repository.find_items(&user, query).unwrap(); } ```
This example will generate exactly 6 tests grouped by 2 different cases:
``` running 6 tests test shouldbeinvalidqueryerror::case1loggeduser::query1 ... ok test shouldbeinvalidqueryerror::case2guest::query2 ... ok test shouldbeinvalidqueryerror::case2guest::query3 ... ok test shouldbeinvalidqueryerror::case1loggeduser::query2 ... ok test shouldbeinvalidqueryerror::case1loggeduser::query3 ... ok test shouldbeinvalidqueryerror::case2guest::query_1 ... ok
test result: ok. 6 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out ```
Is that all? Not yet!
A fixture can be injected by another fixture and they can be called using just some of its arguments.
```rust
fn name() -> &'static str { "Alice" }
fn age() -> u8 { 22 }
fn user(name: &str, age: u8) -> User { User::new(name, age) }
fn isalice(user: User) { asserteq!(user.name(), "Alice") }
fn is22(user: User) { asserteq!(user.age(), 22) }
fn isbob(user: User) { asserteq!(user.name(), "Bob") }
fn is42(user: User) { asserteq!(user.age(), 42) } ```
Currently, using a fixture is required also to just provide default value, but this will change soon with the introduction of a syntax for default values, without the need of the fixture function definition.
Finally if you need tracing the input values you can just
add the trace
attribute to your test to enable the dump of all input
variables.
```rust
number, name, tuple,
case(42, "FortyTwo", ("minus twelve", -12)),
case(24, "TwentyFour", ("minus twentyfour", -24))
::trace //This attribute enable traceing
)] fn should_fail(number: u32, name: &str, tuple: (&str, i32)) { assert!(false); // <- stdout come out just for failed tests } ```
``` running 2 tests test shouldfail::case1 ... FAILED test shouldfail::case2 ... FAILED
failures:
---- shouldfail::case1 stdout ----
------------ TEST ARGUMENTS ------------
number = 42
name = "FortyTwo"
tuple = ("minus twelve", -12)
-------------- TEST START --------------
thread 'shouldfail::case1' panicked at 'assertion failed: false', src/main.rs:64:5
note: run with RUST_BACKTRACE=1
environment variable to display a backtrace.
---- shouldfail::case2 stdout ---- ------------ TEST ARGUMENTS ------------ number = 24 name = "TwentyFour" tuple = ("minus twentyfour", -24) -------------- TEST START -------------- thread 'shouldfail::case2' panicked at 'assertion failed: false', src/main.rs:64:5
failures: shouldfail::case1 shouldfail::case2
test result: FAILED. 0 passed; 2 failed; 0 ignored; 0 measured; 0 filtered out ```
In case one or more variables don't implement the Debug
trait, an error
is raised, but it's also possible to exclude a variable using the
notrace(var,list,that,not,implement,Debug)
attribute.
You can learn more on Docs and find more
examples in resources
directory and in
rs8080
which uses this module in-depth.
See CHANGELOG.md
Licensed under either of
Apache License, Version 2.0, (LICENSE-APACHE or [license-apache-link])
MIT license (LICENSE-MIT or [license-MIT-link]) at your option.