All-purpose assert!(...)
and check!(...)
macros, inspired by Catch2.
There is also a debug_assert!(...)
macro that is disabled on optimized builds by default.
As cherry on top there is a let_assert!(...)
macro that lets you test a pattern while capturing parts of it.
These macros offer some benefits over the assertions from the standard library:
- The macros parse your expression to detect comparisons and adjust the error message accordingly.
No more
assert_eq
orassert_ne
, just writeassert!(1 + 1 == 2)
, or evenassert!(1 + 1 > 1)
! - You can test for pattern matches:
assert!(let Err(_) = File::open("/non/existing/file"))
. - You can capture parts of the pattern for further testing by using the
let_assert!(...)
macro. - The
check
macro can be used to perform multiple checks before panicking. - The macros provide more information when the assertion fails.
- Colored failure messages!
The macros also accept additional arguments for a custom message, so it is fully compatible with std::assert
.
That means you don't have to worry about overwriting the standard assert
with use assert2::assert
.
check!(6 + 1 <= 2 * 3);
check!(true && false);
check!(let Ok(_) = File::open("/non/existing/file"));
let_assert!(Err(e) = File::open("/non/existing/file"));
check!(e.kind() == ErrorKind::PermissionDenied);
The crate provides two macros: check!(...)
and assert!(...)
.
The main difference is that check
is really intended for test cases and doesn't immediately panic.
Instead, it will print the assertion error and fail the test.
This allows you to run multiple checks and can help to determine the reason of a test failure more easily.
The assert
macro on the other hand simply prints the error and panics,
and can be used outside of tests just as well.
Currently, check
uses a scope guard to delay the panic until the current scope ends.
Ideally, check
doesn't panic at all, but only signals that a test case has failed.
If this becomes possible in the future, the check
macro will change, so you should not rely on check
to panic.
You can also use the let_assert!(...)
.
It is very similar to assert!(let ...)
,
but all placeholders will be made available as variables in the calling scope.
This allows you to run additional checks on the captured variables.
For example:
let_assert!(Ok(foo) = Foo::try_new("bar"));
check!(foo.name() == "bar");
let_assert!(Err(Error::InvalidName(e)) = Foo::try_new("bogus name"));
check!(e.name() == "bogus name");
check!(e.to_string() == "invalid name: bogus name");
Colored output can be controlled using environment variables, as per the clicolors spec:
CLICOLOR != 0
: ANSI colors are supported and should be used when the program isn't piped.CLICOLOR == 0
: Don't output ANSI color escape codes.CLICOLOR_FORCE != 0
: ANSI colors should be enabled no matter what.