Testing
Testing is a good subject to learn now that we understand modules. Testing your code is very easy in Rust, because you can write tests right next to your code.
The easiest way to start testing is to add #[test]
above a function. Here is a simple one:
#![allow(unused)] fn main() { #[test] fn two_is_two() { assert_eq!(2, 2); } }
But if you try to run it in the Playground, it gives an error: error[E0601]: `main` function not found in crate `playground
. That's because you don't use Run for tests, you use Test. Also, you don't use a main()
function for tests - they go outside. To run this in the Playground, click on 路路路
next to RUN and change it to Test. Now if you click on it, it will run the test. (If you have Rust installed already, you will type cargo test
to do this)
Here is the output:
running 1 test
test two_is_two ... ok
test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out
Let's change assert_eq!(2, 2)
to assert_eq!(2, 3)
and see what we get. When a test fails you get a lot more information:
running 1 test
test two_is_two ... FAILED
failures:
---- two_is_two stdout ----
thread 'two_is_two' panicked at 'assertion failed: `(left == right)`
left: `2`,
right: `3`', src/lib.rs:3:5
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
failures:
two_is_two
test result: FAILED. 0 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out
assert_eq!(left, right)
is the main way to test a function in Rust. If it doesn't work, it will show the different values: left has 2, but right has 3.
What does RUST_BACKTRACE=1
mean? This is a setting on your computer to give a lot more information about errors. Luckily the Playground has it too: click on 路路路
next to STABLE
and set backtrace to ENABLED
. If you do that, it will give you a lot of information:
running 1 test
test two_is_two ... FAILED
failures:
---- two_is_two stdout ----
thread 'two_is_two' panicked at 'assertion failed: 2 == 3', src/lib.rs:3:5
stack backtrace:
0: backtrace::backtrace::libunwind::trace
at /cargo/registry/src/github.com-1ecc6299db9ec823/backtrace-0.3.46/src/backtrace/libunwind.rs:86
1: backtrace::backtrace::trace_unsynchronized
at /cargo/registry/src/github.com-1ecc6299db9ec823/backtrace-0.3.46/src/backtrace/mod.rs:66
2: std::sys_common::backtrace::_print_fmt
at src/libstd/sys_common/backtrace.rs:78
3: <std::sys_common::backtrace::_print::DisplayBacktrace as core::fmt::Display>::fmt
at src/libstd/sys_common/backtrace.rs:59
4: core::fmt::write
at src/libcore/fmt/mod.rs:1076
5: std::io::Write::write_fmt
at /rustc/c367798cfd3817ca6ae908ce675d1d99242af148/src/libstd/io/mod.rs:1537
6: std::io::impls::<impl std::io::Write for alloc::boxed::Box<W>>::write_fmt
at src/libstd/io/impls.rs:176
7: std::sys_common::backtrace::_print
at src/libstd/sys_common/backtrace.rs:62
8: std::sys_common::backtrace::print
at src/libstd/sys_common/backtrace.rs:49
9: std::panicking::default_hook::{{closure}}
at src/libstd/panicking.rs:198
10: std::panicking::default_hook
at src/libstd/panicking.rs:215
11: std::panicking::rust_panic_with_hook
at src/libstd/panicking.rs:486
12: std::panicking::begin_panic
at /rustc/c367798cfd3817ca6ae908ce675d1d99242af148/src/libstd/panicking.rs:410
13: playground::two_is_two
at src/lib.rs:3
14: playground::two_is_two::{{closure}}
at src/lib.rs:2
15: core::ops::function::FnOnce::call_once
at /rustc/c367798cfd3817ca6ae908ce675d1d99242af148/src/libcore/ops/function.rs:232
16: <alloc::boxed::Box<F> as core::ops::function::FnOnce<A>>::call_once
at /rustc/c367798cfd3817ca6ae908ce675d1d99242af148/src/liballoc/boxed.rs:1076
17: <std::panic::AssertUnwindSafe<F> as core::ops::function::FnOnce<()>>::call_once
at /rustc/c367798cfd3817ca6ae908ce675d1d99242af148/src/libstd/panic.rs:318
18: std::panicking::try::do_call
at /rustc/c367798cfd3817ca6ae908ce675d1d99242af148/src/libstd/panicking.rs:297
19: std::panicking::try
at /rustc/c367798cfd3817ca6ae908ce675d1d99242af148/src/libstd/panicking.rs:274
20: std::panic::catch_unwind
at /rustc/c367798cfd3817ca6ae908ce675d1d99242af148/src/libstd/panic.rs:394
21: test::run_test_in_process
at src/libtest/lib.rs:541
22: test::run_test::run_test_inner::{{closure}}
at src/libtest/lib.rs:450
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
failures:
two_is_two
test result: FAILED. 0 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out
You don't need to use a backtrace unless you really can't find where the problem is. But luckily you don't need to understand it all either. If you keep reading, you will eventually see line 13 where it says playground
- that's where it talks about your code. Everything else is about what Rust is doing in other libraries to run your program. But these two lines show you that it looked at line 2 and line 3 of playground, which is a hint to check there. Here's that part again:
13: playground::two_is_two
at src/lib.rs:3
14: playground::two_is_two::{{closure}}
at src/lib.rs:2
So let's turn backtrace off again and return to regular tests. Now we'll write some other functions, and use test functions to test them. Here are a few:
#![allow(unused)] fn main() { fn return_two() -> i8 { 2 } #[test] fn it_returns_two() { assert_eq!(return_two(), 2); } fn return_six() -> i8 { 4 + return_two() } #[test] fn it_returns_six() { assert_eq!(return_six(), 6) } }
Now it runs both:
running 2 tests
test it_returns_two ... ok
test it_returns_six ... ok
test result: ok. 2 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out
That's not too hard.
Usually you will want to put your tests in their own module. To do this, use the same mod
keyword and add #[cfg(test)]
above it (remember: cfg
means "configure). You also want to continue to write #[test]
above each test. This is because later on when you install Rust, you can do more complicated testing. You will be able to run one test, or all of them, or run a few. Also don't forget to write use super::*;
because the test module needs to use the functions above it. Now it will look like this:
#![allow(unused)] fn main() { fn return_two() -> i8 { 2 } fn return_six() -> i8 { 4 + return_two() } #[cfg(test)] mod tests { use super::*; #[test] fn it_returns_six() { assert_eq!(return_six(), 6) } #[test] fn it_returns_two() { assert_eq!(return_two(), 2); } } }
Test-driven development
You might see the words "test-driven development" when reading about Rust or another language. It's one way to write programs, and some people like it while others prefer something else. "Test-driven development" means "writing tests first, then writing the code". When you do this, you will have a lot of tests for everything you want your code to do. Then you start writing the code, and run the tests to see if you did it right. Then the tests are always there to show you if something goes wrong when you add to and rewrite your code. This is pretty easy in Rust because the compiler gives a lot of information about what to fix. Let's write a small example of test-driven development and see what it looks like.
Let's imagine a calculator that takes user input. It can add (+) and it can subtract (-). If the user writes "5 + 6" it should return 11, if the user writes "5 + 6 - 7" it should return 4, and so on. So we'll start with test functions. You can also see that function names in tests are usually quite long. That is because you might run a lot of tests, and you want to understand which tests have failed.
We'll imagine that a single function called math()
will do everything. It will return an i32
(we won't use floats). Because it needs to return something, we'll just return 6
every time. Then we will write three test functions. They will all fail, of course. Now the code looks like this:
#![allow(unused)] fn main() { fn math(input: &str) -> i32 { 6 } #[cfg(test)] mod tests { use super::*; #[test] fn one_plus_one_is_two() { assert_eq!(math("1 + 1"), 2); } #[test] fn one_minus_two_is_minus_one() { assert_eq!(math("1 - 2"), -1); } #[test] fn one_minus_minus_one_is_two() { assert_eq!(math("1 - -1), 2); } } }
It gives us this information:
running 3 tests
test tests::one_minus_minus_one_is_two ... FAILED
test tests::one_minus_two_is_minus_one ... FAILED
test tests::one_plus_one_is_two ... FAILED
and all the information about thread 'tests::one_plus_one_is_two' panicked at 'assertion failed: `(left == right)`
. We don't need to print it all here.
Now to think about how to make the calculator. We will accept any number, and the symbols +-
. We will allow spaces, but nothing else. So let's start with a const
that contains all the values. Then we will use .chars()
to iterate by character, and .all()
to make sure they are all inside.
Then we will add a test that should panic. To do that, add #[should_panic]
attribute: now if it panics the test will succeed.
Now the code looks like this:
#![allow(unused)] fn main() { const OKAY_CHARACTERS: &str = "1234567890+- "; // Don't forget the space at the end fn math(input: &str) -> i32 { if let false = input.chars().all(|character| OKAY_CHARACTERS.contains(character)) { panic!("Please only input numbers, +-, or spaces"); } 6 // we still return a 6 for now } #[cfg(test)] mod tests { use super::*; #[test] fn one_plus_one_is_two() { assert_eq!(math("1 + 1"), 2); } #[test] fn one_minus_two_is_minus_one() { assert_eq!(math("1 - 2"), -1); } #[test] fn one_minus_minus_one_is_two() { assert_eq!(math("1 - -1"), 2); } #[test] #[should_panic] // Here is our new test - it should panic fn panics_when_characters_not_right() { math("7 + seven"); } } }
Now when we run the tests we get this result:
running 4 tests
test tests::one_minus_two_is_minus_one ... FAILED
test tests::one_minus_minus_one_is_two ... FAILED
test tests::panics_when_characters_not_right ... ok
test tests::one_plus_one_is_two ... FAILED
One succeeded! Our math()
function will only accept good input now.
The next step is to write the actual calculator. This is the interesting part about having tests first: the actual code starts much later. First we will put the logic together for the calculator. We want the following:
- All empty spaces should be removed. This is easy with
.filter()
- The input should turn into a
Vec
with all the inputs.+
doesn't need to be an input, but when the program sees+
it should know that the number is done. For example, the input11+1
should do something like this: 1) See1
, push it into an empty string. 2) See another 1, push it into the string (it is now "11"). 3) See a+
, know the number has ended. It will push the string into the vec, then clear the string. - The program must count the number of
-
. An odd number (1, 3, 5...) will mean subtract, an even number (2, 4, 6...) will mean add. So "1--9" should give 10, not -8. - The program should remove anything after the last number.
5+5+++++----
is made out of all the characters inOKAY_CHARACTERS
, but it should turn to5+5
. This is easy with.trim_end_matches()
, where you remove anything that matches at the end of a&str
.
(By the way, .trim_end_matches()
and .trim_start_matches()
used to be trim_right_matches()
and trim_left_matches()
. But then people noticed that some languages go from right to left (Persian, Hebrew, etc.) so right and left were wrong. You might still see the older names in some code but they are the same thing.)
First we just want to pass all the tests. After we pass the tests, we can "refactor". Refactor means to make code better, usually through things like structs and enums and methods. Here is our code to make the tests pass:
#![allow(unused)] fn main() { const OKAY_CHARACTERS: &str = "1234567890+- "; fn math(input: &str) -> i32 { if let false = input.chars().all(|character| OKAY_CHARACTERS.contains(character)) { panic!("Please only input numbers, +-, or spaces"); } let input = input.trim_end_matches(|x| "+-".contains(x)).chars().filter(|x| *x != ' ').collect::<String>(); // Remove + and - at the end, and all spaces let mut result_vec = vec![]; // Results go in here let mut push_string = String::new(); // This is the string we push in every time. We will keep reusing it in the loop. for character in input.chars() { match character { '+' => { if !push_string.is_empty() { // If the string is empty, we don't want to push "" into result_vec result_vec.push(push_string.clone()); // But if it's not empty, it will be a number. Push it into the vec push_string.clear(); // Then clear the string } }, '-' => { // If we get a -, if push_string.contains('-') || push_string.is_empty() { // check to see if it's empty or has a - push_string.push(character) // if so, then push it in } else { // otherwise, it will contain a number result_vec.push(push_string.clone()); // so push the number into result_vec, clear it and then push - push_string.clear(); push_string.push(character); } }, number => { // number here means "anything else that matches". We selected the name here if push_string.contains('-') { // We might have some - characters to push in first result_vec.push(push_string.clone()); push_string.clear(); push_string.push(number); } else { // But if we don't, that means we can push the number in push_string.push(number); } }, } } result_vec.push(push_string); // Push one last time after the loop is over. Don't need to .clone() because we don't use it anymore let mut total = 0; // Now it's time to do math. Start with a total let mut adds = true; // true = add, false = subtract let mut math_iter = result_vec.into_iter(); while let Some(entry) = math_iter.next() { // Iter through the items if entry.contains('-') { // If it has a - character, check if it's even or odd if entry.chars().count() % 2 == 1 { adds = false; continue; // Go to the next item } else { continue; } } if adds == true { total += entry.parse::<i32>().unwrap(); // If there is no '-', it must be a number. So we are safe to unwrap } else { total -= entry.parse::<i32>().unwrap(); adds = true; // After subtracting, reset adds to true. } } total // Finally, return the total } /// We'll add a few more tests just to make sure #[cfg(test)] mod tests { use super::*; #[test] fn one_plus_one_is_two() { assert_eq!(math("1 + 1"), 2); } #[test] fn one_minus_two_is_minus_one() { assert_eq!(math("1 - 2"), -1); } #[test] fn one_minus_minus_one_is_two() { assert_eq!(math("1 - -1"), 2); } #[test] fn nine_plus_nine_minus_nine_minus_nine_is_zero() { assert_eq!(math("9+9-9-9"), 0); // This is a new test } #[test] fn eight_minus_nine_plus_nine_is_eight_even_with_characters_on_the_end() { assert_eq!(math("8 - 9 +9-----+++++"), 8); // This is a new test } #[test] #[should_panic] fn panics_when_characters_not_right() { math("7 + seven"); } } }
And now the tests pass!
running 6 tests
test tests::one_minus_minus_one_is_two ... ok
test tests::nine_plus_nine_minus_nine_minus_nine_is_zero ... ok
test tests::one_minus_two_is_minus_one ... ok
test tests::eight_minus_nine_plus_nine_is_eight_even_with_characters_on_the_end ... ok
test tests::one_plus_one_is_two ... ok
test tests::panics_when_characters_not_right ... ok
test result: ok. 6 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out
You can see that there is a back and forth process in test-driven development. It's something like this:
- First you write all the tests you can think of.
- Then you start writing the code.
- As you write the code, you get ideas for other tests.
- You add the tests, and your tests grow as you go. The more tests you have, the more times your code gets checked.
Of course, tests don't check everything and it is wrong to think that "passing all tests = the code is perfect". But tests are great for when you change your code. If you change your code later on and run the tests, if one of them doesn't work you will know what to fix.
Now we can rewrite (refactor) the code a bit. One good way to start is with clippy. If you installed Rust then you can type cargo clippy
, and if you're using the Playground then click on TOOLS
and select Clippy. Clippy will look at your code and give you tips to make it simpler. Our code doesn't have any mistakes, but it could be better.
Clippy tells us two things:
warning: this loop could be written as a `for` loop
--> src/lib.rs:44:5
|
44 | while let Some(entry) = math_iter.next() { // Iter through the items
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ help: try: `for entry in math_iter`
|
= note: `#[warn(clippy::while_let_on_iterator)]` on by default
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#while_let_on_iterator
warning: equality checks against true are unnecessary
--> src/lib.rs:53:12
|
53 | if adds == true {
| ^^^^^^^^^^^^ help: try simplifying it as shown: `adds`
|
= note: `#[warn(clippy::bool_comparison)]` on by default
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#bool_comparison
This is true: for entry in math_iter
is much simpler than while let Some(entry) = math_iter.next()
. And a for
loop is actually an iterator so we don't have any reason to write .iter()
. Thanks, clippy! And also we didn't need to make math_iter
: we can just write for entry in result_vec
.
And the second point is true too: if adds == true
can just be if adds
(because adds
= true
).
Now we'll start some real refactoring. Instead of separate variables, we will create a Calculator
struct. This will have all the variables we used together. We will change two names to make it more clear. result_vec
will become results
, and push_string
will become current_input
(current means "now"). And so far it only has one method: new.
#![allow(unused)] fn main() { // 馃毀 #[derive(Clone)] struct Calculator { results: Vec<String>, current_input: String, total: i32, adds: bool, } impl Calculator { fn new() -> Self { Self { results: vec![], current_input: String::new(), total: 0, adds: true, } } } }
Now our code is actually a bit longer, but easier to read. For example, if adds
is now if calculator.adds
, which is exactly like reading English. It looks like this:
#![allow(unused)] fn main() { #[derive(Clone)] struct Calculator { results: Vec<String>, current_input: String, total: i32, adds: bool, } impl Calculator { fn new() -> Self { Self { results: vec![], current_input: String::new(), total: 0, adds: true, } } } const OKAY_CHARACTERS: &str = "1234567890+- "; fn math(input: &str) -> i32 { if let false = input.chars().all(|character| OKAY_CHARACTERS.contains(character)) { panic!("Please only input numbers, +-, or spaces"); } let input = input.trim_end_matches(|x| "+-".contains(x)).chars().filter(|x| *x != ' ').collect::<String>(); let mut calculator = Calculator::new(); for character in input.chars() { match character { '+' => { if !calculator.current_input.is_empty() { calculator.results.push(calculator.current_input.clone()); calculator.current_input.clear(); } }, '-' => { if calculator.current_input.contains('-') || calculator.current_input.is_empty() { calculator.current_input.push(character) } else { calculator.results.push(calculator.current_input.clone()); calculator.current_input.clear(); calculator.current_input.push(character); } }, number => { if calculator.current_input.contains('-') { calculator.results.push(calculator.current_input.clone()); calculator.current_input.clear(); calculator.current_input.push(number); } else { calculator.current_input.push(number); } }, } } calculator.results.push(calculator.current_input); for entry in calculator.results { if entry.contains('-') { if entry.chars().count() % 2 == 1 { calculator.adds = false; continue; } else { continue; } } if calculator.adds { calculator.total += entry.parse::<i32>().unwrap(); } else { calculator.total -= entry.parse::<i32>().unwrap(); calculator.adds = true; } } calculator.total } #[cfg(test)] mod tests { use super::*; #[test] fn one_plus_one_is_two() { assert_eq!(math("1 + 1"), 2); } #[test] fn one_minus_two_is_minus_one() { assert_eq!(math("1 - 2"), -1); } #[test] fn one_minus_minus_one_is_two() { assert_eq!(math("1 - -1"), 2); } #[test] fn nine_plus_nine_minus_nine_minus_nine_is_zero() { assert_eq!(math("9+9-9-9"), 0); } #[test] fn eight_minus_nine_plus_nine_is_eight_even_with_characters_on_the_end() { assert_eq!(math("8 - 9 +9-----+++++"), 8); } #[test] #[should_panic] fn panics_when_characters_not_right() { math("7 + seven"); } } }
Finally we add two new methods. One is called .clear()
and clears the current_input()
. The other one is called push_char()
and pushes the input onto current_input()
. Here is our refactored code:
#![allow(unused)] fn main() { #[derive(Clone)] struct Calculator { results: Vec<String>, current_input: String, total: i32, adds: bool, } impl Calculator { fn new() -> Self { Self { results: vec![], current_input: String::new(), total: 0, adds: true, } } fn clear(&mut self) { self.current_input.clear(); } fn push_char(&mut self, character: char) { self.current_input.push(character); } } const OKAY_CHARACTERS: &str = "1234567890+- "; fn math(input: &str) -> i32 { if let false = input.chars().all(|character| OKAY_CHARACTERS.contains(character)) { panic!("Please only input numbers, +-, or spaces"); } let input = input.trim_end_matches(|x| "+-".contains(x)).chars().filter(|x| *x != ' ').collect::<String>(); let mut calculator = Calculator::new(); for character in input.chars() { match character { '+' => { if !calculator.current_input.is_empty() { calculator.results.push(calculator.current_input.clone()); calculator.clear(); } }, '-' => { if calculator.current_input.contains('-') || calculator.current_input.is_empty() { calculator.push_char(character) } else { calculator.results.push(calculator.current_input.clone()); calculator.clear(); calculator.push_char(character); } }, number => { if calculator.current_input.contains('-') { calculator.results.push(calculator.current_input.clone()); calculator.clear(); calculator.push_char(number); } else { calculator.push_char(number); } }, } } calculator.results.push(calculator.current_input); for entry in calculator.results { if entry.contains('-') { if entry.chars().count() % 2 == 1 { calculator.adds = false; continue; } else { continue; } } if calculator.adds { calculator.total += entry.parse::<i32>().unwrap(); } else { calculator.total -= entry.parse::<i32>().unwrap(); calculator.adds = true; } } calculator.total } #[cfg(test)] mod tests { use super::*; #[test] fn one_plus_one_is_two() { assert_eq!(math("1 + 1"), 2); } #[test] fn one_minus_two_is_minus_one() { assert_eq!(math("1 - 2"), -1); } #[test] fn one_minus_minus_one_is_two() { assert_eq!(math("1 - -1"), 2); } #[test] fn nine_plus_nine_minus_nine_minus_nine_is_zero() { assert_eq!(math("9+9-9-9"), 0); } #[test] fn eight_minus_nine_plus_nine_is_eight_even_with_characters_on_the_end() { assert_eq!(math("8 - 9 +9-----+++++"), 8); } #[test] #[should_panic] fn panics_when_characters_not_right() { math("7 + seven"); } } }
This is probably good enough for now. We could write more methods but lines like calculator.results.push(calculator.current_input.clone());
are already very clear. Refactoring is best when you can still easily read the code after you are done. You don't want to just refactor to make the code short: clc.clr()
is much worse than calculator.clear()
, for example.