21 releases

0.8.0 Oct 4, 2023
0.7.1 Mar 22, 2022
0.7.0 Feb 22, 2022
0.6.3 May 7, 2020
0.2.2 Feb 21, 2019

#3 in #test-framework

Download history 35/week @ 2023-11-26 2/week @ 2024-01-14 1/week @ 2024-02-11 18/week @ 2024-02-18 33/week @ 2024-02-25 7/week @ 2024-03-03

59 downloads per month
Used in 3 crates

MIT/Apache

52KB
825 lines

Datatest: data-driven tests in Rust

crates.io Documentation Build Status

Crate for supporting data-driven tests.

Data-driven tests are tests where individual cases are defined via data rather than in code. This crate implements a custom test runner that adds support for additional test types.

Files-driven test

First type of data-driven tests are "file-driven" tests. These tests define a directory to scan for test data, a pattern (a regular expression) to match and, optionally, a set of templates to derive other file paths based on the matched file name. For each matched file, a new test instance is created, with test function arguments derived based on the specified mappings.

Each argument of the test function must be mapped either to the pattern or to the template. See the example below for the syntax.

The following argument types are supported:

  • &str, String: capture file contents as string and pass it to the test function
  • &[u8], Vec<u8>: capture file contents and pass it to the test function
  • &Path: pass file path as-is

Note

Each test could also be marked with #[test] attribute, to allow running test from IDEs which have built-in support for #[test] tests. However, if such attribute is used, it should go after #[datatest::files] attribute, so datatest attribute is handled earlier and #[test] attribute is removed.

Example

#![feature(custom_test_frameworks)]
#![test_runner(datatest::runner)]

#[datatest::files("tests/test-cases", {
  input in r"^(.*).input\.txt",
  output = r"${1}.output.txt",
})]
fn sample_test(input: &str, output: &str) {
  assert_eq!(format!("Hello, {}!", input), output);
}

Ignoring individual tests

Individual tests could be ignored by specifying a function of signature fn(&std::path::Path) -> bool using the following syntax on the pattern (if !<func_name>):

#![feature(custom_test_frameworks)]
#![test_runner(datatest::runner)]

fn is_ignore(path: &std::path::Path) -> bool {
  true // some condition
}

#[datatest::files("tests/test-cases", {
  input in r"^(.*).input\.txt" if !is_ignore,
  output = r"${1}.output.txt",
})]
fn sample_test(input: &str, output: &str) {
  assert_eq!(format!("Hello, {}!", input), output);
}

Data-driven tests

Second type of tests supported by this crate are "data-driven" tests. These tests define a YAML file with a list of test cases (via #[datatest::data(..)] attribute, see example below). Each test case in this file (the file contents must be an array) is deserialized into the argument type of the test function and a separate test instance is created for it.

Test function must take exactly one argument and the type of this argument must implement serde::Deserialize. Optionally, if this implements ToString (or std::fmt::Display), it's ToString::to_string result is used to generate test name.

#[test] attribute

Each test could also be marked with #[test] attribute, to allow running test from IDEs which have built-in support for #[test] tests. However, if such attribute is used, it should go after #[datatest::files] attribute, so datatest attribute is handled earlier and #[test] attribute is removed.

Example

#![feature(custom_test_frameworks)]
#![test_runner(datatest::runner)]

use serde::Deserialize;

#[derive(Deserialize)]
struct TestCase {
  name: String,
  expected: String,
}

#[datatest::data("tests/tests.yaml")]
fn sample_test(case: TestCase) {
  assert_eq!(case.expected, format!("Hi, {}!", case.name));
}

More examples

For more examples, check the tests.

Notes on Rust channel

Currently this crate targets primarily nightly Rust and can break at any time.

It could be compiled on stable by enabling certain feature (see Cargo.toml), but using this feature would subvert any stability guarantees Rust provides.

License

Licensed under either of

at your option.

Contribution

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.

Dependencies

~4–14MB
~164K SLoC