yunit
is a data driven testing tool for .NET using markdown and YAML.
Data driven tests separate your test data from test logic, making it easy to add, modify and review test cases.
Markdown and YAML are popular formats designed to be writable and readable by humans, using these formats makes testing more enjoyable.
We have some samples to demonstrate the usage of yunit
, or you can follow these steps to scaffold a new test project:
- Create a new class library
dotnet new console
- Add NuGet reference to
yunit
andMicrosoft.NET.Test.Sdk
dotnet add package yunit
dotnet add package Microsoft.NET.Test.Sdk
- Create a serializable data contract to describe your test
public class HelloTestSpec
{
public string Input;
public string Output;
}
- Create a public method that takes the data contract with the corresponding verification code:
- Add
[YamlTest]
attribute to include test cases written in YAML format - Add
[MarkdownTest]
attribute to include test cases written in markdown format
💡 The glob pattern is relative to the current working directory, which could vary based on current build configuration and target platform. Use
~/
to denote path relative to the current git repository.
public class HelloTest
{
[YamlTest("~/*.yml")]
[MarkdownTest("~/README.md")]
public void Hello(HelloTestSpec spec)
{
Assert.Equal(spec.Output, $"Hello {spec.Input}");
}
}
- Run your tests using
dotnet test
or Visual Studio Test Explorer. All test cases run in parallel.
A YAML file can contain multiple YAML documents separated by ---
. Each YAML document is a test case.
The first line of comment appears as test description in Visual Studio test explorer.
# Hello YAML test!
input: YAML test!
output: Hello YAML test!
---
# Hello another YAML test
input: Another YAML test
output: Hello Another YAML test
A markdown test is a specialized YAML code block in an ordinary markdown file. A normal code block starts with 3 backticks (```
). A test code block starts with at least 6 backticks (``````
).
# This is an ordinary markdown file
Using markdown tests, you can enhance your docs with real examples that are correct and up to date.
``````yml
input: Markdown Test!
output: Hello Markdown Test!
``````
You can further customize how tests are extracted from markdown files in the [MarkdownTest]
attributes.
You can use any assertion library to verify your test expectations.
json-diff
is a built-in tool to write assertions in YAML that fits the yunit
testing style.
var jsonDiff = new JsonDiffBuilder().Build();
jsonDiff.Verify(expected, actual, "Test failed");
The idea of json-dff
is to serialize your expected object and actual object into JSON and compare them using a diff algorithm.
The bare metal comparison expect the actual object to match exactly the same as the expected object, but you can custimize the json-diff
pipeline to transform the expected and actual object before doing the text comparison. The built in json-diff
pipeline comes with several handy extensions:
method | description | example |
---|---|---|
UseIgnoreNull |
Ignore the actual result of a property if the expected value is null | { "a": null } matches {"a": xxx} where xxx could be anything |
UseNegate |
Assert the actual result must not be the expected result if the expected value starts with ! |
Given an expectation of { "a": "!value"} , `{"a": "value"} should fail |
UseRegex |
The value must match a regex if the expectation looks like /{regex}/ |
{"a": "/^.*$/" } uses ^.*$ to match the actual value |
UseWildcard |
Use wildcard match if the expectation contains * |
{"a": "a*" } uses wildcard a* to match against actual value |
UseJson |
The value must be a JSON string and its value is verified using the current pipeline, all the above rules will apply if they are configured in the pipeline | |
UseHtml |
The value must be an HTML string and its value is normalized before verification |
json-diff
allows you to write stateful data driven tests. The Shopping Cart Web API sample demonstrates a custom json-diff
extension that preserves states between multiple web API calls for a typical web API testing scenario
.