Installation
Just include htest on HTML pages with the appropriate format (see below).
<link rel="stylesheet" href="https://htest.dev/htest.css" crossorigin />
<script src="https://htest.dev/htest.js" crossorigin></script>
In the wild
Testsuites
Single page tests
Creating new tests
Our primary test format is reftests, i.e. automatic comparison of two things, typically app output with expected (reference) output. When the two match, the test passes (green), otherwise it fails (red).
You can create reftests by using a table with class="reftest"
. Each row is a new test. Typically these tables have two columns: output and expected. However, if your test requires initialization data, you can also have 3 columns, and the first one will be ignored in the matching.
Typically these tables are inside sections with HTML like the following:
<section>
<h1>Heading</h1>
<table class="reftest">
<!-- ...tests as <tr>s -->
</table>
</section>
This structure is not necessary for the reftests to work, but it allows you to isolate specific sections, which is often convenient when debugging. You can also isolate an individual row by Alt + double clicking.
Below is a short description of the syntax we support.
data-test
attribute
This applies to either the whole table or individual tests and controls how the matching is done.
By default the contents of the cells are compared which corresponds to data-test="contents"
.
Other useful values are:
"selector"
which treats the reference cell as a selector (or list of a selectors, if using a<ul>
) that the output HTML must match. You can reverse the matching withclass="not"
."numbers"
which only compares the numbers returned, ignoring all other output. You can specify an epsilon value by using adata-epsilon
attribute, either on the row or an ancestor."dom"
which compares both contents and attributes"attribute"
???
Besides the built-in comparison functions, you can provide your own, by defining a JavaScript function, either as the content of the data-test
attribute, or by defining a global function.
It accepts the 2-3 cells of your test as arguments and should return a truthy value for pass and a falsy value for fail.
data-columns
attribute
2 by default, which means there are two columns: ref and test. Set it to 3 or more for custom tests that need the previous columns for data.
data-click
attribute
Automatic clicking on elements. It can be placed on either the whole table or individual tests. Its location specifies the root for the selector, if one is specified. Its syntax is (angle brackets indicate a parameter, square brackets mean that something is optional):
[<selector>] [wait <delay>s] [after <event name>] [<times> times]
The parameters can be specified in any order.
Examples:
data-click=""
: Clicks the element it's specified on immediately on DOMContentLoadeddata-click=".foo"
: Clicks.foo
elements immediately on DOMContentLoadeddata-click=".foo .bar wait 5s after mv-load"
: Clicks.foo .bar
elements 5 seconds after themv-load
event firesdata-click=".foo 3 times after hashchange"
: Clicks.foo
elements 3 times after thehashchange
eventdata-click="wait 1s after load"
: Clicks the element it's specified on 1 second after theload
event fires.data-click="wait 1s after load 2 times"
: Same as above, but clicks twice.
data-error
attribute
Use on tests that should produce an error to pass.
Use the data-error
attribute on the <tr>
, not the table cell.
Put the expected error type in the "expected" table cell.
$out()
and $outln()
functions
Sometimes what is tested is actual JS output. In that case, use <script>
tags and the $out()
or $outln()
functions.
Their only difference is that $outln()
also prints a line break.
Isolating tests
It is often useful to isolate a single group of tests, or even a single test so you can debug a particular failure.
To isolate a group of tests (<section>
), simply click the link of the section heading.
To isolate a specific test (<tr>
), hold down the Alt/Option key and double click on the table row.