Getting started
Begin by installing {shinytest2} from CRAN:
Recording tests
This is the general procedure for recording tests:
Run
record_test()to launch the app in a test recorder.Create the tests by interacting with the application and telling the recorder to make an expectation on the state at various points.
Quit the test recorder. When you do this, three things will happen:
The test script will be saved in
tests/testthat/test-shinytest2.Rin a subdirectory of the application.If you are running in the RStudio IDE, it will automatically open this file in the editor.
The tests will execute
To record tests, we first need an app.
Here is a simple app that I’ve placed in
simple-app/app.R:
library(shiny)
ui <- fluidPage(
textInput("name", "What is your name?"),
actionButton("greet", "Greet"),
textOutput("greeting")
)
server <- function(input, output, session) {
output$greeting <- renderText({
req(input$greet)
paste0("Hello ", isolate(input$name), "!")
})
}
shinyApp(ui, server)To test it, I now load {shinytest2}, and call
record_test() with the path to my app:
In a separate R process, this launches the Shiny application to be tested. We’ll refer to this as the target app. This also launches a special Shiny application in the current R process which displays the target app in an iframe and has some controls outside the iframe. We’ll refer to this as the recorder app. You will see something like this:
On the left is the target app (in this case, the “What is your name” app), and on the right is the recorder app sidebar. Note that you may need to make the browser window wider because the recorder sidebar occupies some space.
The sidebar on the right displays some controls for the test recorder, as well as a the set of Code for all expectations and actions. As you interact with the target app – in other words, when you set inputs on the app – you will see those actions recorded in the Code output.
For testing a Shiny application, setting inputs is only one part.
It’s also necessary to check that the application produces the correct
outputs. This is accomplished by taking expectations of the
application’s state.
There are two ways to record output values. One way is
to take an expectation of the application’s state. This
will record all input values, output values,
and exported values (more on exported values later). To do
this, click the Expect Shiny values button on the recorder
app.
After modifying some inputs and making an expectation,
it will look something like this:
It is also possible to take a targeted value expectation.
This is a snapshot of one or more specified output (instead
of the default, which is to snapshot the entire application). To do
this, hold down the Control (or Command) button on your keyboard and
click on an output; it will snapshot just that one
output.
When you are done recording a sequence of events, click on the
Save test and exit button. If you are in the RStudio IDE,
it will open the test script
tests/testthat/test-shinytest2.R from a subdirectory of the
application.
Running tests
When you quit the test recorder, it will automatically run the test script. There are three separate components involved in running the tests:
First is the test driver. This is the R process that coordinates the testing and controls the web browser. When working on creating tests interactively, this is the R process that you use.
Next is the Shiny process, also known as the server. This is the R process that runs the target Shiny application.
Finally, there is the web browser, also known as the client, which connects to the server. This is a headless web browser – one which renders the web page internally, but doesn’t display the content to the screen (
{chromote}).
When you exit the test recorder, it will (by default) automatically run the test script and will print something like this:
• Saving test runner: tests/testthat.R
• Saving test file: tests/testthat/test-shinytest2.R
• Running recorded test: tests/testthat/test-shinytest2.R
✔ | F W S OK | Context
✔ | 2 1 | shinytest2 [2.6s]
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Warning (test-shinytest2.R:7:3): {shinytest2} recording: simple-app
Adding new file snapshot: 'tests/testhat/_snaps/simple-app-001_.png'
Warning (test-shinytest2.R:7:3): {shinytest2} recording: simple-app
Adding new file snapshot: 'tests/testhat/_snaps/simple-app-001.json'
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
══ Results ══════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════
Duration: 2.6 s
[ FAIL 0 | WARN 2 | SKIP 0 | PASS 1 ]Behind the scenes, it runs test_app(). You can manually
run the tests with this:
This will play back the interactions and record snapshots, as
specified in the test script. The very first time the tests are run,
they will be saved in a subdirectory of the app called
tests/testthat/_snaps. These are the expected results, and
future test runs will be compared against them.
The directory will contain two types of files for different expectations. For example:
001.json
001_.png
002.png- The
.jsonfile is a JSON representation of the state of the application whenapp$expect_values()was called. - The
.pngfile is a screenshot of the application fromapp$expect_screenshot(), which you can use to insure that the visual state of the application does not change. - The
_.pngfile is a debug screenshot of the application whenapp$expect_values()was called. This file should be tracked usinggitto know how the application visually changes over time. However, if there is a difference found in a new screenshot, the difference will not throw an error.
The JSON file stores the state of all the input, output, and export values at the time of the snapshot:
{
"input": {
"greet": 1,
"name": "Barret"
},
"output": {
"greeting": "Hello Barret!"
},
"export": {
}
}If you are using a source control system (like git), you should check
in the expected results. (And ignore _.new.png files.)
Subsequent test runs
After the initial test run, you can run the tests again in the future to check for changes in your application’s behavior. If there are no changes to the snapshots, you will just see something like this, with no additional output:
> test_app("path/to/app")
✔ | F W S OK | Context
✔ | 1 | shinytest2 [2.2s]
══ Results ══════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════
Duration: 2.2 s
[ FAIL 0 | WARN 0 | SKIP 0 | PASS 1 ]If the results are exactly the same, then the test is considered successful.
If there are any differences between the latest run and the expected results, you’ll see output like this:
✔ | F W S OK | Context
✖ | 1 0 | shinytest2 [2.1s]
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Failure (test-shinytest2.R:7:3): {shinytest2} recording: simple-app
Snapshot of `file` to 'shinytest2/simple-app-001.json' has changed
Run `testthat::snapshot_review('shinytest2/')` to review changes
Backtrace:
1. app$expect_values()
at test-shinytest2.R:7:2
2. shinytest2::app_expect_values(...)
at shinytest2.nosync/R/app-driver.R:651:6
3. shinytest2::app__expect_snapshot_file(...)
at shinytest2.nosync/R/app-driver-expect-values.R:211:2
4. testthat::expect_snapshot_file(...)
at shinytest2.nosync/R/expect-snapshot.R:35:2
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
══ Results ══════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════
Duration: 2.2 s
[ FAIL 1 | WARN 0 | SKIP 0 | PASS 0 ]To view the differences, follow the prompted code
testthat::snapshot_review('shinytest2/') from within your
application directory. (You may need to run
install.packages("diffviewer") for
testthat::snapshot_review() to successfully execute.)
It will look something like this:
After inspecting the changes, you can choose to update the expected results (that is, replace the old expected results with the new current results), or to just quit without updating.
You should update the results if the changes are expected. You should skip/quit without updating if the changes are unexpected.
Multiple test scripts
In the example above, we had a single test script, named
test-shinytest2.R.
If you want to have more than one set of tests for an application, you can record new tests, each with a unique name. You can enter the name when recording your tests:
The saved test can be moved to a different testing file,
e.g. tests/testthat/test-simple-app.R. When you run
test_app("path/to/app"), it will run all the tests scripts
in the application’s tests/testhat directory.