Custom Queries
Starting in Cypress 12, Cypress comes with its own API for creating custom queries. The built in Cypress queries use the very same API that's explained below.
Queries are a type of command, used for querying the state of your application. They are different from other commands in that they follow three important rules:
- Queries are synchronous. They do not return or await promises.
- Queries are retriable. Once you return the inner function, Cypress takes control, handling retries on your behalf.
- Queries are idempotent. Once you return the inner function, Cypress will invoke it repeatedly. Invoking the inner function multiple times must not change the state of your application.
With these rules, queries are simple to write and extremely powerful. They are the building blocks on which Cypress' API is built. To learn more about the differences between commands and queries, see our guide on Retry-ability.
If you want to chain together existing Cypress commands as a shortcut, you probably want to write a custom command instead.
You'll also want to write a command instead of a query if your method needs to be asynchronous, or if it can be called only once.
We recommend defining queries in your cypress/support/commands.js
file, since
it is loaded before any test files are evaluated via an import statement in the
supportFile.
Syntax​
Cypress.Commands.addQuery(name, callbackFn)
Cypress.Commands.overwriteQuery(name, callbackFn)
Usage​
Correct Usage
Cypress.Commands.addQuery('getById', function (id) {
return (subject) => newSubject
})
Cypress.Commands.overwriteQuery('get', function (originalFn, ...args) {
return originalFn.apply(this, args)
})
Arguments​
name (String)
The name of the query you're adding.
callbackFn (Function)
Pass a function that receives the arguments passed to the query.
This outer function is invoked once. It should return a function that takes a subject and returns a new subject; this inner function might be called multiple times.
The query API relies on this
to set timeouts, which means that callbackFn
should always use function () {}
and not be an arrow function (() => {}
).
Examples​
.focused()
​
The callback function can be thought of as two separate parts. The outer function, which is invoked once, where you perform setup and state management, and the query function, which might be called repeatedly.
Let's look at an example. This is actual Cypress code - how .focused()
is
implemented internally, with some small adjustments to make it work from a
support file. The only thing omitted here for simplicity is the TypeScript
definitions.
Cypress.Commands.addQuery('focused2', function focused2(options = {}) {
const log = options.log !== false && Cypress.log({ timeout: options.timeout })
this.set('timeout', options.timeout)
return () => {
let $el = cy.getFocused()
log &&
cy.state('current') === this &&
log.set({
$el,
consoleProps: () => {
return {
Yielded: $el?.length ? $el[0] : '--nothing--',
Elements: $el != null ? $el.length : 0,
}
},
})
if (!$el) {
$el = cy.$$(null)
$el.selector = 'focused'
}
return $el
}
})
The outer function​
The outer function is called once each time test uses the query. It performs setup and state management:
function focused2(options = {}) {
const log = options.log !== false && Cypress.log({ timeout: options.timeout })
this.set('timeout', options.timeout)
return () => { ... } // Inner function
}
Let's look at this piece by piece.
function focused2(options = {}) { ... }
Cypress passes the outer function whatever arguments the user calls it with; no
processing or validation is done on the user's arguments. In our case,
.focused2()
accepts one optional argument, options
.
If you wanted to validate the incoming arguments, you might add something like:
if (options === null || !_.isPlainObject(options)) {
const err = `cy.root() requires an \`options\` object. You passed in: \`{options}\``
throw new TypeError(err)
}
This is a general pattern: when something goes wrong, queries just throw an error. Cypress will handle displaying the error in the Command Log.
const log = options.log !== false && Cypress.log({ timeout: options.timeout })
If the user has not set { log: false }
, we create a new Cypress.log()
instance. See Cypress.log()
for more
information.
This line is setup code, so it lives in the outer function - we only want it to
run once, creating the log message when Cypress first begins executing this
query. We hold onto a reference to Log
instance. We'll update it later with
additional details when the inner function executes.
this.set('timeout', options.timeout)
When defining focused2()
, it's important to note that we used function
,
rather than an arrow function. This gives us access to this
, where we can set
the timeout
. If you don't call this.set('timeout')
, or call it with null
or undefined
, your query will use the
default timeout.
return () => { ... }
The inner function​
The outer function's return value is the inner function.
The inner function is called any number of times. It's first invoked repeatedly until it passes or the query times out; it can then be invoked again later to determine the subject of future commands, or when the user retrieves an alias.
The inner function is called with one argument: the previous subject. Cypress
performs no validation on this - it could be any type, including null
or
undefined
.
.focused2()
ignores any previous subject, but many queries do not - for
example, .contains()
accepts only certain types of subjects. You can use
Cypress' builtin ensures
functions, as .contains()
does:
cy.ensureSubjectByType(subject, ['optional', 'element', 'window', 'document'], this)
or you can perform your own validation and simply throw an error:
if (!_.isString(subject)) { throw new Error('MyCustomCommand only accepts strings as a subject!') }
If the inner function throws an error, Cypress will retry it after a short delay until it either passes or the query times out. This is the core of Cypress' retry-ability, and the guarantees it provides that your tests interact with the page as a user would.
Looking back to our .focused2()
example:
return () => {
let $el = cy.getFocused()
log &&
cy.state('current') === this &&
log.set({
$el,
consoleProps: () => {
return {
Yielded: $el?.length ? $el[0] : '--nothing--',
Elements: $el != null ? $el.length : 0,
}
},
})
if (!$el) {
$el = cy.$$(null)
$el.selector = 'focused'
}
return $el
}
Piece by piece again:
let $el = cy.getFocused()
This is the 'business end' of .focused2()
- finding the element on the page
that's currently focused.
log && cy.state('current') === this && log.set({...})
If log
is defined (ie, the user did not pass in { log: false }
), and this
query is the current command, we update the log message with new information,
such as $el
(the subject we're about to yield from this query), and the
consoleProps
, a function that
returns console output for
the user.
if (!$el) {
$el = cy.$$(null)
$el.selector = 'focused'
}
If there's no focused element on the page, we create an empty jquery object.
return $el
The return value of the inner function becomes the new subject for the next command.
With this return value in hand, Cypress verifies any upcoming assertions, such
as user's .should()
commands, or if there are none, the default implicit
assertions that the subject should exist.
Overwriting Existing Queries​
You can also modify the behavior of existing Cypress queries. This is useful to extend the functionality of builtin commands.
Cypress.Commands.overwriteQuery
can only overwrite queries, not other
commands. If you want to modify the behavior of a non-query command, you'll need
to use Cypress.Commands.overwrite
instead.
Remember that query functions rely on this
- when you invoke originalFn
, be
sure to use .call
or .apply
.
Cypress.Commands.overwriteQuery('get', function (originalFn, ...args) {
console.log('get called with args:', args)
const innerFn = originalFn.apply(this, args)
return (subject) => {
console.log('get inner function called with subject:', subject)
return innerFn(subject)
}
})
The originalFn
is the function originally passed to
Cypress.Commands.addQuery
- it is a function that returns a function. This
gives you access to both the outer arguments (before you call originalFn
) and
the inner function (the return value of originalFn
), giving you a great deal
of control over how the query executes.
Adding alias support to .contains()
​
In this example, cy.contains()
is extended to support querying for aliased
subjects, like cy.contains('@foo')
.
Cypress.Commands.overwriteQuery(
'contains',
function (originalFn, filter, text, userOptions) {
if (_.isString(filter) && filter[0] === '@') {
let alias = cy.state('aliases')[filter.slice(1)]
let subject = cy.getSubjectFromChain(alias?.subjectChain)
filter = subject
}
if (_.isString(text) && text[0] === '@') {
let alias = cy.state('aliases')[text.slice(1)]
let subject = cy.getSubjectFromChain(alias?.subjectChain)
text = subject
}
return originalFn.call(this, filter, text, userOptions)
}
)
cy.wrap('li').as('element')
cy.wrap('asdf 1').as('content')
cy.contains('@element', '@content')
Validation​
As noted in the examples above, Cypress performs very little validation around queries - it is the responsibility of each implementation to ensure that its arguments and subject are of the correct type.
Cypress has several builtin 'ensures' which can be helpful in this regard:
cy.ensureSubjectByType(subject, types, this)
: Accepts an array with any of the stringsoptional
,element
,document
, orwindow
.ensureSubjectByType
is howprevSubject
validation is implemented for commmands.cy.ensureElement(subject, queryName)
: Ensure that the passed insubject
is one or more DOM elements.cy.ensureWindow(subject)
: Ensure that the passed insubject
is awindow
.cy.ensureDocument(subject)
: Ensure that the passed insubject
is adocument
.cy.ensureAttached(subject, queryName)
: Ensure that DOM element(s) are attached to the page.cy.ensureNotDisabled(subject)
: Ensure that form elements aren't disabled.cy.ensureVisibility(subject)
: Ensure that a DOM element is visible on the page.
There's nothing special about these functions - they simply validate their argument and throw an error if the check fails. You can throw errors of any type at any time inside your queries - Cypress will catch and handle it appropriately.
Notes​
Best Practices​
1. Don't make everything a custom query​
Custom queries work well when you're needing to describe behavior that's
desirable across all of your tests. Examples would be cy.findBreadcrumbs()
or cy.getLoginForm()
. These are specific to your application and can be used
everywhere.
However, this pattern can be used and abused. Let's not forget - writing Cypress tests is JavaScript, and it's often more efficient to write a function for repeatable behavior than it is to implement a custom query.
2. Don't overcomplicate things​
Every custom query you write is generally an abstraction for locating elements on the page. That means you and your team members exert much more mental effort to understand what your custom command does.
There's no reason to add this level of complexity when the builtin queries are already quite expressive and powerful.
Don't do things like:
-
cy.getButton()
-
.getFirstTableRow()
Both of these are wrapping cy.get(selector)
. It's completely unnecessary. Just
call .get('button')
or .get('tr:first')
.
Testing in Cypress is all about readability and simplicity. You don't have to do that much actual programming to get a lot done. You also don't need to worry about keeping your code as DRY as possible. Test code serves a different purpose than app code. Understandability and debuggability should be prioritized above all else.
Try not to overcomplicate things and create too many abstractions.
History​
Version | Changes |
---|---|
12.6.0 | overrideQuery API added |
12.0.0 | addQuery API added |