xdotoolify.js is used to simulate clicks and keystrokes in Selenium
Selenium tries to cater to many different browsers and operating systems, but as a result, simulated user input is not the same as if the user did the input themselves. This can make some things very hard to test for.
xdotoolify.js simulates clicks and keystrokes at the operating system level (using xdotool) so the browser behaves exactly the same as when interacting with a real user.
xdotoolify.js only supports Firefox at the moment
xdotoolify.js requires that a window manager (e.g. fluxbox) be installed if using it in a docker container
npm install --save-dev xdotoolify
import Xdotoolify from 'xdotoolify';
import {Builder, By} from 'selenium-webdriver';
import firefox from 'selenium-webdriver/firefox';
var profile = new firefox.Profile('/tmp/profile');
var binary = new firefox.Binary();
binary.addArguments(['-no-remote']);
var firefoxOpts = new firefox.Options();
firefoxOpts.setProfile(profile);
firefoxOpts.setBinary(binary);
var page = await new Builder()
.forBrowser('firefox')
.setFirefoxOptions(firefoxOpts)
.build();
Xdotoolify(page);
await page.manage().window().setSize(1280, 1024);
await page.X.autoClick('#button').do();
await page.X
.autoKey('#email', 'ctrl+a BackSpace')
.type('me@example.com')
.autoClick('#submit')
.do();
page
is a seleniumThenableWebDriver
tied to Firefox
Returns undefined
.
This method must be called immediately after calling selenium's build()
.
It will add a .X
property to page
that exposes methods for manipulating
input in that Firefox instance.
selector
is one of:- a css selector e.g.
'.article > .date'
- an array where the first element is a css selector and the second element is an integer. E.g.
['.question', 2]
would select the third of the elements with classquestion
(note that this is different than the:nth-child
selector) - an object with
x
andy
properties. This is interpreted as an absolute position within the window client area. - and object with
screenx
andscreeny
properties. This is interpreted as an absolute screen position. - an object with
relx
andrely
properties. This is interpreted as a position relative to where the mouse is now.
- a css selector e.g.
relpos
(optional: defaults to'center'
) is one of'center'
,'topleft'
,'topright'
,'bottomleft'
, or'bottomright'
, ignored ifselector
is not a css selector or array. If'center'
, it will move the mouse to the center of the element represented byselector
. If'bottomright'
, it will move the mouse to the bottom right of the element (useful for positioning the cursor at the end of the text of an input element, for example)
Returns page.X
for easy chaining.
This method queues up a mousemove to the selector
. The mousemove will happen
asynchronously when page.X.do()
is next called.
mouseButton
is one of'left'
,'middle'
,'right'
Returns page.X
for easy chaining.
This method queues up a click (without moving the mouse). The click will happen
asynchronously when page.X.do()
is next called.
Same as page.X.click()
, but just holds the mouse down until page.X.mouseup()
is called.
Same as page.X.click()
, but releases the mouse if it was held down by page.X.mousedown()
.
Returns page.X
for easy chaining.
This method queues up a mouse wheel down event. The mouse wheel will be
scrolled asynchronously when page.X.do()
is next called.
Same as page.X.wheeldown()
but going up.
selector
is the same as inpage.X.mousemove()
mouseButton
is the same as inpage.X.click()
Returns page.X
for easy chaining.
This method queues up a mouse drag from where the mouse currently is to selector
. The drag will happen asynchronously when page.X.do()
is next called.
- string of key names (usually just one) to be typed separated by spaces, e.g.
'ctrl+a BackSpace'
. Key names come from here (remove the prefix)
Returns page.X
for easy chaining.
This method queues up a key or set of keys to be typed the next time page.X.do()
is called. This is as if the user just started typing on the keyboard, so it's expected that the element is already focused and the text cursor is in the right position.
- string of characters to type, e.g.
'blah blah lady gaga'
Returns page.X
for easy chaining.
This method queues up a set of characters to be typed the next time page.X.do()
is called. This is as if the user just started typing on the keyboard, so it's expected that the element is already focused and the text cursor is in the right position.
page.X.autoClick(selector, ...), page.X.autoDrag(selector, ...), page.X.autoKey(selector, ...), page.X.autoType(selector, ...)
selector
is the same as inpage.X.mousemove()
These are convenience methods to specify a selector
target on which to apply the operation. The arguments following the first argument are the same as in the non-auto
version of the command. For example, page.X.autoType('#myemail', 'me@example.com')
will first queue up a page.X.click()
to the bottom right of the #myemail
element before typing in me@example.com
.
milliseconds
is the number of milliseconds to sleep for
Returns page.X
for easy chaining.
This method will queue up a pause in between two commands. Useful when you need to wait for the result of an action. For example:
page.X
.autoClick('#login')
.sleep(500) // wait for the page to load
.autoType('#myemail', 'me@example.com')
.do();
predicate
is the async function that should return a booleantimeout
is the number of milliseconds for how long to call the predicate until it'strue
. When exceeded,Timeout exceeded
error is thrown
Returns page.X
for easy chaining.
This method will call the predicate until it returns true or timeout is exceeded. Useful when you need to wait for the result of an action based on some condition. For example:
page.X
.autoClick('#show-toggle-columns') // open an animated dialog
.sleepUntil(dialogAnimationIsCompleted()) // wait until dialog animation is completed
.autoClick('#dialog-button') // click dialog button
.do()
Returns undefined
.
Executes all the commands that have been queued up asynchronously. Before any
commands are executed, the page
will be brought to front and focused so
that input goes to the correct window.
page.X.do()
has not been tested to work in parallel with itself, so it's
currently expected that only one call to page.X.do()
is in process at a time.
This method will put the window manager's focus on the firefox instance
behind page
, and bring that instance in front of any other windows that
may exist. This is similar to bringing a window forward with alt+tab.