We are pleased to introduce the first Python library to fully enable GUI testing of Mac applications via the Apple Accessibility API. This library was created out of desperation. Existing tools such as using appscript to send messages to accessibility objects are painful to write and slow to use. PyATOM has direct access to the API. It's fast and easy to use to write tests.
PyATOM requires a system running OS X and Xcode installed. It has been tested extensively on 10.6. 10.5 may work. If you experience issues with PyATOM on a particular version of OS X, please open a ticket in the issue tracker.
Systemwide accessibility must be enabled. Check the checkbox: System Preferences > Universal Access > Enable access for assistive devices. Failure to enable this will result in ErrorAPIDisabled exceptions during some module usage.
Installation should be as simple as running the following command line, which will download, build and install PyATOM:
$ sudo easy_install atom
Once the pyatom module is installed, you should be able to use it to launch an application:
>>> import pyatom >>> pyatom.launchAppByBundleId('com.apple.Automator')
This should launch Automator. Next, get a reference to the UI Element for the application itself:
>>> automator = pyatom.getAppRefByBundleId('com.apple.Automator') >>> automator <pyatom.AXClasses.NativeUIElement AXApplication u'Automator'>
Now, we can find objects in the accessibility hierarchy:
>>> window = automator.windows()[0] >>> window.AXTitle u'Untitled' >>> sheet = window.sheets()[0]
Note that we retrieved an accessibility attribute from the Window object - AXTitle. PyATOM supports reading and writing of most attributes. Using Xcode's included accessibility inspector can provide a quick way to find these attributes.
There is a shortcut for getting the sheet object which bypasses accessing it through the Window object - PyATOM can search all objects in the hierarchy:
>>> sheet = automator.sheetsR()[0]
There are search methods for most types of accessibility objects. Each search method, such as windows
, has a corresponding recursive search function, such as windowsR
. The recursive search finds items that aren't just direct children, but children of children. These search methods can be given terms to identify specific elements. Note that * and ? can be used as wildcard match characters in all PyATOM search methods:
>>> close = sheet.buttons('Close')[0]
PyATOM has a method to search for UI Elements that match any number of criteria. The criteria are accessibility attributes:
>>> close = sheet.findFirst(AXRole='AXButton', AXTitle='Close')
FindFirst
and FindFirstR
return the first item found to match the criteria or None. FindAll
and FindAllR
return a list of all items that match the criteria or an empty list.
Objects are fairly versatile. You can get a list of supported attributes and actions on an object:
>>> close.getAttributes() [u'AXRole', u'AXRoleDescription', u'AXHelp', u'AXEnabled', u'AXFocused', u'AXParent', u'AXWindow', u'AXTopLevelUIElement', u'AXPosition', u'AXSize', u'AXTitle'] >>> close.AXTitle u'Close' >>> close.getActions() [u'Press']
Performing an action is as natural as:
>>> close.Press()
Any action can be triggered this way.
Although PyATOM is fully functional and drives hundreds of automated test cases at VMware, we have a to-do list to make the project even better.
- Documentation - The docstrings need to be better sphynxified, and the basic doc layout needs to be created.
- Formatting - this code is not currently PEP-8 compliant.
- Better mouse handling - for example, a method to smoothly drag from one UI Element to another.
- Cleanup the search methods - We could use currying to define all the search methods in AXClasses in a cleaner way.
Feel free to submit pull requests against the project on Github. If you're interested in developing PyATOM itself, sign up to the pyatom-dev mailing list.
- The PyATOM home page
- Mailing lists
- Source code on Github
- Issue tracker
PyATOM is released under the GNU General Public License. See COPYING.txt for more details.
James Tatum <[email protected]>, Andrew Wu, Jesse Mendonca, Ken Song