blob: ce34d93bd23bb7d9aa18f22c71df9465b7057285 [file] [log] [blame]
This directory contains semi-automated tests of Chrome with
NVDA (NonVisual Desktop Access), a popular open-source screen reader for
visually impaired users on Windows. It works by launching Chrome in a
subprocess, then launching NVDA in a special environment that simulates
speech rather than actually speaking, and ignores all events coming from
processes other than a specific Chrome process ID. Each test automates
Chrome with a series of actions and asserts that NVDA gives the expected
feedback in response.
Instructions for running these tests:
1. Install Python 2.7, 32-bit:
Note - the version of Python installed by Chrome's depot_tools will not
work, it's 64-bit.
2. Download pywinauto here:
Unzip it, then install it by running this from a cmd shell in that directory:
python install
If you get an error, make sure you're using the 32-bit version of Python.
3. Install the latest NVDA "next" snapshot from:
In the installer, choose "Create Portable copy" rather than "Install...".
From the Browse dialog, create an new folder called nvdaPortable inside
this folder.
Note: after NVDA 2014.3 stable is released, just use the stable version
instead, from
- if you do this, you need to run NVDA, then from the NVDA menu, choose
Tools > Create Portable Copy.
From the Browse dialog, create an new folder called nvdaPortable inside
this folder.
You should now have something like this:
You can now exit NVDA.
4. Install Chrome Canary. The binary is typically installed in:
c:\Users\USERNAME\AppData\Local\Google\Chrome SxS\Application\chrome.exe
...if not, edit to point to it.
5. Clone the nvda-proctest environment into this directory:
git clone
6. Run the tests:
First make sure NVDA is not already running.
Open a cmd console, change to the nvda_chrome_tests directory, and run:
If you get an error, open the Windows task manager and make sure NVDA
isn't running, kill it if necessary.