05231bb254
This adds support for: mw.util.addPortletLink('p-views', '#', 'text', 'fooz',null, null,'#page-actions-overflow' ) and adding multiple edit icons via hooks. Depends-On: I655097769620301143140ded1cde4c7a16879ef4 Depends-On: I401805224c0f387ac85b52b50c1f298b83c03a91 Bug: T346944 Change-Id: I3c9d59f49f1b939981a7b2b450448db6736d5958 |
||
---|---|---|
.. | ||
config | ||
features | ||
pageobjects | ||
specs | ||
.eslintrc.json | ||
README.md | ||
wdio.conf.cucumber.js | ||
wdio.conf.js |
Selenium tests
For more information see https://www.mediawiki.org/wiki/Selenium
Setup
See https://www.mediawiki.org/wiki/MediaWiki-Docker/MinervaNeue
Run all specs
npm run selenium-test
Run specific tests
Filter by file name:
npm run selenium-test -- --spec tests/selenium/specs/[FILE-NAME]
Filter by file name and test name:
npm run selenium-test -- --spec tests/selenium/specs/[FILE-NAME] --mochaOpts.grep [TEST-NAME]
Migrating a test from Ruby to Node.js
Currently we are in the midst of porting our Ruby tests to Node.js. When the tests/browser/features folder is empty, we are done and the whole tests/browser folder can be removed.
This is a slow process (porting a single test can take an entire afternoon).
Step 1 - move feature file
Move the feature you want to port to Node.js
mv tests/browser/features/<name>.feature tests/selenium/features/
Step 2 - add boilerplate for missing steps
Run the feature in cucumber
npm run selenium-test-cucumber -- --spec tests/selenium/features/<name>.feature
You will get warnings such as:
Step "I go to a page that has languages" is not defined. You can ignore this error by setting cucumberOpts.ignoreUndefinedDefinitions as true.
For each missing step define them as one liners inside selenium/features/step_definitions/index.js
Create functions with empty bodies for each step.
Function names should be camel case without space, for example, I go to a page that has languages
becomes iGoToAPageThatHasLanguages
. Each function should be imported from a step file inside the features/step_definitions folder.
Rerun the test. If done correctly this should now pass.
Example: https://gerrit.wikimedia.org/r/#/c/mediawiki/skins/MinervaNeue/+/501792/1..2
Step 3 - copy across Ruby function bodies
Copy across the body of the Ruby equivalent inside each function body in tests/browser/features/step_definitions as comments.
Example: https://gerrit.wikimedia.org/r/#/c/mediawiki/skins/MinervaNeue/+/501792/2..3
Step 4 - rewrite Ruby to Node.js
Sadly there is no shortcut here. Reuse as much as you can. Work with the knowledge that the parts you are adding will aid the next browser test migration.
The docs are your friend: http://v4.webdriver.io/api/utility/waitForVisible.html
Example: https://gerrit.wikimedia.org/r/#/c/mediawiki/skins/MinervaNeue/+/501792/2..4
Step 5 - Make it work without Cucumber
Now the tests should be passing when run the browser tests using wdio.conf.cucumber.js or npm run selenium-test-cucumber
The final step involves making these run with
npm run selenium-test
This is relatively straightforward and mechanical.
- Copy the feature file to the specs folder
cp tests/selenium/features/editor_wikitext_saving.feature tests/selenium/specs/editor_wikitext_saving.js
- Convert indents to tabs
- Add
//
before any tags - Replace
Scenario: <name>
withit( '<name>', () => {
- Add closing braces for all scenarios:
} );
- Replace
Feature: <feature>
withdescribe('<feature>)', () => {
and add closing brace. - Replace
Background:
withbeforeEach( () => {
and add closing brace. - Find and replace
Given
,When
,And
,Then
with empty strings. - At top of file copy and paste imports from
selenium/features/step_definitions/index.js
to top of your new file and rewrite their paths. - Relying on autocomplete (VisualStudio Code works great) replace all the lines with the imports
- Drop unused imports from the file.