* push to excess labels to avoid reaching the limit
* build dist
* never set more than 100 labels
* use splice instead of set
* ignore IDE folders
* install @octokit/plugin-retry
* always setLabels
* fix indentations
* fix specs
* add spec for excess labels
* prettier
* licensed cache
* revert to !!core.getInput('sync-labels')
* better warning for exceeded labels
* keep manually-added labels
* nest the dedupe logic
* rename `removeLabel` to `removeLabelFromList` to avoid confusion
* use Sets, and issue a call only if labels have actually changed
* remove IDE config folders from gitignore
* remove obsolete duplucation check
---------
Co-authored-by: Mark Massoud <mark@unrealcloud.io>
* fix: correct reading of sync-labels input.
Contrary to the assumptions made in the unit tests, core.getInput
always returns a string, and !!'false' is true.
Also updates the unit test to reduce changes of regressions by ensuring
that the mocked getInput returns a string, as it would in production.
Fixes https://github.com/actions/labeler/issues/112
Make sure test catches regressions.
* Rebuild dist/index.js.
This is not intended to be a comprehensive test suite; I'm just trying to get us started by adding tests for some of the most important functionality.
This necessitated some minor refactoring. Previously, `main.ts` just directly invoked the main entrypoint function `run()`, which made it impossible to unit test the module without causing the side-effects of `run`.
As a workaround I created a new module `labeler.ts`, which just exports the functions we want to test without executing the main entrypoint, and simplified `main.ts` so that it only imports the entrypoint and executes it.
It's basically just a re-org. The diff makes it look way more complicated than it is.
