- Coding style is fully defined in [`.eslintrc`](https://github.com/puppeteer/puppeteer/blob/main/.eslintrc.js) and we automatically format our code with [Prettier](https://prettier.io).
- If you're working in a JS file, code should be annotated with [closure annotations](https://github.com/google/closure-compiler/wiki/Annotating-JavaScript-for-the-Closure-Compiler).
- If you're working in a TS file, you should explicitly type all variables and return types. You'll get ESLint warnings if you don't so if you're not sure use them as guidelines, and feel free to ask us for help!
You can check your code (both JS & TS) type-checks by running:
```bash
npm run tsc
```
## TypeScript guidelines
- Try to avoid the use of `any` when possible. Consider `unknown` as a better alternative. You are able to use `any` if needbe, but it will generate an ESLint warning.
The code in Puppeteer is split primarily into two folders:
-`src` contains all source code
-`vendor` contains all dependencies that we've vendored into the codebase. See the [`vendor/README.md`](https://github.com/puppeteer/puppeteer/blob/main/vendor/README.md) for details.
We structure these using TypeScript's project references, which lets us treat each folder like a standalone TypeScript project.
### Shipping CJS and ESM bundles
Currently Puppeteer ships two bundles; a CommonJS version for Node and an ESM bundle for the browser. Therefore we maintain two `tsconfig` files for each project; `tsconfig.esm.json` and `tsconfig.cjs.json`. At build time we compile twice, once outputting to CJS and another time to output to ESM.
We compile into the `lib` directory which is what we publish on the npm repository and it's structured like so:
```
lib
- cjs
- puppeteer <== the output of compiling `src/tsconfig.cjs.json`
- vendor <== the output of compiling `vendor/tsconfig.cjs.json`
- esm
- puppeteer <== the output of compiling `src/tsconfig.esm.json`
- vendor <== the output of compiling `vendor/tsconfig.esm.json`
```
The main entry point for the Node module Puppeteer is `cjs-entry.js`. This imports `lib/cjs/puppeteer/index.js` and exposes it to Node users.
### tsconfig for the tests
We also maintain `test/tsconfig.test.json`. This is **only used to compile the unit test `*.spec.ts` files**. When the tests are run, we first compile Puppeteer as normal before running the unit tests **against the compiled output**. Doing this lets the test run against the compiled code we ship to users so it gives us more confidence in our compiled output being correct.
### Root `tsconfig.json`
The root `tsconfig.json` exists for the API Extractor; it has to find a `tsconfig.json` in the project's root directory. It is _not_ used for anything else.
Commit messages should follow [the Conventional Commits format](https://www.conventionalcommits.org/en/v1.0.0/#summary). This is enforced via `npm run lint`.
All public API should have a descriptive entry in [`docs/api.md`](https://github.com/puppeteer/puppeteer/blob/main/docs/api.md). There's a [documentation linter](https://github.com/puppeteer/puppeteer/tree/main/utils/doclint) which makes sure documentation is aligned with the codebase.
There are additional considerations for dependencies that are environment agonistic. See the [`vendor/README.md`](https://github.com/puppeteer/puppeteer/blob/main/vendor/README.md) for details.
Puppeteer tests are located in [the `test` directory](https://github.com/puppeteer/puppeteer/blob/main/test/) and are written using Mocha. See [`test/README.md`](https://github.com/puppeteer/puppeteer/blob/main/test/README.md) for more details.
- To run Firefox tests, firstly ensure you have Firefox installed locally (you only need to do this once, not on every test run) and then you can run the tests:
Every public API method or event should be called at least once in tests. To ensure this, there's a `coverage` command which tracks calls to public API and reports back if some methods/events were not called.
To do so, run `utils/check_availability.js -rd` to find the latest suitable `dev` Chromium revision (see `utils/check_availability.js -help` for more options).
1. Run `npm run unit` and ensure that all tests pass. If a test fails, [bisect](#bisecting-upstream-changes) the upstream cause of the failure, and either update the test expectations accordingly (if it was an intended change) or work around the changes in Puppeteer (if it’s not desirable to change Puppeteer’s observable behavior).
1. Commit and push your changes and open a pull request.
The commit message must contain the version in `Chromium <version> (<revision>)` format to ensure that [pptr.dev](https://pptr.dev/) can parse it correctly, e.g. `'feat(chromium): roll to Chromium 90.0.4427.0 (r856583)'`.
Sometimes, performing a Chromium roll causes tests to fail. To figure out the cause, you need to bisect Chromium revisions to figure out the earliest possible revision that changed the behavior. The script in `utils/bisect.js` can be helpful here. Given a Node.js script that calls `process.exit(1)` for bad revisions, run this from the Puppeteer repository’s root directory:
1. Run `npm run release`. (This automatically bumps the version number in `package.json`, populates the changelog, updates the docs, and creates a Git commit for the next step.)
1. Send a PR for the commit created in the previous step.
1. Once merged, publish the release notes from `CHANGELOG.md` using [GitHub’s “draft new release tag” option](https://github.com/puppeteer/puppeteer/releases/new).
1. As soon as the Git tag is created by completing the previous step, our CI automatically `npm publish`es the new releases for both the `puppeteer` and `puppeteer-core` packages.
1. Bump `package.json` version to the `-post` version, run `npm run doc` to update the “released APIs” section at the top of `docs/api.md` accordingly, and send a PR titled `'chore: bump version to vXXX.YYY.ZZZ-post'` ([example](https://github.com/puppeteer/puppeteer/pull/6808))