Contributing
First of all, thank you for your interest in Puppeteer! We'd love to accept your patches and contributions!
Contributor License Agreement
Contributions to this project must be accompanied by a Contributor License Agreement. You (or your employer) retain the copyright to your contribution, this simply gives us permission to use and redistribute your contributions as part of the project. Head over to <https://cla.developers.google.com/> to see your current agreements on file or to sign a new one.
You generally only need to submit a CLA once, so if you've already submitted one (even if it was for a different project), you probably don't need to do it again.
Getting started
-
Clone this repository
git clone https://github.com/puppeteer/puppeteer
cd puppeteeror
-
Install the dependencies
npm install
# Or to download Firefox by default
PUPPETEER_BROWSER=firefox npm install -
Build all packages
npm run build
-
Run all tests
npm test
Building a single package
To build a single package, you can run:
npm run build --workspace <package> # e.g. puppeteer
This will build all dependent packages automatically, so specifying a single package is sufficient. This is all possible due to wireit which behaves similar to GNU Make.
Watch mode
To continuously build a package, you can run:
npm run build --watch --workspace <package> # e.g. puppeteer
You have to only specify a single package to watch else things will not work as expected As stated above because of wireit when a change happens all dependencies will be build or rebuild (if needed).
Removing stale artifacts
It's possible some generated artifacts (such as
packages/puppeteer-core/src/types.ts
) can become stale since these artifacts
rely on complex conditions (such as names of distinct files) that cannot be
captured by the build system. To clean artifacts, you can run
npm run clean
# or specify the package
npm run clean --workspace <package>
Comprehensive testing
Outside of npm test
, there are several other
npm
scripts that are
usually check through CI:
test-install
- Tests whetherpuppeteer
andpuppeteer-core
install properly and are functional.test-types
- Tests the TypeScript types inpuppeteer
usingtsd
.test:chrome:**
- Testspuppeteer
on Chromium.test:firefox:**
- Testspuppeteer
on Firefox.unit
- Runs unit tests.
The default npm test
runs test:{chrome,firefox}:headless
which is generally
sufficient.
Puppeteer uses a custom test runner on top of Mocha that consults the
TestExpectations.json
to see if a given test result is expected or not. See more info about the test
runner in
tools/mocha-runner
.
Unit tests
Tests that only test code (without the running browser) are put next to the classes they test and run using the Node test runner (requires Node 20+):
npm run unit
Code reviews
All submissions, including submissions by project members, require review. We use GitHub pull requests for this purpose. Consult GitHub Help for more information on using pull requests.
Code Style
Our coding style is fully defined in
.eslintrc
(ESLint) and
.prettierrc.cjs
(Prettier).
Code is checked for PRs automatically and you can check your code manually by running:
npm run lint
If some errors are returned, you can attempt to fix them using:
npm run format
Project structure
The following is a description of the primary folders in Puppeteer:
packages
contains all public source code.test
contains all test source code.test-d
contains type tests usingtsd
.tools
contains miscellaneous scripts that are used in building and etc.tools/mocha-runner
- contains the source code for our test runner.
API guidelines
When authoring new API methods, consider the following:
- Expose as little information as needed. When in doubt, don’t expose new information.
- Methods are used in favor of getters/setters.
- The only exception is namespaces, e.g.
page.keyboard
andpage.coverage
- The only exception is namespaces, e.g.
- All string literals must be small case. This includes event names and option values.
- Avoid adding "sugar" API (API that is trivially implementable in user-space) unless they're extremely demanded.
Commit messages
Commit messages should follow the Conventional Commits format.
In particular, breaking changes should clearly be noted as “BREAKING CHANGE:” in the commit message footer. Example:
fix(page): fix page.pizza method
This patch fixes page.pizza so that it works with iframes.
Issues: #123, #234
BREAKING CHANGE: page.pizza now delivers pizza at home by default.
To deliver to a different location, use the "deliver" option:
`page.pizza({deliver: 'work'})`.
Writing documentation
Documentation is generated from TSDoc comments via npm run docs
. It is automatically
published to our documentation site on merge and gets versioned on release.
This means that you should not change the markdown in files docs/api
manually.
Writing TSDoc comments
Each change to Puppeteer should be thoroughly documented using TSDoc comments. Refer to the API Extractor documentation for information on the exact syntax.
- Every new method needs to have either
@public
or@internal
added as a tag depending on if it is part of the public API. - Keep each line in a comment to no more than 90 characters (ESLint will warn you if you go over this). If you're a VSCode user the Rewrap plugin is highly recommended!
Running the documentation site locally
- At root, install all dependencies with
npm i --ignore-scripts
. - run
npm run docs
which will generate all the.md
files onpuppeteer/docs/api
. - run
npm i
inpuppeteer/website
. - run
npm start
inpuppeteer/website
.
Adding new dependencies
For all dependencies (both installation and development):
- Do not add a dependency if the desired functionality is easily implementable.
- If adding a dependency, it should be well-maintained and trustworthy.
A barrier for introducing new installation dependencies is especially high:
- Do not add installation dependency unless it's critical to project success.
There are additional considerations for dependencies that are environment
agonistic. See the
third_party/README.md
for details.
Testing tips
- Every feature should be accompanied by a test.
- Every public api event/method should be accompanied by a test.
- Tests should not depend on external services.
- Tests should work on all three platforms: Mac, Linux and Win. This is especially important for screenshot tests.
If a test is expected to fail on certain configurations or became flaky, update
TestExpectations.json
to reflect that. See more info about TestExpectations.json in
tools/mocha-runner
.
API Coverage
Every public API method or event should be called at least once in tests. To
ensure this, the main test
command runs coverage during testing.
Debugging Puppeteer
See Debugging Tips.
Debugging Puppeteer tests via VSCode
Copy the provided default .vscode/launch.template.json
to .vscode/launch.json
and then use the integrated VSCode debugger to debug test.
Remember to build test before launching via:
npm run build --workspace @puppeteer-test/test
For Project Maintainers
Rolling new Chrome version
There is a GitHub action that runs once per day. The action has a manual trigger that can be found on the Actions Tab.
Manual instructions
You can run the tools/update_browser_revision.mjs
locally
and try see if any changes need to be committed.
Note: You may need to run
node --experimental-fetch tools/update_browser_revision.mjs
as the script relies onfetch
The following steps are manual version of the script above.
- Find a suitable Chrome
revision
andversion
via https://googlechromelabs.github.io/chrome-for-testing/ or https://chromiumdash.appspot.com/. - Update
packages/puppeteer-core/src/revisions.ts
with the foundversion
number. - Update
versions.json
with the new Chrome-to-Puppeteerversion
mapping and updatelastMaintainedChromeVersion
with the next one in from the list. - Run
npm run check
. If it fails, updatepackages/puppeteer-core/package.json
with the expecteddevtools-protocol
version and runnpm install
to generate an updatedpackage-lock.json
. - Run
npm run clean
,npm install
andnpm run build
. - Run
npm test
and ensure that all tests pass. If a test fails, bisect 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). - Commit and push your changes and open a pull request. The commit message must
contain the version in
Chrome <version>
format to ensure that pptr.dev can parse it correctly, e.g.feat(chrome): roll to Chrome 90.0.4427.0
.
Bisecting upstream changes
For bisecting Chrome/Chromium changes use https://www.chromium.org/developers/bisect-builds-py/.
Releasing to npm
We use release-please to automate releases. When a release should be done, check for the release PR in our pull requests and merge it.
In case Release Please fails
In the event release-please fails, the following needs to be done:
-
Update anything missing in the CHANGELOG of every package that was supposed to be published. For example, if the header is missing, you may need to add
-
For puppeteer:
## [{NEW_VERSION}](https://github.com/puppeteer/puppeteer/compare/v{PREVIOUS_VERSION}...v{NEW_VERSION}) ({CURRENT_DATE})`
-
For other packages:
## [{NEW_VERSION}](https://github.com/puppeteer/puppeteer/compare/{PACKAGE_FOLDER_NAME}-v{PREVIOUS_VERSION}...{PACKAGE_FOLDER_NAME}-v{NEW_VERSION}) ({CURRENT_DATE})
-
-
Create a GitHub release for each package, following the practice of previous releases.
Bug triage guidelines
Check incoming bug reports that do not have a confirmed
or needs-feedback
label:
- Make sure the issue is labeled as either
bug
orfeature
. - If the issue does not have a clear repro or you cannot repro, ask for the repro and set the
needs-feedback
label. - Follow-up on the issues you previously asked for a feedback on (you should get a notification on GitHub when the user responds).
- If the user does not provide feedback, the issue will be closed by the stale bot eventually.
- If you are able to reproduce the issue, add the label
confirmed
. - If the bug is on the Chromium side, create a corresponding crbug.com issue, label the GitHub issue with the
upstream
label, and post a link to crbug.com in the comments. - If the issue is not related to either Puppeteer or Chromium, close the issue.
- If the issue is about missing/incorrect documentation, label it as
documentation
.
Issues with PDFs:
- If the issue reproduces using the regular print dialog and/or headful, file a crbug.com against the
Blink>Layout
component. - If the issue is specific to Headless mode, file an issue on crbug.com against the
Internals>Headless
component.