Contribute to development
If you would like to contribute to OWASP Juice Shop but need some idea what task to address, the best place to look is in the GitHub issue lists at https://github.com/bkimminich/juice-shop/issues.
- Issues labelled with help wanted indicate tasks where the project team would very much appreciate help from the community
- Issues labelled with good first issue indicate tasks that are isolated and not too hard to implement, so they are well-suited for new contributors
The following sections describe in detail the most important rules and processes when contributing to the OWASP Juice Shop project.
Tips for newcomers
If you are new to application development - particularly with AngularJS and Express.js - it is recommended to read the Codebase 101 to get an overview what belongs where. It will lower the entry barrier for you significantly.
The project uses
git as its version control system and GitHub as the
central server and collaboration platform. OWASP Juice Shop resides in
the following repository:
OWASP Juice Shop is maintained in a simplified
fashion, where all active development happens on the
master is used to deploy stable versions to the
Heroku demo instance and later
create tagged releases from.
Feature branches are only used for long-term tasks that could jeopardize
regular releases from
develop in the meantime. Likewise prototypes and
experiments must be developed on an individual branch or a distinct fork
of the entire project.
Any release from
master is tagged with a unique version in the format
vMAJOR.MINOR.PATCH, for example
Given a version number
MAJOR.MINOR.PATCH, increment the:
MAJORversion when you make incompatible API changes,
MINORversion when you add functionality in a backwards-compatible manner, and
PATCHversion when you make backwards-compatible bug fixes.1
The current version of the project must be manually maintained in the following two places:
All other occurrences of the version (i.e. Docker images, packaged
releases & the menu bar of the application itself) are resolved through
"version" property of
Using Git-Flow means that PRs have the highest chance of getting
accepted and merged when you open them on the
develop branch of your
fork. That allows for some post-merge changes by the team without
directly compromising the
master branch, which is supposed to hold
always be in a release-ready state.
It is usually not a big deal if you accidentally open a PR for the
master branch. GitHub added the possibility to change the target
branch for a PR afterwards some time ago.
The minimum requirements for code contributions are:
- The code must be compliant with the JS Standard Code Style rules
- All new and changed code should have a corresponding unit and/or integration test
- New and changed challenges should have a corresponding e2e test
- All unit, integration and e2e tests must pass locally
As a preliminary step, the
npm test script verifies code compliance with
standard style before running any test. If PRs deviate from this
coding style, they will immediately fail their build and will not be
merged until compliant.
In case your PR is failing from style guide issues try running
--fix over your code - this will fix all syntax or code style issues
automatically without breaking your code. You might need to
npm i -g
npm test # check for code style violations and run all unit tests npm run frisby # run all API integration tests npm run protractor # run all end-to-end tests
Pull Requests are verified to pass all of the following test stages during the continuous integration build. It is recommended that you run these tests on your local computer to verify they pass before submitting a PR. New features should be accompanied by an appropriate number of corresponding tests to verify they behave as intended.
There is a full suite containing isolated unit tests
- for all client-side code in
- for the server-side routes and libraries in
The integration tests in
test/api verify if the backend for all normal
use cases of the application works. All server-side vulnerabilities are
npm run frisby
These tests automatically start a server and run the tests against it. A working internet connection is recommended.
The e2e test suite in
test/e2e verifies if all client- and server-side
vulnerabilities are exploitable. It passes only when all challenges are
solvable on the score board.
npm run protractor
The end-to-end tests require a locally installed Google Chrome browser and internet access to be able to pass.
The mutation tests ensure the quality of the unit test suite by making small changes to the code that should cause one or more tests to fail. If none does this "mutated line" is not properly covered by meaningful assertions and the mutation testing engine that will inform you about this.
npm run stryker
Currently only the client-side unit tests are covered by mutation tests. The integration and end-to-end tests are not suitable for mutation testing because they run against a real server instance with dependencies to the database and an internet connection. The mutation tests are intentionally not executed on Travis-CI due to their significant execution time.
Continuous integration & deployment
The main build and CI server for OWASP Juice Shop is set up on Travis-CI:
On every push to any branch on GitHub, a build is triggered on Travis-CI. A build consists of several jobs: One for each version of Node.js that is officially supported by the application. Each job performs the following actions:
- Clone the repository and checkout the branch to build
- Build the application
- Execute the quality checks consisting of
- Compliance check against the JS Standard Code Style rules
- Unit tests for the Angular client
- Unit tests for the Express routes and server-side libraries
- Integration tests for the server-side API
- End-to-end tests verifying that all challenges can be solved
- Upload of the quality metrics to Code Climate
- Deployment to a Heroku instance
- Trigger some monitoring endpoints about the build result
Pull Requests are built in the same manner (steps 1-3) to assess if they can safely be merged into the codebase. For tag-builds (i.e. versions to be released) an additional step is to package release-artifacts for Linux for each Node.js version and attach these to the release page on GitHub.
AppVeyor is used as a secondary CI server to check if the application can be built on Windows. It also packages and attaches release-artifacts for Windows in case a tag-build is executed:
Testing packaged distrubutions
During releases the application will be packaged into
archives for another easy setup method. When you contribute a change
that impacts what the application needs to include, make sure you test
this manually on your system.
Then take the created archive from
/dist and follow the steps
described above in
to make sure nothing is broken or missing.