package.json

package.json parsing and scripts integration

---

As mentioned in the overview introduction, Pants's approach to enable support for Javascript is to be a thin caching layer on top of your current tooling.

Refer to the example repository for example usage.

Package manager

Pants uses corepack to manage package manager versions and installation. Like corepack, Pants respects the experimental "packageManager" feature in package.json files.

package.json

{
    "name": "@my-company/pkg",
    "packageManager": "[email protected]"
}

this setting will ensure that all scripts invoked for this package.json, and any workspaces managed by this package.json will use this particular version of yarn. It can be more convenient to define a project level package manager.

Choosing between pants.toml or package.json for package manager version configuration

In general, if your team runs all tooling via Pants, using pants.toml reduces boilerplate in cases where you maintain multiple packages. If your team mixes usage of Pants and "bare" package manager invocations, package.json#packageManager is the safer option.

Testing

By default Pants assumes a package_json target mapping a package.json includes a test script, e.g.

package.json

{
    "name": "@my-company/pkg",
    "scripts": {
        "test": "jest"
    },
    "devDependencies": {
        "jest": "^29.5.0"
    }
}

and will use this script to execute your tests when running pants test ::. See Goal arguments for the normal techniques for telling Pants what to run on.

To enable configurability, the build symbol node_test_script contains options for changing the entry point from "test", and to enable coverage reporting.

Packaging

Similarly, build scripts can be introduced to Pants via the node_build_script build symbol. This is intended to be used as a way to introduce artifacts generated via bundlers and/or compilers installed and ran via your package manager. The result can the be consumed by other targets as either resource-targets that can be depended on, or as a package for the docker backend.