webpack-stats-plugin

webpack-stats-plugin

Ingest the webpack stats object, process / transform the object and write out to a file for further consumption. The most common use case is building a hashed bundle and wanting to programmatically refer to the correct bundle path in your Node.js server.


View on Github

Webpack Stats Plugin

Build Status

This plugin will ingest the webpack stats object, process / transform the object and write out to a file for further consumption.

The most common use case is building a hashed bundle and wanting to programmatically refer to the correct bundle path in your Node.js server.

Installation

The plugin is available via npm:

$ npm install --save-dev webpack-stats-plugin
$ yarn add --dev webpack-stats-plugin

Examples

We have example webpack configurations for all versions of webpack. See., e.g. test/webpack4/webpack.config.js.

Basic

const StatsWriterPlugin = require("webpack-stats-plugin").StatsWriterPlugin;

module.exports = {
  plugins: [
    // Everything else **first**.

    // Write out stats file to build directory.
    new StatsWriterPlugin({
      filename: "stats.json" // Default
    })
  ]
}

Custom Transform Function

The transform function has a signature of:

/**
 * Transform skeleton.
 *
 * @param {Object} data           Stats object
 * @param {Object} opts           Options
 * @param {Object} opts.compiler  Current compiler instance
 * @returns {String}              String to emit to file
 */
function (data, opts) {}

which you can use like:

const StatsWriterPlugin = require("webpack-stats-plugin").StatsWriterPlugin;

module.exports = {
  plugins: [
    new StatsWriterPlugin({
      transform(data, opts) {
        return JSON.stringify({
          main: data.assetsByChunkName.main[0],
          css: data.assetsByChunkName.main[1]
        }, null, 2);
      }
    })
  ]
}

Promise transform

You can use an asynchronous promise to transform as well:

const StatsWriterPlugin = require("webpack-stats-plugin").StatsWriterPlugin;

module.exports = {
  plugins: [
    new StatsWriterPlugin({
      filename: "stats-transform-promise.json",
      transform(data) {
        return Promise.resolve().then(() => JSON.stringify({
          main: data.assetsByChunkName.main
        }, null, INDENT));
      }
    })
  ]
}

Plugins

StatsWriterPlugin(opts)

  • opts (Object) options
  • opts.filename (String) output file name (Default: "stat.json")
  • opts.fields (Array) fields of stats obj to keep (Default: ["assetsByChunkName"])
  • opts.transform (Function|Promise) transform stats obj (Default: JSON.stringify())

Stats writer module.

Stats can be a string or array (we"ll have array from using source maps):

"assetsByChunkName": {
  "main": [
    "cd6371d4131fbfbefaa7.bundle.js",
    "../js-map/cd6371d4131fbfbefaa7.bundle.js.map"
  ]
},

Note: The stats object is big. It includes the entire source included in a bundle. Thus, we default opts.fields to ["assetsByChunkName"] to only include those. However, if you want the whole thing (maybe doing an opts.transform function), then you can set fields: null in options to get all of the stats object.

See:

filename: The opts.filename option can be a file name or path relative to output.path in webpack configuration. It should not be absolute.

transform: By default, the retrieved stats object is JSON.stringify'ed but by supplying an alternate transform you can target any output format. See test/webpack4/webpack.config.js for various examples including Markdown output.

  • Warning: The output of transform should be a String, not an object. On Node v4.x if you return a real object in transform, then webpack will break with a TypeError (See #8). Just adding a simple JSON.stringify() around your object is usually what you need to solve any problems.

Contributions

Contributions welcome!

We test against all versions of webpack. For a full explanation of our functional tests, see test/README.md

To get started, first install:

$ yarn

Our tests first do various webpack builds and then run mocha asserts on the real outputted stats files. Inefficient, but for our small sample size efficient enough.

# Lint and tests
$ yarn run lint
$ yarn run test

# All together
$ yarn run check