Skip to content
You're viewing an older version of this GitHub Action. Do you want to see the latest version instead?
archive

GitHub Action

compressed-size-action

2.5.0

compressed-size-action

archive

compressed-size-action

Get compressed size differences for every PR

Installation

Copy and paste the following snippet into your .yml file.

              

- name: compressed-size-action

uses: preactjs/compressed-size-action@2.5.0

Learn more about this action in preactjs/compressed-size-action

Choose a version

compressed-size-action

A GitHub action that reports changes in compressed file sizes on your PRs.

  • Automatically uses yarn, pnpm or npm ci when lockfiles are present
  • Builds your PR, then builds the target and compares between the two
  • Doesn't upload anything or rely on centralized storage
  • Supports custom build scripts and file patterns

Usage:

Add a workflow (.github/workflows/main.yml):

name: Compressed Size

on: [pull_request]

jobs:
  build:
    runs-on: ubuntu-latest

    steps:
      - uses: actions/checkout@v2
      - uses: preactjs/compressed-size-action@v2

Customizing the Build

By default, compressed-size-action will try to build your PR by running the "build" npm script in your package.json.

If you need to perform some tasks after dependencies are installed but before building, you can use a "postinstall" npm script to do so. For example, in Lerna-based monorepo:

{
  "scripts": {
    "postinstall": "lerna bootstrap",
    "build": "lerna run build"
  }
}

It is also possible to define a "prebuild" npm script, which runs after "postinstall" but before "build".

You can also specify a completely different npm script to run instead of the default ("build"). To do this, add a build-script option to your yml workflow:

name: Compressed Size

on: [pull_request]

jobs:
  build:

    runs-on: ubuntu-latest

    steps:
    - uses: actions/checkout@v2
    - uses: preactjs/compressed-size-action@v2
      with:
+       build-script: "ci"

Clean up state between builds

For repositories or custom monorepo setups where files are modified in ways that are not reset by npm ci && npm run build, it may be necessary to define a custom "clean" script. This script resets any file modifications after the upstream (target) build ends and your PR code (HEAD) is checked out, but before installation of npm dependencies for HEAD:

name: Compressed Size
on: [pull_request]
jobs:
  build:
    runs-on: ubuntu-latest
    steps:
    - uses: actions/checkout@v2
    - uses: preactjs/compressed-size-action@v2
      with:
        repo-token: "${{ secrets.GITHUB_TOKEN }}"
+       clean-script: "clean"
// package.json
{
  "scripts": {
    // example - a simple nested node_modules setup:
    "postinstall": "cd packages && npm i",
    // between the two builds, we need to delete the inner node_modules:
    "clean": "rm -rf packages/node_modules"
  }
}

Customizing the list of files

compressed-size-action defaults to tracking the size of all JavaScript files within dist/ directories - anywhere in your repository, not just at the root. You can change the list of files to be tracked and reported using the pattern and exclude options, both of which are minimatch patterns:

name: Compressed Size
on: [pull_request]
jobs:
  build:
    runs-on: ubuntu-latest
    steps:
    - uses: actions/checkout@v2
    - uses: preactjs/compressed-size-action@v2
      with:
+       pattern: "./build-output/**/*.{js,css,html,json}"
+       exclude: "{./build-output/manifest.json,**/*.map,**/node_modules/**}"

Files are collected by finding matches for pattern, then any of those that match exclude are ignored. For that reason, most project don't need to modify exclude. The default values for pattern and exclude are as follows:

with:
  # Any JS files anywhere within a dist directory:
  pattern: "**/dist/**/*.js"

  # Always ignore SourceMaps and node_modules:
  exclude: "{**/*.map,**/node_modules/**}"

Dealing with hashed filenames

A strip-hash option was added in v2 that allows passing a custom Regular Expression pattern that will be used to remove hashes from filenames. The un-hashed filenames are used both for size comparison and display purposes.

By default, the characters matched by the regex are removed from filenames. In the example below, a filename foo.abcde.js will be converted to foo.js:

strip-hash: "\\b\\w{5}\\."

This can be customized further using parens to create submatches, which mark where a hash occurs. When a submatch is detected, it will be replaced with asterisks. This is particularly useful when mix of hashed and unhashed filenames are present. In the example below, a filename foo.abcde.chunk.js will be converted to foo.*****.chunk.js:

strip-hash: "\\.(\\w{5})\\.chunk\\.js$"

Increasing the required threshold

By default, a file that's been changed by a single byte will be reported as changed. If you'd prefer to require a certain minimum threshold for a file to be changed, you can specify minimum-change-threshold in bytes:

minimum-change-threshold: 100

In the above example, a file with a delta of less than 100 bytes will be reported as unchanged.

Compression

By default, files are compared after gzip compression, but it's possible to use other compression algorithms (gzip/brotli/none) or disable the compression.

compression: "none"