feat(puppeteer): Cleanup node_modules (#1497)

## Problem

- The Puppeteer `node_modules` directory is huge

## Solution

- Delete the not needed content from `node_modules`
- Inspired by [node-prune](https://github.com/tj/node-prune/) tool
(written in Go, not much suitable to run during RPM build, I rewrote
that to shell)

## Testing

- Tested manually

## Numbers

- The `node_modules` directory size went down from 39MB to 15MB (-62%)
- The built RPM size went down from 3.5MB to 1.9MB (-46%)
- I expect the Live ISO size reduction similar to the RPM reduction

puppeteer/agama-integration-tests

@@ -19,6 +19,12 @@ MOCHA_OPTIONS=(--bail --slow 10000)
 
 if [ -e "$MYDIR/../.git/" ]; then
   npm install --omit=optional
+
+  # do the same node_modules cleanup as in the RPM package to have the very same
+  # environment and have consistent results
+  "$MYDIR/node-prune.sh"
+  "$MYDIR/node-puppeteer-prune.sh"
+
   npx mocha "${MOCHA_OPTIONS[@]}" "$@"
 else
   # set the default load path

puppeteer/node-prune.sh

@@ -0,0 +1,159 @@
+#! /bin/bash
+
+# This script clean up the node_modules directory which is usually huge and
+# contains a lot of not needed files. It was inspired by the node-prune tool
+# (https://github.com/tj/node-prune).
+#
+# Usage:
+#
+#     node-prune.sh [path]
+#
+# The optional [path] argument is a path to the node_modules directory, if it is
+# not specified it uses node_modules in the current directory.
+#
+# This is a generic tool, you might run it against any node_modules directory,
+# not only in Agama Puppeteer tests.
+
+MODULES_PATH="${1:-./node_modules}"
+
+# The list of names/patterns comes from 
+# https://github.com/tj/node-prune/blob/master/internal/prune/prune.go
+
+# files to delete
+FILES=(
+  Jenkinsfile
+  Makefile
+  Gulpfile.js
+  Gruntfile.js
+  gulpfile.js
+  .DS_Store
+  .tern-project
+  .gitattributes
+  .editorconfig
+  .eslintrc
+  eslint
+  .eslintrc.js
+  .eslintrc.json
+  .eslintrc.yml
+  .eslintignore
+  .stylelintrc
+  stylelint.config.js
+  .stylelintrc.json
+  .stylelintrc.yaml
+  .stylelintrc.yml
+  .stylelintrc.js
+  .htmllintrc
+  htmllint.js
+  .lint
+  .npmrc
+  .npmignore
+  .jshintrc
+  .flowconfig
+  .documentup.json
+  .yarn-metadata.json
+  .travis.yml
+  appveyor.yml
+  .gitlab-ci.yml
+  circle.yml
+  .coveralls.yml
+  CHANGES
+  changelog
+  # keep the package licenses, it's unclear if we can legally delete them...
+  # LICENSE.txt
+  # LICENSE
+  # LICENSE-MIT
+  # LICENSE.BSD
+  # license
+  # LICENCE.txt
+  # LICENCE
+  # LICENCE-MIT
+  # LICENCE.BSD
+  # licence
+  AUTHORS
+  CONTRIBUTORS
+  .yarn-integrity
+  .yarnclean
+  _config.yml
+  .babelrc
+  .yo-rc.json
+  jest.config.js
+  karma.conf.js
+  wallaby.js
+  wallaby.conf.js
+  .prettierrc
+  .prettierrc.yml
+  .prettierrc.toml
+  .prettierrc.js
+  .prettierrc.json
+  prettier.config.js
+  .appveyor.yml
+  tsconfig.json
+  tslint.json
+)
+
+# directories to delete
+DIRECTORIES=(
+  test
+  tests
+  powered-test
+  docs
+  doc
+  .idea
+  .vscode
+  website
+  images
+  assets
+  example
+  examples
+  coverage
+  .nyc_output
+  .circleci
+  .github
+)
+
+# delete files with specific extensions
+EXTENSIONS=(
+  markdown
+  md
+  mkd
+  ts
+  jst
+  coffee
+  tgz
+  swp
+)
+
+# delete additional files with specific extensions (not deleted by the original
+# node-prune tool)
+EXTRA_EXTENSIONS=(
+  # The map files take almost half of the node_modules content! An they would be
+  # useful only for reporting bugs in Puppeteer itself or in some dependent
+  # library.
+  map
+)
+
+echo -n "Before cleanup: "
+du -h -s "$MODULES_PATH" | cut -f1
+
+# delete files
+for F in "${FILES[@]}"; do
+  find "$MODULES_PATH" -type f -name "$F" -delete
+done
+
+# delete directories recursively
+for D in "${DIRECTORIES[@]}"; do
+  find "$MODULES_PATH" -type d -name "$D" -prune -exec rm -rf \{\} \;
+done
+
+# delete files with specific extenstions
+for E in "${EXTENSIONS[@]}"; do
+  find "$MODULES_PATH" -type f -name "*.$E" -delete
+done
+
+# delete additional files with extensions
+for EE in "${EXTRA_EXTENSIONS[@]}"; do
+  find "$MODULES_PATH" -type f -name "*.$EE" -delete
+done
+
+echo -n "After cleanup:  "
+du -h -s "$MODULES_PATH" | cut -f1

puppeteer/node-puppeteer-prune.sh

@@ -0,0 +1,9 @@
+#! /bin/sh
+
+# This is a helper script which deletes some not needed files from NPM packages.
+# This script is specific for Puppeteer installations.
+
+MODULES_PATH="${1:-./node_modules}"
+
+# delete Puppeteer CommonJS modules, we use the ES modules (in lib/esm)
+rm -rf "$MODULES_PATH/puppeteer-core/lib/cjs"

puppeteer/package/agama-integration-tests.changes

@@ -1,3 +1,9 @@
+-------------------------------------------------------------------
+Wed Jul 24 15:03:25 UTC 2024 - Ladislav Slezák <[email protected]>
+
+- Reduce the node_modules size, delete not needed files
+  (gh#openSUSE/agama#1497)
+
 -------------------------------------------------------------------
 Fri Jul 19 10:18:37 UTC 2024 - Ladislav Slezák <[email protected]>
 

puppeteer/package/agama-integration-tests.spec

@@ -47,6 +47,12 @@ outside.
 rm -f package-lock.json
 local-npm-registry %{_sourcedir} install --omit=optional --with=dev --legacy-peer-deps || ( find ~/.npm/_logs -name '*-debug.log' -print0 | xargs -0 cat; false)
 
+# node_modules cleanup
+%{_builddir}/agama/node-prune.sh
+
+# extra cleanup for the Puppeteer NPM packages
+%{_builddir}/agama/node-puppeteer-prune.sh
+
 %install
 install -D -d -m 0755 %{buildroot}%{_datadir}/agama/integration-tests
 cp -aR node_modules %{buildroot}%{_datadir}/agama/integration-tests