make PR about deletable orphan assets (github#26484)

* warn on orphan images

* oops

* temporarily adding orphan image

* fix the bug

* permission

* better comment

* try this

* try this

* debugging

* debugging

* using fs

* cjs

* remove experiment image

* debugging

* debugging

* try that

* try that

* git config

* cleaning

* token

* body

* label

* try that

* try that

* try that

* not upstream

* try that

* token twice

* try again

* cleaning up

* fix unit test

* one more exception

* feedbacked

Author: peterbe

.github/workflows/orphaned-assets-check.yml

@@ -5,20 +5,24 @@ name: 'Orphaned assets check'
 # **Who does it impact**: Docs content.
 
 on:
-  pull_request:
-  push:
-    branches:
-      - gh-readonly-queue/main/**
+  workflow_dispatch:
+  schedule:
+    - cron: '13 10 * * *' # Once a day at 10:13 UTC
 
 permissions:
   contents: read
 
 jobs:
   orphaned-assets-check:
+    if: ${{ github.repository == 'github/docs-internal' }}
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
         uses: actions/checkout@ec3a7ce113134d7a93b817d10a8272cb61118579
+        with:
+          # Using a PAT is necessary so that the new commit will trigger the
+          # CI in the PR. (Events from GITHUB_TOKEN don't trigger new workflows.)
+          token: ${{ secrets.DOCUBOT_REPO_PAT }}
 
       - name: Setup node
         uses: actions/setup-node@1f8c6b94b26d0feae1e387ca63ccbdc44d27b561
@@ -30,4 +34,30 @@ jobs:
         run: npm ci
 
       - name: Check for orphaned assets
-        run: ./script/find-orphaned-assets.mjs --verbose --exit
+        env:
+          # Needed for gh
+          GITHUB_TOKEN: ${{ secrets.DOCUBOT_REPO_PAT }}
+        run: |
+          set -e
+
+          ./script/find-orphaned-assets.mjs | xargs git rm
+
+          # If nothing to commit, exit now. It's fine. No orphans.
+          git status | grep 'nothing to commit' && exit 0
+
+          # Replicated from the translation pipeline PR-maker Action
+          git config --global user.name "docubot"
+          git config --global user.email "[email protected]"
+
+          date=$(date '+%Y-%m-%d-%H-%M')
+          branchname=orphaned-assets-$date-$GITHUB_RUN_ID
+
+          git checkout -b $branchname
+          git commit -m "Delete orphaned assets $date"
+          git push origin $branchname
+
+          gh pr create \
+            --title "Delete orphaned assets ($date)" \
+            --body "Found with the find-orphaned-assets.mjs script" \
+            --repo github/docs-internal \
+            --label docs-content-fr

script/find-orphaned-assets.mjs

@@ -12,15 +12,22 @@ import path from 'path'
 import program from 'commander'
 import walk from 'walk-sync'
 
-const EXCEPTIONS = new Set(['assets/images/site/favicon.ico'])
+const EXCEPTIONS = new Set([
+  'assets/images/site/favicon.ico',
+  'assets/images/site/apple-touch-icon.png',
+])
 
 function isExceptionPath(imagePath) {
   // We also check for .DS_Store because any macOS user that has opened
   // a folder with images will have this on disk. It won't get added
   // to git anyway thanks to our .DS_Store.
   // But if we don't make it a valid exception, it can become inconvenient
   // to run this script locally.
-  return EXCEPTIONS.has(imagePath) || path.basename('.DS_Store')
+  return (
+    EXCEPTIONS.has(imagePath) ||
+    path.basename(imagePath) === '.DS_Store' ||
+    imagePath.split(path.sep).includes('early-access')
+  )
 }
 
 program
@@ -61,7 +68,7 @@ async function main(opts) {
         root,
         Object.assign(
           {
-            globs: ['!**/*.+(png|csv|graphql|json|svg)'],
+            globs: ['!**/*.+(png|jpe?g|csv|graphql|json|svg)'],
           },
           walkOptions
         )