docs: add clean up images job DOC-1232 (#4409)

* docs: add clean up images job  DOC-1232

* docs: add on push for testing

* docs: add npm ci step

* docs: add fetch in script

* docs: add debug lines

* docs: adjust github action flow

* docs: add backport labels

* docs: remove results files from commit

* docs: add PR environment variable

* docs: add slack notification

* docs: remove on push trigger

* docs: adjust slack step if check

* docs: add pus for final testing

* docs: remove push trigger before merge

(cherry picked from commit c272eaf)

.github/workflows/clean-up-unused-images.yaml

@@ -0,0 +1,112 @@
+name: Clean Up Unused Images
+
+on:
+  schedule:
+    # On the first of every month at 2 am
+    - cron: '0 2 1 * *'
+  workflow_dispatch:
+
+concurrency:
+  group: clean-up-images-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  find_unused_images:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Retrieve Credentials
+        id: import-secrets
+        uses: hashicorp/[email protected]
+        with:
+          url: https://vault.prism.spectrocloud.com
+          method: approle
+          roleId: ${{ secrets.VAULT_ROLE_ID }}
+          secretId: ${{ secrets.VAULT_SECRET_ID }}
+          secrets: /providers/github/organizations/spectrocloud/token?org_name=spectrocloud token | VAULT_GITHUB_TOKEN
+
+      - id: checkout
+        name: Checkout Repository
+        uses: actions/checkout@v4
+        with:
+          token: ${{ steps.import-secrets.outputs.VAULT_GITHUB_TOKEN }}
+
+      - name: Setup Nodejs
+        uses: actions/setup-node@v4
+        with:
+          node-version: "20"
+
+      - run: npm ci
+
+      - name: Find unused images
+        run: make find-unused-images
+
+      - name: Set Git User
+        # see: https://github.com/actions/checkout/issues/13#issuecomment-724415212
+        run: |
+          git config --global user.name "github-actions[bot]"
+          git config --global user.email "41898282+github-actions[bot]@users.noreply.github.com"
+
+      - name: Create PR with unused images
+        run: |
+          unused_image_count=$(wc -l < unused_images.json)
+          if unused_image_count == 0; then
+            "No images found to remove."
+            exit 0
+          fi
+
+          # Ensure that we are on master.
+          git checkout master
+
+          # Create a new branch.
+          branch_name="clean-up-unused-images-$(date +%Y%m%d%H%M%S)"
+          git checkout -b "$branch_name"
+
+          # Remove all the images identified as unused.
+          for img in $(cat unused_images.json); do
+            rm static/assets/docs/images/$img
+          done
+
+          # Construct backport labels.
+          backport_labels="auto-backport"
+          for branch in $(cat evaluated_branches.json); do
+            if [[ $branch =~ version-[0-9]+(-[0-9]+)*$ ]]; then
+              backport_labels+=",backport-$branch"
+            fi 
+          done
+
+          # Clean up results file.
+          rm unused_images.json
+          rm evaluated_branches.json
+
+          # Commit and push branch
+          git add .
+          git commit -m "docs: clean up unused images"
+          git push origin $branch_name
+
+          # Create the pull request
+          pr_body='
+          ## Describe the Change
+          This PR removes images identified as unused across all our branches. 
+          The images are identified using `scripts/find-unused-images.sh` script.
+          Please review this PR carefully before merging it.'
+
+          output=$(gh pr create --base master --title "docs: clean up librarium unused images " --body "$pr_body" --label "$backport_labels")
+          pr_url=$(echo "$output" | grep -o "https://[^ ]*")
+          echo "PR successfully created $pr_url."
+
+          echo "GITHUB_CREATED_CLEANUP_PR=$pr_url" >> $GITHUB_ENV
+        env:
+          GH_TOKEN: ${{ steps.import-secrets.outputs.VAULT_GITHUB_TOKEN }}
+
+      - name: Slack Notification
+        if: ${{ env.GITHUB_CREATED_CLEANUP_PR  != ''}}
+        uses: rtCamp/[email protected]
+        env:
+          SLACK_WEBHOOK: ${{ secrets.SLACK_PRIVATE_TEAM_WEBHOOK }}
+          SLACK_USERNAME: "spectromate"
+          SLACK_ICON_EMOJI: ":ok_hand:"
+          SLACK_COLOR: ${{ job.status }}
+          SLACKIFY_MARKDOWN: true
+          ENABLE_ESCAPES: true
+          SLACK_MESSAGE: 'A new PR with unused images to clean up was created. Please review ${{env.GITHUB_CREATED_CLEANUP_PR}} for more details.'

.gitignore

@@ -77,3 +77,5 @@ static/img/packs
 vale/styles/spectrocloud/
 vale/styles/spectrocloud-docs-internal/
 vale/styles/config/vocabularies/spectrocloud-vocab
+
+unused_images.json

Makefile

@@ -65,7 +65,6 @@ clean-visuals:
 	@echo "Cleaning visual regression tests"
 
 	rm -rf test-results/  playwright-report/  screenshots/
-
 
 ##@ npm Targets
 
@@ -221,6 +220,12 @@ format-images: ## Format images
 	@echo "formatting images in /static/assets/docs/images/ folder"
 	./scripts/compress-convert-images.sh
 
+###@ Find unused images assets
+
+find-unused-images:
+	@echo "Find unused image assets"
+	./scripts/find-unused-images.sh	
+
 ###@ Generate _partials/index.ts required to automatic partials usage.
 
 generate-partials: ## Generate

scripts/find-unused-images.sh

@@ -0,0 +1,53 @@
+#!/bin/bash
+
+# Enable error handling
+set -e
+
+# Create a list of all the images we have and save it to a json.
+# Trim the path static/assets/docs/images.
+find static/assets/docs/images -type f \( -name "*.gif" -o -name "*.webp" \) ! -name ".DS_STORE" ! -name ".DS_Store" | sed 's|static/assets/docs/images||g'  > all_images.json
+image_count=$(wc -l < all_images.json)
+echo "Detected $image_count .webp and .gif assets in static/assets/docs/images..." 
+
+# Fetch all branches
+git fetch --all
+
+# List all the version branches
+branches="master"
+version_branches=$(git branch -a | grep -E 'version-[0-9]+(-[0-9]+)*$')
+for version_branch in $version_branches; do
+    # Remove leading spaces and remote prefix (if any)
+    version_branch=$(echo $version_branch | sed 's/ *//;s/remotes\/origin\///')
+
+    branches+=" $version_branch"
+done
+
+echo "Evaluating the following branches for image usage: { $branches }"
+echo "$branches" > evaluated_branches.json
+
+for current_branch in $branches; do    
+    git checkout $current_branch
+
+    find docs -type f -name "*.md" -exec grep -Hn -E "\.webp|\.gif" {} \; > docs_used_images.json
+    find _partials -type f -name "*.mdx" -exec grep -Hn -E "\.webp|\.gif" {} \; > partials_used_images.json
+    cat docs_used_images.json partials_used_images.json > used_images.json
+
+    line_number=1
+    for img in $(cat all_images.json); do
+        if grep -q $img used_images.json; then
+            sed -i "${line_number}s|.*|${img},FOUND_USED|" all_images.json
+        fi
+        ((line_number++))
+    done
+done
+
+# Remove all marked used files to make up the list 
+sed '/FOUND_USED/d' all_images.json > unused_images.json
+unused_image_count=$(wc -l < unused_images.json)
+echo "Detected $unused_image_count unused webp assets in static/assets/docs/images that can be safely removed." 
+
+# Clean up files
+rm all_images.json
+rm docs_used_images.json
+rm partials_used_images.json
+rm used_images.json