upstream-rocketchat/.github/actions/docker-image-size-tracker/action.yml

name: 'Docker Image Size Tracker'
description: 'Track and report Docker image sizes in Pull Requests'
author: 'Rocket.Chat'

inputs:
  github-token:
    description: 'GitHub token for commenting on PRs'
    required: true
  ci-pat:
    description: 'GitHub token for committing to history branch'
    required: true
  registry:
    description: 'Container registry (e.g., ghcr.io)'
    required: false
    default: 'ghcr.io'
  repository:
    description: 'Repository name (e.g., rocketchat)'
    required: true
  tag:
    description: 'Image tag to measure'
    required: true
  baseline-tag:
    description: 'Baseline tag to compare against'
    required: false
    default: 'develop'
  platform:
    description: 'Platform architecture to compare (amd64 or arm64)'
    required: false
    default: 'amd64'
  size-thresholds:
    description: 'Optional JSON: per-image trigger thresholds. Only comment when an image increase exceeds its threshold. Example: {"rocketchat":{"mb":50,"percent":5},"omnichannel":{"mb":10,"percent":2}}'
    required: false
    default: ''
  fail-thresholds:
    description: 'Optional JSON: per-image fail thresholds. Task fails when an image increase exceeds its threshold (mb and/or percent). Example: {"rocketchat":{"mb":100,"percent":15}}'
    required: false
    default: ''

outputs:
  total-size:
    description: 'Total size in bytes'
    value: ${{ steps.measure.outputs.total-size }}
  size-diff:
    description: 'Size difference in bytes'
    value: ${{ steps.compare.outputs.size-diff }}
  size-diff-percent:
    description: 'Size difference percentage'
    value: ${{ steps.compare.outputs.size-diff-percent }}
  comment-triggered:
    description: 'Whether to post PR comment (only when size is bigger and thresholds met)'
    value: ${{ steps.compare.outputs.comment-triggered }}
  failed:
    description: 'True if image size exceeded fail-thresholds'
    value: ${{ steps.compare.outputs.failed }}

runs:
  using: 'composite'
  steps:
    - name: Download manifests
      uses: actions/download-artifact@018cc2cf5baa6db3ef3c5f8a56943fffe632ef53 # v6.0.0
      with:
        pattern: manifests-*
        path: /tmp/manifests
        merge-multiple: true

    - name: Measure image sizes from artifacts
      id: measure
      shell: bash
      env:
        PLATFORM: ${{ inputs.platform }}
        TAG: ${{ inputs.tag }}
      run: |
        echo "Reading image sizes from build artifacts for platform: $PLATFORM"

        declare -A sizes
        declare -a services_list
        total=0

        # Loop through service directories (same as publish workflow)
        shopt -s nullglob
        for service_dir in /tmp/manifests/*; do
          [[ -d "$service_dir" ]] || continue
          service="$(basename "$service_dir")"

          echo "Processing service: $service"
          services_list+=("$service")

          size=0
          # Read only the specified platform architecture
          manifest_file="$service_dir/$PLATFORM/manifest.json"
          if [[ -f "$manifest_file" ]]; then
            # docker manifest inspect -v puts the manifest under SchemaV2Manifest
            # (Docker schema2 images) or OCIManifest (OCI images, which buildx
            # produces by default). Coalesce so both formats are read.
            config_size=$(jq -r '(.SchemaV2Manifest // .OCIManifest).config.size // 0' "$manifest_file")
            layers_size=$(jq '[(.SchemaV2Manifest // .OCIManifest).layers[]?.size // 0] | add // 0' "$manifest_file")
            size=$((config_size + layers_size))

            echo "  → Found $manifest_file: $size bytes (config: $config_size, layers: $layers_size)"
          else
            echo "  ⚠ Manifest not found for platform $PLATFORM: $manifest_file"
          fi

          sizes[$service]=$size
          total=$((total + size))
        done

        echo "Total size (all services, $PLATFORM only): $total bytes"
        echo "total-size=$total" >> $GITHUB_OUTPUT

        # Save to JSON
        echo "{" > current-sizes.json
        echo "  \"timestamp\": \"$(date -u +%Y-%m-%dT%H:%M:%SZ)\"," >> current-sizes.json
        echo "  \"tag\": \"$TAG\"," >> current-sizes.json
        echo "  \"total\": $total," >> current-sizes.json
        echo "  \"services\": {" >> current-sizes.json

        first=true
        for service in "${services_list[@]}"; do
          if [[ "$first" == "true" ]]; then
            first=false
          else
            echo "," >> current-sizes.json
          fi
          echo "    \"$service\": ${sizes[$service]}" >> current-sizes.json
        done

        echo "  }" >> current-sizes.json
        echo "}" >> current-sizes.json

        echo "Current sizes saved:"
        cat current-sizes.json

        # Save services list for baseline measurement
        printf '%s\n' "${services_list[@]}" > /tmp/services-list.txt

    - name: Measure baseline
      id: baseline
      shell: bash
      continue-on-error: true
      env:
        REGISTRY: ${{ inputs.registry }}
        ORG: ${{ inputs.repository }}
        TAG: ${{ inputs.baseline-tag }}
        PLATFORM: ${{ inputs.platform }}
      run: |
        echo "Measuring baseline: $REGISTRY/$ORG/*:$TAG (platform: $PLATFORM)"

        declare -A sizes
        declare -a services_list
        total=0

        # Read services list from current measurement
        while IFS= read -r service; do
          services_list+=("$service")

          # Map service name to image name (handle rocketchat -> rocket.chat)
          if [[ "$service" == "rocketchat" ]] || [[ "$service" == "rocketchat-cov" ]]; then
            image_name="rocket.chat"
            [[ "$service" == "rocketchat-cov" ]] && image_name="rocket.chat-cov"
          else
            image_name="$service"
          fi

          image="$REGISTRY/$ORG/$image_name:$TAG"
          echo "  → Inspecting $image"

          size=0
          if manifest=$(docker manifest inspect "$image" 2>/dev/null); then
            # Check if it's a manifest list (multi-arch)
            if echo "$manifest" | jq -e '.manifests' > /dev/null 2>&1; then
              # Manifest list - find the specified platform
              echo "    → Multi-arch manifest detected, filtering for $PLATFORM"
              platform_digest=$(echo "$manifest" | jq -r --arg arch "$PLATFORM" '.manifests[] | select(.platform.architecture == $arch) | .digest' | head -1)

              if [[ -n "$platform_digest" ]]; then
                echo "      → Inspecting $PLATFORM platform: $platform_digest"
                if platform_manifest=$(docker manifest inspect "$REGISTRY/$ORG/$image_name@$platform_digest" 2>/dev/null); then
                  config_size=$(echo "$platform_manifest" | jq -r '.config.size // 0')
                  layers_size=$(echo "$platform_manifest" | jq '[.layers[]?.size // 0] | add // 0')
                  size=$((config_size + layers_size))
                  echo "        → Size: $size bytes"
                fi
              else
                echo "      ⚠ Platform $PLATFORM not found in manifest"
              fi
            else
              # Single arch manifest
              config_size=$(echo "$manifest" | jq -r '.config.size // 0')
              layers_size=$(echo "$manifest" | jq '[.layers[]?.size // 0] | add // 0')
              size=$((config_size + layers_size))
            fi
          fi

          sizes[$service]=$size
          total=$((total + size))
        done < /tmp/services-list.txt

        echo "baseline-total=$total" >> $GITHUB_OUTPUT

        echo "{" > baseline-sizes.json
        echo "  \"timestamp\": \"$(date -u +%Y-%m-%dT%H:%M:%SZ)\"," >> baseline-sizes.json
        echo "  \"tag\": \"$TAG\"," >> baseline-sizes.json
        echo "  \"total\": $total," >> baseline-sizes.json
        echo "  \"services\": {" >> baseline-sizes.json

        first=true
        for service in "${services_list[@]}"; do
          if [[ "$first" == "true" ]]; then
            first=false
          else
            echo "," >> baseline-sizes.json
          fi
          echo "    \"$service\": ${sizes[$service]}" >> baseline-sizes.json
        done

        echo "  }" >> baseline-sizes.json
        echo "}" >> baseline-sizes.json

    - name: Setup history storage
      id: history
      shell: bash
      run: |
        git config --global user.name "github-actions[bot]"
        git config --global user.email "github-actions[bot]@users.noreply.github.com"

        # Create a separate worktree for history branch
        mkdir -p /tmp/history-worktree

        # Try to fetch history branch
        if git ls-remote --heads origin image-size-history | grep -q image-size-history; then
          git fetch origin image-size-history
          git worktree add /tmp/history-worktree image-size-history
        else
          # Create orphan branch for history in worktree
          git worktree add --detach /tmp/history-worktree
          cd /tmp/history-worktree
          git checkout --orphan image-size-history
          git rm -rf . 2>/dev/null || true
          mkdir -p history
          echo "# Image Size History" > README.md
          echo "This branch stores historical image size data for tracking" >> README.md
          git add README.md
          git commit -m "Initialize image size history"
          git push origin image-size-history
          cd -
        fi

        mkdir -p /tmp/history-worktree/history

    - name: Load historical data
      shell: bash
      run: |
        # Load last 30 measurements and group by day (keep only last entry per day)
        echo "[]" > history-data.json
        if [[ -d /tmp/history-worktree/history ]]; then
          jq -s '.' /tmp/history-worktree/history/*.json 2>/dev/null | jq '
            sort_by(.timestamp) |
            group_by(.timestamp[0:10]) |
            map(.[-1]) |
            .[-30:]
          ' > history-data.json || echo "[]" > history-data.json
        fi

        count=$(jq 'length' history-data.json)
        echo "Loaded $count historical measurements (one per day)"

    - name: Save current measurement to history
      if: github.ref == 'refs/heads/develop'
      shell: bash
      env:
        CI_PAT: ${{ inputs.ci-pat }}
      run: |
        timestamp=$(date -u +%Y%m%d-%H%M%S)
        commit_sha="${{ github.sha }}"

        # Add commit info to current measurement
        jq --arg sha "$commit_sha" '. + {commit: $sha}' current-sizes.json > "/tmp/history-worktree/history/${timestamp}.json"

        cd /tmp/history-worktree
        git add "history/${timestamp}.json"
        git commit -m "Add measurement for ${timestamp} (${commit_sha:0:7})"
        git config --global user.email "ci@rocket.chat"
        git config --global user.name "rocketchat-ci[bot]"
        git config --global url.https://$CI_PAT@github.com/.insteadOf https://github.com/
        git push origin image-size-history
        cd -

        echo "Saved measurement to history"

    - name: Compare and generate report
      id: compare
      shell: bash
      env:
        SIZE_THRESHOLDS: ${{ inputs.size-thresholds }}
        FAIL_THRESHOLDS: ${{ inputs.fail-thresholds }}
        TAG: ${{ inputs.tag }}
        BASELINE_TAG: ${{ inputs.baseline-tag }}
      run: |
        current_total=$(jq -r '.total' current-sizes.json)

        if [[ ! -f baseline-sizes.json ]]; then
          echo "No baseline available"
          echo "size-diff=0" >> $GITHUB_OUTPUT
          echo "size-diff-percent=0" >> $GITHUB_OUTPUT
          echo "comment-triggered=false" >> $GITHUB_OUTPUT

          cat > report.md << 'EOF'
        # 📦 Docker Image Size Report

        **Status:** First measurement - no baseline for comparison

        **Total Size:** $(numfmt --to=iec-i --suffix=B $current_total)
        EOF
          exit 0
        fi

        baseline_total=$(jq -r '.total' baseline-sizes.json)
        diff=$((current_total - baseline_total))

        if [[ $baseline_total -gt 0 ]]; then
          percent=$(awk "BEGIN {printf \"%.2f\", ($diff / $baseline_total) * 100}")
        else
          percent=0
        fi

        echo "size-diff=$diff" >> $GITHUB_OUTPUT
        echo "size-diff-percent=$percent" >> $GITHUB_OUTPUT

        # Only comment when size is bigger than baseline; optionally require per-image thresholds
        THRESHOLDS="$SIZE_THRESHOLDS"
        FAIL_THRESHOLDS="$FAIL_THRESHOLDS"
        comment_triggered=false
        fail_triggered=false
        if [[ $diff -gt 0 ]]; then
          if [[ -z "$THRESHOLDS" ]] || [[ "$THRESHOLDS" == "{}" ]]; then
            comment_triggered=true
          fi
        fi

        color="gray"
        if (( $(awk "BEGIN {print ($percent > 0.01)}") )); then
          color="red"
        elif (( $(awk "BEGIN {print ($percent < -0.01)}") )); then
          color="green"
        fi

        # Generate report
        if [[ $diff -gt 0 ]]; then
          emoji="📈"
          badge="![](https://img.shields.io/badge/size-+${percent}%25-${color})"
          sign="+"
        elif [[ $diff -lt 0 ]]; then
          emoji="📉"
          badge="![](https://img.shields.io/badge/size--${percent}%25-${color})"
          sign="-"
        else
          emoji="➡️"
          badge="![](https://img.shields.io/badge/size-unchanged-${color})"
          sign=""
        fi

        cat > report.md << EOF
        # 📦 Docker Image Size Report

        ## $emoji Changes $badge

        | Service | Current | Baseline | Change | Percent |
        |---------|---------|----------|--------|---------|
        | **sum of all images** | **$(numfmt --to=iec-i --suffix=B $current_total)** | **$(numfmt --to=iec-i --suffix=B $baseline_total)** | **${sign}$(numfmt --to=iec-i --suffix=B ${diff#-})** | $badge |
        EOF

        # Get services dynamically from current-sizes.json, sorted by size (largest first)
        for service in $(jq -r '.services | to_entries | sort_by(-.value) | .[].key' current-sizes.json); do
          current=$(jq -r ".services.\"$service\"" current-sizes.json)
          baseline=$(jq -r ".services.\"$service\" // 0" baseline-sizes.json)
          service_diff=$((current - baseline))

          if [[ $baseline -gt 0 ]]; then
            service_percent=$(awk "BEGIN {printf \"%.2f\", ($service_diff / $baseline) * 100}")
          else
            service_percent=0
          fi

          # Check per-image thresholds for comment trigger (only when size increased)
          if [[ $diff -gt 0 ]] && [[ -n "$THRESHOLDS" ]] && [[ "$THRESHOLDS" != "{}" ]]; then
            threshold_mb=$(echo "$THRESHOLDS" | jq -r ".\"$service\".mb // empty")
            threshold_pct=$(echo "$THRESHOLDS" | jq -r ".\"$service\".percent // empty")
            if [[ -n "$threshold_mb" ]] || [[ -n "$threshold_pct" ]]; then
              exceeded=false
              if [[ -n "$threshold_mb" ]] && [[ $service_diff -ge $(awk "BEGIN {printf \"%.0f\", $threshold_mb * 1048576}") ]]; then
                exceeded=true
              fi
              if [[ -n "$threshold_pct" ]] && [[ $service_percent != "0.00" ]] && (( $(awk "BEGIN {print ($service_percent >= $threshold_pct)}") )); then
                exceeded=true
              fi
              [[ "$exceeded" == "true" ]] && comment_triggered=true
            fi
          fi

          # Check per-image fail thresholds (task fails when exceeded)
          if [[ $diff -gt 0 ]] && [[ -n "$FAIL_THRESHOLDS" ]] && [[ "$FAIL_THRESHOLDS" != "{}" ]]; then
            fail_mb=$(echo "$FAIL_THRESHOLDS" | jq -r ".\"$service\".mb // empty")
            fail_pct=$(echo "$FAIL_THRESHOLDS" | jq -r ".\"$service\".percent // empty")
            if [[ -n "$fail_mb" ]] || [[ -n "$fail_pct" ]]; then
              exceeded=false
              if [[ -n "$fail_mb" ]] && [[ $service_diff -ge $(awk "BEGIN {printf \"%.0f\", $fail_mb * 1048576}") ]]; then
                exceeded=true
              fi
              if [[ -n "$fail_pct" ]] && [[ $service_percent != "0.00" ]] && (( $(awk "BEGIN {print ($service_percent >= $fail_pct)}") )); then
                exceeded=true
              fi
              [[ "$exceeded" == "true" ]] && fail_triggered=true
            fi
          fi

          color="gray"
          if (( $(awk "BEGIN {print ($service_percent > 0.01)}") )); then
            color="red"
          elif (( $(awk "BEGIN {print ($service_percent < -0.01)}") )); then
            color="green"
          fi

          if [[ $service_diff -gt 0 ]]; then
            badge="![](https://img.shields.io/badge/size-+${service_percent}%25-${color})"
            sign="+"
          elif [[ $service_diff -lt 0 ]]; then
            badge="![](https://img.shields.io/badge/size--${service_percent}%25-${color})"
            sign="-"
          else
            badge="![](https://img.shields.io/badge/size-unchanged-${color})"
            sign=""
          fi

          echo "| $service | $(numfmt --to=iec-i --suffix=B $current) | $(numfmt --to=iec-i --suffix=B $baseline) | ${sign}$(numfmt --to=iec-i --suffix=B ${service_diff#-}) | $badge |" >> report.md
        done

        # Generate historical trend chart
        echo "" >> report.md
        echo "## 📊 Historical Trend" >> report.md
        echo "" >> report.md

        # Load history and generate mermaid chart
        history_count=$(jq 'length' history-data.json)

        if [[ $history_count -gt 0 ]]; then
          # Get all services from current build
          all_services=$(jq -r '.services | keys | .[]' current-sizes.json | sort)

          # Generate Mermaid chart data
          x_labels=""
          declare -A service_data

          # Initialize service data arrays
          for service in $all_services; do
            service_data[$service]=""
          done

          # Process historical data
          while IFS= read -r line; do
            timestamp=$(echo "$line" | jq -r '.timestamp')
            date_label=$(date -d "$timestamp" +"%m/%d %H:%M")

            if [[ -z "$x_labels" ]]; then
              x_labels="\"$date_label\""
            else
              x_labels="$x_labels, \"$date_label\""
            fi

            # Add data point for each service
            for service in $all_services; do
              size=$(echo "$line" | jq -r ".services.\"$service\" // 0")
              size_gb=$(awk "BEGIN {printf \"%.2f\", $size / 1073741824}")

              if [[ -z "${service_data[$service]}" ]]; then
                service_data[$service]="$size_gb"
              else
                service_data[$service]="${service_data[$service]}, $size_gb"
              fi
            done
          done < <(jq -c '.[]' history-data.json)

          # Add current PR as last point
          current_date=$(date -u +"%m/%d %H:%M")
          x_labels="$x_labels, \"$current_date (PR)\""

          for service in $all_services; do
            size=$(jq -r ".services.\"$service\" // 0" current-sizes.json)
            size_gb=$(awk "BEGIN {printf \"%.2f\", $size / 1073741824}")
            service_data[$service]="${service_data[$service]}, $size_gb"
          done

          # Generate mermaid chart with multiple lines
          cat >> report.md << EOF
        \`\`\`mermaid
        ---
        config:
          theme: "dark"
          xyChart:
            width: 900
            height: 400
        ---
        xychart
          title "Image Size Evolution by Service (Last 30 Days + This PR)"
          x-axis [$x_labels]
          y-axis "Size (GB)" 0 --> 0.5
        EOF

          # Add a line for each service
          for service in $all_services; do
            echo "  line \"$service\" [${service_data[$service]}]" >> report.md
          done

          cat >> report.md << 'EOF'
        ```

        EOF

          # Add summary stats
          min_size=$(jq '[.[].total] | min' history-data.json)
          max_size=$(jq '[.[].total] | max' history-data.json)
          avg_size=$(jq '[.[].total] | add / length' history-data.json)

          cat >> report.md << EOF
        **Statistics (last $history_count days):**
        - 📊 Average: $(numfmt --to=iec-i --suffix=B ${avg_size%.*})
        - ⬇️ Minimum: $(numfmt --to=iec-i --suffix=B $min_size)
        - ⬆️ Maximum: $(numfmt --to=iec-i --suffix=B $max_size)
        - 🎯 Current PR: $(numfmt --to=iec-i --suffix=B $current_total)

        EOF

        else
          cat >> report.md << 'EOF'
        *No historical data available yet. History tracking starts after merging to develop.*

        EOF
        fi

        cat >> report.md << EOF

        <details>
        <summary>ℹ️ About this report</summary>

        This report compares Docker image sizes from this build against the \`$BASELINE_TAG\` baseline.

        - **Tag:** \`$TAG\`
        - **Baseline:** \`$BASELINE_TAG\`
        - **Timestamp:** $(date -u +"%Y-%m-%d %H:%M:%S UTC")
        - **Historical data points:** $history_count

        </details>
        EOF

        echo "comment-triggered=$comment_triggered" >> $GITHUB_OUTPUT
        echo "failed=$fail_triggered" >> $GITHUB_OUTPUT

        if [[ "$fail_triggered" == "true" ]]; then
          echo "::error::Docker image size exceeded fail-thresholds (mb/percent). Check the report for details."
          exit 1
        fi

    - name: Add report to Job Summary
      if: always()
      shell: bash
      run: |
        if [[ -f report.md ]]; then
          cat report.md >> $GITHUB_STEP_SUMMARY
        fi

    - name: Cleanup worktree
      if: always()
      shell: bash
      run: |
        if [[ -d /tmp/history-worktree ]]; then
          git worktree remove /tmp/history-worktree --force 2>/dev/null || true
        fi

branding:
  icon: 'package'
  color: 'blue'