feat: add Playwright last-run composites for failed-only re-runs

[Ntnelbaba]

Summary

Adds lean composite actions under actions/playwright-last-run/ so Elementor Core and Pro can share Playwright failed-only on Actions re-run logic instead of duplicating inline workflow steps.

• restore: when github.run_attempt > 1, downloads the tiny prior artifact, restores .last-run.json, sets extra_args / PLAYWRIGHT_EXTRA_ARGS to --last-failed only when failedTests is non-empty (jq).
• save: when failures exist, uploads only .last-run.json with retention-days: 1, include-hidden-files: true, and if-no-files-found: ignore.
• First attempt / new commits: restore is a no-op (full shard). No test-results trees, traces, videos, or WP bootstrap in this action.

Jira

https://elementor.atlassian.net/browse/ED-25653

Inputs / outputs

See actions/playwright-last-run/README.md. Summary:

Action	Key inputs	Outputs
restore	artifact-name, last-run-paths, optional github-token, download-path	extra_args ("" or --last-failed)
save	artifact-name, last-run-paths, optional staging-dir	uploaded (true / false)

Example uses: (Core)

- name: Restore Playwright last-run
  uses: elementor/elementor-editor-github-actions/actions/playwright-last-run/restore@2fcc90f64bb91129cca56c0b80f4ae51fab2e0ba
  id: pw_last_run
  with:
    artifact-name: playwright-last-run-${{ matrix.shardIndex }}-${{ matrix.browser }}
    last-run-paths: |
      tests/playwright/test-results/.last-run.json

- name: Run Playwright
  run: npx playwright test --shard=${{ matrix.shardIndex }}/${{ matrix.shardTotal }} ${{ steps.pw_last_run.outputs.extra_args }}

- name: Save Playwright last-run
  if: always()
  uses: elementor/elementor-editor-github-actions/actions/playwright-last-run/save@2fcc90f64bb91129cca56c0b80f4ae51fab2e0ba
  with:
    artifact-name: playwright-last-run-${{ matrix.shardIndex }}-${{ matrix.browser }}
    last-run-paths: |
      tests/playwright/test-results/.last-run.json

Elements regression path: tests/elements-regression/test-results/.last-run.json.

Example uses: (Pro)

- uses: elementor/elementor-editor-github-actions/actions/playwright-last-run/restore@2fcc90f64bb91129cca56c0b80f4ae51fab2e0ba
  id: pw_last_run
  with:
    artifact-name: playwright-last-run-${{ matrix.shardIndex }}
    last-run-paths: |
      tests/playwright/test-results/.last-run.json

- uses: elementor/elementor-editor-github-actions/actions/playwright-last-run/save@2fcc90f64bb91129cca56c0b80f4ae51fab2e0ba
  if: always()
  with:
    artifact-name: playwright-last-run-${{ matrix.shardIndex }}
    last-run-paths: |
      tests/playwright/test-results/.last-run.json
    staging-dir: playwright-last-run-ci

After merge, pin workflows to the merge commit SHA instead of @main.

QA

• Bash syntax check on restore.sh / save.sh
• CI (Lint, PR Name Linter, tests)
• save sets include-hidden-files: true for .last-run.json (Pro #7675 validation)
• Consumer wiring in Core/Pro (re-pin to latest head SHA before merge)

  

✨ PR Description

1. Problem & Context

Implements optimized Playwright re-runs by persisting only the .last-run.json state between GitHub Action attempts, avoiding full shard execution on failures.

2. What Changed (Where)

• actions/playwright-last-run/restore/: New composite action and script to restore failure state.
• actions/playwright-last-run/save/: New composite action and script to stage failure state.
• actions/playwright-last-run/README.md: Documentation for implementation and inputs.

3. How It Works

save.sh uses jq to detect failed tests and uploads .last-run.json as a 1-day artifact. On re-runs (run_attempt > 1), restore.sh downloads this artifact, restores it to specified paths, and outputs --last-failed to trigger Playwright's targeted re-run mode.

4. Risks

Dependency on jq availability on the runner; mitigated by targeting ubuntu-latest where it is pre-installed.

Generated by LinearB AI and added by gitStream.
_{AI-generated content may contain inaccuracies. Please verify before using.
💡 Tip: You can customize your AI Description using Guidelines Learn how}