@startuml
  title Создание отчёта о рабочем времени (face_work_time_json)

  actor Client
  participant "face/api" as API
  database "MongoDB\nuser_reports" as ReportsDB
  participant "report_checker\n(Worker)" as Checker
  participant "reporter.py" as Reporter
  participant "WorkTimeJsonReport" as WorkTime
  participant "data_loader" as DataLoader
  database "MongoDB\nface_events" as EventsDB
  database "MongoDB\nfaces" as FacesDB
  participant "tools.py" as Tools
  database "Redis" as Redis

  == 1. Создание задачи на отчёт ==
  Client -> API: POST /reports\n(type=face_work_time_json)
  API -> API: FaceReport.create()\nvalidate options, build query
  API -> ReportsDB: insert(status='in_queue')
  API --> Client: {success: true, id: report_id}

  == 2. Обработка отчёта воркером ==
  loop каждые SLEEP_INTERVAL секунд
      Checker -> ReportsDB: find_one_and_update\n(status='in_queue' → 'in_progress')
  end

  Checker -> Checker: ProcessPoolExecutor\ncreate_report_sync(report)
  note right: Новый процесс\nс asyncio.run()

  Checker -> Reporter: create_report(report)
  Reporter -> Reporter: @with_context\nустановить контекст пользователя
  Reporter -> Reporter: Выбрать класс из REGISTRY\n(face_work_time_json → WorkTimeJsonReport)
  Reporter -> WorkTime: make_report()

  == 3. Загрузка данных ==
  WorkTime -> DataLoader: get_events(query)
  DataLoader -> EventsDB: find(query)\n[best_shot_time, face_id, camera_id]
  EventsDB --> DataLoader: events[]
  DataLoader --> WorkTime: events

  WorkTime -> DataLoader: get_faces(galleries, events,\ncameras_in, cameras_out, tz)

  loop для каждого события
      DataLoader -> DataLoader: result[face_id] = GroupedFace(face_id)
      DataLoader -> DataLoader: group.handle(best_shot_time, direction)
  end

  DataLoader -> DataLoader: _get_faces_without_events()\n(лица из галерей без событий)
  DataLoader -> FacesDB: find(_id NOT IN face_ids,\ngallery IN galleries)
  FacesDB --> DataLoader: faces_without_events

  DataLoader -> DataLoader: face_ids = events + faces_without_events

  == 4. Обогащение данных фотографиями ==
  DataLoader -> DataLoader: _get_faces_data(face_ids)
  DataLoader -> FacesDB: find(_id IN face_ids)\n[person, photos, description, gallery_id]
  FacesDB --> DataLoader: faces_data[]

  loop для каждого face
      DataLoader -> DataLoader: _get_photo(face_data)
      alt photos пустой или None
          DataLoader --> DataLoader: return None
          note right #pink: **ПРИЧИНА 1**\nphotos отсутствует
      else photos есть
          DataLoader -> DataLoader: найти default фото\nили первое из списка
          DataLoader -> Tools: get_thumb_url(photo)
          alt есть thumbnails.thumbnail_200.url
              Tools --> DataLoader: url
          else есть thumbnails.thumbnail_200.obj_ref
              Tools -> Tools: storage.generate_presigned_url()
              Tools --> DataLoader: presigned_url
          else нет thumbnail_200
              Tools --> DataLoader: None
              note right #pink: **ПРИЧИНА 2**\nнет thumbnail_200
          end
      end
      DataLoader -> DataLoader: face.photo = photo_url
  end

  DataLoader --> WorkTime: List[GroupedFace]

  == 5. Генерация и сохранение ==
  WorkTime -> WorkTime: _generate(query, faces, tz, schedule)
  note right: Формирует JSON:\n[name, photo, face_id, ...]

  WorkTime -> Redis: setex(key, json, TTL=1 day)
  WorkTime -> ReportsDB: update(status='done',\n_redis_key=key)

  == 6. Получение результата ==
  Client -> API: GET /reports/{id}/json
  API -> ReportsDB: find(_id=id)
  API -> Redis: get(_redis_key)
  Redis --> API: json_data
  API --> Client: {summary: {...}, details: {...}}

  @enduml