btrc-hub/backend/app/services/material_sync_runner.rb

class MaterialSyncRunner
  Result = Struct.new(:imported, :updated, :unchanged, :suppressed, :failed,
                      :errors, keyword_init: true)

  def self.sync_enabled!
    results = MaterialSyncSource.enabled.order(:id).map { |source| new(source).sync! }

    Result.new(imported: results.sum(&:imported),
               updated: results.sum(&:updated),
               unchanged: results.sum(&:unchanged),
               suppressed: results.sum(&:suppressed),
               failed: results.sum(&:failed),
               errors: results.flat_map(&:errors))
  end

  def initialize source
    @source = source
  end

  def sync!
    result = Result.new(imported: 0, updated: 0, unchanged: 0,
                        suppressed: 0, failed: 0, errors: [])

    if @source.source_kind == 'google_drive_path'
      sync_google_drive_path!(result)
    else
      candidates.each do |candidate|
        next if candidate.blank?

        sync_candidate!(candidate, result)
      end
    end

    @source.update!(last_synced_at: Time.current)
    log_result(result)
    result
  rescue NotImplementedError, StandardError => e
    result.failed += 1
    result.errors << { source_id: @source.id, error: e.message }
    log_result(result)
    result
  end

  private

  def candidates
    case @source.source_kind
    when 'uri'
      [uri_candidate]
    when 'google_drive_file'
      [google_drive_file_candidate]
    when 'legacy_drive_path'
      raise NotImplementedError, 'legacy_drive_path material sync is not implemented'
    else
      raise NotImplementedError, "Unsupported material sync source_kind: #{ @source.source_kind }"
    end
  end

  def uri_candidate
    { source_kind: @source.source_kind,
      source_uri: @source.source_uri,
      source_path: @source.source_path,
      source_file_id: @source.source_file_id,
      url: @source.source_uri,
      tag: nil,
      created_by_user: @source.created_by_user,
      updated_by_user: @source.updated_by_user || @source.created_by_user }
  end

  def sync_candidate! candidate, result
    block = MaterialImportBlockMatcher.match_for_sha256(candidate[:file_sha256])
    if block
      result.suppressed += 1
      Rails.logger.info(
          material_sync_log(action: 'suppressed',
                            reason: block.reason,
                            normalized_source_key: candidate_normalized_source_key(candidate)))
      return
    end

    suppression = suppression_for(candidate)

    if suppression
      result.suppressed += 1
      Rails.logger.info(
          material_sync_log(action: 'suppressed',
                            normalized_source_key: suppression.normalized_source_key))
      return
    end

    import = MaterialSyncImporter.import!(candidate)
    result.public_send("#{ import.action }=", result.public_send(import.action) + 1)
  rescue StandardError => e
    result.failed += 1
    result.errors << { source_id: @source.id,
                       normalized_source_key: candidate_normalized_source_key(candidate),
                       error: e.message }
    Rails.logger.warn(
        material_sync_log(action: 'failed',
                          error: e.message,
                          normalized_source_key: candidate_normalized_source_key(candidate)))
  ensure
    close_candidate_tempfile(candidate)
  end

  def google_drive_path_candidates
    folder_id = google_drive_folder_id
    Enumerator.new do |entries|
      drive_client.each_material_file_under_folder(folder_id).each do |entry|
        entries << build_google_drive_candidate(entry)
      end
    end
  end

  def sync_google_drive_path! result
    folder_id = google_drive_folder_id
    scanned_count = 0

    drive_client.each_material_file_under_folder(folder_id) do |entry|
      scanned_count += 1
      sync_candidate!(build_google_drive_candidate(entry), result)
      log_google_drive_progress(folder_id, scanned_count, result) if progress_log_scan_count?(scanned_count)
    end

    log_google_drive_progress(folder_id, scanned_count, result, summary: true)
  end

  def google_drive_file_candidate
    entry = drive_client.fetch_material_file(google_drive_file_id)
    return nil unless entry

    build_google_drive_candidate(entry, single_file: true)
  end

  def build_google_drive_candidate entry, single_file: false
    relative_path =
        if single_file
          nil
        else
          entry[:relative_path]
        end
    export_path =
        MaterialSyncExportPath.build(prefix: @source.export_path_prefix,
                                     relative_path:,
                                     filename: entry[:name],
                                     source_file_id: entry[:id])

    { source_kind: 'google_drive_file',
      source_uri: entry[:web_view_link] || google_drive_file_url(entry[:id]),
      source_path: relative_path.presence || MaterialSyncExportPath.normalize_path(entry[:name]),
      source_file_id: entry[:id],
      filename: entry[:name],
      content_type: entry[:mime_type],
      file_downloader: lambda {
        drive_client.download_to_tempfile(entry[:id], filename: entry[:name])
      },
      file_sha256: entry[:sha256_checksum],
      export_path:,
      profile: @source.profile,
      tag: nil,
      url: nil,
      created_by_user: @source.created_by_user,
      updated_by_user: @source.updated_by_user || @source.created_by_user }
  end

  def candidate_normalized_source_key candidate
    MaterialSyncSuppression.normalize_source_key(source_kind: candidate[:source_kind],
                                                 source_uri: candidate[:source_uri],
                                                 source_path: candidate[:source_path],
                                                 source_file_id: candidate[:source_file_id])
  end

  def suppression_for candidate
    if candidate[:source_kind] == 'google_drive_file' && candidate[:source_path].present?
      return MaterialSyncSuppressionMatcher.match_google_drive_candidate(
          drive_file_id: candidate[:source_file_id],
          relative_path: candidate[:source_path])
    end

    MaterialSyncSuppressionMatcher.match(source_kind: candidate[:source_kind],
                                         source_uri: candidate[:source_uri],
                                         source_path: candidate[:source_path],
                                         source_file_id: candidate[:source_file_id])
  end

  def drive_client
    @drive_client ||= GoogleDrive::ApiClient.new
  end

  def google_drive_folder_id
    google_drive_target_id.tap do |id|
      if id.blank?
        raise ArgumentError, 'google_drive_path source_file_id or source_uri is required'
      end
    end
  end

  def google_drive_file_id
    google_drive_target_id.tap do |id|
      if id.blank?
        raise ArgumentError, 'google_drive_file source_file_id or source_uri is required'
      end
    end
  end

  def google_drive_target_id
    @source.source_file_id.presence || drive_client.extract_file_id(@source.source_uri)
  end

  def google_drive_file_url file_id
    "https://drive.google.com/file/d/#{ file_id }/view"
  end

  def close_candidate_tempfile candidate
    tempfile = candidate[:file_tempfile]
    return unless tempfile

    tempfile.close! unless tempfile.closed?
  rescue StandardError
    nil
  end

  def log_result result
    Rails.logger.info(material_sync_log(imported: result.imported,
                                        updated: result.updated,
                                        unchanged: result.unchanged,
                                        suppressed: result.suppressed,
                                        failed: result.failed))
  end

  def progress_log_scan_count? scanned_count
    scanned_count == 1 || (scanned_count % 50).zero?
  end

  def log_google_drive_progress folder_id, scanned_count, result, summary: false
    Rails.logger.info(
        material_sync_log(folder_id:,
                          scanned_count:,
                          imported: result.imported,
                          updated: result.updated,
                          unchanged: result.unchanged,
                          suppressed: result.suppressed,
                          failed: result.failed,
                          progress: summary ? 'summary' : 'scan'))
  end

  def material_sync_log fields
    { material_sync_source_id: @source.id,
      material_sync_source_name: @source.name }.merge(fields).to_json
  end
end