Orbit/lib/tasks/ntu_mongo_files.rake

# encoding: utf-8 

namespace :mongo_files do

  MODELS = {  'ad_image' => 'file',
              'archive_file_multiple' => 'file',
              'asset' => 'data',
              'bulletin' => 'image',
              'bulletin_file' => 'file',
              'design' => 'zip_file',
              'design_file' => 'file',
              'gallery_image' => 'file',
              'image' => 'file',
              'lab_file' => 'file',
              'location_info' => 'file',
              'preview_file' => 'file',
              'project_file' => 'file',
              'research_file' => 'file',
              'site' => 'default_image',
              'stylesheet' => 'file_orig',
              'user' => 'avatar',
              'writing_book_file' => 'file',
              'writing_conference_file' => 'file',
              'writing_journal_file' => 'file',
              'writing_patent_file' => 'file' }

  IMAGE_UPLOADER_MODELS = %w[ad_image bulletin image site]

  task :clean => :environment do
    @files = Mongoid.database['fs.files']
    @chunks = Mongoid.database['fs.chunks']
    clean_duplicates
    clean_unused
    remove_objects
  end

  # Remove unused files from db
  def clean_unused
    db_array = @files.find().inject([]) do |db_files, file|
      db_files << file['filename']
      db_files
    end

    p "# of files in db: #{db_array.size}"

    name_array = MODELS.inject([]) do |model_files, (model, type)|
      model.classify.constantize.all.each do |item|
        url = item.send(type).url
        thumb_url = item.send(type).thumb.url rescue nil
        if url && !url.eql?('sign-in-logo.png')
          url = url.gsub('/gridfs/', '')
          model_files << url
        end
        if thumb_url && !thumb_url.eql?('sign-in-logo.png')
          thumb_url = thumb_url.gsub('/gridfs/', '')
          model_files << thumb_url
        end
      end
      model_files
    end

    p "# of files from uploaders: #{name_array.size}"
    useless_files = db_array - name_array
    p "# of unmatched files: #{useless_files.size}"

    useless_files.each do |file|
      id = @files.find_one('filename' => file)['_id']
      @files.remove('_id' => id)
      @chunks.remove('files_id' => id)
    end
    # p files.remove('filename' => { "$in" => useless_files})
  end


  # Remove duplicate files from db
  def clean_duplicates
    p "# of files in db: #{@files.count}"
    file_hash = @files.find().sort({uploadDate: -1}).inject({}) do |db_files, file|
      name = file['filename']
      if db_files[name]
        db_files[name] += [file['_id']]
      else
        db_files[name] = [file['_id']]
      end
      db_files
    end
    file_array = file_hash.inject([]) do |files, (key, value)|
      files += value.drop(1) if value.size > 1
      files
    end
    p "# of duplicate ids to delete: #{file_array.size}"
    file_array.each do |id|
      @files.remove('_id' => id)
      @chunks.remove('files_id' => id)
    end
  end

  task :file_size => :environment do
    files = Mongoid.database['fs.files']
    size_array = files.find().inject([]) do |size, file|
      # size << [file['length'], file['filename']]
      size << [file['filename'], file['length']]
      size
    end
    size_array.sort.each do |pair|
      p "#{pair[0]} - #{pair[1]}"
    end
    p size_array.size
  end

  def remove_objects
    # Destroy writing_journals and writing_journal_files
    Mongoid.database['writing_journals'].remove()
    Mongoid.database['writing_journal_files'].remove()

    db_array = @files.find().inject([]) do |db_files, file|
      db_files << file['filename'] if file['filename'] =~ /writing_journal_file|news_bulletin_file|image\/image/
      db_files
    end
    p db_array.size
    db_array.each do |file|
      id = @files.find_one('filename' => file)['_id']
      @files.remove('_id' => id)
      @chunks.remove('files_id' => id)
    end
  end

  def remove_unlinked
    # name_array = AdBanner.all.inject([]) do |files, banner|
    #   banner.ad_images.inject(files) do |image_files, image|
    #     url = image.file.url
    #     thumb_url = image.file.thumb.url rescue nil
    #     if url && !url.eql?('sign-in-logo.png')
    #       url = url.gsub('/gridfs/', '')
    #       image_files << url
    #     end
    #     if thumb_url && !thumb_url.eql?('sign-in-logo.png')
    #       thumb_url = thumb_url.gsub('/gridfs/', '')
    #       image_files << thumb_url
    #     end
    #     image_files
    #   end
    #   files
    # end
    name_array = Design.all.inject([]) do |files, design|
      files += design.images.inject(files) do |image_files, image|
        p image_files += get_url_and_thumb(image.file)
        image_files
      end
      files
    end
    p name_array.size
  end

  def get_url_and_thumb(file)
    url = file.url
    thumb_url = file.thumb.url rescue nil
    files = [url, thumb_url].inject([]) do |urls, current|
      urls << current.gsub('/gridfs/', '') if current && !current.eql?('sign-in-logo.png')
      urls
    end
    files
  end

  task :delete_rss_ann => :environment do
    Bulletin.where(create_user_id: nil).destroy
  end
end