Search ES

gistfile1.rb

# Searches ElasticSearch fulltext engine via Tire
  #
  # @param term     Search term to be used
  #                 It has to start with a prefix specifying the field(s) to be searched:
  #                 e.g.
  #                   _all:Apple
  #                   converted_text:Apple
  #                   field_summary:Apple, etc.
  #
  #                 Each Lucene search term can be used
  #
  # @param options  Hash with the following keys:
  #                   :page         page number [1..]
  #                   :per          how many items per page [1..]
  #                   :days         limit search to N days [1..]
  #                   :min_length   min size of article in converted_text field [1..]
  #                   :all          show all articles [true/false]
  #                   :sort         sort field to be used ['id', 'feed_timestamp']
  #                   :fields       array of fields that should be returned. If empty: return all available fields
  #
  def self.es_search(term, options={})
    n_days = options[:days] || 0
    n_days_ago = Time.now - n_days * 60 * 60 * 24
    min_length = options[:min_length] || 500
    if term.nil? or true == options[:all]
      use_all = true
    else
      use_all = false
    end

    sort_field = options[:sort]
    selected_fields =options[:fields]

    s = Tire.search 'posts' do

      fields(selected_fields) unless selected_fields.nil?
      n_millis = n_days_ago.tv_sec * 1000

      query do
        custom_score :script => "_score  * Math.pow( (doc['feed_timestamp'].date.getMillis() - #{n_millis}) / (time()-#{n_millis}), 2 )" do
          boolean do
            unless use_all
              # XXX make this configurable: - exact match/ OR / AND
              should   { string term, :default_operator => 'AND' }
              #should   { text '_all', term.split(':')[1], :operator => 'AND', :type => 'phrase' }

            else
              should   { all }
            end
            # don't show texts that were not converted
            must_not { string 'converted_text:N/A' }
          end
        end
      end

      # filter by time range
      filter :range, :feed_timestamp => {:from => n_days_ago, :to => Time.now} if n_days != 0

      # filter out articles with less than a specific length
      filter :numeric_range, :converted_text_size => {:from => min_length }

      # XXX make sort field parametrized
      # sort results, paginate
      sort { by sort_field, 'desc' } unless sort_field.nil?
      search_size = options[:per] || 10
      page = (((options[:page] || 1).to_i) -1) * search_size
      from page
      size search_size
    end
    puts s.to_curl
    s.results
  end