Xapian high offset performance

simple_xapian_search.rb

require 'xapian'

def setup_xapian(database_path)
  $db_path = File.join(database_path)

  begin
      $db = Xapian::Database.new($db_path)
      $enquire = Xapian::Enquire.new($db)
  rescue IOError => e
      raise "Failed to open Xapian database #{$db_path}: #{e.message}"
  end

  $query_parser = Xapian::QueryParser.new
  $query_parser.database = $db
  $query_parser.default_op = Xapian::Query::OP_AND


  $query_parser.add_boolean_prefix("model", "M")
  $query_parser.add_boolean_prefix("modelid", "I")
end

class SimpleXapianSearch
  def initialize(model_classes, query_string, options = {}, user_query = nil)
    setup_xapian(options[:database_path])

    model_query = Xapian::Query.new(Xapian::Query::OP_OR, model_classes.map { |mc| "M#{ mc }" })
    user_query = $query_parser.parse_query(
                           query_string,
                           Xapian::QueryParser::FLAG_BOOLEAN | Xapian::QueryParser::FLAG_PHRASE |
                           Xapian::QueryParser::FLAG_LOVEHATE |
                           Xapian::QueryParser::FLAG_SPELLING_CORRECTION)

    query = Xapian::Query.new(Xapian::Query::OP_AND, model_query, user_query)

    limit = options[:limit] || 25
    offset = options[:offset] || 0
    sort_by_ascending = true

    $enquire.query = query

    if ENV.key?('SORT_BY_VALUE_THEN_RELEVANCE')
      $enquire.sort_by_value_then_relevance!(0, sort_by_ascending)
    end

    $enquire.collapse_key = 1

    matches = $enquire.mset(offset, limit, 100)
  end
end

$xapian_database_path = "#{ Dir.pwd }/lib/acts_as_xapian/xapiandbs/#{Rails.env}"

Benchmark.bmbm do |x|
  x.report("offset-0") {
    low_offset  = SimpleXapianSearch.new(['InfoRequestEvent'], 'information', :offset => 0, :database_path => $xapian_database_path)
  }
  x.report("offset-500") {
    high_offset = SimpleXapianSearch.new(['InfoRequestEvent'], 'information', :offset => 500, :database_path => $xapian_database_path)
  }
end

# Rehearsal ----------------------------------------------
# offset-0     0.400000   0.010000   0.410000 (  0.411031)
# offset-500   0.430000   0.020000   0.450000 (  0.448809)
# ------------------------------------- total: 0.860000sec
#
#                  user     system      total        real
# offset-0     0.410000   0.000000   0.410000 (  0.416388)
# offset-500   0.390000   0.020000   0.410000 (  0.420391)