FQL vs Graph

TA8522-benchmarks.rb

#!/usr/bin/env ruby

require 'rubygems'
require 'koala'
require 'benchmark'
require 'optparse'
require 'yaml'
require 'json'

@session_token = nil
@count         = 10
@notes         = []

optparse = OptionParser.new do |options|
  options.banner = "Usage: #{$0} [options]"
  options.on "-c", "--count COUNT", "How many iterations to run for benchmark" do |@count| end
  options.on "-t", "--token TOKEN", "Facebook Application Session Token" do |@session_token| end
  options.on "-i", "--ids ID_LIST", "Comma-separated list of FB page_ids" do |@ids| end
  options.on "-b", "--begin BEGINTIME", "Unix timestamp for beginning of report time" do |@begin_time| end
  options.on "-e", "--end ENDTIME", "Unix timestamp for ending of report time" do |@end_time| end
  options.on "-h", "--help"
end

optparse.parse!

raise "No --ids option provided." unless @ids
raise "No --token option provided." unless @session_token
raise "No --begin option provided." unless @begin_time
raise "No --end option provided." unless @end_time

@count = @count.to_i
@ids = @ids.to_s.split(',').map { |x| x.strip.to_i }

api = Koala::Facebook::API.new(@session_token)
posts = nil
insights = nil
fql = nil

class BatchBatch
  MAXBATCHSIZE = 50
  def initialize(api)
    @api     = api
    @queries = []
  end
  def method_missing(method_id, *arguments)
    @queries << [method_id, arguments]
  end
  def results
    @results = []
    until @queries.empty?
      query_count = 1
      @results.concat(@api.batch do |batch|
        until query_count > MAXBATCHSIZE or @queries.empty?
          query_method, query_args = @queries.shift
          batch.send(query_method, *query_args)
          query_count += 1
        end
      end.to_a)
    end
    @results
  end
  def self.for(api, &block)
    bb = BatchBatch.new(api)
    yield bb
    bb.results
  end
end

Benchmark.bm(60) do |bm|

  bm.report "'posts' from fql for #{@ids.size} pages" do
    @count.times do |n|
      posts = BatchBatch.for(api) do |batch|
        @ids.each do |id|
          fql = <<-end
            SELECT post_id, message
            FROM stream
            WHERE source_id = #{id}
          end
          batch.graph_call("method/fql.query", {:query => fql}, "post")
        end
      end
      posts.inspect
    end
  end

  @notes << "'posts' from fql batch for #{@ids.size} pages returned #{posts.flatten.size} posts."

  bm.report "'posts' from graph batch for #{@ids.size} pages" do
    @count.times do |n|
      posts = BatchBatch.for(api) do |batch|
        @ids.each do |id|
          batch.get_connections(id, 'posts', 'fields' => %w[from message type created_time], 'since' => @begin_time, 'until' => @end_time)
        end
      end
      posts.inspect
    end
  end

  @notes << "'posts' from graph batch for #{@ids.size} pages returned #{posts.flatten.size} posts."

  all_posts = posts.flatten

  bm.report "'posts' from graph non-batched for #{@ids.size} pages" do
    @ids.each do |id|
      @count.times do |n|
        posts = api.get_connections(id, 'posts', 'fields' => %w[from message type created_time], 'since' => @begin_time, 'until' => @end_time)
        posts.inspect
      end
      @notes << "'posts' from graph non-batched for #{id} returned #{posts.flatten.size} posts."
    end
  end

  post_ids = all_posts.flatten.map { |post| post['id'] }

  bm.report "'insights' from fql single query" do
    @count.times do |n|
      fql = <<-end
        SELECT object_id, value, period, end_time
        FROM insights
        WHERE object_id IN (#{ post_ids.map{ |id| id.to_s.inspect }.join(',') })
        AND metric IN (
          'post_impressions_organic_unique',
          'post_impressions_unique',
          'post_engaged_users',
          'post_story_adds_unique'
        )
        AND period = 0
      end
      insights = api.fql_query(fql)
      insights.inspect
    end
  end

  @notes << "'insights' from fql returned #{insights.flatten.size} data points."

  bm.report "'insights' from graph batch for #{post_ids.size} posts" do
    @count.times do |n|
      insights = BatchBatch.for(api) do |batch|
        post_ids.each do |post_id|
          batch.get_connections(post_id, 'insights/post_impressions_organic_unique/lifetime')
          batch.get_connections(post_id, 'insights/post_impressions_unique/lifetime')
          batch.get_connections(post_id, 'insights/post_engaged_users/lifetime')
          batch.get_connections(post_id, 'insights/post_story_adds_unique/lifetime')
        end
      end
      insights.inspect
    end
  end

  @notes << "'insights' from graph batch for #{post_ids.size} posts returned #{insights.flatten.size} data points."

end

unless @notes.empty?
  puts
  puts "NOTES"
  puts @notes
end