load_sample.rb

#!/usr/bin/env ruby
#
# ruby load_sample.rb biosample/annotated_json dev 

require 'elasticsearch'
require 'json'
require 'pp'

#client = Elasticsearch::Client.new({log: true, hosts: { host: 'localhost', user: 'user', password: 'password' }})
client = Elasticsearch::Client.new({log: true, hosts: { host: 'localhost'}})


dir = ARGV[0]
index = ARGV[1] || 'genome_public_sample'
type = 'dm5'

#Dir.glob(dir + "/**/*.json").first(1).each do |file|
Dir.glob(dir + "/**/*.json").each do |file|
  puts file
  json_data = open(file) do |io|
      JSON.load(io)
  end
  #pp json_data.keys
  #pp json_data['@context']
  ary =[]

  json_data.each do |g|
    #next if g['@id'].match(/^http:\/\/purl.jp\/bio\/11\/meo\/MEO_/)
    #next if g['@type'].match(/(owl:Ontology|owl:AnnotationProperty)/)
    #puts g.to_json
    #meos = get_hierarchical_meo( g['@id'])
    #sleep(5)
    #id =  g['@id'].gsub('http://purl.jp/bio/11/meo/','')

    puts id = g['identifier']

    ary.push("{ \"create\": { \"_index\": \"#{index}\", \"_type\": \"#{type}\", \"_id\": \"#{id}\" }}") 
    ary.push(g.to_json)

    begin
      #client.create index: index, type: type, id: id, body: g.to_json
      #pp g.to_json
    rescue Exception => ex
      puts ex.backtrace.first + ": #{ex.message} (#{ex.class})"
    end
  end
  begin
    client.bulk body: ary
    #  pp g.to_json
  rescue Exception => ex
      puts ex.backtrace.first + ": #{ex.message} (#{ex.class})"
  end
end

#client.cluster.health
#puts client.cat.indices