Example of Bulk load Ntriples into Kasabi using kasabi.rb

bulk_load.rb

# Bulk load N-Triples into Kasabi
#
# Simple script to iterate through a large n-triples file and load it into Kasabi
#
# Be sure to set two environment variables: KASABI_DATASET and KASABI_API_KEY
#
# Script accepts a single command-line parameter which is the location of the file to upload

require 'rubygems'
require 'kasabi'

#URI of dataset api
dataset = ENV["KASABI_DATASET"]

#Developer API key
apikey = ENV["KASABI_API_KEY"]
  
#base url of API
base = ENV["KASABI_BASE_URI"] || "http://api.kasabi.com"
  
#create our client
dataset = Kasabi::Dataset.new("#{base}/dataset/#{dataset}", {:apikey => apikey})

#Uncomment for HTTP level debugging
#dataset.client.debug_dev = $stderr

puts "Dataset: #{dataset.endpoint}"
puts "API key: #{apikey}"

to_add = File.read(ARGV[0])

count = 0
lines = ""
File.open(ARGV[0], "r").each do |line|
  lines << line + "\n"
    
  count = count + 1
  if count % 9000 == 0
    
    puts "Submitting data to Kasabi dataset"
    begin  
      client = dataset.store_api_client()
      resp = client.store_data( lines, "text/turtle")
      puts "Data submitted. Update URI: #{resp}"
    rescue => e
      puts "FAIL"
      puts e
    end

    
    lines = ""    
    
  end
    
end
if lines != ""
  puts "Submitting test data to Kasabi dataset"  
  begin  
    client = dataset.store_api_client()
    resp = client.store_data( lines, "text/turtle")
    puts "Data submitted. Update URI: #{resp}"
  rescue => e
    puts "FAIL"
    puts e
  end
end
puts "Data loaded"