A script to parce a salesforce generated CSV into JSON and pass to Keen IO

salesforce-csv-process.rb

# Salesforce CSV import
#
# Author:        Jevon Wild <[email protected]>
# Creation Date: 15 DEC 2016
#
# This file contains one method for processing multiple CSVs in the same
# directory and returning a block of bulk events for immediate ingestion. It
# does assume the events are in a particular format (for the Addons) so it would
# probably need to be modified for other use

require 'csv'
require 'keen'
require 'time'

def process
  filenames =  Dir.glob("*.csv")

  events = {}

  filenames.each do |filename|
    collection_name = filename.split('_').last.split('.').first
    events[collection_name] = []


    file = File.read(filename)
    csv = CSV.parse(file, :headers => true)

    csv.each do |row|
      event = {
        keen: {}
      }
      row.each_with_index do |column, idx|
        key = column[0].downcase
        value = column[1]
        if key == 'timestamp_derived'
          event[:keen][:timestamp] = Time.parse(value).iso8601
        elsif key == 'timestamp'
        elsif ["run_time", "cpu_time", "db_total_time", "db_blocks", "db_cpu_time", "response_size", "rows_processed", "request_size"].include?(key)
          if value
            event[key] = value.to_i
          else
            event[key] = nil
          end
        else
          event[key] = value
        end
      end

      event[:keen][:addons] = [
        {
          name: "keen:date_time_parser",
          input: {
            date_time: "keen.timestamp"
          },
          output: "timestamp_info"
        }
      ]

      if event["client_ip"] =~ /^\d*\.\d*\.\d*\.\d*$/
        event[:keen][:addons].push({
          name: "keen:ip_to_geo",
          input: {
            ip: "client_ip"
          },
          output: "ip_geo_info"
        })
      end

      events[collection_name].push(event)
    end
  end
  return events
end


# Keen.project_id = ""
# Keen.write_key  = ""

# events = process
# puts JSON.pretty_generate(events)
# Keen.publish_batch(events)