Identify conflicting field mappings in ElasticSearch 1.x > 2.0 upgrade

es_index_conflict_mapping

#!/usr/bin/env ruby

require 'json'

if ARGV.size < 1
  puts "usage: #{__FILE__} <FILE1> [FILE2 ...]"
  puts "   eg: ./es_index_conflict_mapping applogs-2016.01.10.json applogs-2016.01.13.json applogs-2016.01.08.json applogs-2016.01.11.json applogs-2016.01.27.json"
  abort
end

class PropertyMapping < Struct.new(:path, :property, :type)
end

def parse_node(node, path = [], property_mappings = [])
  node.each do |key, value|
    if value.is_a?(Hash)
      properties = value['properties']
      if properties
        properties.each do |property, definition|
          type = definition['type']
          if type
            property_mappings << PropertyMapping.new(path + [key], property, type)
          elsif definition.has_key?('properties')
            property_mappings << PropertyMapping.new(path + [key], property, 'object')
          end
        end
      end
      parse_node(value, path + [key], property_mappings)
    end
  end

  property_mappings
end

document = {}

paths = ARGV.map { |path| File.expand_path(path.strip) }
paths.each do |path|
  begin
    document.merge!(JSON.parse(File.read(path)))
  rescue JSON::ParserError
    puts "malformed JSON in file."
    abort
  end
end

property_mappings = []
document.values.each do |index|
  property_mappings += parse_node(index['mappings'])
end

conflicts = {}
property_mappings.each do |property_mapping|
  conflicts[property_mapping.property] ||= {}
  conflicts[property_mapping.property][property_mapping.type] ||= []
  conflicts[property_mapping.property][property_mapping.type] << property_mapping.path.join('.')
  conflicts[property_mapping.property][property_mapping.type].uniq!
end

conflicts.each do |property, types|
  if types.size > 1
    puts property
    types.each do |type, paths|
      puts "  #{type}"
      puts "    #{paths.join(', ')}"
    end
    puts "---"
  end
end