cleanup spikes from rrdtool xml dumps

clean_rrdb.rb

require 'rubygems'
require 'nokogiri'

class RRow

  attr_reader :exponent, :coefficient, :xml, :value

  def initialize(xml)
    @xml = xml
    @value_xml = xml.children.first
    @value = @value_xml.text
    @coefficient, @exponent = @value.strip.split(/e\+|-/)
    @sign = @value.match(/e([+\-])/)[1] if @exponent
    if @coefficient == 'NaN'
      @exponent = nil
    else
      @exponent = @exponent.to_i
      @exponent = - @exponent if @sign == '-'
      @coefficient = @coefficient.to_f
    end
  end

  def number
    return 0 unless exponent
    coefficient * 10 ** exponent
  end

  def <=>(other)
    number <=> other.number
  end

  def nullify
    @value_xml.content = " NaN "
  end

end


class Rra

  attr_reader :xml
  
  def initialize(xml)
    @xml = xml
  end

  def values
    @values ||= xml.search('value')
  end

  def rows
    @rows ||= xml.search('v').map { |r| RRow.new(r) }.sort.reverse
  end

  def anomaly_index
    max_exp = nil
    max_exp = ARGV[1].to_i if ARGV[1]

    position = nil
    candidates = rows
    candidates.each_with_index do |candidate, index|
      if max_exp && candidate.exponent && candidate.exponent >= max_exp
        position = index
        next
      end
      next_exponent = candidates[index + 1] ? candidates[index + 1].exponent : nil
      next if !(next_exponent.is_a?(Numeric)) || !candidate.exponent.is_a?(Numeric)
      position = index if (candidate.exponent - next_exponent >= 2)
    end
    position
  end

  def normalize
    return nil unless anomaly_index
    anomalies = rows[0..anomaly_index]
    p anomalies.map { |a| a.value }
    #sane = rows[anomaly_index + 1].value
    #to_replace = anomalies.first.value
    anomalies.each { |a| a.nullify }
    #[to_replace, sane]
  end

end

to_clean = File.open(ARGV[0]) { |f| f.read }
document = Nokogiri::XML.parse(to_clean)
databases = document.search('rra')
replacements = databases.map { |rra| results = Rra.new(rra).normalize }.compact
xml = document.to_xml
#replacements.each do |replacement|
#  xml.gsub!(replacement[0], replacement[1])
#end
File.open('/tmp/xml_foo', 'w') do |f|
  f.puts xml
end