heapprof.rb

# encoding: utf-8
require 'set'
require 'json'
require 'pp'

class HeapDump
  def initialize(data)
    @objects = data.split("\n").map.with_index{ |line, idx|
      begin
        JSON.parse(line)
      rescue JSON::ParserError => e
        STDERR.puts "*** Parse error on line #{idx+1}: #{e.inspect}"
      end
    }.compact

    by_addr = by_address
    @objects.each do |o|
      addr = o['address']
      next unless o['references']

      o['references'].each do |ref|
        if refd = by_addr[ref]
          (refd['parents'] ||= []) << addr
        end
      end
    end
  end
  attr_reader :objects

  def count_files(with_line=false)
    objects.inject(Hash.new(0)) do |hash, o|
      key = with_line ? "#{o['file']}:#{o['line']}" : o['file']
      hash[key] += 1
      hash
    end.sort_by{ |k,v| -v }
  end

  def count_objects
    objects.inject(Hash.new(0)) do |hash, o|
      hash[o['type']] += 1
      hash
    end
  end

  def classes
    @classes ||= objects.inject(Hash.new) do |hash, o|
      hash[o['address']] = o['name'] if o.key?('name')
      hash
    end
  end

  def count_instances
    Hash[
      *objects.inject(Hash.new(0)) do |hash, o|
        class_name = classes[o['class']]
        hash[class_name] += 1
        hash
      end.select{ |k,v| v > 1 }.sort_by{ |k,v| v }.flatten(1)
    ]
  end

  def by_address
    @by_address ||= objects.inject(Hash.new) do |hash, obj|
      hash[obj['address']] = obj
      hash
    end
  end

  def display_object(obj, level)
    addr, type = obj.values_at('address', 'type')
    addr, type = obj.values_at('type', 'root') if type == 'ROOT'
    type = obj['node_type'] if type == 'NODE'

    printf "%s (%s %s)", " "*(level*2), addr, type
    case type
    when 'STRING'
      printf " %s: %s", classes[obj['class']], obj['value'].inspect
    when 'CLASS', 'MODULE'
      printf " %s", obj['name']
    when 'ARRAY', 'HASH'
      printf " %s: length:%d", classes[obj['class']] || '(hidden)', obj['size'] || obj['length']
    when 'DATA'
      printf " %s (struct %s)", classes[obj['class']] || '(hidden)', obj['struct'] || 'unknown'
    end
    puts
  end

  def traverse_up(obj, level=0, seen=Set.new)
    addr = obj['address']
    display_object(obj, level)
    seen << addr

    parents = (obj['parents'] || []).map{ |ref| by_address[ref] }
    parents = parents.select{ |o| o['type'] == 'ROOT' } if toplevel?(obj)
    parents.each do |o|
      next if seen.include?(o['address'])
      traverse_up(o, level+1, seen)
    end
    #seen.merge parents.map{ |o| o['address'] }
  end

  def toplevel?(obj)
    obj['type'] == 'CLASS' && %w[Object Hash Array String].include?(obj['name'])
  end

  def traverse_down(obj, level=0, seen=Set.new)
    return if seen.include?(obj['address'])
    display_object(obj, level)
    seen << obj['address']

    (obj['references'] || []).each do |addr|
      if ref = by_address[addr]
        traverse_down(ref, level+1, seen)
      end
    end
  end
end

dump = HeapDump.new(ARGF.read)
pp dump.count_files(true)
#p dump.count_objects
#p dump.count_instances
#dump.objects.select{ |o| o['type'] == "STRING" && o['value'] == '[' }.each do |obj|
#  dump.traverse_up(obj)
#end
#dump.traverse_down(dump.by_address['0x7fbc8b02a0e0'])