Breadth-first search of memprof heap dump

bfs.rb

require 'progressbar'
require 'shellwords'
require 'json'
require 'active_support/core_ext/hash/keys'
require 'set'
require 'thread' # queue

def main
  objs = load_objs
  $objs = objs
  # These object_ids identify ActiveRecord::Relations from the article
  # table which I'm presuming are the ones that are leaked.  I had to find
  # them in the memprof dump because when I compare
  # ObjectSpace.each_object(ActiveRecord::Relation) before and after
  # app.get("/"), the object_ids have a different format than those in the
  # dump.
  oids = Set.new ["0x508d418", "0x5413830"]
  path = search_for(objs, oids)
  p path
end

def load_objs
  filename = "myapp_heap.json"
  line_count = `wc -l #{filename.shellescape}`.strip.to_i
  bar = ProgressBar.new "Reading json", line_count
  objs = {}
  File.foreach("myapp_heap.json") do |line|
    obj = JSON.parse(line)
    obj.symbolize_keys!
    objs[obj[:_id]] = obj
    bar.inc
  end
  bar.finish
  objs
end

def search_for(objs, target_oids)
  queue = Queue.new
  $queue = queue
  seen = Set.new
  queue.enq ["globals"] # initial
  until queue.empty?
    path = queue.deq
    oid = path.last
    $stdout.write "#{oid} depth: #{path.length} / queued: #{queue.length} / seen: #{seen.length}  \r"
    next if seen.include? oid
    seen.add oid
    if target_oids.include? oid
      return path
    end
    (ch=child_oids(objs[oid])).each do |child|
      child_path = path.dup
      child_path.push child
      queue.enq child_path
    end
  end
  puts
end

def child_oids(obj)
  # Perhaps I'm missing some cases here... ? 
  case obj
  when String
    if /\A0x[0-9a-f]{1,16}\Z/.match obj
      [obj]
    else
      []
    end
  when Array
    obj.map {|v| child_oids(v) }.inject([], &:+)
  when Hash
    obj.map {|k, v| child_oids(v) }.inject([], &:+)
  else
    []
  end
end

if File.identical?(__FILE__, $0)
  main
end