jeremy-less-pathological-gcdata.rb

gistfile1.rb

HTML_ESCAPE = { '&' => '&amp;',  '>' => '&gt;',   '<' => '&lt;', '"' => '&quot;' }

def original(s)
  s.gsub(/[&"><]/n) { |special| HTML_ESCAPE[special] }
end

def new_range(s)
  s.encode(s.encoding, :xml => :attr)[1...-1]
end

def new_tr(s)
  s.encode(s.encoding, :xml => :attr).tr!('"', '')
end

def new_slice(s)
  t = s.encode(s.encoding, :xml => :attr)
  t[1, t.size - 1]
end

def new_delete(s)
  s.encode(s.encoding, :xml => :attr).delete!('"')
end

def new_gsub(s)
  s.encode(s.encoding, :xml => :attr).gsub!(/^"|"$/o, '')
end

def bench(method, string, times = 1000)
  GC.start
  GC.disable
  t0 = Time.now
  counts = ObjectSpace.count_objects
  o0 = counts[:TOTAL] - counts[:FREE]
  s = GC.malloc_allocated_size if GC.respond_to?(:malloc_allocated_size)
  c = GC.count
  times.times do
    send method, string
  end
  elapsed = Time.now - t0
  counts = ObjectSpace.count_objects
  allocated = counts[:TOTAL] - counts[:FREE]
  allocated_size = (GC.malloc_allocated_size - s).to_f/(1024.0*1024.0) if GC.respond_to?(:malloc_allocated_size)
  allocated_size ||= -1.0
  garbage_collections = GC.count - c
  GC.enable
  [elapsed, allocated, allocated_size, garbage_collections]
end

s = '<script>alert("foo & bar & baz");</script> Laborum next level organic nihil cardigan.'

[s, s * 10, s * 100, s * 1000].each do |string|
  %w(original new_range new_tr new_slice new_delete new_gsub).each do |method|
    elapsed, allocated, allocated_size, garbage_collections = bench method, string
    puts "%10s %5d chars:  %7.1fms elapsed %10d objects %.3f MB %5d collections" % [method, string.bytesize, elapsed * 1000, allocated, allocated_size, garbage_collections]
  end
end