benchmark read lines and scan lines

bm_read_lines.rb

# frozen_string_literal: true

require "benchmark"
require "strscan"

def open_file(filename = "development.log")
  File.open(filename, "r") do |io|
    yield io
  end
end

def action(line)
  # noop
end

Benchmark.bmbm do |bm|
  bm.report("#read [all]") do
    open_file do |io|
      io.read
    end
  end

  bm.report("#gets") do
    open_file do |io|
      while line = io.gets
        action(line)
      end
    end
  end

  bm.report("#each_line") do
    open_file do |io|
      io.each_line do |line|
        action(line)
      end
    end
  end

  [128, 512, 1024, 4096, 8192].each do |size|
    bm.report("#sysread [#{size}]") do
      open_file do |io|
        buf = +""
        rest = nil
        while buf << io.sysread(size)
          buf.each_line do |line|
            unless line[-1] == "\n"
              rest = line
              break
            end

            action(line)
            rest = nil
          end

          buf = rest || +""
        end
      rescue EOFError
        action(rest) if rest
      end
    end
  end

  [128, 512, 1024, 4096, 8192].each do |size|
    bm.report("#read [#{size}]") do
      open_file do |io|
        buf = +""
        rest = nil
        while buf << io.read(size)
          buf.each_line do |line|
            unless line[-1] == "\n"
              rest = line
              break
            end

            action(line)
            rest = nil
          end

          if io.eof?
            action(rest) if rest
            break
          end

          buf = rest || +""
        end
      end
    end
  end

  [128, 512, 1024, 4096, 8192].each do |size|
    bm.report("#readpartial [#{size}]") do
      open_file do |io|
        buf = +""
        rest = nil
        while buf << io.readpartial(size)
          buf.each_line do |line|
            unless line[-1] == "\n"
              rest = line
              break
            end

            action(line)
            rest = nil
          end

          if io.eof?
            action(rest) if rest
            break
          end

          buf = rest || +""
        end
      end
    end
  end

  [128, 512, 1024, 4096, 8192].each do |size|
    bm.report("StringScanner [#{size}]") do
      s = StringScanner.new(+"")
      open_file do |io|
        while s << io.readpartial(size)
          while line = s.scan_until(/\n/)
            action(line)
          end

          if io.eof?
            action(s.rest) if s.rest?
            break
          end
        end
      end
    end
  end
end

bm_scan_lines.rb

# frozen_string_literal: true

require "benchmark"
require "strscan"

def open_file(filename = "development.log")
  File.open(filename, "r") do |io|
    yield io
  end
end

def time(year, month, day, hour, min, sec, usec)
  Time.local(year.to_i, month.to_i, day.to_i, hour.to_i, min.to_i, sec.to_i, usec.to_i)
end

def action(*args)
  $result << args if $result.size < 100
end

def setup
  $result = []
end

def check
  return if $expected == $result
  raise unless $expected.size == $result.size

  $expected.each_with_index do |expected, idx|
    result = $result[idx]
    unless expected == result
      puts
      p expected
      p result
      raise
    end
  end
end

pattern = "(.), \\[(\\d{4})-(\\d{2})-(\\d{2})T(\\d{2}):(\\d{2}):(\\d{2}).(\\d+) #(\\d+)\\] \\S+ -- : "

setup
open_file do |io|
  io.read.scan(/^#{pattern}(.*)\n/o) do |severity, *time_segs, pid, message|
    time = time(*time_segs)
    action(severity, time, pid.to_i, message.chomp)
  end
end
$expected = $result

Benchmark.bmbm do |bm|
  bm.report("#read [all]") do
    setup
    open_file do |io|
      io.read.scan(/^#{pattern}(.*)\n/o) do |severity, *time_segs, pid, message|
        time = time(*time_segs)
        action(severity, time, pid.to_i, message.chomp)
      end
    end
    check
  end

  bm.report("#gets") do
    setup
    open_file do |io|
      while line = io.gets
        next unless /\A#{pattern}/o =~ line
        time = time($2, $3, $4, $5, $6, $7, $8)
        action($1, time, $9.to_i, $'.chomp)
      end
    end
    check
  end

  bm.report("#each_line") do
    setup
    open_file do |io|
      io.each_line do |line|
        next unless /\A#{pattern}/o =~ line
        time = time($2, $3, $4, $5, $6, $7, $8)
        action($1, time, $9.to_i, $'.chomp)
      end
    end
    check
  end

  [4096, 8192, 1024 * 16, 1024 * 64, 1024 * 256].each do |size|
    bm.report("#readpartial [#{size}]") do
      setup
      open_file do |io|
        buf = +""
        while buf << io.readpartial(size)
          buf.scan(/^#{pattern}(.*)\n/o) do |severity, *time_segs, pid, message|
            time = time(*time_segs)
            action(severity, time, pid.to_i, message)
          end
          break if io.eof?

          buf = $'.empty? ? +"" : $' if $'
        end
      end
      check
    end
  end

  [4096, 8192, 1024 * 16, 1024 * 64, 1024 * 256].each do |size|
    bm.report("StringScanner [#{size}]") do
      setup
      s = StringScanner.new(+"")
      open_file do |io|
        while s << io.readpartial(size)
          while line = s.skip_until(/^#{pattern}(.*)\n/o)
            time = time(s[2], s[3], s[4], s[5], s[6], s[7], s[8])
            action(s[1], time, s[9].to_i, s[10])
          end
          break if io.eof?
        end
      end
      check
    end
  end
end