#!/usr/bin/ruby

$:.push './html-parser'

require 'htmlscan'
require 'sgml-parser'

class LooseHTMLScanner
  def parse_ary(a)
    parse a
  end
end

class SGMLParser
  alias parse feed
  def parse_ary(a)
    for i in a
      feed i
    end
  end
end

Entries = [ SGMLParser, LooseHTMLScanner ]



require 'nkf'
#$KCODE = 'E'


def benchmark(klass, method, arg)
  begin
    scanner = klass.new
    t1 = Time.times.utime
    scanner.send method, arg
    t2 = Time.times.utime
    sprintf "%.2f", t2 - t1
  rescue Exception
    raise if $!.is_a? Interrupt
    sprintf "=%s=", $!.type
  end
end


if ARGV[0] == '-f' then
  files = File.open(ARGV[1]).readlines
  files.each { |i| i.chomp! }
else
  files = ARGV
end

STDOUT.sync = true

print Entries.collect{ |i| i.name }.join("\t"), "\n"
print "at once (read)\tby line (gets)\n"

files.each { |i|
  print i
  src = File.open(i.chomp) { |f|
    s = f.gets("\r\n\r\n")  # skip http header
    f.read or s
  }
  src = NKF.nkf('-dexm0', src)
  print "\t", src.size
  src_a = src.to_a
  print "\t", src_a.size

  Entries.each { |klass|
    print "\t", benchmark(klass, :parse, src)
    print "\t", benchmark(klass, :parse_ary, src_a)
  }
  print "\n"
}