From: Brian Candler <b.candler@...>
Date: 2008-11-11T21:56:11+09:00
Subject: Re: How to optimize my ruby code

Valentino Lun wrote:
> ps=ps.to_a
> for idx in (0...ps.length)

This is fine since you use idx later on to skip the first line, although 
an alternative is

  ps.to_a.each_with_index do |line, idx|

>    next if idx==0   #skip the header
>    ps_user=ps[idx].split[0]
>    ps_pid=ps[idx].split[1]
>    ps_cpu=ps[idx].split[2]
>    ps_mem=ps[idx].split[3]
>    ps_sz=ps[idx].split[4]
>    ps_rss=ps[idx].split[5]
>    ps_tty=ps[idx].split[6]
>    ps_stat=ps[idx].split[7]

  user, pid, cpu, mem, sz, rss, tty, stat, *rest = line.split

>    # The STIME field may equal to May 29 or 13:27:12, and the COMMAND
> has unknown number of space.
>    if (ps[idx].split[8].include?(":"))
>       ps_stime=ps[idx].split[8]
>       ps_cmd=ps[idx].split[9]+" "+ps[idx].split[10]+" "+ ... +
> ps[idx].split[ps[idx].length]
>    else
>       ps_stime=ps[idx].split[8]+" "+ps[idx].split[9]
>       ps_cmd=ps[idx].split[10]+" "+ps[idx].split[11]+" "+ ... +
> ps[idx].split[ps[idx].length]
>    end

  if rest[0].include?(":")
    stime, cmd = rest[0], rest[1..-1].join(" ")
  else
    stime, cmd = rest[0]+" "+rest[1], rest[2..-1].join(" ")
  end

But a shorter (and still very efficient) way is to write a big Regexp 
for matching all this:

  PS_LINE = %r{^
    (\S+) \s+  # user
    (\S+) \s+  # pid
    (\S+) \s+  # cpu
    (\S+) \s+  # mem
    (\S+) \s+  # sz
    (\S+) \s+  # rss
    (\S+) \s+  # tty
    (\S+) \s+  # stat
    (\d\d:\d\d:\d\d | \w\w\w\s\d\d) \s+  # stime
    (.*)       # cmd
  $}x

  ps.each_line do |line|
    if PS_LINE =~ line
      MyTable.create(:ps_user => $1, :ps_pid => $2, ...SKIP...
                     :ps_cmd => $10)
    end
  end

The 'x' flag to a Regexp lets you embed spaces and comments to make it 
more readable. You can tweak it to match the data more precisely, e.g. 
use (\d+) instead of (\S+) to match the pid, so that any unexpected 
lines are ignored.
-- 
Posted via http://www.ruby-forum.com/.