gistfile1

#!/usr/bin/env ruby19
#
# Total time loading the reference genome: 0 hour, 3 minutes and 2 seconds.
# Total time loading and deleting indexes: 6 hour, 28 minutes and 36 seconds.
# Total time searching indexes: 1 hour, 30 minutes and 2 seconds.
# Total time merging and writing output: 0 hour, 12 minutes and 25 seconds.
# Total time elapsed: 8 hours, 20 minutes and 12 seconds.
#
require 'find'

csv = "/stornext/snfs1/next-gen/solid/csv_dump/csv.dump.latest.csv"
@in_order = [:load_ref, :load_indexes, :searching, :merging, :elapsed]
@res = {
  :load_ref     => %r{Total time loading the reference genome: (\d+) hour, (\d+) minutes and (\d+) seconds.},
  :load_indexes => %r{Total time loading and deleting indexes: (\d+) hour, (\d+) minutes and (\d+) seconds.},
  :searching    => %r{Total time searching indexes: (\d+) hour, (\d+) minutes and (\d+) seconds.},
  :merging      => %r{Total time merging and writing output: (\d+) hour, (\d+) minutes and (\d+) seconds.},
  :elapsed      => %r{Total time elapsed: (\d+) hours, (\d+) minutes and (\d+) seconds.}
}

# Total time searching indexes: 1 hour, 14 minutes and 51 seconds. 
def hours_searching_indexes(f)
  return nil unless File.exists?(f) and File.readable?(f)
  text  = File.open(f).read
  line = ""
  @in_order.each do |s|
    if ma = text.match(@res[s])
      h, m, s = ma[1..3]
      line << "#{h.to_f + m.to_f/60 + s.to_f/3600}, "
    else
      line << "0, "
    end
  end
  line
end

# Main
#
i=0
File.open(csv).each_line do |l|
  next if l =~ /^name/
  log_dir = l.split(",")[1].split('/')[0..9].join("/") + "/lsf_logs"
  Find.find(log_dir) do |path|
    if path =~ %r{.*match.split\d+.\w+.err$}
      time  = File.stat(path).mtime.strftime("%Y-%m-%d %H:%M")
      wtime = hours_searching_indexes(path)
      puts time + ", " + wtime if time 
    end
    $stderr.printf "# of files processed: #{i}\r"
    i = i + 1
  end
end

$stderr.puts "\n"