[ruby-talk:19907] Re: how to speed up hash script?

Thanks for all the help. I've learned something from all the postings.
Joseph McDonald came up with the fastest script which clocked in at a
little over 3 seconds as opposed to the original script at 14 seconds.

--------------------------------------------------------------------
class Climo
  
  DATE_FORMAT = "YYMMDDHH"
  DATE_LEN = DATE_FORMAT.length
  
  attr_reader :start_date, :end_date, :header
  
  def initialize(in_file)
    
    f = File.open(in_file)
    @header = f.readline
    @data = f.readline
    @rec_len = @data.length
    @data << f.read
    
    @dict = Hash.new()
    start = 0
    @num_recs = (@data.length/@rec_len) -1
    GC.disable
    0.upto(@num_recs) do 
      @dict[@data[start,DATE_LEN].to_i] = start
      start += @rec_len
    end
    GC.enable

    @start_date = getbyline(0)[0,DATE_LEN]
    @end_date = getbyline(@num_recs)[0,DATE_LEN]
    puts "Start Date: #{@start_date}"
    puts "End Date: #{@end_date}"
    puts "By date: #{getbydate(@start_date.to_i)}"
  end
  
  def getbyline(lineno)
    @data[lineno*@rec_len,@rec_len]
  end
  
  def getbydate(date)
    @data[@dict[date],@rec_len]
  end
  
end
  
def main()
    in_file = ARGV[0]
        
    climo_obj = Climo.new(in_file)
end

main()
--------------------------------------------------------------------

After studying Joe's script, I learned that 

    for i in 1...@num_lines
      @dict[@lines.at(i)[0, @len_date].to_i] = i
    end

is faster than

    for i in 1...@num_lines
      @dict[@lines[i][0...@len_date].to_i] = i
    end

when modifying the original lines-based script.

In particular,

The change from ... to , saved approximately 3.5 seconds and the
switch to at() save another 0.3 seconds. The following lines-based
script now clocks in around 4 seconds. The difference in time between
this script and the fastest script can be attributed to the fact that
read() is about 1 second faster than readlines() on the file being
tested.

class Climo
  DATE_FORMAT = "YYMMDDHH"
  DATE_LEN = DATE_FORMAT.length

  def initialize(in_file)
    # Read in data
    fp = File.new(in_file, "r")
    @lines = fp.readlines()
    fp.close()

    @header = @lines.shift()
    @num_lines = @lines.size()
    
    # Store line indices in dictionary based on date field
    @dict = Hash.new()

    GC.disable
    for i in 0...@num_lines
      @dict[@lines.at(i)[0,DATE_LEN].to_i] = i
    end
    GC.enable

    first_date_line = @lines[@dict[@lines[0][0,DATE_LEN].to_i]]
    last_date_line =
@lines[@dict[@lines[@num_lines-1][0,DATE_LEN].to_i]]
    puts "first_date_line: ", first_date_line
    puts "last_date_line: ", last_date_line
  end
  
end  

def main()
    in_file = ARGV[0]

    climo_obj = Climo.new(in_file)
end

main()