From: ahoward <ahoward@...>
Date: 2003-01-29T01:41:55+09:00
Subject: Re: Use mulit-dim. Arrays? [Questions from a Ruby Newbie (file io and d ata structures)...]

On Wed, 29 Jan 2003 christopher.j.meisenzahl@citicorp.com wrote:

> I've got the file IO hammered out. I think I can read each line into a string
> and then parse into an array using:
<snip>

one of the things that i LOVE, LOVE, LOVE about ruby is that it's so easy to
write an object to encapsulate common scripting tasks.  there were some slides
floating around somewhere in which matz said something to the effect that
'what's good for programming is good for scripting', refering to
object-oriented programming.  for example, i have a similar problem to yours,
where we have all these silly config files floating around, each with a
slightly different syntax - you _could_ parse each one out in every script, or
you could write a class which does this for you, for example (overly simplistic)

say you have two config files

FILE : 'in.csv'
----CUT----
# a comment
col1,col2,col3
a,b,c
d,e,f
g,h # another comment

j

----CUT----

and

FILE : 'printers.txt'
----CUT----
# type name
hp  bugs
xerox fg-color
----CUT----

'in.csv' is a comma-separated values file, while 'printers.txt' is a
tab-delimited file.  both could contain comments and empty lines.  here's a
silly parser which handles both types


#!/usr/bin/env ruby

class Cfg < Array

  DEFAULT_DELIM = ','
  DEFAULT_COMMENT = '#'

  attr_reader :max_cols
  attr_accessor :comment
  attr_accessor :delim

  def initialize path, delim = DEFAULT_DELIM, comment = DEFAULT_COMMENT
    @max_cols = 0
    @delim = delim
    @comment = comment
    comment_pat = %r{#{comment}.*$}
    lines = IO.readlines path

    lines.each do |line|
      line.strip!
      next if line.size == 0
      line[comment_pat] = ''
      row = line.split delim
      next unless row.size > 0
      @max_cols = [row.size, @max_cols].max
      self << row
    end
  end

  def to_s
    (map {|r| r.join (delim)}).join ($/) << $/
  end

  def pad value = nil
    map! {|r| r.fill value, r.size...max_cols}
  end
end


if $0 == __FILE__
  csv = Cfg.new 'in.csv'
  print csv
  csv.pad '!'
  print csv

  printers = Cfg.new 'printers.txt', "\t"
  print printers
end


the program outputs this :

/usr/home/howardat/eg/ruby > ./Csv.rb

col1,col2,col3
a,b,c
d,e,f
g,h
j
col1,col2,col3
a,b,c
d,e,f
g,h ,!
j,!,!
hp  bugs
xerox fg-color


my point is simply that, although you can do things really tersely and slickly
in ruby (as you can in perl) i think it's ability to use OO programming for
every day tasks is where it really shines.  if you wrote a class similar to
the one above, and made it recognize which type of config file it was reading
based on file extension (for example), and saved this class into your
site_ruby/ directory, all your scripts which require cfg information start
looking really short and sweet, something like :

  require 'cfg.rb'

  cfgpath = ARGV.shift

  cfg = Cfg.new cfgpath

  cfg.each do |row|
    ... do something
  end

which is exactly what i've done at our site.

just thought i'd point all that out since, IMHO, it's the OO aspect of ruby
which is the most usefull, and yet also the aspect which requires the biggest
paradigm shift.

-a


-- 

 ====================================
 | Ara Howard
 | NOAA Forecast Systems Laboratory
 | Information and Technology Services
 | Data Systems Group
 | R/FST 325 Broadway
 | Boulder, CO 80305-3328
 | Email: ahoward@fsl.noaa.gov
 | Phone:  303-497-7238
 | Fax:    303-497-7259
 ====================================