From: Matt Maycock <ummaycoc@...>
Date: 2004-12-24T04:03:43+09:00
Subject: Re: Inconsistency with IO.readlines

# ========== ext/input_reader.rb ==========
# ext/input_reader.rb
# accept some name argument.  if name is nil || '-',
#   then use $stdin and send that to the block
#   otherwise, use File.open(name, *restargs)

require 'delegate'

def input_reader(fname, *fargs)
  block = Proc.new {|source|
    getter = SimpleDelegator.new(source)
    yield getter
    getter.__setobj__(nil)
  }

  if fname.nil? || fname == '-' then
    block[$stdin]
  else
    File.open(fname, *fargs) {|f| block[f]}
  end
end
# ========== ext/input_reader.rb ==========



# ========== ext/text2lines.rb ==========
# ext/text2lines.rb
#  Take character input from a source at a time, and
#  if we've struct a sequence \n|\r gold, then replace
#  print an argument-defined eoln marker, instead.
#
#  endls is the array of eoln marker characters.
#    valid integers (string or actual) are valid, along
#    with the string `newline' - which reduces to the
#    system newline $/
#
#  source is anything that responds to getc()
#
#  sink is anything that responds to print(string)
#

require 'ext/input_reader'

def text2lines(endls, source, sink=nil)
  separator, marker = *endls.inject([[], []]) {|(sep, mark), new|
    case new.downcase
      when /^(-)?newline$/ then ($1.nil? ? mark : sep) << $/
      when /^(-)?\d+$/     then ($1.nil? ? mark : sep) << new.to_i.abs.chr
    end
    
    [sep, mark]
  }
  marker    = marker.empty? ? $/ : marker.join('')
  separator = [10.chr, 13.chr] if separator.empty?

  char, prev, lastp = nil, nil, true
  lline, splitter   = nil, ''
  counts = Hash.new(0)

  pchar = Proc.new {
    if block_given? then
      lline ||= ''
      lline << char.chr
    else
      sink.print char.chr
    end
  }
  pmark = Proc.new {
    if block_given? then
      yield(lline || '', marker)
      lline = nil
    else
      sink.print marker
    end
  }
  domark = Proc.new {
    pmark[]
    counts = Hash.new(0)
  }  
  pnull = Proc.new {}

  if separator.include?(char.chr) then
    domark[] if counts[char] > 0
    counts[char] += 1
  else
    domark[] if counts.values.include?(1)
    pchar[]
  end while (char = source.getc)

  domark[]
end

def read_text2lines(file, *args)
  lines, chomper = [], args.delete('-c') {false}
  input_reader(file) {|source|
    text2lines(args.map {|i| i.to_s}, source) {|l,t|
      lines << l
      lines[-1] << t unless chomper || t.nil?
    }
  }
  lines
end
# ========== ext/text2lines.rb ==========


# ========== ~/local/bin/text2lines ==========
#!/usr/bin/env ruby

require 'ext/text2lines'

def usage(out=$stdout)
out.puts <<-END_USAGE
Usage:  #{$0} [newline | ascii-code]+
  Replaces all instances of \\r and \\n with new end of line
  markers.  \\r\\n and \\n\\r are treated as one unit.  \\r\\r and
  \\n\\n are treated as two.

  The new markers are formed from command-line arguments.  If
  no arguments are given, then the system's end of line marker
  is used.  Otherwise, the sequence of ascii-codes / newlines
  are used, with newline representing the system's end of line
  marker.  Characters are read from stdin.

  EXAMPLES:
     #{$0} 13 10
  replaces all `standard' end of line markers with \\r\\n.
  
     #{$0} newline
  replaces all `standard' end of line markers with the system
  end of line marker.
END_USAGE
  exit(-1)
end

args = ARGV.map {|arg|
  case arg
    when /^--+h(e(l(p)?)?)?$/i then usage
    when /^newline$/           then arg
    when /^\d+$/               then arg
    else
      $stderr.puts "Error - bad argument #{arg}"
      usage[$stderr]
  end
}

text2lines(args, $stdin, $stdout)
# ========== ~/local/bin/text2lines ==========


[ummaycoc@localhost ummaycoc]$ echo 'hello
my
ruby
loving
friends' | text2lines 65
helloAmyArubyAlovingAfriendsA[ummaycoc@localhost ummaycoc]$

[ummaycoc@localhost ummaycoc]$ echo 'hello
my
ruby
loving
friends' | text2lines 13 > rubytmp

[ummaycoc@localhost ummaycoc]$ more rubytmp
friends
[ummaycoc@localhost ummaycoc]$

so, obviously, if this doesn't work for you - getc will :-)

-- 
There's no word in the English language for what you do to a dead
thing to make it stop chasing you.