add or del blanks between Chinese and English words.

add_del_blank.rb

#!/usr/bin/env ruby
# Usage: deal_blanks.rb input.txt >out.txt

def isEn(char)
  /\w/.match(char) != nil
end

File.open(ARGV[0], "r") do |file|

  blanks_del = 0
  blanks_add = 0
  
  file.each do |line|
    buf = Array.new
    arr = line.chars.to_a
    skips = ['（', '）', '。', '，', '；', '？', '(', ')']

    for i in 1...arr.size
      prev, curr = arr[i-1], arr[i]

      if prev != ' ' then
        buf << prev
      else
        $stderr.puts '+ ' + prev + '/' + curr
        blanks_del += 1
        next
      end

      if skips.include?(prev) or skips.include?(curr) then
        next
      end

      if not isEn(prev) and not isEn(curr) then
        next
      end

      if isEn(prev) and isEn(curr) then
        next
      end 

      $stderr.puts '- ' + prev + '/' + curr
      blanks_add += 1
      buf << ' '
    end
    $stdout.puts buf.join('') 
  end
  $stderr.puts "blanks add: " + blanks_add.to_s
  $stderr.puts "blanks del: " + blanks_del.to_s
end