#!/usr/bin/env ruby def ngrams_it(s, n, fix=false) a = s.strip.split a.each_with_index { |tok, i| tok.strip! 0.upto([n-1, a.size-i-1].min) { |m| yield a[i..i+m] if !(fix^(a[i..i+m].size==n)) } } end def main(n, fix, sep) STDIN.set_encoding 'utf-8' STDOUT.set_encoding 'utf-8' while line = STDIN.gets a = [] ngrams_it(line, n, fix) {|ng| a << ng.join(' ')} a.reject! {|i| i.strip.size==0 } puts a.join sep if a.size > 0 end end def usage STDERR.write "./ng [-n ] [--fix] [--separator ] < \n" exit 1 end if __FILE__ == $0 require 'trollop' opts = Trollop::options do opt :n, "Ngrams", :type => :int, :default => 4 opt :fix, "Don't output lower order Ngrams.", :type => :bool, :default => true opt :separator, "separte ngrams of a line by this string", :type => :string, :default => "\n" end usage if not [0,2,4,6].include? ARGV.size main(opts[:n], opts[:fix], opts[:separator]) end