query.rb

#!/usr/bin/ruby
require 'rubygems'
require 'activesupport'
require 'couchrest'

$KCODE='u'
require 'jcode'

NGRAM_SIZE=3
@db = CouchRest.database("http://localhost:5984/geo_#{NGRAM_SIZE}")

terms=Hash.new(0.0)
word = (ARGV[0] || 'кришатик').mb_chars
puts word.to_s.inspect

(0..word.size-3).each do |i|
  s=word.slice(i,3)
  rows = @db.view('z/trgm', :key => s)['rows']
  next if rows.empty?
  rows.each do |r|
    k=r["value"]
    terms[k]+=1
  end
end

# normalize by length
terms.keys.each do |k|
  d = k.mb_chars.size-word.size
  d *= -1 if d<0
  next if d.zero?

  terms[k] /= d.to_f
end

puts terms.sort{|a,b| (a[1]<=>b[1])}[-4..-1].inspect
puts terms.sort{|a,b| a[1] <=> b[1]}.last.inspect

trigrams

// output
// "abc" -> ["cabc", "abck", "jabcr"]

// map
function(doc) {
  var l=doc.title.length;
  for(var i=0; i<(l-2); i++) {
    var s=doc.title.substr(i,3);
    emit(s, doc.title);
  };
}

// reduce?