gistfile1.rb

# Basic text search with relevancy for MongoDB. 
# See http://blog.tty.nl/2010/02/08/simple-ranked-text-search-for-mongodb/
# Copythingie 2010 - Ward Bekker - [email protected]

#create (or empty) a docs collection
doc_col = MongoMapper.connection.db('example_db').collection('docs')
doc_col.remove({})

#add some sample data
doc_col.insert({ "txt" => "it is what it is"})
doc_col.insert({ "txt" => "what is it"})
doc_col.insert({ "txt" => "it is a banana"})

#The invix creation map function. Splits the texts in individual words
map_index =<<JS
  function() {
    var words = this.txt.split(' ');
    
    for ( var i=0; i<words.length; i++ ) {
      emit(words[i], { docs: [this._id] });
    }
  }
JS

# Groups the doc id's for every unique word
reduce_index =<<JS
  function(key, values) {
      var docs = [];
      
      values.forEach ( function(val) { docs = docs.concat(val.docs); })
      
      return { docs: docs };
  }
JS

# Every document counts as one
map_relevance =<<JS
  function() {
   for ( var i=0; i< this.value.docs.length; i++ ) {
     emit(this.value.docs[i], { count: 1 });
   }
 }
JS

# And calculate the amount of occurrences for every unique document id
reduce_relevance=<<JS
  function(key, values) {
      var sum = 0;
      
      values.forEach ( function(val) { sum += val.count; })
      
      return { count: sum };
  }
JS

#calculate the inverted index
invix_col = doc_col.map_reduce(map_index, reduce_index)
#calculate the # occcurances of each searchterm
query = ["what", "is", "it"]
ranked_result = invix_col.map_reduce(map_relevance, reduce_relevance, { :query => { "_id" => { "$in" =>  query} }  } )

#output the results, most relevant on top
ranked_result.find().sort("count", :desc).each do |result|
  puts "document with id #{result["_id"]} has rank #{result["value"]["count"]} : #{doc_col.find_one("_id" => result["_id"]).inspect}"
end