lib/lda-ruby/vocabulary.rb in ealdent-lda-ruby-0.3.0 vs lib/lda-ruby/vocabulary.rb in ealdent-lda-ruby-0.3.1
- old
+ new
@@ -1,8 +1,8 @@
module Lda
class Vocabulary
- attr_reader :words
+ attr_reader :words, :indexes
def initialize(words = nil)
@words = Hash.new do |hash, key|
if hash.member?(:MAX_VALUE)
hash[:MAX_VALUE] = hash[:MAX_VALUE] + 1
@@ -11,13 +11,20 @@
end
hash[key] = hash[:MAX_VALUE]
end
words.each { |w| @words[w] } if words
+ @indexes = Hash.new
+
+ @words.each_pair do |w, i|
+ @indexes[i] = w
+ end
end
def check_word(word)
- @words[word.dup]
+ w = @words[word.dup]
+ @indexes[w] = word.dup
+ w
end
def load_file(filename)
txt = File.open(filename, 'r') { |f| f.read }
txt.split(/[\n\r]+/).each { |word| check_word(word) }
\ No newline at end of file