Class: Corpus

Inherits:
Object
  • Object
show all
Defined in:
lib/engine/corpus.rb

Instance Method Summary collapse

Constructor Details

#initialize ⇒ Corpus

Returns a new instance of Corpus.



5
6
7
 
# File 'lib/engine/corpus.rb', line 5

def initialize
  @tokens = {}
end

Instance Method Details

#add(document) ⇒ Object 



13
14
15
16
17
 
# File 'lib/engine/corpus.rb', line 13

def add document
  document.each_token do |token|
    @tokens[token] = token_count(token) + 1
  end
end

#entry_count ⇒ Object 



9
10
11
 
# File 'lib/engine/corpus.rb', line 9

def entry_count
  @tokens.values.inject(0, :+)
end

#load_from_directory(directory) ⇒ Object 



19
20
21
22
23
24
25
 
# File 'lib/engine/corpus.rb', line 19

def load_from_directory directory
  Dir.glob("#{directory}/*.txt") do |entry|
    IO.foreach(entry, encoding: Encoding::UTF_8) do |line|
      add Document.new(line)
    end
  end
end

#token_count(token) ⇒ Object 



27
28
29
 
# File 'lib/engine/corpus.rb', line 27

def token_count token
  @tokens[token] || 0
end