Module: Taxamatch::Normalizer

Defined in:
lib/taxamatch_rb/normalizer.rb

Class Method Summary collapse

Class Method Details

.normalize(string) ⇒ Object 



6
7
8
 
# File 'lib/taxamatch_rb/normalizer.rb', line 6

def self.normalize(string)
  utf8_to_ascii(string.strip.upcase).gsub(/[^\x00-\x7F]/,'?')
end

.normalize_author(string) ⇒ Object 



14
15
16
 
# File 'lib/taxamatch_rb/normalizer.rb', line 14

def self.normalize_author(string)
  self.normalize(string).gsub(/[^A-Z]/, ' ').gsub(/[\s]{2,}/, ' ').strip
end

.normalize_word(word) ⇒ Object 



10
11
12
 
# File 'lib/taxamatch_rb/normalizer.rb', line 10

def self.normalize_word(word)
  self.normalize(word).gsub(/[^A-Z0-9\-]/, '').strip
end

.normalize_year(year_string) ⇒ Object 



18
19
20
21
22
 
# File 'lib/taxamatch_rb/normalizer.rb', line 18

def self.normalize_year(year_string)
  year_int = year_string.gsub(/[^\d]/, '').to_i
  year_int = nil unless year_int.between?(1757, Time.now.year + 1)
  year_int
end