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).upcase
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

.utf8_to_ascii(string) ⇒ Object 



19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
 
# File 'lib/taxamatch_rb/normalizer.rb', line 19

def self.utf8_to_ascii(string)
  string = string.gsub(/[ÀÂÅÃÄÁẤẠ]/, "A")
  string = string.gsub(/[ÉÈÊË]/, "E")
  string = string.gsub(/[ÍÌÎÏ]/, "I")
  string = string.gsub(/[ÓÒÔØÕÖỚỔ]/, "O")
  string = string.gsub(/[ÚÙÛÜ]/, "U")
  string = string.gsub(/[Ý]/, "Y")
  string = string.gsub(/Æ/, "AE")
  string = string.gsub(/[ČÇ]/, "C")
  string = string.gsub(/[ŠŞ]/, "S")
  string = string.gsub(/[Đ]/, "D")
  string = string.gsub(/Ž/, "Z")
  string = string.gsub(/Ñ/, "N")
  string = string.gsub(/Œ/, "OE")
  string = string.gsub(/ß/, "B")
  string = string.gsub(/Ķ/, "K")
  string = string.gsub(/[áàâåãäăãắảạậầằ]/, "a")
  string = string.gsub(/[éèêëĕěếệểễềẻ]/, "e")
  string = string.gsub(/[íìîïǐĭīĩỉï]/, "i")
  string = string.gsub(/[óòôøõöŏỏỗộơọỡốơồờớổ]/, "o")
  string = string.gsub(/[úùûüůưừựủứụ]/, "u")
  string = string.gsub(/[žź]/, "z")
  string = string.gsub(/[ýÿỹ]/, "y")
  string = string.gsub(/[đ]/, "d")
  string = string.gsub(/æ/, "ae")
  string = string.gsub(/[čćç]/, "c")
  string = string.gsub(/[ñńň]/, "n")
  string = string.gsub(/œ/, "oe")
  string = string.gsub(/[śšş]/, "s")
  string = string.gsub(/ř/, "r")
  string = string.gsub(/ğ/, "g")
  string = string.gsub(/Ř/, "R")
end