Module: AnyStyle::StringUtils

Included in:
Document, Document::Page, Feature, ParserCore
Defined in:
lib/anystyle/utils.rb

Class Method Summary collapse

Class Method Details

.canonize(string) ⇒ Object 



24
25
26
 
# File 'lib/anystyle/utils.rb', line 24

def canonize(string)
  scrub(transliterate(string)).downcase
end

.count(string, pattern) ⇒ Object 



42
43
44
 
# File 'lib/anystyle/utils.rb', line 42

def count(string, pattern)
  string.to_enum(:scan, pattern).inject(0) { |c| c + 1 }
end

.display_chars(string) ⇒ Object 



36
37
38
39
40
 
# File 'lib/anystyle/utils.rb', line 36

def display_chars(string)
  string
    .gsub(/\p{Mn}|\p{Me}|\p{Cc}/, '')
    .gsub(/\p{Zs}/, ' ')
end

.display_width(string) ⇒ Object 



32
33
34
 
# File 'lib/anystyle/utils.rb', line 32

def display_width(string)
  display_chars(string).rstrip.length
end

.indent(token) ⇒ Object 



46
47
48
 
# File 'lib/anystyle/utils.rb', line 46

def indent(token)
  display_chars(token).rstrip[/^(\s*)/].length
end

.page_break?(string) ⇒ Boolean

Returns:

  • (Boolean) 


28
29
30
 
# File 'lib/anystyle/utils.rb', line 28

def page_break?(string)
  string =~ /\f/
end

.scrub(string, blacklist: /[\p{^Alnum}\p{Lm}]/) ⇒ Object 



14
15
16
 
# File 'lib/anystyle/utils.rb', line 14

def scrub(string, blacklist: /[\p{^Alnum}\p{Lm}]/)
  string.scrub.gsub(blacklist, '')
end

.transliterate(string, form: :nfkd) ⇒ Object 



18
19
20
21
22
 
# File 'lib/anystyle/utils.rb', line 18

def transliterate(string, form: :nfkd)
  string
    .unicode_normalize(form)
    .gsub(/\p{Mark}/, '')
end