[ruby] cyrillic detection becnh

gistfile1.rb

class TestCyrillic
  class << self
    RUSSIAN_CODES = (1040..1103).to_a + (32..64).to_a + (91..96).to_a + (123..126).to_a + [1025, 1105, 8470]
    
    def cyrillic?(string)
      result = true
      string.force_encoding("UTF-8").each_char{|c| result &&= RUSSIAN_CODES.include?(c.ord)}
      result
    end

    # \u2116 - №
    # \u0410 - \u044f - cyrillics themself
    # ... - other common symbols 

    RXP = /^[\u2116\u0020-\u0040\u005b-\u0060\u007b-\u007e\u00a2-\u00bf\u0100-\u0171\u0401\u0451\u0410-\u044f]+$/

    def cyr?(string)
      !!(string.force_encoding("UTF-8") =~ RXP)
    end
  end
end

class String
  def cyrillic?
    TestCyrillic.cyrillic?(self)
  end

  def cyr?
    TestCyrillic.cyr?(self)
  end
end

def run_cyrillic(*args)
  args.each { |s| s.cyrillic? }
end

def run_cyr(*args)
  args.each { |s| s.cyr? }
end


def assert(&block)
  raise "assert failed" unless block.call
end

def assertNot(&block)
  raise "assert failed" if block.call
end

assert{"Русские буквы".cyrillic?}                            
assert{"Русские буквы и !№%:,.;()_+=-0986654".cyrillic?}     
assert{"!№%:,.;()_+=-0986654".cyrillic?}                     
assertNot{"Русские буквы и English letters".cyrillic?}          
assertNot{"Only English letters".cyrillic?}                     
assertNot{"English letters and !№%:,.;()_+=-0986654".cyrillic?}

assert{"Русские буквы".cyr? }                           
assert{"Русские буквы и !№%:,.;()_+=-0986654".cyr? }    
assert{"!№%:,.;()_+=-0986654".cyr? }                    
assertNot{"Русские буквы и English letters".cyr? }         
assertNot{"Only English letters".cyr? }                    
assertNot{"English letters and !№%:,.;()_+=-0986654".cyr? }

require 'benchmark'

test_data = [
  "Русские буквы", 
  "Русские буквы и !№%:,.;()_+=-0986654", 
  "!№%:,.;()_+=-0986654", 
  "Русские буквы и English letters",
  "Only English letters",
  "English letters and !№%:,.;()_+=-0986654"
]

n = 50000
Benchmark.bm do |x|
  x.report { n.times { run_cyrillic(*test_data) } }
  x.report { n.times { run_cyr(*test_data) }}
end