Sha256: 985c98d1f29fa90bc3f8b75481c0181217546b078235e5cf3ba0cf93808af002
Contents?: true
Size: 1.18 KB
Versions: 13
Compression:
Stored size: 1.18 KB
Contents
module DataMaker
module Validators
class ChineseCharacters
CJK_CHARACTER_RANGES = [ { min: "4E00", max: "9FFF", comment: "Common" },
{ min: "3400", max: "4DFF", comment: "Rare" },
{ min: "20000", max: "2A6DF", comment: "Rare, historic" },
{ min: "F900", max: "FAFF", comment: "Duplicates, unifiable variants" },
{ min: "2F800", max: "2FA1F", comment: "unifiable variants" },
{ min: "AC00", max: "D7AF", comment: "Hangul syllables" } ]
def initialize(characters)
@characters = characters
end
def valid?
valid_characters = true
@characters.each_char do |character|
hex_chinese = "%04x" % character[0].ord
valid_characters = true unless within_cjk_range(hex_chinese)
end
end
private
def within_cjk_range(hex_string)
val = hex_string.hex
CJK_CHARACTER_RANGES.each do |valid_range|
return true if val >= valid_range[:min].hex && val <= valid_range[:max].hex
end
return false
end
end
end
end
Version data entries
13 entries across 13 versions & 1 rubygems