Sha256: 868b160349a7dfbd2ac759954ec2ae714451aacf591df02c6047317da3fda6e3
Contents?: true
Size: 1.42 KB
Versions: 24
Compression:
Stored size: 1.42 KB
Contents
require "htmlentities"
# -- decode HTML entities -----------------------------------------------------
class String
module HtmlDecoder
def self.instance
@htmldecoder ||= HTMLEntities.new
end
end
def unhtml
HtmlDecoder.instance.decode self
end
end
# -- convert to UTF8 ----------------------------------------------------------
require 'iconv'
class String
def to_utf8
# require "charguess"
# encoding = CharGuess.guess(self)
# puts "encoding: #{encoding.inspect}"
# return self if !encoding || encoding == "UTF-8"
encoding = 'ISO-8859-1'
Iconv.conv('utf-8', encoding, self)
end
end
# -- remove accents -----------------------------------------------------------
require_relative "./string_without_accents"
class String
include WithoutAccents
end
class String
def sortkey
# Convert the key into an sortable ascii string
self.without_accents. # remove accents
downcase.
gsub(/^\s*(der|die|das|the|a|ein)\b\s*/, ""). # remove leading stop words
gsub(/[0-9]+/) { |s| "%03d" % s.to_i }. # fill in leading zeroes
gsub(/[^a-z0-9]/, "") # keep only letters and digits
end
end
class String
def starts_with?(other)
length >= other.length && self[0, other.length] == other
end
def ends_with?(other)
length >= other.length && self[length - other.length .. -1] == other
end
end
Version data entries
24 entries across 24 versions & 2 rubygems