Module: DR::Encoding

Defined in:
lib/dr/base/encoding.rb

Class Method Summary collapse

Class Method Details

.fix_utf8(s = nil) ⇒ Object

if a mostly utf8 has some mixed in latin1 characters, replace the invalid characters



8
9
10
11
12
13
14
15
16
17
 
# File 'lib/dr/base/encoding.rb', line 8

def fix_utf8(s=nil)
  s=self if s.nil? #if we are included
  if String.method_defined?(:scrub)
    #Ruby 2.1
    #cf http://ruby-doc.org/core-2.1.0/String.html#method-i-scrub
    return s.scrub {|bytes| '<'+bytes.unpack('H*')[0]+'>' }
  else
    return DR::Encoding.to_utf8(s)
  end
end

.to_utf8(s = nil, from: nil) ⇒ Object 



19
20
21
22
23
24
25
26
27
28
29
30
31
32
 
# File 'lib/dr/base/encoding.rb', line 19

def to_utf8(s=nil,from:nil)
  s=self if s.nil? #if we are included
  from=s.encoding if from.nil?
  if String.method_defined?(:encode)
    #Ruby 1.9
    return s.encode('UTF-8',from, :invalid => :replace, :undef => :replace,
             :fallback => Proc.new { |bytes| '<'+bytes.unpack('H*')[0]+'>' }
           )
  else
    #Ruby 1.8
    ic = Iconv.new(from, 'UTF-8//IGNORE')
    return ic.iconv(s)
  end
end

.to_utf8!(s = nil, from: nil) ⇒ Object

assume ruby>=1.9 here



35
36
37
38
39
40
41
 
# File 'lib/dr/base/encoding.rb', line 35

def to_utf8!(s=nil,from:nil)
  s=self if s.nil? #if we are included
  from=s.encoding if from.nil?
  return s.encode!('UTF-8',from, :invalid => :replace, :undef => :replace,
           :fallback => Proc.new { |bytes| '<'+bytes.unpack('H*')[0]+'>' }
         )
end