Module: HBLText::Setter

Defined in:
lib/hbl_text/setter.rb

Class Method Summary collapse

Class Method Details

.convert_encoding(str, from_charset = nil) ⇒ Object

convert_encoding



54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
 
# File 'lib/hbl_text/setter.rb', line 54

def convert_encoding( str, from_charset=nil )
  
  encoding = from_charset || get_charset( str )
  
  if encoding.present?
    return str.encode( 'UTF-8', encoding,
      :invalid => :replace, 
      :undef   => :replace, 
      :replace => "?"
    )
  else 
    return str.encode( 'UTF-8',
      :invalid => :replace, 
      :undef   => :replace, 
      :replace => "?"
    )
  end #if
end

.get_charset(str) ⇒ Object

get_charset



76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
 
# File 'lib/hbl_text/setter.rb', line 76

def get_charset( str )

  ###########################################################
  #  work on local copy                                     #
  ###########################################################
  str = str.dup
  
  ###########################################################
  #  if string is valid return immediately                  #
  ###########################################################
  if str.force_encoding("UTF-8").valid_encoding?
    return "UTF-8"
  end 
  
  ###########################################################
  #  now, try the most often string encodings               #
  ###########################################################
  encodings = %w(Windows-1252 Windows-1251 ISO-8859-1 Macintosh)
  encodings.each do |enc|
    return enc if str.force_encoding( enc ).valid_encoding?
  end
  
  ###########################################################
  #  if we are here, then some other encoding is used       #
  #  Encoding list: run through all available encodings     #
  ###########################################################
  Encoding.list.each do |enc|
    return enc if str.force_encoding( enc ).valid_encoding?
  end
  
  ###########################################################
  # this code is only reached if everything else failed     #
  ###########################################################
  # here we give up and set to no encoding 
  return nil
  
end

.get_charset_from_tag(str) ⇒ Object

get_charset_from_tag



117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
 
# File 'lib/hbl_text/setter.rb', line 117

def get_charset_from_tag( str )

  # create HTML from fragment
  html  = Nokogiri::HTML::fragment(to_utf8(str))
  
  # create Nokokiris XPath expression to match 'meta', 'charset', 'content-type'
  #      and 'content' in a case insensitive way
  values         = []
  downcased_name = "translate(local-name(), 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz')='meta'"
  downcased_chrs = "translate(local-name(), 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz')='charset'"
  downcased_attr = "translate(@http-equiv,  'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz')='content-type'"
  downcased_getr = "translate(local-name(), 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz')='content'"
  
  # case insensitive fetch <meta charset="whatever">
  values << html.xpath("./*[#{downcased_name}]/@*[#{downcased_chrs}]").last.try(:value)
  
  # case insensitive fetch <meta http-equiv="Content-Type" content="anything; charset=whatever>"  
  value   = html.xpath("./*[#{downcased_name}][#{downcased_attr}]/@*[#{downcased_getr}]").last.try(:value)
  value   = value.to_s.split(/;/).
    select{|s| s =~ /\s*(?<=charset=).*/}.
    map{|s| s.match(/\s*(?<=charset=)(.*)/)}.
    compact.last.try(:[],1)
    
  values << value
  values.compact.last
end

.to_utf8(arg, from_charset = nil) ⇒ Object

to_utf8



34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
 
# File 'lib/hbl_text/setter.rb', line 34

def to_utf8(arg, from_charset=nil)
  case arg
  when String
    convert_encoding(arg, from_charset)
  when Array
    arg.map do |val|
      to_utf8(val, from_charset)
    end
  when Hash
    arg.map do |key, val|
      [key, to_utf8(val, from_charset)]
    end.to_h
  else
    arg
  end
end