1 #!/usr/bin/env ruby
2
3 # The XChar library is provided courtesy of Sam Ruby (See
4 # http://intertwingly.net/stories/2005/09/28/xchar.rb)
5
6 # --------------------------------------------------------------------
7
8 # If the Builder::XChar module is not currently defined, fail on any
9 # name clashes in standard library classes.
10
11 module Builder
12 def self.check_for_name_collision(klass, method_name, defined_constant=nil)
13 if klass.instance_methods.include?(method_name.to_s)
14 fail RuntimeError,
15 "Name Collision: Method '#{method_name}' is already defined in #{klass}"
16 end
17 end
18 end
19
20 if ! defined?(Builder::XChar)
21 Builder.check_for_name_collision(String, "to_xs")
22 Builder.check_for_name_collision(Fixnum, "xchr")
23 end
24
25 ######################################################################
26 module Builder
27
28 ####################################################################
29 # XML Character converter, from Sam Ruby:
30 # (see http://intertwingly.net/stories/2005/09/28/xchar.rb).
31 #
32 module XChar # :nodoc:
33
34 # See
35 # http://intertwingly.net/stories/2004/04/14/i18n.html#CleaningWindows
36 # for details.
37 CP1252 = { # :nodoc:
38 128 => 8364, # euro sign
39 130 => 8218, # single low-9 quotation mark
40 131 => 402, # latin small letter f with hook
41 132 => 8222, # double low-9 quotation mark
42 133 => 8230, # horizontal ellipsis
43 134 => 8224, # dagger
44 135 => 8225, # double dagger
45 136 => 710, # modifier letter circumflex accent
46 137 => 8240, # per mille sign
47 138 => 352, # latin capital letter s with caron
48 139 => 8249, # single left-pointing angle quotation mark
49 140 => 338, # latin capital ligature oe
50 142 => 381, # latin capital letter z with caron
51 145 => 8216, # left single quotation mark
52 146 => 8217, # right single quotation mark
53 147 => 8220, # left double quotation mark
54 148 => 8221, # right double quotation mark
55 149 => 8226, # bullet
56 150 => 8211, # en dash
57 151 => 8212, # em dash
58 152 => 732, # small tilde
59 153 => 8482, # trade mark sign
60 154 => 353, # latin small letter s with caron
61 155 => 8250, # single right-pointing angle quotation mark
62 156 => 339, # latin small ligature oe
63 158 => 382, # latin small letter z with caron
64 159 => 376, # latin capital letter y with diaeresis
65 }
66
67 # See http://www.w3.org/TR/REC-xml/#dt-chardata for details.
68 PREDEFINED = {
69 38 => '&', # ampersand
70 60 => '<', # left angle bracket
71 62 => '>', # right angle bracket
72 }
73
74 # See http://www.w3.org/TR/REC-xml/#charsets for details.
75 VALID = [
76 0x9, 0xA, 0xD,
77 (0x20..0xD7FF),
78 (0xE000..0xFFFD),
79 (0x10000..0x10FFFF)
80 ]
81 end
82
83 end
84
85
86 ######################################################################
87 # Enhance the Fixnum class with a XML escaped character conversion.
88 #
89 class Fixnum
90 XChar = Builder::XChar if ! defined?(XChar)
91
92 # XML escaped version of chr
93 def xchr
94 n = XChar::CP1252[self] || self
95 case n when *XChar::VALID
96 XChar::PREDEFINED[n] or (n<128 ? n.chr : "&##{n};")
97 else
98 '*'
99 end
100 end
101 end
102
103
104 ######################################################################
105 # Enhance the String class with a XML escaped character version of
106 # to_s.
107 #
108 class String
109 # XML escaped version of to_s
110 def to_xs
111 unpack('U*').map {|n| n.xchr}.join # ASCII, UTF-8
112 rescue
113 unpack('C*').map {|n| n.xchr}.join # ISO-8859-1, WIN-1252
114 end
115 end