1
by Sebastien Delafond
Import upstream version 1.3.1 |
1 |
#
|
2 |
# This class was contributed by Mikko Tiihonen mikko DOT tiihonen AT hut DOT fi
|
|
3 |
#
|
|
4 |
module REXML |
|
5 |
module Encoding |
|
6 |
register( "CP-1252" ) do |o| |
|
7 |
class << o |
|
8 |
alias encode encode_cp1252 |
|
9 |
alias decode decode_cp1252 |
|
10 |
end
|
|
11 |
end
|
|
12 |
||
13 |
# Convert from UTF-8
|
|
14 |
def encode_cp1252(content) |
|
15 |
array_utf8 = content.unpack('U*') |
|
16 |
array_enc = [] |
|
17 |
array_utf8.each do |num| |
|
18 |
case num |
|
19 |
# shortcut first bunch basic characters
|
|
20 |
when 0..0xFF; array_enc << num |
|
21 |
# characters added compared to iso-8859-1
|
|
22 |
when 0x20AC; array_enc << 0x80 # 0xe2 0x82 0xac |
|
23 |
when 0x201A; array_enc << 0x82 # 0xe2 0x82 0x9a |
|
24 |
when 0x0192; array_enc << 0x83 # 0xc6 0x92 |
|
25 |
when 0x201E; array_enc << 0x84 # 0xe2 0x82 0x9e |
|
26 |
when 0x2026; array_enc << 0x85 # 0xe2 0x80 0xa6 |
|
27 |
when 0x2020; array_enc << 0x86 # 0xe2 0x80 0xa0 |
|
28 |
when 0x2021; array_enc << 0x87 # 0xe2 0x80 0xa1 |
|
29 |
when 0x02C6; array_enc << 0x88 # 0xcb 0x86 |
|
30 |
when 0x2030; array_enc << 0x89 # 0xe2 0x80 0xb0 |
|
31 |
when 0x0160; array_enc << 0x8A # 0xc5 0xa0 |
|
32 |
when 0x2039; array_enc << 0x8B # 0xe2 0x80 0xb9 |
|
33 |
when 0x0152; array_enc << 0x8C # 0xc5 0x92 |
|
34 |
when 0x017D; array_enc << 0x8E # 0xc5 0xbd |
|
35 |
when 0x2018; array_enc << 0x91 # 0xe2 0x80 0x98 |
|
36 |
when 0x2019; array_enc << 0x92 # 0xe2 0x80 0x99 |
|
37 |
when 0x201C; array_enc << 0x93 # 0xe2 0x80 0x9c |
|
38 |
when 0x201D; array_enc << 0x94 # 0xe2 0x80 0x9d |
|
39 |
when 0x2022; array_enc << 0x95 # 0xe2 0x80 0xa2 |
|
40 |
when 0x2013; array_enc << 0x96 # 0xe2 0x80 0x93 |
|
41 |
when 0x2014; array_enc << 0x97 # 0xe2 0x80 0x94 |
|
42 |
when 0x02DC; array_enc << 0x98 # 0xcb 0x9c |
|
43 |
when 0x2122; array_enc << 0x99 # 0xe2 0x84 0xa2 |
|
44 |
when 0x0161; array_enc << 0x9A # 0xc5 0xa1 |
|
45 |
when 0x203A; array_enc << 0x9B # 0xe2 0x80 0xba |
|
46 |
when 0x0152; array_enc << 0x9C # 0xc5 0x93 |
|
47 |
when 0x017E; array_enc << 0x9E # 0xc5 0xbe |
|
48 |
when 0x0178; array_enc << 0x9F # 0xc5 0xb8 |
|
49 |
else
|
|
50 |
# all remaining basic characters can be used directly
|
|
51 |
if num <= 0xFF |
|
52 |
array_enc << num |
|
53 |
else
|
|
54 |
# Numeric entity (&#nnnn;); shard by Stefan Scholl
|
|
55 |
array_enc.concat "&\##{num};".unpack('C*') |
|
56 |
end
|
|
57 |
end
|
|
58 |
end
|
|
59 |
array_enc.pack('C*') |
|
60 |
end
|
|
61 |
||
62 |
# Convert to UTF-8
|
|
63 |
def decode_cp1252(str) |
|
64 |
array_latin9 = str.unpack('C*') |
|
65 |
array_enc = [] |
|
66 |
array_latin9.each do |num| |
|
67 |
case num |
|
68 |
# characters that added compared to iso-8859-1
|
|
69 |
when 0x80; array_enc << 0x20AC # 0xe2 0x82 0xac |
|
70 |
when 0x82; array_enc << 0x201A # 0xe2 0x82 0x9a |
|
71 |
when 0x83; array_enc << 0x0192 # 0xc6 0x92 |
|
72 |
when 0x84; array_enc << 0x201E # 0xe2 0x82 0x9e |
|
73 |
when 0x85; array_enc << 0x2026 # 0xe2 0x80 0xa6 |
|
74 |
when 0x86; array_enc << 0x2020 # 0xe2 0x80 0xa0 |
|
75 |
when 0x87; array_enc << 0x2021 # 0xe2 0x80 0xa1 |
|
76 |
when 0x88; array_enc << 0x02C6 # 0xcb 0x86 |
|
77 |
when 0x89; array_enc << 0x2030 # 0xe2 0x80 0xb0 |
|
78 |
when 0x8A; array_enc << 0x0160 # 0xc5 0xa0 |
|
79 |
when 0x8B; array_enc << 0x2039 # 0xe2 0x80 0xb9 |
|
80 |
when 0x8C; array_enc << 0x0152 # 0xc5 0x92 |
|
81 |
when 0x8E; array_enc << 0x017D # 0xc5 0xbd |
|
82 |
when 0x91; array_enc << 0x2018 # 0xe2 0x80 0x98 |
|
83 |
when 0x92; array_enc << 0x2019 # 0xe2 0x80 0x99 |
|
84 |
when 0x93; array_enc << 0x201C # 0xe2 0x80 0x9c |
|
85 |
when 0x94; array_enc << 0x201D # 0xe2 0x80 0x9d |
|
86 |
when 0x95; array_enc << 0x2022 # 0xe2 0x80 0xa2 |
|
87 |
when 0x96; array_enc << 0x2013 # 0xe2 0x80 0x93 |
|
88 |
when 0x97; array_enc << 0x2014 # 0xe2 0x80 0x94 |
|
89 |
when 0x98; array_enc << 0x02DC # 0xcb 0x9c |
|
90 |
when 0x99; array_enc << 0x2122 # 0xe2 0x84 0xa2 |
|
91 |
when 0x9A; array_enc << 0x0161 # 0xc5 0xa1 |
|
92 |
when 0x9B; array_enc << 0x203A # 0xe2 0x80 0xba |
|
93 |
when 0x9C; array_enc << 0x0152 # 0xc5 0x93 |
|
94 |
when 0x9E; array_enc << 0x017E # 0xc5 0xbe |
|
95 |
when 0x9F; array_enc << 0x0178 # 0xc5 0xb8 |
|
96 |
else
|
|
97 |
array_enc << num |
|
98 |
end
|
|
99 |
end
|
|
100 |
array_enc.pack('U*') |
|
101 |
end
|
|
102 |
end
|
|
103 |
end
|