line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
# $Id$ |
2
|
|
|
|
|
|
|
|
3
|
|
|
|
|
|
|
package XML::SAX::PurePerl; # NB, not ::EncodingDetect! |
4
|
|
|
|
|
|
|
|
5
|
12
|
|
|
12
|
|
70
|
use strict; |
|
12
|
|
|
|
|
25
|
|
|
12
|
|
|
|
|
8132
|
|
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
sub encoding_detect { |
8
|
22
|
|
|
22
|
0
|
45
|
my ($parser, $reader) = @_; |
9
|
|
|
|
|
|
|
|
10
|
22
|
|
|
|
|
57
|
my $error = "Invalid byte sequence at start of file"; |
11
|
|
|
|
|
|
|
|
12
|
22
|
|
|
|
|
133
|
my $data = $reader->data; |
13
|
22
|
50
|
|
|
|
868
|
if ($data =~ /^\x00\x00\xFE\xFF/) { |
|
|
50
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
14
|
|
|
|
|
|
|
# BO-UCS4-be |
15
|
0
|
|
|
|
|
0
|
$reader->move_along(4); |
16
|
0
|
|
|
|
|
0
|
$reader->set_encoding('UCS-4BE'); |
17
|
0
|
|
|
|
|
0
|
return; |
18
|
|
|
|
|
|
|
} |
19
|
|
|
|
|
|
|
elsif ($data =~ /^\x00\x00\xFF\xFE/) { |
20
|
|
|
|
|
|
|
# BO-UCS-4-2143 |
21
|
0
|
|
|
|
|
0
|
$reader->move_along(4); |
22
|
0
|
|
|
|
|
0
|
$reader->set_encoding('UCS-4-2143'); |
23
|
0
|
|
|
|
|
0
|
return; |
24
|
|
|
|
|
|
|
} |
25
|
|
|
|
|
|
|
elsif ($data =~ /^\x00\x00\x00\x3C/) { |
26
|
0
|
|
|
|
|
0
|
$reader->set_encoding('UCS-4BE'); |
27
|
0
|
|
|
|
|
0
|
return; |
28
|
|
|
|
|
|
|
} |
29
|
|
|
|
|
|
|
elsif ($data =~ /^\x00\x00\x3C\x00/) { |
30
|
0
|
|
|
|
|
0
|
$reader->set_encoding('UCS-4-2143'); |
31
|
0
|
|
|
|
|
0
|
return; |
32
|
|
|
|
|
|
|
} |
33
|
|
|
|
|
|
|
elsif ($data =~ /^\x00\x3C\x00\x00/) { |
34
|
0
|
|
|
|
|
0
|
$reader->set_encoding('UCS-4-3412'); |
35
|
0
|
|
|
|
|
0
|
return; |
36
|
|
|
|
|
|
|
} |
37
|
|
|
|
|
|
|
elsif ($data =~ /^\x00\x3C\x00\x3F/) { |
38
|
0
|
|
|
|
|
0
|
$reader->set_encoding('UTF-16BE'); |
39
|
0
|
|
|
|
|
0
|
return; |
40
|
|
|
|
|
|
|
} |
41
|
|
|
|
|
|
|
elsif ($data =~ /^\xFF\xFE\x00\x00/) { |
42
|
|
|
|
|
|
|
# BO-UCS-4LE |
43
|
0
|
|
|
|
|
0
|
$reader->move_along(4); |
44
|
0
|
|
|
|
|
0
|
$reader->set_encoding('UCS-4LE'); |
45
|
0
|
|
|
|
|
0
|
return; |
46
|
|
|
|
|
|
|
} |
47
|
|
|
|
|
|
|
elsif ($data =~ /^\xFF\xFE/) { |
48
|
1
|
|
|
|
|
5
|
$reader->move_along(2); |
49
|
1
|
|
|
|
|
4
|
$reader->set_encoding('UTF-16LE'); |
50
|
1
|
|
|
|
|
3
|
return; |
51
|
|
|
|
|
|
|
} |
52
|
|
|
|
|
|
|
elsif ($data =~ /^\xFE\xFF\x00\x00/) { |
53
|
0
|
|
|
|
|
0
|
$reader->move_along(4); |
54
|
0
|
|
|
|
|
0
|
$reader->set_encoding('UCS-4-3412'); |
55
|
0
|
|
|
|
|
0
|
return; |
56
|
|
|
|
|
|
|
} |
57
|
|
|
|
|
|
|
elsif ($data =~ /^\xFE\xFF/) { |
58
|
1
|
|
|
|
|
7
|
$reader->move_along(2); |
59
|
1
|
|
|
|
|
4
|
$reader->set_encoding('UTF-16BE'); |
60
|
1
|
|
|
|
|
6
|
return; |
61
|
|
|
|
|
|
|
} |
62
|
|
|
|
|
|
|
elsif ($data =~ /^\xEF\xBB\xBF/) { # UTF-8 BOM |
63
|
0
|
|
|
|
|
0
|
$reader->move_along(3); |
64
|
0
|
|
|
|
|
0
|
$reader->set_encoding('UTF-8'); |
65
|
0
|
|
|
|
|
0
|
return; |
66
|
|
|
|
|
|
|
} |
67
|
|
|
|
|
|
|
elsif ($data =~ /^\x3C\x00\x00\x00/) { |
68
|
0
|
|
|
|
|
0
|
$reader->set_encoding('UCS-4LE'); |
69
|
0
|
|
|
|
|
0
|
return; |
70
|
|
|
|
|
|
|
} |
71
|
|
|
|
|
|
|
elsif ($data =~ /^\x3C\x00\x3F\x00/) { |
72
|
0
|
|
|
|
|
0
|
$reader->set_encoding('UTF-16LE'); |
73
|
0
|
|
|
|
|
0
|
return; |
74
|
|
|
|
|
|
|
} |
75
|
|
|
|
|
|
|
elsif ($data =~ /^\x3C\x3F\x78\x6D/) { |
76
|
|
|
|
|
|
|
# $reader->set_encoding('UTF-8'); |
77
|
9
|
|
|
|
|
59
|
return; |
78
|
|
|
|
|
|
|
} |
79
|
|
|
|
|
|
|
elsif ($data =~ /^\x3C\x3F\x78/) { |
80
|
|
|
|
|
|
|
# $reader->set_encoding('UTF-8'); |
81
|
0
|
|
|
|
|
0
|
return; |
82
|
|
|
|
|
|
|
} |
83
|
|
|
|
|
|
|
elsif ($data =~ /^\x3C\x3F/) { |
84
|
|
|
|
|
|
|
# $reader->set_encoding('UTF-8'); |
85
|
0
|
|
|
|
|
0
|
return; |
86
|
|
|
|
|
|
|
} |
87
|
|
|
|
|
|
|
elsif ($data =~ /^\x3C/) { |
88
|
|
|
|
|
|
|
# $reader->set_encoding('UTF-8'); |
89
|
11
|
|
|
|
|
51
|
return; |
90
|
|
|
|
|
|
|
} |
91
|
|
|
|
|
|
|
elsif ($data =~ /^[\x20\x09\x0A\x0D]+\x3C[^\x3F]/) { |
92
|
|
|
|
|
|
|
# $reader->set_encoding('UTF-8'); |
93
|
0
|
|
|
|
|
|
return; |
94
|
|
|
|
|
|
|
} |
95
|
|
|
|
|
|
|
elsif ($data =~ /^\x4C\x6F\xA7\x94/) { |
96
|
0
|
|
|
|
|
|
$reader->set_encoding('EBCDIC'); |
97
|
0
|
|
|
|
|
|
return; |
98
|
|
|
|
|
|
|
} |
99
|
|
|
|
|
|
|
|
100
|
0
|
|
|
|
|
|
warn("Unable to recognise encoding of this document"); |
101
|
0
|
|
|
|
|
|
return; |
102
|
|
|
|
|
|
|
} |
103
|
|
|
|
|
|
|
|
104
|
|
|
|
|
|
|
1; |
105
|
|
|
|
|
|
|
|