Message/Charset/Info.pm

package Message::Charset::Info;
use strict;
our $VERSION=do{my @r=(q$Revision: 1.1 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};

sub UNREGISTERED_CHARSET_NAME () { 0b1 }
sub REGISTERED_CHARSET_NAME () { 0b10 }
sub PRIMARY_CHARSET_NAME () { 0b100 | REGISTERED_CHARSET_NAME }
    ## "Name:" field for IANA names
sub PREFERRED_CHARSET_NAME () { 0b1000 | REGISTERED_CHARSET_NAME }
    ## "preferred MIME name" for IANA names

## iana_status
sub STATUS_COMMON () { 0b1 }
sub STATUS_LIMITED_USE () { 0b10 }
sub STATUS_OBSOLETE () { 0b100 }

## iana_names
## is_html_ascii_superset: "superset of US-ASCII (specifically, ANSI_X3.4-1968)
##     for bytes in the range 0x09 - 0x0D, 0x20, 0x21, 0x22, 0x26, 0x27,
##     0x2C - 0x3F, 0x41 - 0x5A, and 0x61 - 0x7A" [HTML5]
## is_ebcdic_based

## ISSUE: Shift_JIS is a superset of US-ASCII?  ISO-2022-JP is?
## ISSUE: 0x5F (_) should be added to the range?

my $Charset;

our $IANACharset;

$Charset->{'us-ascii'}
= $IANACharset->{'ansi_x3.4-1968'}
= $IANACharset->{'iso-ir-6'}
= $IANACharset->{'ansi_x3.4-1986'}
= $IANACharset->{'iso_646.irv:1991'}
= $IANACharset->{'ascii'}
= $IANACharset->{'iso646-us'}
= $IANACharset->{'us-ascii'}
= $IANACharset->{'us'}
= $IANACharset->{'ibm367'}
= $IANACharset->{'cp367'}
= $IANACharset->{'csascii'}
= {
  iana_names => {
    'ansi_x3.4-1968' => PRIMARY_CHARSET_NAME,
    'iso-ir-6' => REGISTERED_CHARSET_NAME,
    'ansi_x3.4-1986' => REGISTERED_CHARSET_NAME,
    'iso_646.irv:1991' => REGISTERED_CHARSET_NAME,
    'ascii' => REGISTERED_CHARSET_NAME,
    'iso646-us' => REGISTERED_CHARSET_NAME,
    'us-ascii' => PREFERRED_CHARSET_NAME,
    'us' => REGISTERED_CHARSET_NAME,
    'ibm367' => REGISTERED_CHARSET_NAME,
    'cp367' => REGISTERED_CHARSET_NAME,
    'csascii' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'iso-8859-1'}
= $IANACharset->{'iso_8859-1:1987'}
= $IANACharset->{'iso-ir-100'}
= $IANACharset->{'iso_8859-1'}
= $IANACharset->{'iso-8859-1'}
= $IANACharset->{'latin1'}
= $IANACharset->{'l1'}
= $IANACharset->{'ibm819'}
= $IANACharset->{'cp819'}
= $IANACharset->{'csisolatin1'}
= {
  iana_names => {
    'iso_8859-1:1987' => PRIMARY_CHARSET_NAME,
    'iso-ir-100' => REGISTERED_CHARSET_NAME,
    'iso_8859-1' => REGISTERED_CHARSET_NAME,
    'iso-8859-1' => PREFERRED_CHARSET_NAME,
    'latin1' => REGISTERED_CHARSET_NAME,
    'l1' => REGISTERED_CHARSET_NAME,
    'ibm819' => REGISTERED_CHARSET_NAME,
    'cp819' => REGISTERED_CHARSET_NAME,
    'csisolatin1' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'iso-8859-2'}
= $IANACharset->{'iso_8859-2:1987'}
= $IANACharset->{'iso-ir-101'}
= $IANACharset->{'iso_8859-2'}
= $IANACharset->{'iso-8859-2'}
= $IANACharset->{'latin2'}
= $IANACharset->{'l2'}
= $IANACharset->{'csisolatin2'}
= {
  iana_names => {
    'iso_8859-2:1987' => PRIMARY_CHARSET_NAME,
    'iso-ir-101' => REGISTERED_CHARSET_NAME,
    'iso_8859-2' => REGISTERED_CHARSET_NAME,
    'iso-8859-2' => PREFERRED_CHARSET_NAME,
    'latin2' => REGISTERED_CHARSET_NAME,
    'l2' => REGISTERED_CHARSET_NAME,
    'csisolatin2' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'iso-8859-3'}
= $IANACharset->{'iso_8859-3:1988'}
= $IANACharset->{'iso-ir-109'}
= $IANACharset->{'iso_8859-3'}
= $IANACharset->{'iso-8859-3'}
= $IANACharset->{'latin3'}
= $IANACharset->{'l3'}
= $IANACharset->{'csisolatin3'}
= {
  iana_names => {
    'iso_8859-3:1988' => PRIMARY_CHARSET_NAME,
    'iso-ir-109' => REGISTERED_CHARSET_NAME,
    'iso_8859-3' => REGISTERED_CHARSET_NAME,
    'iso-8859-3' => PREFERRED_CHARSET_NAME,
    'latin3' => REGISTERED_CHARSET_NAME,
    'l3' => REGISTERED_CHARSET_NAME,
    'csisolatin3' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'iso-8859-4'}
= $IANACharset->{'iso_8859-4:1988'}
= $IANACharset->{'iso-ir-110'}
= $IANACharset->{'iso_8859-4'}
= $IANACharset->{'iso-8859-4'}
= $IANACharset->{'latin4'}
= $IANACharset->{'l4'}
= $IANACharset->{'csisolatin4'}
= {
  iana_names => {
    'iso_8859-4:1988' => PRIMARY_CHARSET_NAME,
    'iso-ir-110' => REGISTERED_CHARSET_NAME,
    'iso_8859-4' => REGISTERED_CHARSET_NAME,
    'iso-8859-4' => PREFERRED_CHARSET_NAME,
    'latin4' => REGISTERED_CHARSET_NAME,
    'l4' => REGISTERED_CHARSET_NAME,
    'csisolatin4' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'iso-8859-5'}
= $IANACharset->{'iso_8859-5:1988'}
= $IANACharset->{'iso-ir-144'}
= $IANACharset->{'iso_8859-5'}
= $IANACharset->{'iso-8859-5'}
= $IANACharset->{'cyrillic'}
= $IANACharset->{'csisolatincyrillic'}
= {
  iana_names => {
    'iso_8859-5:1988' => PRIMARY_CHARSET_NAME,
    'iso-ir-144' => REGISTERED_CHARSET_NAME,
    'iso_8859-5' => REGISTERED_CHARSET_NAME,
    'iso-8859-5' => PREFERRED_CHARSET_NAME,
    'cyrillic' => REGISTERED_CHARSET_NAME,
    'csisolatincyrillic' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'iso-8859-6'}
= $IANACharset->{'iso_8859-6:1987'}
= $IANACharset->{'iso-ir-127'}
= $IANACharset->{'iso_8859-6'}
= $IANACharset->{'iso-8859-6'}
= $IANACharset->{'ecma-114'}
= $IANACharset->{'asmo-708'}
= $IANACharset->{'arabic'}
= $IANACharset->{'csisolatinarabic'}
= {
  iana_names => {
    'iso_8859-6:1987' => PRIMARY_CHARSET_NAME,
    'iso-ir-127' => REGISTERED_CHARSET_NAME,
    'iso_8859-6' => REGISTERED_CHARSET_NAME,
    'iso-8859-6' => PREFERRED_CHARSET_NAME,
    'ecma-114' => REGISTERED_CHARSET_NAME,
    'asmo-708' => REGISTERED_CHARSET_NAME,
    'arabic' => REGISTERED_CHARSET_NAME,
    'csisolatinarabic' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
      ## NOTE: 3/0..3/9 have different semantics from U+0030..0039,
      ## but have same character names (maybe).
};

$Charset->{'iso-8859-7'}
= $IANACharset->{'iso_8859-7:1987'}
= $IANACharset->{'iso-ir-126'}
= $IANACharset->{'iso_8859-7'}
= $IANACharset->{'iso-8859-7'}
= $IANACharset->{'elot_928'}
= $IANACharset->{'ecma-118'}
= $IANACharset->{'greek'}
= $IANACharset->{'greek8'}
= $IANACharset->{'csisolatingreek'}
= {
  iana_names => {
    'iso_8859-7:1987' => PRIMARY_CHARSET_NAME,
    'iso-ir-126' => REGISTERED_CHARSET_NAME,
    'iso_8859-7' => REGISTERED_CHARSET_NAME,
    'iso-8859-7' => PREFERRED_CHARSET_NAME,
    'elot_928' => REGISTERED_CHARSET_NAME,
    'ecma-118' => REGISTERED_CHARSET_NAME,
    'greek' => REGISTERED_CHARSET_NAME,
    'greek8' => REGISTERED_CHARSET_NAME,
    'csisolatingreek' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'iso-8859-8'}
= $IANACharset->{'iso_8859-8:1988'}
= $IANACharset->{'iso-ir-138'}
= $IANACharset->{'iso_8859-8'}
= $IANACharset->{'iso-8859-8'}
= $IANACharset->{'hebrew'}
= $IANACharset->{'csisolatinhebrew'}
= {
  iana_names => {
    'iso_8859-8:1988' => PRIMARY_CHARSET_NAME,
    'iso-ir-138' => REGISTERED_CHARSET_NAME,
    'iso_8859-8' => REGISTERED_CHARSET_NAME,
    'iso-8859-8' => PREFERRED_CHARSET_NAME,
    'hebrew' => REGISTERED_CHARSET_NAME,
    'csisolatinhebrew' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'iso-8859-9'}
= $IANACharset->{'iso_8859-9:1989'}
= $IANACharset->{'iso-ir-148'}
= $IANACharset->{'iso_8859-9'}
= $IANACharset->{'iso-8859-9'}
= $IANACharset->{'latin5'}
= $IANACharset->{'l5'}
= $IANACharset->{'csisolatin5'}
= {
  iana_names => {
    'iso_8859-9:1989' => PRIMARY_CHARSET_NAME,
    'iso-ir-148' => REGISTERED_CHARSET_NAME,
    'iso_8859-9' => REGISTERED_CHARSET_NAME,
    'iso-8859-9' => PREFERRED_CHARSET_NAME,
    'latin5' => REGISTERED_CHARSET_NAME,
    'l5' => REGISTERED_CHARSET_NAME,
    'csisolatin5' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'iso-8859-10'}
= $IANACharset->{'iso-8859-10'}
= $IANACharset->{'iso-ir-157'}
= $IANACharset->{'l6'}
= $IANACharset->{'iso_8859-10:1992'}
= $IANACharset->{'csisolatin6'}
= $IANACharset->{'latin6'}
= {
  iana_names => {
    'iso-8859-10' => PRIMARY_CHARSET_NAME | PREFERRED_CHARSET_NAME,
    'iso-ir-157' => REGISTERED_CHARSET_NAME,
    'l6' => REGISTERED_CHARSET_NAME,
    'iso_8859-10:1992' => REGISTERED_CHARSET_NAME,
    'csisolatin6' => REGISTERED_CHARSET_NAME,
    'latin6' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'iso_6937-2-add'}
= $IANACharset->{'iso_6937-2-add'}
= $IANACharset->{'iso-ir-142'}
= $IANACharset->{'csisotextcomm'}
= {
  iana_names => {
    'iso_6937-2-add' => PRIMARY_CHARSET_NAME,
    'iso-ir-142' => REGISTERED_CHARSET_NAME,
    'csisotextcomm' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'jis_x0201'}
= $IANACharset->{'jis_x0201'}
= $IANACharset->{'x0201'}
= $IANACharset->{'cshalfwidthkatakana'}
= {
  iana_names => {
    'jis_x0201' => PRIMARY_CHARSET_NAME,
    'x0201' => REGISTERED_CHARSET_NAME,
    'cshalfwidthkatakana' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'jis_encoding'}
= $IANACharset->{'jis_encoding'}
= $IANACharset->{'csjisencoding'}
= {
  iana_names => {
    'jis_encoding' => PRIMARY_CHARSET_NAME,
    'csjisencoding' => REGISTERED_CHARSET_NAME,
  },
  ## NOTE: What is this?
};

$Charset->{'shift_jis'}
= $IANACharset->{'shift_jis'}
= $IANACharset->{'ms_kanji'}
= $IANACharset->{'csshiftjis'}
= {
  iana_names => {
    'shift_jis' => PREFERRED_CHARSET_NAME | PRIMARY_CHARSET_NAME,
    'ms_kanji' => REGISTERED_CHARSET_NAME,
    'csshiftjis' => REGISTERED_CHARSET_NAME,
  },
  mime_text_suitable => 1,
};

$Charset->{'euc-jp'}
= $IANACharset->{'extended_unix_code_packed_format_for_japanese'}
= $IANACharset->{'cseucpkdfmtjapanese'}
= $IANACharset->{'euc-jp'}
= {
  iana_names => {
    'extended_unix_code_packed_format_for_japanese' => PRIMARY_CHARSET_NAME,
    'cseucpkdfmtjapanese' => REGISTERED_CHARSET_NAME,
    'euc-jp' => PREFERRED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'extended_unix_code_fixed_width_for_japanese'}
= $IANACharset->{'extended_unix_code_fixed_width_for_japanese'}
= $IANACharset->{'cseucfixwidjapanese'}
= {
  iana_names => {
    'extended_unix_code_fixed_width_for_japanese' => PRIMARY_CHARSET_NAME,
    'cseucfixwidjapanese' => REGISTERED_CHARSET_NAME,
  },
};

## TODO: ...

$Charset->{'euc-kr'}
= $IANACharset->{'euc-kr'}
= $IANACharset->{'cseuckr'}
= {
  iana_names => {
    'euc-kr' => PRIMARY_CHARSET_NAME | PREFERRED_CHARSET_NAME,
    'cseuckr' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

$Charset->{'iso-2022-jp'}
= $IANACharset->{'iso-2022-jp'}
= $IANACharset->{'csiso2022jp'}
= {
  iana_names => {
    'iso-2022-jp' => PREFERRED_CHARSET_NAME | PRIMARY_CHARSET_NAME,
    'csiso2022jp' => REGISTERED_CHARSET_NAME,
  },
  mime_text_suitable => 1,
};

$Charset->{'iso-2022-jp-2'}
= $IANACharset->{'iso-2022-jp-2'}
= $IANACharset->{'csiso2022jp2'}
= {
  iana_names => {
    'iso-2022-jp-2' => PREFERRED_CHARSET_NAME | PRIMARY_CHARSET_NAME,
    'csiso2022jp2' => REGISTERED_CHARSET_NAME,
  },
  mime_text_suitable => 1,
};

## TODO: ...

$Charset->{'utf-8'}
= $IANACharset->{'utf-8'}
= {
  iana_names => {
    'utf-8' => PRIMARY_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

## TODO: ...

$Charset->{'gbk'}
= $IANACharset->{'gbk'}
= $IANACharset->{'cp936'}
= $IANACharset->{'ms936'}
= $IANACharset->{'windows-936'}
= {
  iana_names => {
    'gbk' => PRIMARY_CHARSET_NAME,
    'cp936' => REGISTERED_CHARSET_NAME,
    'ms936' => REGISTERED_CHARSET_NAME,
    'windows-936' => REGISTERED_CHARSET_NAME,
  },
  iana_status => STATUS_COMMON | STATUS_OBSOLETE,
  mime_text_suitable => 1,
};

$Charset->{'gb18030'}
= $IANACharset->{'gb18030'}
= {
  iana_names => {
    'gb18030' => PRIMARY_CHARSET_NAME,
  },
  iana_status => STATUS_COMMON,
  mime_text_suitable => 1,
};

## TODO: ...

$Charset->{'utf-16be'}
= $IANACharset->{'utf-16be'}
= {
  iana_names => {
    'utf-16be' => PRIMARY_CHARSET_NAME,
  },
};

$Charset->{'utf-16le'}
= $IANACharset->{'utf-16le'}
= {
  iana_names => {
    'utf-16le' => PRIMARY_CHARSET_NAME,
  },
};

$Charset->{'utf-16'}
= $IANACharset->{'utf-16'}
= {
  iana_names => {
    'utf-16' => PRIMARY_CHARSET_NAME,
  },
};

## TODO: ...

$Charset->{'windows-31j'}
= $IANACharset->{'windows-31j'}
= $IANACharset->{'cswindows31j'}
= {
  iana_names => {
    'windows-31j' => PRIMARY_CHARSET_NAME,
    'cswindows31j' => REGISTERED_CHARSET_NAME,
  },
  iana_status => STATUS_LIMITED_USE, # maybe
  mime_text_suitable => 1,
};

$Charset->{'gb2312'}
= $IANACharset->{'gb2312'}
= $IANACharset->{'csgb2312'}
= {
  iana_names => {
    'gb2312' => PRIMARY_CHARSET_NAME | PREFERRED_CHARSET_NAME,
    'csgb2312' => REGISTERED_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
  mime_text_suitable => 1,
};

$Charset->{'big5'}
= $IANACharset->{'big5'}
= $IANACharset->{'csbig5'}
= {
  iana_names => {
    'big5' => PRIMARY_CHARSET_NAME | PREFERRED_CHARSET_NAME,
    'csbig5' => REGISTERED_CHARSET_NAME,
  },
  mime_text_suitable => 1,
};

## TODO: ...

$Charset->{'big5-hkscs'}
= $IANACharset->{'big5-hkscs'}
= {
  iana_names => {
    'big5-hkscs' => PRIMARY_CHARSET_NAME,
  },
  mime_text_suitable => 1,
};

## TODO: ...

$Charset->{'windows-1252'}
= $IANACharset->{'windows-1252'}
= {
  iana_names => {
    'windows-1252' => PRIMARY_CHARSET_NAME,
  },
  is_html_ascii_superset => 1,
};

## TODO: ...

sub is_syntactically_iana_charset_name ($) {
  my $name = shift;
  return $name =~ /\A[\x20-\x7E]{1,40}\z/;
} # is_suntactically_valid_iana_charset_name

1;
## $Date: 2007/11/18 11:08:40 $

1	wakaba	1.1	package Message::Charset::Info;
2			use strict;
3	wakaba	1.2	our $VERSION=do{my @r=(q$Revision: 1.1 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};
4	wakaba	1.1
5			sub UNREGISTERED_CHARSET_NAME () { 0b1 }
6			sub REGISTERED_CHARSET_NAME () { 0b10 }
7			sub PRIMARY_CHARSET_NAME () { 0b100 \| REGISTERED_CHARSET_NAME }
8			## "Name:" field for IANA names
9			sub PREFERRED_CHARSET_NAME () { 0b1000 \| REGISTERED_CHARSET_NAME }
10			## "preferred MIME name" for IANA names
11
12	wakaba	1.2	## iana_status
13			sub STATUS_COMMON () { 0b1 }
14			sub STATUS_LIMITED_USE () { 0b10 }
15			sub STATUS_OBSOLETE () { 0b100 }
16
17	wakaba	1.1	## iana_names
18			## is_html_ascii_superset: "superset of US-ASCII (specifically, ANSI_X3.4-1968)
19			## for bytes in the range 0x09 - 0x0D, 0x20, 0x21, 0x22, 0x26, 0x27,
20			## 0x2C - 0x3F, 0x41 - 0x5A, and 0x61 - 0x7A" [HTML5]
21			## is_ebcdic_based
22
23			## ISSUE: Shift_JIS is a superset of US-ASCII? ISO-2022-JP is?
24			## ISSUE: 0x5F (_) should be added to the range?
25
26			my $Charset;
27
28			our $IANACharset;
29
30			$Charset->{'us-ascii'}
31			= $IANACharset->{'ansi_x3.4-1968'}
32			= $IANACharset->{'iso-ir-6'}
33			= $IANACharset->{'ansi_x3.4-1986'}
34			= $IANACharset->{'iso_646.irv:1991'}
35			= $IANACharset->{'ascii'}
36			= $IANACharset->{'iso646-us'}
37			= $IANACharset->{'us-ascii'}
38			= $IANACharset->{'us'}
39			= $IANACharset->{'ibm367'}
40			= $IANACharset->{'cp367'}
41			= $IANACharset->{'csascii'}
42			= {
43			iana_names => {
44			'ansi_x3.4-1968' => PRIMARY_CHARSET_NAME,
45			'iso-ir-6' => REGISTERED_CHARSET_NAME,
46			'ansi_x3.4-1986' => REGISTERED_CHARSET_NAME,
47			'iso_646.irv:1991' => REGISTERED_CHARSET_NAME,
48			'ascii' => REGISTERED_CHARSET_NAME,
49			'iso646-us' => REGISTERED_CHARSET_NAME,
50			'us-ascii' => PREFERRED_CHARSET_NAME,
51			'us' => REGISTERED_CHARSET_NAME,
52			'ibm367' => REGISTERED_CHARSET_NAME,
53			'cp367' => REGISTERED_CHARSET_NAME,
54			'csascii' => REGISTERED_CHARSET_NAME,
55			},
56			is_html_ascii_superset => 1,
57			};
58
59			$Charset->{'iso-8859-1'}
60			= $IANACharset->{'iso_8859-1:1987'}
61			= $IANACharset->{'iso-ir-100'}
62			= $IANACharset->{'iso_8859-1'}
63			= $IANACharset->{'iso-8859-1'}
64			= $IANACharset->{'latin1'}
65			= $IANACharset->{'l1'}
66			= $IANACharset->{'ibm819'}
67			= $IANACharset->{'cp819'}
68			= $IANACharset->{'csisolatin1'}
69			= {
70			iana_names => {
71			'iso_8859-1:1987' => PRIMARY_CHARSET_NAME,
72			'iso-ir-100' => REGISTERED_CHARSET_NAME,
73			'iso_8859-1' => REGISTERED_CHARSET_NAME,
74			'iso-8859-1' => PREFERRED_CHARSET_NAME,
75			'latin1' => REGISTERED_CHARSET_NAME,
76			'l1' => REGISTERED_CHARSET_NAME,
77			'ibm819' => REGISTERED_CHARSET_NAME,
78			'cp819' => REGISTERED_CHARSET_NAME,
79			'csisolatin1' => REGISTERED_CHARSET_NAME,
80			},
81			is_html_ascii_superset => 1,
82			};
83
84	wakaba	1.2	$Charset->{'iso-8859-2'}
85			= $IANACharset->{'iso_8859-2:1987'}
86			= $IANACharset->{'iso-ir-101'}
87			= $IANACharset->{'iso_8859-2'}
88			= $IANACharset->{'iso-8859-2'}
89			= $IANACharset->{'latin2'}
90			= $IANACharset->{'l2'}
91			= $IANACharset->{'csisolatin2'}
92			= {
93			iana_names => {
94			'iso_8859-2:1987' => PRIMARY_CHARSET_NAME,
95			'iso-ir-101' => REGISTERED_CHARSET_NAME,
96			'iso_8859-2' => REGISTERED_CHARSET_NAME,
97			'iso-8859-2' => PREFERRED_CHARSET_NAME,
98			'latin2' => REGISTERED_CHARSET_NAME,
99			'l2' => REGISTERED_CHARSET_NAME,
100			'csisolatin2' => REGISTERED_CHARSET_NAME,
101			},
102			is_html_ascii_superset => 1,
103			};
104
105			$Charset->{'iso-8859-3'}
106			= $IANACharset->{'iso_8859-3:1988'}
107			= $IANACharset->{'iso-ir-109'}
108			= $IANACharset->{'iso_8859-3'}
109			= $IANACharset->{'iso-8859-3'}
110			= $IANACharset->{'latin3'}
111			= $IANACharset->{'l3'}
112			= $IANACharset->{'csisolatin3'}
113			= {
114			iana_names => {
115			'iso_8859-3:1988' => PRIMARY_CHARSET_NAME,
116			'iso-ir-109' => REGISTERED_CHARSET_NAME,
117			'iso_8859-3' => REGISTERED_CHARSET_NAME,
118			'iso-8859-3' => PREFERRED_CHARSET_NAME,
119			'latin3' => REGISTERED_CHARSET_NAME,
120			'l3' => REGISTERED_CHARSET_NAME,
121			'csisolatin3' => REGISTERED_CHARSET_NAME,
122			},
123			is_html_ascii_superset => 1,
124			};
125
126			$Charset->{'iso-8859-4'}
127			= $IANACharset->{'iso_8859-4:1988'}
128			= $IANACharset->{'iso-ir-110'}
129			= $IANACharset->{'iso_8859-4'}
130			= $IANACharset->{'iso-8859-4'}
131			= $IANACharset->{'latin4'}
132			= $IANACharset->{'l4'}
133			= $IANACharset->{'csisolatin4'}
134			= {
135			iana_names => {
136			'iso_8859-4:1988' => PRIMARY_CHARSET_NAME,
137			'iso-ir-110' => REGISTERED_CHARSET_NAME,
138			'iso_8859-4' => REGISTERED_CHARSET_NAME,
139			'iso-8859-4' => PREFERRED_CHARSET_NAME,
140			'latin4' => REGISTERED_CHARSET_NAME,
141			'l4' => REGISTERED_CHARSET_NAME,
142			'csisolatin4' => REGISTERED_CHARSET_NAME,
143			},
144			is_html_ascii_superset => 1,
145			};
146
147			$Charset->{'iso-8859-5'}
148			= $IANACharset->{'iso_8859-5:1988'}
149			= $IANACharset->{'iso-ir-144'}
150			= $IANACharset->{'iso_8859-5'}
151			= $IANACharset->{'iso-8859-5'}
152			= $IANACharset->{'cyrillic'}
153			= $IANACharset->{'csisolatincyrillic'}
154			= {
155			iana_names => {
156			'iso_8859-5:1988' => PRIMARY_CHARSET_NAME,
157			'iso-ir-144' => REGISTERED_CHARSET_NAME,
158			'iso_8859-5' => REGISTERED_CHARSET_NAME,
159			'iso-8859-5' => PREFERRED_CHARSET_NAME,
160			'cyrillic' => REGISTERED_CHARSET_NAME,
161			'csisolatincyrillic' => REGISTERED_CHARSET_NAME,
162			},
163			is_html_ascii_superset => 1,
164			};
165
166			$Charset->{'iso-8859-6'}
167			= $IANACharset->{'iso_8859-6:1987'}
168			= $IANACharset->{'iso-ir-127'}
169			= $IANACharset->{'iso_8859-6'}
170			= $IANACharset->{'iso-8859-6'}
171			= $IANACharset->{'ecma-114'}
172			= $IANACharset->{'asmo-708'}
173			= $IANACharset->{'arabic'}
174			= $IANACharset->{'csisolatinarabic'}
175			= {
176			iana_names => {
177			'iso_8859-6:1987' => PRIMARY_CHARSET_NAME,
178			'iso-ir-127' => REGISTERED_CHARSET_NAME,
179			'iso_8859-6' => REGISTERED_CHARSET_NAME,
180			'iso-8859-6' => PREFERRED_CHARSET_NAME,
181			'ecma-114' => REGISTERED_CHARSET_NAME,
182			'asmo-708' => REGISTERED_CHARSET_NAME,
183			'arabic' => REGISTERED_CHARSET_NAME,
184			'csisolatinarabic' => REGISTERED_CHARSET_NAME,
185			},
186			is_html_ascii_superset => 1,
187			## NOTE: 3/0..3/9 have different semantics from U+0030..0039,
188			## but have same character names (maybe).
189			};
190
191			$Charset->{'iso-8859-7'}
192			= $IANACharset->{'iso_8859-7:1987'}
193			= $IANACharset->{'iso-ir-126'}
194			= $IANACharset->{'iso_8859-7'}
195			= $IANACharset->{'iso-8859-7'}
196			= $IANACharset->{'elot_928'}
197			= $IANACharset->{'ecma-118'}
198			= $IANACharset->{'greek'}
199			= $IANACharset->{'greek8'}
200			= $IANACharset->{'csisolatingreek'}
201			= {
202			iana_names => {
203			'iso_8859-7:1987' => PRIMARY_CHARSET_NAME,
204			'iso-ir-126' => REGISTERED_CHARSET_NAME,
205			'iso_8859-7' => REGISTERED_CHARSET_NAME,
206			'iso-8859-7' => PREFERRED_CHARSET_NAME,
207			'elot_928' => REGISTERED_CHARSET_NAME,
208			'ecma-118' => REGISTERED_CHARSET_NAME,
209			'greek' => REGISTERED_CHARSET_NAME,
210			'greek8' => REGISTERED_CHARSET_NAME,
211			'csisolatingreek' => REGISTERED_CHARSET_NAME,
212			},
213			is_html_ascii_superset => 1,
214			};
215
216			$Charset->{'iso-8859-8'}
217			= $IANACharset->{'iso_8859-8:1988'}
218			= $IANACharset->{'iso-ir-138'}
219			= $IANACharset->{'iso_8859-8'}
220			= $IANACharset->{'iso-8859-8'}
221			= $IANACharset->{'hebrew'}
222			= $IANACharset->{'csisolatinhebrew'}
223			= {
224			iana_names => {
225			'iso_8859-8:1988' => PRIMARY_CHARSET_NAME,
226			'iso-ir-138' => REGISTERED_CHARSET_NAME,
227			'iso_8859-8' => REGISTERED_CHARSET_NAME,
228			'iso-8859-8' => PREFERRED_CHARSET_NAME,
229			'hebrew' => REGISTERED_CHARSET_NAME,
230			'csisolatinhebrew' => REGISTERED_CHARSET_NAME,
231			},
232			is_html_ascii_superset => 1,
233			};
234
235			$Charset->{'iso-8859-9'}
236			= $IANACharset->{'iso_8859-9:1989'}
237			= $IANACharset->{'iso-ir-148'}
238			= $IANACharset->{'iso_8859-9'}
239			= $IANACharset->{'iso-8859-9'}
240			= $IANACharset->{'latin5'}
241			= $IANACharset->{'l5'}
242			= $IANACharset->{'csisolatin5'}
243			= {
244			iana_names => {
245			'iso_8859-9:1989' => PRIMARY_CHARSET_NAME,
246			'iso-ir-148' => REGISTERED_CHARSET_NAME,
247			'iso_8859-9' => REGISTERED_CHARSET_NAME,
248			'iso-8859-9' => PREFERRED_CHARSET_NAME,
249			'latin5' => REGISTERED_CHARSET_NAME,
250			'l5' => REGISTERED_CHARSET_NAME,
251			'csisolatin5' => REGISTERED_CHARSET_NAME,
252			},
253			is_html_ascii_superset => 1,
254			};
255
256			$Charset->{'iso-8859-10'}
257			= $IANACharset->{'iso-8859-10'}
258			= $IANACharset->{'iso-ir-157'}
259			= $IANACharset->{'l6'}
260			= $IANACharset->{'iso_8859-10:1992'}
261			= $IANACharset->{'csisolatin6'}
262			= $IANACharset->{'latin6'}
263			= {
264			iana_names => {
265			'iso-8859-10' => PRIMARY_CHARSET_NAME \| PREFERRED_CHARSET_NAME,
266			'iso-ir-157' => REGISTERED_CHARSET_NAME,
267			'l6' => REGISTERED_CHARSET_NAME,
268			'iso_8859-10:1992' => REGISTERED_CHARSET_NAME,
269			'csisolatin6' => REGISTERED_CHARSET_NAME,
270			'latin6' => REGISTERED_CHARSET_NAME,
271			},
272			is_html_ascii_superset => 1,
273			};
274
275			$Charset->{'iso_6937-2-add'}
276			= $IANACharset->{'iso_6937-2-add'}
277			= $IANACharset->{'iso-ir-142'}
278			= $IANACharset->{'csisotextcomm'}
279			= {
280			iana_names => {
281			'iso_6937-2-add' => PRIMARY_CHARSET_NAME,
282			'iso-ir-142' => REGISTERED_CHARSET_NAME,
283			'csisotextcomm' => REGISTERED_CHARSET_NAME,
284			},
285			is_html_ascii_superset => 1,
286			};
287
288			$Charset->{'jis_x0201'}
289			= $IANACharset->{'jis_x0201'}
290			= $IANACharset->{'x0201'}
291			= $IANACharset->{'cshalfwidthkatakana'}
292			= {
293			iana_names => {
294			'jis_x0201' => PRIMARY_CHARSET_NAME,
295			'x0201' => REGISTERED_CHARSET_NAME,
296			'cshalfwidthkatakana' => REGISTERED_CHARSET_NAME,
297			},
298			is_html_ascii_superset => 1,
299			};
300
301			$Charset->{'jis_encoding'}
302			= $IANACharset->{'jis_encoding'}
303			= $IANACharset->{'csjisencoding'}
304			= {
305			iana_names => {
306			'jis_encoding' => PRIMARY_CHARSET_NAME,
307			'csjisencoding' => REGISTERED_CHARSET_NAME,
308			},
309			## NOTE: What is this?
310			};
311	wakaba	1.1
312			$Charset->{'shift_jis'}
313			= $IANACharset->{'shift_jis'}
314			= $IANACharset->{'ms_kanji'}
315			= $IANACharset->{'csshiftjis'}
316			= {
317			iana_names => {
318			'shift_jis' => PREFERRED_CHARSET_NAME \| PRIMARY_CHARSET_NAME,
319			'ms_kanji' => REGISTERED_CHARSET_NAME,
320			'csshiftjis' => REGISTERED_CHARSET_NAME,
321			},
322	wakaba	1.2	mime_text_suitable => 1,
323	wakaba	1.1	};
324
325			$Charset->{'euc-jp'}
326			= $IANACharset->{'extended_unix_code_packed_format_for_japanese'}
327			= $IANACharset->{'cseucpkdfmtjapanese'}
328			= $IANACharset->{'euc-jp'}
329			= {
330			iana_names => {
331			'extended_unix_code_packed_format_for_japanese' => PRIMARY_CHARSET_NAME,
332			'cseucpkdfmtjapanese' => REGISTERED_CHARSET_NAME,
333			'euc-jp' => PREFERRED_CHARSET_NAME,
334			},
335			is_html_ascii_superset => 1,
336			};
337
338	wakaba	1.2	$Charset->{'extended_unix_code_fixed_width_for_japanese'}
339			= $IANACharset->{'extended_unix_code_fixed_width_for_japanese'}
340			= $IANACharset->{'cseucfixwidjapanese'}
341			= {
342			iana_names => {
343			'extended_unix_code_fixed_width_for_japanese' => PRIMARY_CHARSET_NAME,
344			'cseucfixwidjapanese' => REGISTERED_CHARSET_NAME,
345			},
346			};
347
348	wakaba	1.1	## TODO: ...
349
350	wakaba	1.2	$Charset->{'euc-kr'}
351			= $IANACharset->{'euc-kr'}
352			= $IANACharset->{'cseuckr'}
353			= {
354			iana_names => {
355			'euc-kr' => PRIMARY_CHARSET_NAME \| PREFERRED_CHARSET_NAME,
356			'cseuckr' => REGISTERED_CHARSET_NAME,
357			},
358			is_html_ascii_superset => 1,
359			};
360
361	wakaba	1.1	$Charset->{'iso-2022-jp'}
362			= $IANACharset->{'iso-2022-jp'}
363			= $IANACharset->{'csiso2022jp'}
364			= {
365			iana_names => {
366			'iso-2022-jp' => PREFERRED_CHARSET_NAME \| PRIMARY_CHARSET_NAME,
367			'csiso2022jp' => REGISTERED_CHARSET_NAME,
368			},
369	wakaba	1.2	mime_text_suitable => 1,
370			};
371
372			$Charset->{'iso-2022-jp-2'}
373			= $IANACharset->{'iso-2022-jp-2'}
374			= $IANACharset->{'csiso2022jp2'}
375			= {
376			iana_names => {
377			'iso-2022-jp-2' => PREFERRED_CHARSET_NAME \| PRIMARY_CHARSET_NAME,
378			'csiso2022jp2' => REGISTERED_CHARSET_NAME,
379			},
380			mime_text_suitable => 1,
381	wakaba	1.1	};
382
383			## TODO: ...
384
385			$Charset->{'utf-8'}
386			= $IANACharset->{'utf-8'}
387			= {
388			iana_names => {
389			'utf-8' => PRIMARY_CHARSET_NAME,
390			},
391			is_html_ascii_superset => 1,
392			};
393
394			## TODO: ...
395
396	wakaba	1.2	$Charset->{'gbk'}
397			= $IANACharset->{'gbk'}
398			= $IANACharset->{'cp936'}
399			= $IANACharset->{'ms936'}
400			= $IANACharset->{'windows-936'}
401			= {
402			iana_names => {
403			'gbk' => PRIMARY_CHARSET_NAME,
404			'cp936' => REGISTERED_CHARSET_NAME,
405			'ms936' => REGISTERED_CHARSET_NAME,
406			'windows-936' => REGISTERED_CHARSET_NAME,
407			},
408			iana_status => STATUS_COMMON \| STATUS_OBSOLETE,
409			mime_text_suitable => 1,
410			};
411
412			$Charset->{'gb18030'}
413			= $IANACharset->{'gb18030'}
414			= {
415			iana_names => {
416			'gb18030' => PRIMARY_CHARSET_NAME,
417			},
418			iana_status => STATUS_COMMON,
419			mime_text_suitable => 1,
420			};
421
422			## TODO: ...
423
424	wakaba	1.1	$Charset->{'utf-16be'}
425			= $IANACharset->{'utf-16be'}
426			= {
427			iana_names => {
428			'utf-16be' => PRIMARY_CHARSET_NAME,
429			},
430			};
431
432			$Charset->{'utf-16le'}
433			= $IANACharset->{'utf-16le'}
434			= {
435			iana_names => {
436			'utf-16le' => PRIMARY_CHARSET_NAME,
437			},
438			};
439
440			$Charset->{'utf-16'}
441			= $IANACharset->{'utf-16'}
442			= {
443			iana_names => {
444			'utf-16' => PRIMARY_CHARSET_NAME,
445			},
446			};
447
448			## TODO: ...
449
450	wakaba	1.2	$Charset->{'windows-31j'}
451			= $IANACharset->{'windows-31j'}
452			= $IANACharset->{'cswindows31j'}
453			= {
454			iana_names => {
455			'windows-31j' => PRIMARY_CHARSET_NAME,
456			'cswindows31j' => REGISTERED_CHARSET_NAME,
457			},
458			iana_status => STATUS_LIMITED_USE, # maybe
459			mime_text_suitable => 1,
460			};
461
462			$Charset->{'gb2312'}
463			= $IANACharset->{'gb2312'}
464			= $IANACharset->{'csgb2312'}
465			= {
466			iana_names => {
467			'gb2312' => PRIMARY_CHARSET_NAME \| PREFERRED_CHARSET_NAME,
468			'csgb2312' => REGISTERED_CHARSET_NAME,
469			},
470			is_html_ascii_superset => 1,
471			mime_text_suitable => 1,
472			};
473
474			$Charset->{'big5'}
475			= $IANACharset->{'big5'}
476			= $IANACharset->{'csbig5'}
477			= {
478			iana_names => {
479			'big5' => PRIMARY_CHARSET_NAME \| PREFERRED_CHARSET_NAME,
480			'csbig5' => REGISTERED_CHARSET_NAME,
481			},
482			mime_text_suitable => 1,
483			};
484
485			## TODO: ...
486
487			$Charset->{'big5-hkscs'}
488			= $IANACharset->{'big5-hkscs'}
489			= {
490			iana_names => {
491			'big5-hkscs' => PRIMARY_CHARSET_NAME,
492			},
493			mime_text_suitable => 1,
494			};
495
496			## TODO: ...
497
498	wakaba	1.1	$Charset->{'windows-1252'}
499			= $IANACharset->{'windows-1252'}
500			= {
501			iana_names => {
502			'windows-1252' => PRIMARY_CHARSET_NAME,
503			},
504			is_html_ascii_superset => 1,
505			};
506
507			## TODO: ...
508
509			sub is_syntactically_iana_charset_name ($) {
510			my $name = shift;
511			return $name =~ /\A[\x20-\x7E]{1,40}\z/;
512			} # is_suntactically_valid_iana_charset_name
513
514			1;
515	wakaba	1.2	## $Date: 2007/11/18 11:08:40 $
516	wakaba	1.1