mirror of
git://git.sv.gnu.org/emacs.git
synced 2025-12-27 07:41:28 -08:00
New file.
This commit is contained in:
Kenichi Handa
parent
c90e1d1c1b
commit
c341fc3dae
3 changed files with 22111 additions and 0 deletions
206
etc/charsets/gb18030-bmp.map
Normal file
206
etc/charsets/gb18030-bmp.map
Normal file
|
|
@ -0,0 +1,206 @@
|
|||
0x81308130-0x81308435 U+0000-U+0023
|
||||
0x81308436-0x81308437 U+00A5-U+00A6
|
||||
0x81308438-0x81308534 U+00A9-U+00AF
|
||||
0x81308535-0x81308539 U+00B2-U+00B6
|
||||
0x81308630-0x81308930 U+00B8-U+00D6
|
||||
0x81308931-0x81308938 U+00D8-U+00DF
|
||||
0x81308939-0x81308A34 U+00E2-U+00E7
|
||||
0x81308A35 U+00EB
|
||||
0x81308A36-0x81308A39 U+00EE-U+00F1
|
||||
0x81308B30-0x81308B32 U+00F4-U+00F6
|
||||
0x81308B33 U+00F8
|
||||
0x81308B34 U+00FB
|
||||
0x81308B35-0x81308B38 U+00FD-U+0100
|
||||
0x81308B39-0x81308D35 U+0102-U+0112
|
||||
0x81308D36-0x81308E32 U+0114-U+011A
|
||||
0x81308E33-0x81308F37 U+011C-U+012A
|
||||
0x81308F38-0x81309231 U+012C-U+0143
|
||||
0x81309232-0x81309234 U+0145-U+0147
|
||||
0x81309235-0x81309238 U+0149-U+014C
|
||||
0x81309239-0x81309537 U+014E-U+016A
|
||||
0x81309538-0x81309F35 U+016C-U+01CD
|
||||
0x81309F36 U+01CF
|
||||
0x81309F37 U+01D1
|
||||
0x81309F38 U+01D3
|
||||
0x81309F39 U+01D5
|
||||
0x8130A030 U+01D7
|
||||
0x8130A031 U+01D9
|
||||
0x8130A032 U+01DB
|
||||
0x8130A033-0x8130A330 U+01DD-U+01F8
|
||||
0x8130A331-0x8130AB37 U+01FA-U+0250
|
||||
0x8130AB38-0x8130AD32 U+0252-U+0260
|
||||
0x8130AD33-0x8130B733 U+0262-U+02C6
|
||||
0x8130B734 U+02C8
|
||||
0x8130B735-0x8130B837 U+02CC-U+02D8
|
||||
0x8130B838-0x8130CB30 U+02DA-U+0390
|
||||
0x8130CB31 U+03A2
|
||||
0x8130CB32-0x8130CB38 U+03AA-U+03B0
|
||||
0x8130CB39 U+03C2
|
||||
0x8130CC30-0x8130D134 U+03CA-U+0400
|
||||
0x8130D135-0x8130D238 U+0402-U+040F
|
||||
0x8130D239 U+0450
|
||||
0x8130D330-0x8136A531 U+0452-U+200F
|
||||
0x8136A532-0x8136A533 U+2011-U+2012
|
||||
0x8136A534 U+2017
|
||||
0x8136A535-0x8136A536 U+201A-U+201B
|
||||
0x8136A537-0x8136A633 U+201E-U+2024
|
||||
0x8136A634-0x8136A732 U+2027-U+202F
|
||||
0x8136A733 U+2031
|
||||
0x8136A734 U+2034
|
||||
0x8136A735-0x8136A739 U+2036-U+203A
|
||||
0x8136A830-0x8136B331 U+203C-U+20AB
|
||||
0x8136B332-0x8136BB37 U+20AD-U+2102
|
||||
0x8136BB38 U+2104
|
||||
0x8136BB39-0x8136BC31 U+2106-U+2108
|
||||
0x8136BC32-0x8136BD33 U+210A-U+2115
|
||||
0x8136BD34-0x8136BE33 U+2117-U+2120
|
||||
0x8136BE34-0x8136C435 U+2122-U+215F
|
||||
0x8136C436-0x8136C439 U+216C-U+216F
|
||||
0x8136C530-0x8136C731 U+217A-U+218F
|
||||
0x8136C732-0x8136C733 U+2194-U+2195
|
||||
0x8136C734-0x8136D233 U+219A-U+2207
|
||||
0x8136D234-0x8136D239 U+2209-U+220E
|
||||
0x8136D330 U+2210
|
||||
0x8136D331-0x8136D333 U+2212-U+2214
|
||||
0x8136D334-0x8136D337 U+2216-U+2219
|
||||
0x8136D338-0x8136D339 U+221B-U+221C
|
||||
0x8136D430-0x8136D431 U+2221-U+2222
|
||||
0x8136D432 U+2224
|
||||
0x8136D433 U+2226
|
||||
0x8136D434-0x8136D435 U+222C-U+222D
|
||||
0x8136D436-0x8136D530 U+222F-U+2233
|
||||
0x8136D531-0x8136D535 U+2238-U+223C
|
||||
0x8136D536-0x8136D635 U+223E-U+2247
|
||||
0x8136D636-0x8136D638 U+2249-U+224B
|
||||
0x8136D639-0x8136D733 U+224D-U+2251
|
||||
0x8136D734-0x8136D836 U+2253-U+225F
|
||||
0x8136D837-0x8136D838 U+2262-U+2263
|
||||
0x8136D839-0x8136D934 U+2268-U+226D
|
||||
0x8136D935-0x8136DD31 U+2270-U+2294
|
||||
0x8136DD32-0x8136DD34 U+2296-U+2298
|
||||
0x8136DD35-0x8136DE35 U+229A-U+22A4
|
||||
0x8136DE36-0x8136E130 U+22A6-U+22BE
|
||||
0x8136E131-0x8136E932 U+22C0-U+2311
|
||||
0x8136E933-0x81378C35 U+2313-U+245F
|
||||
0x81378C36-0x81378D35 U+246A-U+2473
|
||||
0x81378D36-0x81379735 U+249C-U+24FF
|
||||
0x81379736-0x81379739 U+254C-U+254F
|
||||
0x81379830-0x81379932 U+2574-U+2580
|
||||
0x81379933-0x81379935 U+2590-U+2592
|
||||
0x81379936-0x81379A35 U+2596-U+259F
|
||||
0x81379A36-0x81379C31 U+25A2-U+25B1
|
||||
0x81379C32-0x81379C39 U+25B4-U+25BB
|
||||
0x81379D30-0x81379D37 U+25BE-U+25C5
|
||||
0x81379D38-0x81379E30 U+25C8-U+25CA
|
||||
0x81379E31-0x81379E32 U+25CC-U+25CD
|
||||
0x81379E33-0x8137A030 U+25D0-U+25E1
|
||||
0x8137A031-0x8137A331 U+25E6-U+2604
|
||||
0x8137A332-0x8137A333 U+2607-U+2608
|
||||
0x8137A334-0x8137A837 U+260A-U+263F
|
||||
0x8137A838 U+2641
|
||||
0x8137A839-0x8138FD38 U+2643-U+2E80
|
||||
0x8138FD39-0x8138FE30 U+2E82-U+2E83
|
||||
0x8138FE31-0x8138FE33 U+2E85-U+2E87
|
||||
0x8138FE34-0x8138FE35 U+2E89-U+2E8A
|
||||
0x8138FE36-0x81398135 U+2E8D-U+2E96
|
||||
0x81398136-0x81398330 U+2E98-U+2EA6
|
||||
0x81398331-0x81398332 U+2EA8-U+2EA9
|
||||
0x81398333-0x81398335 U+2EAB-U+2EAD
|
||||
0x81398336-0x81398339 U+2EAF-U+2EB2
|
||||
0x81398430-0x81398431 U+2EB4-U+2EB5
|
||||
0x81398432-0x81398434 U+2EB8-U+2EBA
|
||||
0x81398435-0x81398538 U+2EBC-U+2EC9
|
||||
0x81398539-0x8139A331 U+2ECB-U+2FEF
|
||||
0x8139A332-0x8139A335 U+2FFC-U+2FFF
|
||||
0x8139A336 U+3004
|
||||
0x8139A337-0x8139A431 U+3018-U+301C
|
||||
0x8139A432-0x8139A433 U+301F-U+3020
|
||||
0x8139A434-0x8139A633 U+302A-U+303D
|
||||
0x8139A634-0x8139A635 U+303F-U+3040
|
||||
0x8139A636-0x8139A732 U+3094-U+309A
|
||||
0x8139A733-0x8139A734 U+309F-U+30A0
|
||||
0x8139A735-0x8139A739 U+30F7-U+30FB
|
||||
0x8139A830-0x8139A835 U+30FF-U+3104
|
||||
0x8139A836-0x8139C131 U+312A-U+321F
|
||||
0x8139C132-0x8139C138 U+322A-U+3230
|
||||
0x8139C139-0x8139CD31 U+3232-U+32A2
|
||||
0x8139CD32-0x8139E435 U+32A4-U+338D
|
||||
0x8139E436-0x8139E537 U+3390-U+339B
|
||||
0x8139E538-0x8139E539 U+339F-U+33A0
|
||||
0x8139E630-0x8139E933 U+33A2-U+33C3
|
||||
0x8139E934-0x8139EA32 U+33C5-U+33CD
|
||||
0x8139EA33-0x8139EA34 U+33CF-U+33D0
|
||||
0x8139EA35-0x8139EA36 U+33D3-U+33D4
|
||||
0x8139EA37-0x8139F539 U+33D6-U+3446
|
||||
0x8139F630-0x8139FA32 U+3448-U+3472
|
||||
0x8139FA33-0x82309A30 U+3474-U+359D
|
||||
0x82309A31-0x8230A531 U+359F-U+360D
|
||||
0x8230A532-0x8230A632 U+360F-U+3619
|
||||
0x8230A633-0x8230F237 U+361B-U+3917
|
||||
0x8230F238-0x8230FB32 U+3919-U+396D
|
||||
0x8230FB33-0x82318638 U+396F-U+39CE
|
||||
0x82318639-0x82318832 U+39D1-U+39DE
|
||||
0x82318833-0x82319639 U+39E0-U+3A72
|
||||
0x82319730-0x8231AC37 U+3A74-U+3B4D
|
||||
0x8231AC38-0x8231C934 U+3B4F-U+3C6D
|
||||
0x8231C935-0x8231D437 U+3C6F-U+3CDF
|
||||
0x8231D438-0x8232AF32 U+3CE1-U+4055
|
||||
0x8232AF33-0x8232C936 U+4057-U+415E
|
||||
0x8232C937-0x8232F837 U+4160-U+4336
|
||||
0x8232F838-0x82338633 U+4338-U+43AB
|
||||
0x82338634-0x82338637 U+43AD-U+43B0
|
||||
0x82338638-0x82338B30 U+43B2-U+43DC
|
||||
0x82338B31-0x8233A338 U+43DE-U+44D5
|
||||
0x8233A339-0x8233C931 U+44D7-U+464B
|
||||
0x8233C932-0x8233CB31 U+464D-U+4660
|
||||
0x8233CB32-0x8233DE34 U+4662-U+4722
|
||||
0x8233DE35-0x8233DE39 U+4724-U+4728
|
||||
0x8233DF30-0x8233E731 U+472A-U+477B
|
||||
0x8233E732-0x8233E837 U+477D-U+478C
|
||||
0x8233E838-0x82349638 U+478E-U+4946
|
||||
0x82349639-0x82349B38 U+4948-U+4979
|
||||
0x82349B39-0x82349C30 U+497B-U+497C
|
||||
0x82349C31-0x82349C34 U+497E-U+4981
|
||||
0x82349C35 U+4984
|
||||
0x82349C36-0x82349E35 U+4987-U+499A
|
||||
0x82349E36-0x82349E38 U+499C-U+499E
|
||||
0x82349E39-0x8234A130 U+49A0-U+49B5
|
||||
0x8234A131-0x8234E733 U+49B8-U+4C76
|
||||
0x8234E734-0x8234EB32 U+4C78-U+4C9E
|
||||
0x8234EB33-0x8234F633 U+4CA4-U+4D12
|
||||
0x8234F634-0x82358731 U+4D1A-U+4DAD
|
||||
0x82358732-0x82358F32 U+4DAF-U+4DFF
|
||||
0x82358F33-0x8336C738 U+9FA6-U+D7FF
|
||||
0x8336C739 U+E76C
|
||||
0x8336C830 U+E7C8
|
||||
0x8336C831-0x8336C933 U+E7E7-U+E7F3
|
||||
0x8336C934 U+E815
|
||||
0x8336C935-0x8336C939 U+E819-U+E81D
|
||||
0x8336CA30-0x8336CA36 U+E81F-U+E825
|
||||
0x8336CA37-0x8336CB30 U+E827-U+E82A
|
||||
0x8336CB31-0x8336CB34 U+E82D-U+E830
|
||||
0x8336CB35-0x8336CC32 U+E833-U+E83A
|
||||
0x8336CC33-0x8336CC39 U+E83C-U+E842
|
||||
0x8336CD30-0x8336CE35 U+E844-U+E853
|
||||
0x8336CE36-0x8336CF39 U+E856-U+E863
|
||||
0x8336D030-0x84308534 U+E865-U+F92B
|
||||
0x84308535-0x84308D30 U+F92D-U+F978
|
||||
0x84308D31-0x84308F37 U+F97A-U+F994
|
||||
0x84308F38-0x84309738 U+F996-U+F9E6
|
||||
0x84309739-0x84309837 U+F9E8-U+F9F0
|
||||
0x84309838-0x84309B33 U+F9F2-U+FA0B
|
||||
0x84309B34 U+FA10
|
||||
0x84309B35 U+FA12
|
||||
0x84309B36-0x84309B38 U+FA15-U+FA17
|
||||
0x84309B39-0x84309C34 U+FA19-U+FA1E
|
||||
0x84309C35 U+FA22
|
||||
0x84309C36-0x84309C37 U+FA25-U+FA26
|
||||
0x84309C38-0x84318537 U+FA2A-U+FE2F
|
||||
0x84318538 U+FE32
|
||||
0x84318539-0x84318632 U+FE45-U+FE48
|
||||
0x84318633 U+FE53
|
||||
0x84318634 U+FE58
|
||||
0x84318635 U+FE67
|
||||
0x84318636-0x84319534 U+FE6C-U+FF00
|
||||
0x84319535-0x8431A233 U+FF5F-U+FFDF
|
||||
0x8431A234-0x8431A439 U+FFE6-U+FFFF
|
||||
114
etc/charsets/gb18030.awk
Normal file
114
etc/charsets/gb18030.awk
Normal file
|
|
@ -0,0 +1,114 @@
|
|||
BEGIN {
|
||||
tohex["A"] = 10;
|
||||
tohex["B"] = 11;
|
||||
tohex["C"] = 12;
|
||||
tohex["D"] = 13;
|
||||
tohex["E"] = 14;
|
||||
tohex["F"] = 15;
|
||||
tohex["a"] = 10;
|
||||
tohex["b"] = 11;
|
||||
tohex["c"] = 12;
|
||||
tohex["d"] = 13;
|
||||
tohex["e"] = 14;
|
||||
tohex["f"] = 15;
|
||||
}
|
||||
|
||||
function decode_hex(str) {
|
||||
n = 0;
|
||||
len = length(str);
|
||||
for (i = 1; i <= len; i++)
|
||||
{
|
||||
c = substr (str, i, 1);
|
||||
if (c >= "0" && c <= "9")
|
||||
n = n * 16 + (c - "0");
|
||||
else
|
||||
n = n * 16 + tohex[c];
|
||||
}
|
||||
return n;
|
||||
}
|
||||
|
||||
function gb_to_index(b0,b1,b2,b3) {
|
||||
return ((((b0 - 129) * 10 + (b1 - 48)) * 126 + (b2 - 129)) * 10 + b3 - 48);
|
||||
}
|
||||
|
||||
function index_to_gb(idx) {
|
||||
b3 = (idx % 10) + 48;
|
||||
idx /= 10;
|
||||
b2 = (idx % 126) + 129;
|
||||
idx /= 126;
|
||||
b1 = (idx % 10) + 48;
|
||||
b0 = (idx / 10) + 129;
|
||||
return sprintf("%02X%02X%02X%02X", b0, b1, b2, b3);
|
||||
}
|
||||
|
||||
function decode_gb(str) {
|
||||
b0 = decode_hex(substr(str, 3, 2));
|
||||
b1 = decode_hex(substr(str, 7, 2));
|
||||
b2 = decode_hex(substr(str, 11, 2));
|
||||
b3 = decode_hex(substr(str, 15, 2));
|
||||
return gb_to_index(b0, b1, b2, b3);
|
||||
}
|
||||
|
||||
function printline(from, to) {
|
||||
fromgb = index_to_gb(from);
|
||||
fromuni = gbtable[from];
|
||||
if (from == to)
|
||||
printf ("0x%s U+%04X\n", fromgb, fromuni);
|
||||
else
|
||||
printf ("0x%s-0x%s U+%04X-U+%04X\n", fromgb, index_to_gb(to),
|
||||
fromuni, fromuni + (to - from));
|
||||
}
|
||||
|
||||
/^<U[0-9A-F][0-9A-F][0-9A-F][0-9A-F]>/ {
|
||||
unicode = decode_hex(substr($1, 3, 4));
|
||||
if ($2 ~ /\\x8[1-4]\\x3[0-9]\\x[8-9A-F][0-9A-F]\\x3[0-9]/)
|
||||
unitable[unicode] = decode_gb($2);
|
||||
else
|
||||
unitable[unicode] = -1;
|
||||
}
|
||||
|
||||
END {
|
||||
lastgb = 0;
|
||||
surrogate_min = decode_hex("D800")
|
||||
surrogate_max = decode_hex("DFFF")
|
||||
for (i = 129; i < 65536; i++)
|
||||
{
|
||||
if (unitable[i] == 0 && (i < surrogate_min || i > surrogate_max))
|
||||
{
|
||||
lastgb++;
|
||||
gbtable[lastgb] = i;
|
||||
unitable[i] = lastgb;
|
||||
}
|
||||
else if (unitable[i] > 0)
|
||||
{
|
||||
lastgb = unitable[i];
|
||||
gbtable[lastgb] = i;
|
||||
}
|
||||
}
|
||||
|
||||
print "You blew it again!" > "/dev/stderr"
|
||||
fromgb = lastgb = unitable[128];
|
||||
for (i = 129; i < 65536; i++)
|
||||
{
|
||||
if (unitable[i] > 0)
|
||||
{
|
||||
if (lastgb + 1 == unitable[i])
|
||||
{
|
||||
lastgb++;
|
||||
}
|
||||
else
|
||||
{
|
||||
if (lastgb >= 0)
|
||||
printline(fromgb, lastgb);
|
||||
fromgb = lastgb = unitable[i];
|
||||
}
|
||||
}
|
||||
else # i.e. (unitable[i] < 0)
|
||||
{
|
||||
if (lastgb >= 0)
|
||||
printline(fromgb, lastgb);
|
||||
lastgb = -1;
|
||||
}
|
||||
}
|
||||
printline(fromgb, unitable[65535]);
|
||||
}
|
||||
21791
etc/charsets/gbk.map
Normal file
21791
etc/charsets/gbk.map
Normal file
File diff suppressed because it is too large
Load diff
Loading…
Add table
Add a link
Reference in a new issue