X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/138b47f5f5ac5364dbfc15075e21b0990ab919d9..98699865db67c9772d2a56571e8ddcaee5a5e7bd:/charset.plp
diff --git a/charset.plp b/charset.plp
index 777e0ee..2dc6964 100644
--- a/charset.plp
+++ b/charset.plp
@@ -32,7 +32,7 @@ use Encode qw(decode resolve_alias);
# substr strings is twice as fast as splitting to an array)
my %ALIAS = (
# default => [qw(unicode utf-8 iso-8859-1 cp437 -cp1252- --iso-8859-15- -koi8-f)],
- default => [qw(unicode utf-8 iso-8859-1 -cp1252- --iso-8859-15- cp437 -cp850)],
+ default => [qw(unicode- utf-8 iso-8859-1 -cp1252- --iso-8859-15- cp437 -cp850)],
0 => [qw(cp437 cp863)],
1 => [qw(iso-8859-1 cp1252 MacRoman cp850)],
2 => [qw(iso-8859-2 cp1250 cp852 MacCentralEurRoman MacCroatian MacRumanian)],
@@ -55,7 +55,7 @@ my @request = map {
}
if ($row{set} = resolve_alias($input)) {
if ($row{set} eq 'Internal') {
- $row{table} = ' 'x640;
+ $row{table} = ' ' x ($endpoint < 255 ? 640 : 4096);
$row{set} = 'Unicode BMP';
}
elsif ($row{set} eq 'utf-8-strict') {
@@ -96,32 +96,38 @@ sub quote {
sub printcell_unicode {
my ($value) = @_;
- if ($value > 0x27F) {
+ if ($value > 0xFFF) {
print "\n".'| ?';
}
elsif ($value == 0) {
- print ' | control';
+ print ' | control';
}
elsif ($value == 2) {
- print ' | latin';
+ print ' | comn';
+ }
+ elsif ($value == 4) {
+ print ' | basic latin';
}
elsif ($value == 8) {
- print ' | control';
+ print ' | control';
}
elsif ($value == 10) {
- print ' | latin supplement';
+ print ' | comn';
+ }
+ elsif ($value == 12) {
+ print ' | latin1';
}
elsif ($value == 0x10) {
- print ' | latin ext-A';
+ print ' | latin extended-A';
}
elsif ($value == 0x18) {
- print ' | latin ext-B';
+ print ' | latin extended-B';
}
elsif ($value == 0x20) {
- print ' | latin ext-B';
+ print ' | latin ext-B';
}
elsif ($value == 0x25) {
- print ' | IPA';
+ print ' | IPA';
}
elsif ($value == 0x2B) {
print ' | spacing modifier';
@@ -130,199 +136,199 @@ sub printcell_unicode {
print ' | diacritics';
}
elsif ($value == 0x38) {
- print ' | greek';
+ print ' | greek';
}
elsif ($value == 0x40) {
- print ' | cyrillic';
+ print ' | cyrillic';
}
elsif ($value == 0x50) {
- print ' | cyrillic+';
+ print ' | cyrillic+';
}
elsif ($value == 0x53) {
- print ' | armenian';
+ print ' | armenian';
}
elsif ($value == 0x58) {
- print ' | hebrew';
+ print ' | hebrew';
}
elsif ($value == 0x60) {
- print ' | arabic';
+ print ' | arabic';
}
elsif ($value == 0x70) {
- print ' | syriac';
+ print ' | syriac';
}
elsif ($value == 0x75) {
- print ' | arabic+';
+ print ' | arabic+';
}
elsif ($value == 0x78) {
- print ' | thaana';
+ print ' | thaana';
}
elsif ($value == 0x7C) {
- print ' | nko';
+ print ' | n\'ko';
}
elsif ($value == 0x80) {
- print ' | samaritan';
+ print ' | samaritan';
}
elsif ($value == 0x84) {
- print ' | manda';
+ print ' | manda';
}
elsif ($value == 0x86) {
- print ' | reserved';
+ print ' | reserved';
}
elsif ($value == 0x90) {
- print ' | devanagari';
+ print ' | devanagari';
}
elsif ($value == 0x98) {
- print ' | bengali';
+ print ' | bengali';
}
elsif ($value == 0xA0) {
- print ' | gurmukhi';
+ print ' | gurmukhi';
}
elsif ($value == 0xA8) {
- print ' | gujarati';
+ print ' | gujarati';
}
elsif ($value == 0xB0) {
- print ' | oriya';
+ print ' | oriya';
}
elsif ($value == 0xB8) {
- print ' | tamil';
+ print ' | tamil';
}
elsif ($value == 0xC0) {
- print ' | telugu';
+ print ' | telugu';
}
elsif ($value == 0xC8) {
- print ' | kannada';
+ print ' | kannada';
}
elsif ($value == 0xD0) {
- print ' | malayalam';
+ print ' | malayalam';
}
elsif ($value == 0xD8) {
- print ' | sinhala';
+ print ' | sinhala';
}
elsif ($value == 0xE0) {
- print ' | thai';
+ print ' | thai';
}
elsif ($value == 0xE8) {
- print ' | lao';
+ print ' | lao';
}
elsif ($value == 0xF0) {
- print ' | tibetan';
+ print ' | tibetan';
}
elsif ($value == 0x100) {
- print ' | myanmar';
+ print ' | myanmar';
}
elsif ($value == 0x10A) {
- print ' | georgian';
+ print ' | georgian';
}
elsif ($value == 0x110) {
- print ' | hangeul jamo';
+ print ' | hangeul jamo';
}
elsif ($value == 0x120) {
- print ' | ethiopic';
+ print ' | ethiopic';
}
elsif ($value == 0x130) {
- print ' | ethiopic';
+ print ' | ethiopic';
}
elsif ($value == 0x138) {
- print ' | eth+';
+ print ' | eth+';
}
elsif ($value == 0x13A) {
- print ' | cherokee';
+ print ' | cherokee';
}
elsif ($value == 0x140) {
- print ' | unified canadian aboriginal syllabics';
+ print ' | unified canadian aboriginal syllabics';
}
elsif ($value == 0x160) {
- print ' | unified canadian syllabics';
+ print ' | unified canadian syllabics';
}
elsif ($value == 0x168) {
- print ' | ogham';
+ print ' | ogham';
}
elsif ($value == 0x16A) {
- print ' | runic';
+ print ' | runic';
}
elsif ($value == 0x170) {
- print ' | tagalog';
+ print ' | tagalog';
}
elsif ($value == 0x172) {
- print ' | hanun';
+ print ' | hanun';
}
elsif ($value == 0x174) {
- print ' | buhid';
+ print ' | buhid';
}
elsif ($value == 0x176) {
- print ' | tagb';
+ print ' | tagb';
}
elsif ($value == 0x178) {
- print ' | khmer';
+ print ' | khmer';
}
elsif ($value == 0x180) {
- print ' | mongolian';
+ print ' | mongolian';
}
elsif ($value == 0x18B) {
- print ' | canadian+';
+ print ' | canadian+';
}
elsif ($value == 0x190) {
- print ' | limbu';
+ print ' | limbu';
}
elsif ($value == 0x195) {
- print ' | tai le';
+ print ' | tai le';
}
elsif ($value == 0x198) {
- print ' | new tai lue';
+ print ' | new tai lue';
}
elsif ($value == 0x19E) {
- print ' | km';
+ print ' | km';
}
elsif ($value == 0x1A0) {
- print ' | lontara';
+ print ' | lontara';
}
elsif ($value == 0x1A2) {
- print ' | tai tham';
+ print ' | tai tham';
}
elsif ($value == 0x1AB) {
- print ' | reserved';
+ print ' | reserved';
}
elsif ($value == 0x1B0) {
- print ' | balinese';
+ print ' | balinese';
}
elsif ($value == 0x1B8) {
- print ' | sundanese';
+ print ' | sundanese';
}
elsif ($value == 0x1BC) {
- print ' | batak';
+ print ' | batak';
}
elsif ($value == 0x1C0) {
- print ' | lepcha';
+ print ' | lepcha';
}
elsif ($value == 0x1C5) {
- print ' | ol chiki';
+ print ' | ol chiki';
}
elsif ($value == 0x1C8) {
- print ' | reserved';
+ print ' | reserved';
}
elsif ($value == 0x1CD) {
- print ' | vedic';
+ print ' | vedic';
}
elsif ($value == 0x1D0) {
- print ' | phonetic';
+ print ' | phonetic';
}
elsif ($value == 0x1D8) {
- print ' | phonetic+';
+ print ' | phonetic+';
}
elsif ($value == 0x1DC) {
print ' | combining';
}
elsif ($value == 0x1E0) {
- print ' | latin extended additional';
+ print ' | latin extended additional';
}
elsif ($value == 0x1F0) {
- print ' | greek+';
+ print ' | greek+';
}
elsif ($value == 0x200) {
- print ' | general punctuation';
+ print ' | general punctuation';
}
elsif ($value == 0x207) {
- print ' | su[bp]script';
+ print ' | suþscript'; # suth now means "sub and/or sup"
}
elsif ($value == 0x20A) {
print ' | currency';
@@ -375,23 +381,260 @@ sub printcell_unicode {
elsif ($value == 0x27F) {
print ' | arr';
}
+ elsif ($value == 0x280) {
+ print ' | braille';
+ }
+ elsif ($value == 0x290) {
+ print ' | supplemental arrows-B';
+ }
+ elsif ($value == 0x298) {
+ print ' | mathematical symbols-B';
+ }
+ elsif ($value == 0x2A0) {
+ print ' | supplemental mathematical operators';
+ }
+ elsif ($value == 0x2B0) {
+ print ' | miscellaneous symbols and arrows';
+ }
+ elsif ($value == 0x2C0) {
+ print ' | glagolitic';
+ }
+ elsif ($value == 0x2C6) {
+ print ' | latin-C';
+ }
+ elsif ($value == 0x2C8) {
+ print ' | coptic';
+ }
+ elsif ($value == 0x2D0) {
+ print ' | georgian+';
+ }
+ elsif ($value == 0x2D3) {
+ print ' | tifinagh'; #TODO: proto-canaanite
+ }
+ elsif ($value == 0x2D8) {
+ print ' | ethiopic+';
+ }
+ elsif ($value == 0x2DE) {
+ print ' | cyrl-A';
+ }
+ elsif ($value == 0x2E0) {
+ print ' | punctuation+';
+ }
+ elsif ($value == 0x2E8) {
+ print ' | cjk radicals';
+ }
+ elsif ($value == 0x2F0) {
+ print ' | kangxi radicals';
+ }
+ elsif ($value == 0x2FE) {
+ print ' | ';
+ }
+ elsif ($value == 0x2FF) {
+ print ' | idc';
+ }
+ elsif ($value == 0x300) {
+ print ' | cjk misc';
+ }
+ elsif ($value == 0x304) {
+ print ' | hiragana';
+ }
+ elsif ($value == 0x30A) {
+ print ' | katakana';
+ }
+ elsif ($value == 0x310) {
+ print ' | bopomofo';
+ }
+ elsif ($value == 0x313) {
+ print ' | hangeul compat';
+ }
+ elsif ($value == 0x319) {
+ print ' | kbn';
+ }
+ elsif ($value == 0x31A) {
+ print ' | bpmf';
+ }
+ elsif ($value == 0x31C) {
+ print ' | strokes';
+ }
+ elsif ($value == 0x31F) {
+ print ' | k+';
+ }
+ elsif ($value == 0x320) {
+ print ' | enclosed cjk characters';
+ }
+ elsif ($value == 0x330) {
+ print ' | cjk compatibility';
+ }
+ elsif ($value == 0x340) {
+ print ' | cjk unified ideographs extension A';
+ }
+ elsif ($value == 0x4D0) {
+ print ' | cjk unified ideographs extension A';
+ }
+ elsif ($value == 0x4DC) {
+ print ' | hexagrams';
+ }
+ elsif ($value == 0x4E0) {
+ print ' | cjk unified ideographs';
+ }
+ elsif ($value == 0xA00) {
+ print ' | yi';
+ }
+ elsif ($value == 0xA40) {
+ print ' | yi';
+ }
+ elsif ($value == 0xA49) {
+ print ' | yi radicals';
+ }
+ elsif ($value == 0xA4D) {
+ print ' | lisu';
+ }
+ elsif ($value == 0xA50) {
+ print ' | vai';
+ }
+ elsif ($value == 0xA60) {
+ print ' | vai';
+ }
+ elsif ($value == 0xA64) {
+ print ' | cyrillic extended-B';
+ }
+ elsif ($value == 0xA6A) {
+ print ' | bamum';
+ }
+ elsif ($value == 0xA70) {
+ print ' | tones';
+ }
+ elsif ($value == 0xA72) {
+ print ' | latin extended-D';
+ }
+ elsif ($value == 0xA80) {
+ print ' | sylheti';
+ }
+ elsif ($value == 0xA83) {
+ print ' | in';
+ }
+ elsif ($value == 0xA84) {
+ print ' | phags-pa';
+ }
+ elsif ($value == 0xA88) {
+ print ' | saurashtra';
+ }
+ elsif ($value == 0xA8E) {
+ print ' | deva+';
+ }
+ elsif ($value == 0xA90) {
+ print ' | kayah li';
+ }
+ elsif ($value == 0xA93) {
+ print ' | rejang';
+ }
+ elsif ($value == 0xA96) {
+ print ' | jamo-A';
+ }
+ elsif ($value == 0xA98) {
+ print ' | javanese';
+ }
+ elsif ($value == 0xA9E) {
+ print ' | res';
+ }
+ elsif ($value == 0xAA0) {
+ print ' | cham';
+ }
+ elsif ($value == 0xAA6) {
+ print ' | mym-A';
+ }
+ elsif ($value == 0xAA8) {
+ print ' | tai viet';
+ }
+ elsif ($value == 0xAAE) {
+ print ' | mtei+';
+ }
+ elsif ($value == 0xAB0) {
+ print ' | reserved';
+ }
+ elsif ($value == 0xABC) {
+ print ' | manipuri';
+ }
+ elsif ($value == 0xAC0) {
+ print ' | hangeul syllables';
+ }
+ elsif ($value == 0xD70) {
+ print ' | hangeul syllables';
+ }
+ elsif ($value == 0xD7B) {
+ print ' | haungeul jamo-B';
+ }
+ elsif ($value == 0xD80) {
+ print ' | high surrogates';
+ }
+ elsif ($value == 0xDC0) {
+ print ' | low surrogates';
+ }
+ elsif ($value == 0xE00) {
+ print ' | private use';
+ }
+ elsif ($value == 0xF90) {
+ print ' | cjk compatibility ideographs';
+ }
+ elsif ($value == 0xFB0) {
+ print ' | presentation';
+ }
+ elsif ($value == 0xFB5) {
+ print ' | ';
+ }
+ elsif ($value == 0xFC0) {
+ print ' | arabic presentation forms A';
+ }
+ elsif ($value == 0xFD0) {
+ print ' | ';
+ }
+ elsif ($value == 0xFDD) {
+ print ' | ?';
+ }
+ elsif ($value == 0xFDF) {
+ print ' | ';
+ }
+ elsif ($value == 0xFE0) {
+ print ' | var';
+ }
+ elsif ($value == 0xFE1) {
+ print ' | ver';
+ }
+ elsif ($value == 0xFE2) {
+ print ' | ½';
+ }
+ elsif ($value == 0xFE3) {
+ print ' | comp';
+ }
+ elsif ($value == 0xFE5) {
+ print ' | small';
+ }
+ elsif ($value == 0xFE7) {
+ print ' | arabic presentation B';
+ }
+ elsif ($value == 0xFF0) {
+ print ' | halfwidth & fullwidth forms';
+ }
+ elsif ($value == 0xFFF) {
+ print ' | sp';
+ }
}
sub printcell_utf8 {
my ($value) = @_;
if ($value <= 0x7F) {
print ' | Single byte ASCII'
+ ' title="U+0000 â U+007F">single byte ASCII'
if $value == 0;
}
elsif ($value <= 0xBF) {
print ' | Multi-byte continuation'
+ '>multi-byte continuation'
if $value == 0x80;
}
elsif ($value <= 0xC1) {
print ' | (Overl.)'
+ ' title="U+0000 â U+007F">(overl.)'
if $value == 0xC0;
}
elsif ($value <= 0xDF) {
@@ -414,7 +657,7 @@ sub printcell_utf8 {
}
elsif ($value <= 0xF7) {
print ' | (Overflow)'
+ ' title="U+11·0000 â U+1FF·FFFF">(overflow)'
if $value == 0xF5;
}
elsif ($value <= 0xFB) {
@@ -428,7 +671,7 @@ sub printcell_utf8 {
if $value == 0xFC;
}
elsif ($value <= 0xFF) {
- print ' | Invalid'
+ print ' | invalid'
if $value == 0xFE;
}
else {
|