Removed the decoding_map from the codecs where this is possible.

Replaced the tis_620, cp1140 and koi8_u codecs with new ones
based on custom mapping files.

1 files changed, 513 insertions, 603 deletions

diff --git a/Lib/encodings/cp1256.py b/Lib/encodings/cp1256.py
index a913f9c57cf..17fa9302ef2 100644
--- a/Lib/encodings/cp1256.py
+++ b/Lib/encodings/cp1256.py
@@ -1,4 +1,4 @@
-""" Python Character Mapping Codec generated from 'VENDORS/MICSFT/WINDOWS/CP1256.TXT' with gencodec.py.
+""" Python Character Mapping Codec generated from 'MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1256.TXT' with gencodec.py.
 
 """#"
 
@@ -28,615 +28,525 @@ def getregentry():
 
     return (Codec().encode,Codec().decode,StreamReader,StreamWriter)
 
-### Decoding Map
-
-decoding_map = codecs.make_identity_dict(range(256))
-decoding_map.update({
-    0x0080: 0x20ac,	#  EURO SIGN
-    0x0081: 0x067e,	#  ARABIC LETTER PEH
-    0x0082: 0x201a,	#  SINGLE LOW-9 QUOTATION MARK
-    0x0083: 0x0192,	#  LATIN SMALL LETTER F WITH HOOK
-    0x0084: 0x201e,	#  DOUBLE LOW-9 QUOTATION MARK
-    0x0085: 0x2026,	#  HORIZONTAL ELLIPSIS
-    0x0086: 0x2020,	#  DAGGER
-    0x0087: 0x2021,	#  DOUBLE DAGGER
-    0x0088: 0x02c6,	#  MODIFIER LETTER CIRCUMFLEX ACCENT
-    0x0089: 0x2030,	#  PER MILLE SIGN
-    0x008a: 0x0679,	#  ARABIC LETTER TTEH
-    0x008b: 0x2039,	#  SINGLE LEFT-POINTING ANGLE QUOTATION MARK
-    0x008c: 0x0152,	#  LATIN CAPITAL LIGATURE OE
-    0x008d: 0x0686,	#  ARABIC LETTER TCHEH
-    0x008e: 0x0698,	#  ARABIC LETTER JEH
-    0x008f: 0x0688,	#  ARABIC LETTER DDAL
-    0x0090: 0x06af,	#  ARABIC LETTER GAF
-    0x0091: 0x2018,	#  LEFT SINGLE QUOTATION MARK
-    0x0092: 0x2019,	#  RIGHT SINGLE QUOTATION MARK
-    0x0093: 0x201c,	#  LEFT DOUBLE QUOTATION MARK
-    0x0094: 0x201d,	#  RIGHT DOUBLE QUOTATION MARK
-    0x0095: 0x2022,	#  BULLET
-    0x0096: 0x2013,	#  EN DASH
-    0x0097: 0x2014,	#  EM DASH
-    0x0098: 0x06a9,	#  ARABIC LETTER KEHEH
-    0x0099: 0x2122,	#  TRADE MARK SIGN
-    0x009a: 0x0691,	#  ARABIC LETTER RREH
-    0x009b: 0x203a,	#  SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
-    0x009c: 0x0153,	#  LATIN SMALL LIGATURE OE
-    0x009d: 0x200c,	#  ZERO WIDTH NON-JOINER
-    0x009e: 0x200d,	#  ZERO WIDTH JOINER
-    0x009f: 0x06ba,	#  ARABIC LETTER NOON GHUNNA
-    0x00a1: 0x060c,	#  ARABIC COMMA
-    0x00aa: 0x06be,	#  ARABIC LETTER HEH DOACHASHMEE
-    0x00ba: 0x061b,	#  ARABIC SEMICOLON
-    0x00bf: 0x061f,	#  ARABIC QUESTION MARK
-    0x00c0: 0x06c1,	#  ARABIC LETTER HEH GOAL
-    0x00c1: 0x0621,	#  ARABIC LETTER HAMZA
-    0x00c2: 0x0622,	#  ARABIC LETTER ALEF WITH MADDA ABOVE
-    0x00c3: 0x0623,	#  ARABIC LETTER ALEF WITH HAMZA ABOVE
-    0x00c4: 0x0624,	#  ARABIC LETTER WAW WITH HAMZA ABOVE
-    0x00c5: 0x0625,	#  ARABIC LETTER ALEF WITH HAMZA BELOW
-    0x00c6: 0x0626,	#  ARABIC LETTER YEH WITH HAMZA ABOVE
-    0x00c7: 0x0627,	#  ARABIC LETTER ALEF
-    0x00c8: 0x0628,	#  ARABIC LETTER BEH
-    0x00c9: 0x0629,	#  ARABIC LETTER TEH MARBUTA
-    0x00ca: 0x062a,	#  ARABIC LETTER TEH
-    0x00cb: 0x062b,	#  ARABIC LETTER THEH
-    0x00cc: 0x062c,	#  ARABIC LETTER JEEM
-    0x00cd: 0x062d,	#  ARABIC LETTER HAH
-    0x00ce: 0x062e,	#  ARABIC LETTER KHAH
-    0x00cf: 0x062f,	#  ARABIC LETTER DAL
-    0x00d0: 0x0630,	#  ARABIC LETTER THAL
-    0x00d1: 0x0631,	#  ARABIC LETTER REH
-    0x00d2: 0x0632,	#  ARABIC LETTER ZAIN
-    0x00d3: 0x0633,	#  ARABIC LETTER SEEN
-    0x00d4: 0x0634,	#  ARABIC LETTER SHEEN
-    0x00d5: 0x0635,	#  ARABIC LETTER SAD
-    0x00d6: 0x0636,	#  ARABIC LETTER DAD
-    0x00d8: 0x0637,	#  ARABIC LETTER TAH
-    0x00d9: 0x0638,	#  ARABIC LETTER ZAH
-    0x00da: 0x0639,	#  ARABIC LETTER AIN
-    0x00db: 0x063a,	#  ARABIC LETTER GHAIN
-    0x00dc: 0x0640,	#  ARABIC TATWEEL
-    0x00dd: 0x0641,	#  ARABIC LETTER FEH
-    0x00de: 0x0642,	#  ARABIC LETTER QAF
-    0x00df: 0x0643,	#  ARABIC LETTER KAF
-    0x00e1: 0x0644,	#  ARABIC LETTER LAM
-    0x00e3: 0x0645,	#  ARABIC LETTER MEEM
-    0x00e4: 0x0646,	#  ARABIC LETTER NOON
-    0x00e5: 0x0647,	#  ARABIC LETTER HEH
-    0x00e6: 0x0648,	#  ARABIC LETTER WAW
-    0x00ec: 0x0649,	#  ARABIC LETTER ALEF MAKSURA
-    0x00ed: 0x064a,	#  ARABIC LETTER YEH
-    0x00f0: 0x064b,	#  ARABIC FATHATAN
-    0x00f1: 0x064c,	#  ARABIC DAMMATAN
-    0x00f2: 0x064d,	#  ARABIC KASRATAN
-    0x00f3: 0x064e,	#  ARABIC FATHA
-    0x00f5: 0x064f,	#  ARABIC DAMMA
-    0x00f6: 0x0650,	#  ARABIC KASRA
-    0x00f8: 0x0651,	#  ARABIC SHADDA
-    0x00fa: 0x0652,	#  ARABIC SUKUN
-    0x00fd: 0x200e,	#  LEFT-TO-RIGHT MARK
-    0x00fe: 0x200f,	#  RIGHT-TO-LEFT MARK
-    0x00ff: 0x06d2,	#  ARABIC LETTER YEH BARREE
-})
 
 ### Decoding Table
 
 decoding_table = (
-    u'\x00'	#  0x0000 -> NULL
-    u'\x01'	#  0x0001 -> START OF HEADING
-    u'\x02'	#  0x0002 -> START OF TEXT
-    u'\x03'	#  0x0003 -> END OF TEXT
-    u'\x04'	#  0x0004 -> END OF TRANSMISSION
-    u'\x05'	#  0x0005 -> ENQUIRY
-    u'\x06'	#  0x0006 -> ACKNOWLEDGE
-    u'\x07'	#  0x0007 -> BELL
-    u'\x08'	#  0x0008 -> BACKSPACE
-    u'\t'	#  0x0009 -> HORIZONTAL TABULATION
-    u'\n'	#  0x000a -> LINE FEED
-    u'\x0b'	#  0x000b -> VERTICAL TABULATION
-    u'\x0c'	#  0x000c -> FORM FEED
-    u'\r'	#  0x000d -> CARRIAGE RETURN
-    u'\x0e'	#  0x000e -> SHIFT OUT
-    u'\x0f'	#  0x000f -> SHIFT IN
-    u'\x10'	#  0x0010 -> DATA LINK ESCAPE
-    u'\x11'	#  0x0011 -> DEVICE CONTROL ONE
-    u'\x12'	#  0x0012 -> DEVICE CONTROL TWO
-    u'\x13'	#  0x0013 -> DEVICE CONTROL THREE
-    u'\x14'	#  0x0014 -> DEVICE CONTROL FOUR
-    u'\x15'	#  0x0015 -> NEGATIVE ACKNOWLEDGE
-    u'\x16'	#  0x0016 -> SYNCHRONOUS IDLE
-    u'\x17'	#  0x0017 -> END OF TRANSMISSION BLOCK
-    u'\x18'	#  0x0018 -> CANCEL
-    u'\x19'	#  0x0019 -> END OF MEDIUM
-    u'\x1a'	#  0x001a -> SUBSTITUTE
-    u'\x1b'	#  0x001b -> ESCAPE
-    u'\x1c'	#  0x001c -> FILE SEPARATOR
-    u'\x1d'	#  0x001d -> GROUP SEPARATOR
-    u'\x1e'	#  0x001e -> RECORD SEPARATOR
-    u'\x1f'	#  0x001f -> UNIT SEPARATOR
-    u' '	#  0x0020 -> SPACE
-    u'!'	#  0x0021 -> EXCLAMATION MARK
-    u'"'	#  0x0022 -> QUOTATION MARK
-    u'#'	#  0x0023 -> NUMBER SIGN
-    u'$'	#  0x0024 -> DOLLAR SIGN
-    u'%'	#  0x0025 -> PERCENT SIGN
-    u'&'	#  0x0026 -> AMPERSAND
-    u"'"	#  0x0027 -> APOSTROPHE
-    u'('	#  0x0028 -> LEFT PARENTHESIS
-    u')'	#  0x0029 -> RIGHT PARENTHESIS
-    u'*'	#  0x002a -> ASTERISK
-    u'+'	#  0x002b -> PLUS SIGN
-    u','	#  0x002c -> COMMA
-    u'-'	#  0x002d -> HYPHEN-MINUS
-    u'.'	#  0x002e -> FULL STOP
-    u'/'	#  0x002f -> SOLIDUS
-    u'0'	#  0x0030 -> DIGIT ZERO
-    u'1'	#  0x0031 -> DIGIT ONE
-    u'2'	#  0x0032 -> DIGIT TWO
-    u'3'	#  0x0033 -> DIGIT THREE
-    u'4'	#  0x0034 -> DIGIT FOUR
-    u'5'	#  0x0035 -> DIGIT FIVE
-    u'6'	#  0x0036 -> DIGIT SIX
-    u'7'	#  0x0037 -> DIGIT SEVEN
-    u'8'	#  0x0038 -> DIGIT EIGHT
-    u'9'	#  0x0039 -> DIGIT NINE
-    u':'	#  0x003a -> COLON
-    u';'	#  0x003b -> SEMICOLON
-    u'<'	#  0x003c -> LESS-THAN SIGN
-    u'='	#  0x003d -> EQUALS SIGN
-    u'>'	#  0x003e -> GREATER-THAN SIGN
-    u'?'	#  0x003f -> QUESTION MARK
-    u'@'	#  0x0040 -> COMMERCIAL AT
-    u'A'	#  0x0041 -> LATIN CAPITAL LETTER A
-    u'B'	#  0x0042 -> LATIN CAPITAL LETTER B
-    u'C'	#  0x0043 -> LATIN CAPITAL LETTER C
-    u'D'	#  0x0044 -> LATIN CAPITAL LETTER D
-    u'E'	#  0x0045 -> LATIN CAPITAL LETTER E
-    u'F'	#  0x0046 -> LATIN CAPITAL LETTER F
-    u'G'	#  0x0047 -> LATIN CAPITAL LETTER G
-    u'H'	#  0x0048 -> LATIN CAPITAL LETTER H
-    u'I'	#  0x0049 -> LATIN CAPITAL LETTER I
-    u'J'	#  0x004a -> LATIN CAPITAL LETTER J
-    u'K'	#  0x004b -> LATIN CAPITAL LETTER K
-    u'L'	#  0x004c -> LATIN CAPITAL LETTER L
-    u'M'	#  0x004d -> LATIN CAPITAL LETTER M
-    u'N'	#  0x004e -> LATIN CAPITAL LETTER N
-    u'O'	#  0x004f -> LATIN CAPITAL LETTER O
-    u'P'	#  0x0050 -> LATIN CAPITAL LETTER P
-    u'Q'	#  0x0051 -> LATIN CAPITAL LETTER Q
-    u'R'	#  0x0052 -> LATIN CAPITAL LETTER R
-    u'S'	#  0x0053 -> LATIN CAPITAL LETTER S
-    u'T'	#  0x0054 -> LATIN CAPITAL LETTER T
-    u'U'	#  0x0055 -> LATIN CAPITAL LETTER U
-    u'V'	#  0x0056 -> LATIN CAPITAL LETTER V
-    u'W'	#  0x0057 -> LATIN CAPITAL LETTER W
-    u'X'	#  0x0058 -> LATIN CAPITAL LETTER X
-    u'Y'	#  0x0059 -> LATIN CAPITAL LETTER Y
-    u'Z'	#  0x005a -> LATIN CAPITAL LETTER Z
-    u'['	#  0x005b -> LEFT SQUARE BRACKET
-    u'\\'	#  0x005c -> REVERSE SOLIDUS
-    u']'	#  0x005d -> RIGHT SQUARE BRACKET
-    u'^'	#  0x005e -> CIRCUMFLEX ACCENT
-    u'_'	#  0x005f -> LOW LINE
-    u'`'	#  0x0060 -> GRAVE ACCENT
-    u'a'	#  0x0061 -> LATIN SMALL LETTER A
-    u'b'	#  0x0062 -> LATIN SMALL LETTER B
-    u'c'	#  0x0063 -> LATIN SMALL LETTER C
-    u'd'	#  0x0064 -> LATIN SMALL LETTER D
-    u'e'	#  0x0065 -> LATIN SMALL LETTER E
-    u'f'	#  0x0066 -> LATIN SMALL LETTER F
-    u'g'	#  0x0067 -> LATIN SMALL LETTER G
-    u'h'	#  0x0068 -> LATIN SMALL LETTER H
-    u'i'	#  0x0069 -> LATIN SMALL LETTER I
-    u'j'	#  0x006a -> LATIN SMALL LETTER J
-    u'k'	#  0x006b -> LATIN SMALL LETTER K
-    u'l'	#  0x006c -> LATIN SMALL LETTER L
-    u'm'	#  0x006d -> LATIN SMALL LETTER M
-    u'n'	#  0x006e -> LATIN SMALL LETTER N
-    u'o'	#  0x006f -> LATIN SMALL LETTER O
-    u'p'	#  0x0070 -> LATIN SMALL LETTER P
-    u'q'	#  0x0071 -> LATIN SMALL LETTER Q
-    u'r'	#  0x0072 -> LATIN SMALL LETTER R
-    u's'	#  0x0073 -> LATIN SMALL LETTER S
-    u't'	#  0x0074 -> LATIN SMALL LETTER T
-    u'u'	#  0x0075 -> LATIN SMALL LETTER U
-    u'v'	#  0x0076 -> LATIN SMALL LETTER V
-    u'w'	#  0x0077 -> LATIN SMALL LETTER W
-    u'x'	#  0x0078 -> LATIN SMALL LETTER X
-    u'y'	#  0x0079 -> LATIN SMALL LETTER Y
-    u'z'	#  0x007a -> LATIN SMALL LETTER Z
-    u'{'	#  0x007b -> LEFT CURLY BRACKET
-    u'|'	#  0x007c -> VERTICAL LINE
-    u'}'	#  0x007d -> RIGHT CURLY BRACKET
-    u'~'	#  0x007e -> TILDE
-    u'\x7f'	#  0x007f -> DELETE
-    u'\u20ac'	#  0x0080 -> EURO SIGN
-    u'\u067e'	#  0x0081 -> ARABIC LETTER PEH
-    u'\u201a'	#  0x0082 -> SINGLE LOW-9 QUOTATION MARK
-    u'\u0192'	#  0x0083 -> LATIN SMALL LETTER F WITH HOOK
-    u'\u201e'	#  0x0084 -> DOUBLE LOW-9 QUOTATION MARK
-    u'\u2026'	#  0x0085 -> HORIZONTAL ELLIPSIS
-    u'\u2020'	#  0x0086 -> DAGGER
-    u'\u2021'	#  0x0087 -> DOUBLE DAGGER
-    u'\u02c6'	#  0x0088 -> MODIFIER LETTER CIRCUMFLEX ACCENT
-    u'\u2030'	#  0x0089 -> PER MILLE SIGN
-    u'\u0679'	#  0x008a -> ARABIC LETTER TTEH
-    u'\u2039'	#  0x008b -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK
-    u'\u0152'	#  0x008c -> LATIN CAPITAL LIGATURE OE
-    u'\u0686'	#  0x008d -> ARABIC LETTER TCHEH
-    u'\u0698'	#  0x008e -> ARABIC LETTER JEH
-    u'\u0688'	#  0x008f -> ARABIC LETTER DDAL
-    u'\u06af'	#  0x0090 -> ARABIC LETTER GAF
-    u'\u2018'	#  0x0091 -> LEFT SINGLE QUOTATION MARK
-    u'\u2019'	#  0x0092 -> RIGHT SINGLE QUOTATION MARK
-    u'\u201c'	#  0x0093 -> LEFT DOUBLE QUOTATION MARK
-    u'\u201d'	#  0x0094 -> RIGHT DOUBLE QUOTATION MARK
-    u'\u2022'	#  0x0095 -> BULLET
-    u'\u2013'	#  0x0096 -> EN DASH
-    u'\u2014'	#  0x0097 -> EM DASH
-    u'\u06a9'	#  0x0098 -> ARABIC LETTER KEHEH
-    u'\u2122'	#  0x0099 -> TRADE MARK SIGN
-    u'\u0691'	#  0x009a -> ARABIC LETTER RREH
-    u'\u203a'	#  0x009b -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
-    u'\u0153'	#  0x009c -> LATIN SMALL LIGATURE OE
-    u'\u200c'	#  0x009d -> ZERO WIDTH NON-JOINER
-    u'\u200d'	#  0x009e -> ZERO WIDTH JOINER
-    u'\u06ba'	#  0x009f -> ARABIC LETTER NOON GHUNNA
-    u'\xa0'	#  0x00a0 -> NO-BREAK SPACE
-    u'\u060c'	#  0x00a1 -> ARABIC COMMA
-    u'\xa2'	#  0x00a2 -> CENT SIGN
-    u'\xa3'	#  0x00a3 -> POUND SIGN
-    u'\xa4'	#  0x00a4 -> CURRENCY SIGN
-    u'\xa5'	#  0x00a5 -> YEN SIGN
-    u'\xa6'	#  0x00a6 -> BROKEN BAR
-    u'\xa7'	#  0x00a7 -> SECTION SIGN
-    u'\xa8'	#  0x00a8 -> DIAERESIS
-    u'\xa9'	#  0x00a9 -> COPYRIGHT SIGN
-    u'\u06be'	#  0x00aa -> ARABIC LETTER HEH DOACHASHMEE
-    u'\xab'	#  0x00ab -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
-    u'\xac'	#  0x00ac -> NOT SIGN
-    u'\xad'	#  0x00ad -> SOFT HYPHEN
-    u'\xae'	#  0x00ae -> REGISTERED SIGN
-    u'\xaf'	#  0x00af -> MACRON
-    u'\xb0'	#  0x00b0 -> DEGREE SIGN
-    u'\xb1'	#  0x00b1 -> PLUS-MINUS SIGN
-    u'\xb2'	#  0x00b2 -> SUPERSCRIPT TWO
-    u'\xb3'	#  0x00b3 -> SUPERSCRIPT THREE
-    u'\xb4'	#  0x00b4 -> ACUTE ACCENT
-    u'\xb5'	#  0x00b5 -> MICRO SIGN
-    u'\xb6'	#  0x00b6 -> PILCROW SIGN
-    u'\xb7'	#  0x00b7 -> MIDDLE DOT
-    u'\xb8'	#  0x00b8 -> CEDILLA
-    u'\xb9'	#  0x00b9 -> SUPERSCRIPT ONE
-    u'\u061b'	#  0x00ba -> ARABIC SEMICOLON
-    u'\xbb'	#  0x00bb -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
-    u'\xbc'	#  0x00bc -> VULGAR FRACTION ONE QUARTER
-    u'\xbd'	#  0x00bd -> VULGAR FRACTION ONE HALF
-    u'\xbe'	#  0x00be -> VULGAR FRACTION THREE QUARTERS
-    u'\u061f'	#  0x00bf -> ARABIC QUESTION MARK
-    u'\u06c1'	#  0x00c0 -> ARABIC LETTER HEH GOAL
-    u'\u0621'	#  0x00c1 -> ARABIC LETTER HAMZA
-    u'\u0622'	#  0x00c2 -> ARABIC LETTER ALEF WITH MADDA ABOVE
-    u'\u0623'	#  0x00c3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE
-    u'\u0624'	#  0x00c4 -> ARABIC LETTER WAW WITH HAMZA ABOVE
-    u'\u0625'	#  0x00c5 -> ARABIC LETTER ALEF WITH HAMZA BELOW
-    u'\u0626'	#  0x00c6 -> ARABIC LETTER YEH WITH HAMZA ABOVE
-    u'\u0627'	#  0x00c7 -> ARABIC LETTER ALEF
-    u'\u0628'	#  0x00c8 -> ARABIC LETTER BEH
-    u'\u0629'	#  0x00c9 -> ARABIC LETTER TEH MARBUTA
-    u'\u062a'	#  0x00ca -> ARABIC LETTER TEH
-    u'\u062b'	#  0x00cb -> ARABIC LETTER THEH
-    u'\u062c'	#  0x00cc -> ARABIC LETTER JEEM
-    u'\u062d'	#  0x00cd -> ARABIC LETTER HAH
-    u'\u062e'	#  0x00ce -> ARABIC LETTER KHAH
-    u'\u062f'	#  0x00cf -> ARABIC LETTER DAL
-    u'\u0630'	#  0x00d0 -> ARABIC LETTER THAL
-    u'\u0631'	#  0x00d1 -> ARABIC LETTER REH
-    u'\u0632'	#  0x00d2 -> ARABIC LETTER ZAIN
-    u'\u0633'	#  0x00d3 -> ARABIC LETTER SEEN
-    u'\u0634'	#  0x00d4 -> ARABIC LETTER SHEEN
-    u'\u0635'	#  0x00d5 -> ARABIC LETTER SAD
-    u'\u0636'	#  0x00d6 -> ARABIC LETTER DAD
-    u'\xd7'	#  0x00d7 -> MULTIPLICATION SIGN
-    u'\u0637'	#  0x00d8 -> ARABIC LETTER TAH
-    u'\u0638'	#  0x00d9 -> ARABIC LETTER ZAH
-    u'\u0639'	#  0x00da -> ARABIC LETTER AIN
-    u'\u063a'	#  0x00db -> ARABIC LETTER GHAIN
-    u'\u0640'	#  0x00dc -> ARABIC TATWEEL
-    u'\u0641'	#  0x00dd -> ARABIC LETTER FEH
-    u'\u0642'	#  0x00de -> ARABIC LETTER QAF
-    u'\u0643'	#  0x00df -> ARABIC LETTER KAF
-    u'\xe0'	#  0x00e0 -> LATIN SMALL LETTER A WITH GRAVE
-    u'\u0644'	#  0x00e1 -> ARABIC LETTER LAM
-    u'\xe2'	#  0x00e2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX
-    u'\u0645'	#  0x00e3 -> ARABIC LETTER MEEM
-    u'\u0646'	#  0x00e4 -> ARABIC LETTER NOON
-    u'\u0647'	#  0x00e5 -> ARABIC LETTER HEH
-    u'\u0648'	#  0x00e6 -> ARABIC LETTER WAW
-    u'\xe7'	#  0x00e7 -> LATIN SMALL LETTER C WITH CEDILLA
-    u'\xe8'	#  0x00e8 -> LATIN SMALL LETTER E WITH GRAVE
-    u'\xe9'	#  0x00e9 -> LATIN SMALL LETTER E WITH ACUTE
-    u'\xea'	#  0x00ea -> LATIN SMALL LETTER E WITH CIRCUMFLEX
-    u'\xeb'	#  0x00eb -> LATIN SMALL LETTER E WITH DIAERESIS
-    u'\u0649'	#  0x00ec -> ARABIC LETTER ALEF MAKSURA
-    u'\u064a'	#  0x00ed -> ARABIC LETTER YEH
-    u'\xee'	#  0x00ee -> LATIN SMALL LETTER I WITH CIRCUMFLEX
-    u'\xef'	#  0x00ef -> LATIN SMALL LETTER I WITH DIAERESIS
-    u'\u064b'	#  0x00f0 -> ARABIC FATHATAN
-    u'\u064c'	#  0x00f1 -> ARABIC DAMMATAN
-    u'\u064d'	#  0x00f2 -> ARABIC KASRATAN
-    u'\u064e'	#  0x00f3 -> ARABIC FATHA
-    u'\xf4'	#  0x00f4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX
-    u'\u064f'	#  0x00f5 -> ARABIC DAMMA
-    u'\u0650'	#  0x00f6 -> ARABIC KASRA
-    u'\xf7'	#  0x00f7 -> DIVISION SIGN
-    u'\u0651'	#  0x00f8 -> ARABIC SHADDA
-    u'\xf9'	#  0x00f9 -> LATIN SMALL LETTER U WITH GRAVE
-    u'\u0652'	#  0x00fa -> ARABIC SUKUN
-    u'\xfb'	#  0x00fb -> LATIN SMALL LETTER U WITH CIRCUMFLEX
-    u'\xfc'	#  0x00fc -> LATIN SMALL LETTER U WITH DIAERESIS
-    u'\u200e'	#  0x00fd -> LEFT-TO-RIGHT MARK
-    u'\u200f'	#  0x00fe -> RIGHT-TO-LEFT MARK
-    u'\u06d2'	#  0x00ff -> ARABIC LETTER YEH BARREE
+    u'\x00'	#  0x00 -> NULL
+    u'\x01'	#  0x01 -> START OF HEADING
+    u'\x02'	#  0x02 -> START OF TEXT
+    u'\x03'	#  0x03 -> END OF TEXT
+    u'\x04'	#  0x04 -> END OF TRANSMISSION
+    u'\x05'	#  0x05 -> ENQUIRY
+    u'\x06'	#  0x06 -> ACKNOWLEDGE
+    u'\x07'	#  0x07 -> BELL
+    u'\x08'	#  0x08 -> BACKSPACE
+    u'\t'	#  0x09 -> HORIZONTAL TABULATION
+    u'\n'	#  0x0a -> LINE FEED
+    u'\x0b'	#  0x0b -> VERTICAL TABULATION
+    u'\x0c'	#  0x0c -> FORM FEED
+    u'\r'	#  0x0d -> CARRIAGE RETURN
+    u'\x0e'	#  0x0e -> SHIFT OUT
+    u'\x0f'	#  0x0f -> SHIFT IN
+    u'\x10'	#  0x10 -> DATA LINK ESCAPE
+    u'\x11'	#  0x11 -> DEVICE CONTROL ONE
+    u'\x12'	#  0x12 -> DEVICE CONTROL TWO
+    u'\x13'	#  0x13 -> DEVICE CONTROL THREE
+    u'\x14'	#  0x14 -> DEVICE CONTROL FOUR
+    u'\x15'	#  0x15 -> NEGATIVE ACKNOWLEDGE
+    u'\x16'	#  0x16 -> SYNCHRONOUS IDLE
+    u'\x17'	#  0x17 -> END OF TRANSMISSION BLOCK
+    u'\x18'	#  0x18 -> CANCEL
+    u'\x19'	#  0x19 -> END OF MEDIUM
+    u'\x1a'	#  0x1a -> SUBSTITUTE
+    u'\x1b'	#  0x1b -> ESCAPE
+    u'\x1c'	#  0x1c -> FILE SEPARATOR
+    u'\x1d'	#  0x1d -> GROUP SEPARATOR
+    u'\x1e'	#  0x1e -> RECORD SEPARATOR
+    u'\x1f'	#  0x1f -> UNIT SEPARATOR
+    u' '	#  0x20 -> SPACE
+    u'!'	#  0x21 -> EXCLAMATION MARK
+    u'"'	#  0x22 -> QUOTATION MARK
+    u'#'	#  0x23 -> NUMBER SIGN
+    u'$'	#  0x24 -> DOLLAR SIGN
+    u'%'	#  0x25 -> PERCENT SIGN
+    u'&'	#  0x26 -> AMPERSAND
+    u"'"	#  0x27 -> APOSTROPHE
+    u'('	#  0x28 -> LEFT PARENTHESIS
+    u')'	#  0x29 -> RIGHT PARENTHESIS
+    u'*'	#  0x2a -> ASTERISK
+    u'+'	#  0x2b -> PLUS SIGN
+    u','	#  0x2c -> COMMA
+    u'-'	#  0x2d -> HYPHEN-MINUS
+    u'.'	#  0x2e -> FULL STOP
+    u'/'	#  0x2f -> SOLIDUS
+    u'0'	#  0x30 -> DIGIT ZERO
+    u'1'	#  0x31 -> DIGIT ONE
+    u'2'	#  0x32 -> DIGIT TWO
+    u'3'	#  0x33 -> DIGIT THREE
+    u'4'	#  0x34 -> DIGIT FOUR
+    u'5'	#  0x35 -> DIGIT FIVE
+    u'6'	#  0x36 -> DIGIT SIX
+    u'7'	#  0x37 -> DIGIT SEVEN
+    u'8'	#  0x38 -> DIGIT EIGHT
+    u'9'	#  0x39 -> DIGIT NINE
+    u':'	#  0x3a -> COLON
+    u';'	#  0x3b -> SEMICOLON
+    u'<'	#  0x3c -> LESS-THAN SIGN
+    u'='	#  0x3d -> EQUALS SIGN
+    u'>'	#  0x3e -> GREATER-THAN SIGN
+    u'?'	#  0x3f -> QUESTION MARK
+    u'@'	#  0x40 -> COMMERCIAL AT
+    u'A'	#  0x41 -> LATIN CAPITAL LETTER A
+    u'B'	#  0x42 -> LATIN CAPITAL LETTER B
+    u'C'	#  0x43 -> LATIN CAPITAL LETTER C
+    u'D'	#  0x44 -> LATIN CAPITAL LETTER D
+    u'E'	#  0x45 -> LATIN CAPITAL LETTER E
+    u'F'	#  0x46 -> LATIN CAPITAL LETTER F
+    u'G'	#  0x47 -> LATIN CAPITAL LETTER G
+    u'H'	#  0x48 -> LATIN CAPITAL LETTER H
+    u'I'	#  0x49 -> LATIN CAPITAL LETTER I
+    u'J'	#  0x4a -> LATIN CAPITAL LETTER J
+    u'K'	#  0x4b -> LATIN CAPITAL LETTER K
+    u'L'	#  0x4c -> LATIN CAPITAL LETTER L
+    u'M'	#  0x4d -> LATIN CAPITAL LETTER M
+    u'N'	#  0x4e -> LATIN CAPITAL LETTER N
+    u'O'	#  0x4f -> LATIN CAPITAL LETTER O
+    u'P'	#  0x50 -> LATIN CAPITAL LETTER P
+    u'Q'	#  0x51 -> LATIN CAPITAL LETTER Q
+    u'R'	#  0x52 -> LATIN CAPITAL LETTER R
+    u'S'	#  0x53 -> LATIN CAPITAL LETTER S
+    u'T'	#  0x54 -> LATIN CAPITAL LETTER T
+    u'U'	#  0x55 -> LATIN CAPITAL LETTER U
+    u'V'	#  0x56 -> LATIN CAPITAL LETTER V
+    u'W'	#  0x57 -> LATIN CAPITAL LETTER W
+    u'X'	#  0x58 -> LATIN CAPITAL LETTER X
+    u'Y'	#  0x59 -> LATIN CAPITAL LETTER Y
+    u'Z'	#  0x5a -> LATIN CAPITAL LETTER Z
+    u'['	#  0x5b -> LEFT SQUARE BRACKET
+    u'\\'	#  0x5c -> REVERSE SOLIDUS
+    u']'	#  0x5d -> RIGHT SQUARE BRACKET
+    u'^'	#  0x5e -> CIRCUMFLEX ACCENT
+    u'_'	#  0x5f -> LOW LINE
+    u'`'	#  0x60 -> GRAVE ACCENT
+    u'a'	#  0x61 -> LATIN SMALL LETTER A
+    u'b'	#  0x62 -> LATIN SMALL LETTER B
+    u'c'	#  0x63 -> LATIN SMALL LETTER C
+    u'd'	#  0x64 -> LATIN SMALL LETTER D
+    u'e'	#  0x65 -> LATIN SMALL LETTER E
+    u'f'	#  0x66 -> LATIN SMALL LETTER F
+    u'g'	#  0x67 -> LATIN SMALL LETTER G
+    u'h'	#  0x68 -> LATIN SMALL LETTER H
+    u'i'	#  0x69 -> LATIN SMALL LETTER I
+    u'j'	#  0x6a -> LATIN SMALL LETTER J
+    u'k'	#  0x6b -> LATIN SMALL LETTER K
+    u'l'	#  0x6c -> LATIN SMALL LETTER L
+    u'm'	#  0x6d -> LATIN SMALL LETTER M
+    u'n'	#  0x6e -> LATIN SMALL LETTER N
+    u'o'	#  0x6f -> LATIN SMALL LETTER O
+    u'p'	#  0x70 -> LATIN SMALL LETTER P
+    u'q'	#  0x71 -> LATIN SMALL LETTER Q
+    u'r'	#  0x72 -> LATIN SMALL LETTER R
+    u's'	#  0x73 -> LATIN SMALL LETTER S
+    u't'	#  0x74 -> LATIN SMALL LETTER T
+    u'u'	#  0x75 -> LATIN SMALL LETTER U
+    u'v'	#  0x76 -> LATIN SMALL LETTER V
+    u'w'	#  0x77 -> LATIN SMALL LETTER W
+    u'x'	#  0x78 -> LATIN SMALL LETTER X
+    u'y'	#  0x79 -> LATIN SMALL LETTER Y
+    u'z'	#  0x7a -> LATIN SMALL LETTER Z
+    u'{'	#  0x7b -> LEFT CURLY BRACKET
+    u'|'	#  0x7c -> VERTICAL LINE
+    u'}'	#  0x7d -> RIGHT CURLY BRACKET
+    u'~'	#  0x7e -> TILDE
+    u'\x7f'	#  0x7f -> DELETE
+    u'\u20ac'	#  0x80 -> EURO SIGN
+    u'\u067e'	#  0x81 -> ARABIC LETTER PEH
+    u'\u201a'	#  0x82 -> SINGLE LOW-9 QUOTATION MARK
+    u'\u0192'	#  0x83 -> LATIN SMALL LETTER F WITH HOOK
+    u'\u201e'	#  0x84 -> DOUBLE LOW-9 QUOTATION MARK
+    u'\u2026'	#  0x85 -> HORIZONTAL ELLIPSIS
+    u'\u2020'	#  0x86 -> DAGGER
+    u'\u2021'	#  0x87 -> DOUBLE DAGGER
+    u'\u02c6'	#  0x88 -> MODIFIER LETTER CIRCUMFLEX ACCENT
+    u'\u2030'	#  0x89 -> PER MILLE SIGN
+    u'\u0679'	#  0x8a -> ARABIC LETTER TTEH
+    u'\u2039'	#  0x8b -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK
+    u'\u0152'	#  0x8c -> LATIN CAPITAL LIGATURE OE
+    u'\u0686'	#  0x8d -> ARABIC LETTER TCHEH
+    u'\u0698'	#  0x8e -> ARABIC LETTER JEH
+    u'\u0688'	#  0x8f -> ARABIC LETTER DDAL
+    u'\u06af'	#  0x90 -> ARABIC LETTER GAF
+    u'\u2018'	#  0x91 -> LEFT SINGLE QUOTATION MARK
+    u'\u2019'	#  0x92 -> RIGHT SINGLE QUOTATION MARK
+    u'\u201c'	#  0x93 -> LEFT DOUBLE QUOTATION MARK
+    u'\u201d'	#  0x94 -> RIGHT DOUBLE QUOTATION MARK
+    u'\u2022'	#  0x95 -> BULLET
+    u'\u2013'	#  0x96 -> EN DASH
+    u'\u2014'	#  0x97 -> EM DASH
+    u'\u06a9'	#  0x98 -> ARABIC LETTER KEHEH
+    u'\u2122'	#  0x99 -> TRADE MARK SIGN
+    u'\u0691'	#  0x9a -> ARABIC LETTER RREH
+    u'\u203a'	#  0x9b -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
+    u'\u0153'	#  0x9c -> LATIN SMALL LIGATURE OE
+    u'\u200c'	#  0x9d -> ZERO WIDTH NON-JOINER
+    u'\u200d'	#  0x9e -> ZERO WIDTH JOINER
+    u'\u06ba'	#  0x9f -> ARABIC LETTER NOON GHUNNA
+    u'\xa0'	#  0xa0 -> NO-BREAK SPACE
+    u'\u060c'	#  0xa1 -> ARABIC COMMA
+    u'\xa2'	#  0xa2 -> CENT SIGN
+    u'\xa3'	#  0xa3 -> POUND SIGN
+    u'\xa4'	#  0xa4 -> CURRENCY SIGN
+    u'\xa5'	#  0xa5 -> YEN SIGN
+    u'\xa6'	#  0xa6 -> BROKEN BAR
+    u'\xa7'	#  0xa7 -> SECTION SIGN
+    u'\xa8'	#  0xa8 -> DIAERESIS
+    u'\xa9'	#  0xa9 -> COPYRIGHT SIGN
+    u'\u06be'	#  0xaa -> ARABIC LETTER HEH DOACHASHMEE
+    u'\xab'	#  0xab -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
+    u'\xac'	#  0xac -> NOT SIGN
+    u'\xad'	#  0xad -> SOFT HYPHEN
+    u'\xae'	#  0xae -> REGISTERED SIGN
+    u'\xaf'	#  0xaf -> MACRON
+    u'\xb0'	#  0xb0 -> DEGREE SIGN
+    u'\xb1'	#  0xb1 -> PLUS-MINUS SIGN
+    u'\xb2'	#  0xb2 -> SUPERSCRIPT TWO
+    u'\xb3'	#  0xb3 -> SUPERSCRIPT THREE
+    u'\xb4'	#  0xb4 -> ACUTE ACCENT
+    u'\xb5'	#  0xb5 -> MICRO SIGN
+    u'\xb6'	#  0xb6 -> PILCROW SIGN
+    u'\xb7'	#  0xb7 -> MIDDLE DOT
+    u'\xb8'	#  0xb8 -> CEDILLA
+    u'\xb9'	#  0xb9 -> SUPERSCRIPT ONE
+    u'\u061b'	#  0xba -> ARABIC SEMICOLON
+    u'\xbb'	#  0xbb -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
+    u'\xbc'	#  0xbc -> VULGAR FRACTION ONE QUARTER
+    u'\xbd'	#  0xbd -> VULGAR FRACTION ONE HALF
+    u'\xbe'	#  0xbe -> VULGAR FRACTION THREE QUARTERS
+    u'\u061f'	#  0xbf -> ARABIC QUESTION MARK
+    u'\u06c1'	#  0xc0 -> ARABIC LETTER HEH GOAL
+    u'\u0621'	#  0xc1 -> ARABIC LETTER HAMZA
+    u'\u0622'	#  0xc2 -> ARABIC LETTER ALEF WITH MADDA ABOVE
+    u'\u0623'	#  0xc3 -> ARABIC LETTER ALEF WITH HAMZA ABOVE
+    u'\u0624'	#  0xc4 -> ARABIC LETTER WAW WITH HAMZA ABOVE
+    u'\u0625'	#  0xc5 -> ARABIC LETTER ALEF WITH HAMZA BELOW
+    u'\u0626'	#  0xc6 -> ARABIC LETTER YEH WITH HAMZA ABOVE
+    u'\u0627'	#  0xc7 -> ARABIC LETTER ALEF
+    u'\u0628'	#  0xc8 -> ARABIC LETTER BEH
+    u'\u0629'	#  0xc9 -> ARABIC LETTER TEH MARBUTA
+    u'\u062a'	#  0xca -> ARABIC LETTER TEH
+    u'\u062b'	#  0xcb -> ARABIC LETTER THEH
+    u'\u062c'	#  0xcc -> ARABIC LETTER JEEM
+    u'\u062d'	#  0xcd -> ARABIC LETTER HAH
+    u'\u062e'	#  0xce -> ARABIC LETTER KHAH
+    u'\u062f'	#  0xcf -> ARABIC LETTER DAL
+    u'\u0630'	#  0xd0 -> ARABIC LETTER THAL
+    u'\u0631'	#  0xd1 -> ARABIC LETTER REH
+    u'\u0632'	#  0xd2 -> ARABIC LETTER ZAIN
+    u'\u0633'	#  0xd3 -> ARABIC LETTER SEEN
+    u'\u0634'	#  0xd4 -> ARABIC LETTER SHEEN
+    u'\u0635'	#  0xd5 -> ARABIC LETTER SAD
+    u'\u0636'	#  0xd6 -> ARABIC LETTER DAD
+    u'\xd7'	#  0xd7 -> MULTIPLICATION SIGN
+    u'\u0637'	#  0xd8 -> ARABIC LETTER TAH
+    u'\u0638'	#  0xd9 -> ARABIC LETTER ZAH
+    u'\u0639'	#  0xda -> ARABIC LETTER AIN
+    u'\u063a'	#  0xdb -> ARABIC LETTER GHAIN
+    u'\u0640'	#  0xdc -> ARABIC TATWEEL
+    u'\u0641'	#  0xdd -> ARABIC LETTER FEH
+    u'\u0642'	#  0xde -> ARABIC LETTER QAF
+    u'\u0643'	#  0xdf -> ARABIC LETTER KAF
+    u'\xe0'	#  0xe0 -> LATIN SMALL LETTER A WITH GRAVE
+    u'\u0644'	#  0xe1 -> ARABIC LETTER LAM
+    u'\xe2'	#  0xe2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX
+    u'\u0645'	#  0xe3 -> ARABIC LETTER MEEM
+    u'\u0646'	#  0xe4 -> ARABIC LETTER NOON
+    u'\u0647'	#  0xe5 -> ARABIC LETTER HEH
+    u'\u0648'	#  0xe6 -> ARABIC LETTER WAW
+    u'\xe7'	#  0xe7 -> LATIN SMALL LETTER C WITH CEDILLA
+    u'\xe8'	#  0xe8 -> LATIN SMALL LETTER E WITH GRAVE
+    u'\xe9'	#  0xe9 -> LATIN SMALL LETTER E WITH ACUTE
+    u'\xea'	#  0xea -> LATIN SMALL LETTER E WITH CIRCUMFLEX
+    u'\xeb'	#  0xeb -> LATIN SMALL LETTER E WITH DIAERESIS
+    u'\u0649'	#  0xec -> ARABIC LETTER ALEF MAKSURA
+    u'\u064a'	#  0xed -> ARABIC LETTER YEH
+    u'\xee'	#  0xee -> LATIN SMALL LETTER I WITH CIRCUMFLEX
+    u'\xef'	#  0xef -> LATIN SMALL LETTER I WITH DIAERESIS
+    u'\u064b'	#  0xf0 -> ARABIC FATHATAN
+    u'\u064c'	#  0xf1 -> ARABIC DAMMATAN
+    u'\u064d'	#  0xf2 -> ARABIC KASRATAN
+    u'\u064e'	#  0xf3 -> ARABIC FATHA
+    u'\xf4'	#  0xf4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX
+    u'\u064f'	#  0xf5 -> ARABIC DAMMA
+    u'\u0650'	#  0xf6 -> ARABIC KASRA
+    u'\xf7'	#  0xf7 -> DIVISION SIGN
+    u'\u0651'	#  0xf8 -> ARABIC SHADDA
+    u'\xf9'	#  0xf9 -> LATIN SMALL LETTER U WITH GRAVE
+    u'\u0652'	#  0xfa -> ARABIC SUKUN
+    u'\xfb'	#  0xfb -> LATIN SMALL LETTER U WITH CIRCUMFLEX
+    u'\xfc'	#  0xfc -> LATIN SMALL LETTER U WITH DIAERESIS
+    u'\u200e'	#  0xfd -> LEFT-TO-RIGHT MARK
+    u'\u200f'	#  0xfe -> RIGHT-TO-LEFT MARK
+    u'\u06d2'	#  0xff -> ARABIC LETTER YEH BARREE
 )
 
 ### Encoding Map
 
 encoding_map = {
-    0x0000: 0x0000,	#  NULL
-    0x0001: 0x0001,	#  START OF HEADING
-    0x0002: 0x0002,	#  START OF TEXT
-    0x0003: 0x0003,	#  END OF TEXT
-    0x0004: 0x0004,	#  END OF TRANSMISSION
-    0x0005: 0x0005,	#  ENQUIRY
-    0x0006: 0x0006,	#  ACKNOWLEDGE
-    0x0007: 0x0007,	#  BELL
-    0x0008: 0x0008,	#  BACKSPACE
-    0x0009: 0x0009,	#  HORIZONTAL TABULATION
-    0x000a: 0x000a,	#  LINE FEED
-    0x000b: 0x000b,	#  VERTICAL TABULATION
-    0x000c: 0x000c,	#  FORM FEED
-    0x000d: 0x000d,	#  CARRIAGE RETURN
-    0x000e: 0x000e,	#  SHIFT OUT
-    0x000f: 0x000f,	#  SHIFT IN
-    0x0010: 0x0010,	#  DATA LINK ESCAPE
-    0x0011: 0x0011,	#  DEVICE CONTROL ONE
-    0x0012: 0x0012,	#  DEVICE CONTROL TWO
-    0x0013: 0x0013,	#  DEVICE CONTROL THREE
-    0x0014: 0x0014,	#  DEVICE CONTROL FOUR
-    0x0015: 0x0015,	#  NEGATIVE ACKNOWLEDGE
-    0x0016: 0x0016,	#  SYNCHRONOUS IDLE
-    0x0017: 0x0017,	#  END OF TRANSMISSION BLOCK
-    0x0018: 0x0018,	#  CANCEL
-    0x0019: 0x0019,	#  END OF MEDIUM
-    0x001a: 0x001a,	#  SUBSTITUTE
-    0x001b: 0x001b,	#  ESCAPE
-    0x001c: 0x001c,	#  FILE SEPARATOR
-    0x001d: 0x001d,	#  GROUP SEPARATOR
-    0x001e: 0x001e,	#  RECORD SEPARATOR
-    0x001f: 0x001f,	#  UNIT SEPARATOR
-    0x0020: 0x0020,	#  SPACE
-    0x0021: 0x0021,	#  EXCLAMATION MARK
-    0x0022: 0x0022,	#  QUOTATION MARK
-    0x0023: 0x0023,	#  NUMBER SIGN
-    0x0024: 0x0024,	#  DOLLAR SIGN
-    0x0025: 0x0025,	#  PERCENT SIGN
-    0x0026: 0x0026,	#  AMPERSAND
-    0x0027: 0x0027,	#  APOSTROPHE
-    0x0028: 0x0028,	#  LEFT PARENTHESIS
-    0x0029: 0x0029,	#  RIGHT PARENTHESIS
-    0x002a: 0x002a,	#  ASTERISK
-    0x002b: 0x002b,	#  PLUS SIGN
-    0x002c: 0x002c,	#  COMMA
-    0x002d: 0x002d,	#  HYPHEN-MINUS
-    0x002e: 0x002e,	#  FULL STOP
-    0x002f: 0x002f,	#  SOLIDUS
-    0x0030: 0x0030,	#  DIGIT ZERO
-    0x0031: 0x0031,	#  DIGIT ONE
-    0x0032: 0x0032,	#  DIGIT TWO
-    0x0033: 0x0033,	#  DIGIT THREE
-    0x0034: 0x0034,	#  DIGIT FOUR
-    0x0035: 0x0035,	#  DIGIT FIVE
-    0x0036: 0x0036,	#  DIGIT SIX
-    0x0037: 0x0037,	#  DIGIT SEVEN
-    0x0038: 0x0038,	#  DIGIT EIGHT
-    0x0039: 0x0039,	#  DIGIT NINE
-    0x003a: 0x003a,	#  COLON
-    0x003b: 0x003b,	#  SEMICOLON
-    0x003c: 0x003c,	#  LESS-THAN SIGN
-    0x003d: 0x003d,	#  EQUALS SIGN
-    0x003e: 0x003e,	#  GREATER-THAN SIGN
-    0x003f: 0x003f,	#  QUESTION MARK
-    0x0040: 0x0040,	#  COMMERCIAL AT
-    0x0041: 0x0041,	#  LATIN CAPITAL LETTER A
-    0x0042: 0x0042,	#  LATIN CAPITAL LETTER B
-    0x0043: 0x0043,	#  LATIN CAPITAL LETTER C
-    0x0044: 0x0044,	#  LATIN CAPITAL LETTER D
-    0x0045: 0x0045,	#  LATIN CAPITAL LETTER E
-    0x0046: 0x0046,	#  LATIN CAPITAL LETTER F
-    0x0047: 0x0047,	#  LATIN CAPITAL LETTER G
-    0x0048: 0x0048,	#  LATIN CAPITAL LETTER H
-    0x0049: 0x0049,	#  LATIN CAPITAL LETTER I
-    0x004a: 0x004a,	#  LATIN CAPITAL LETTER J
-    0x004b: 0x004b,	#  LATIN CAPITAL LETTER K
-    0x004c: 0x004c,	#  LATIN CAPITAL LETTER L
-    0x004d: 0x004d,	#  LATIN CAPITAL LETTER M
-    0x004e: 0x004e,	#  LATIN CAPITAL LETTER N
-    0x004f: 0x004f,	#  LATIN CAPITAL LETTER O
-    0x0050: 0x0050,	#  LATIN CAPITAL LETTER P
-    0x0051: 0x0051,	#  LATIN CAPITAL LETTER Q
-    0x0052: 0x0052,	#  LATIN CAPITAL LETTER R
-    0x0053: 0x0053,	#  LATIN CAPITAL LETTER S
-    0x0054: 0x0054,	#  LATIN CAPITAL LETTER T
-    0x0055: 0x0055,	#  LATIN CAPITAL LETTER U
-    0x0056: 0x0056,	#  LATIN CAPITAL LETTER V
-    0x0057: 0x0057,	#  LATIN CAPITAL LETTER W
-    0x0058: 0x0058,	#  LATIN CAPITAL LETTER X
-    0x0059: 0x0059,	#  LATIN CAPITAL LETTER Y
-    0x005a: 0x005a,	#  LATIN CAPITAL LETTER Z
-    0x005b: 0x005b,	#  LEFT SQUARE BRACKET
-    0x005c: 0x005c,	#  REVERSE SOLIDUS
-    0x005d: 0x005d,	#  RIGHT SQUARE BRACKET
-    0x005e: 0x005e,	#  CIRCUMFLEX ACCENT
-    0x005f: 0x005f,	#  LOW LINE
-    0x0060: 0x0060,	#  GRAVE ACCENT
-    0x0061: 0x0061,	#  LATIN SMALL LETTER A
-    0x0062: 0x0062,	#  LATIN SMALL LETTER B
-    0x0063: 0x0063,	#  LATIN SMALL LETTER C
-    0x0064: 0x0064,	#  LATIN SMALL LETTER D
-    0x0065: 0x0065,	#  LATIN SMALL LETTER E
-    0x0066: 0x0066,	#  LATIN SMALL LETTER F
-    0x0067: 0x0067,	#  LATIN SMALL LETTER G
-    0x0068: 0x0068,	#  LATIN SMALL LETTER H
-    0x0069: 0x0069,	#  LATIN SMALL LETTER I
-    0x006a: 0x006a,	#  LATIN SMALL LETTER J
-    0x006b: 0x006b,	#  LATIN SMALL LETTER K
-    0x006c: 0x006c,	#  LATIN SMALL LETTER L
-    0x006d: 0x006d,	#  LATIN SMALL LETTER M
-    0x006e: 0x006e,	#  LATIN SMALL LETTER N
-    0x006f: 0x006f,	#  LATIN SMALL LETTER O
-    0x0070: 0x0070,	#  LATIN SMALL LETTER P
-    0x0071: 0x0071,	#  LATIN SMALL LETTER Q
-    0x0072: 0x0072,	#  LATIN SMALL LETTER R
-    0x0073: 0x0073,	#  LATIN SMALL LETTER S
-    0x0074: 0x0074,	#  LATIN SMALL LETTER T
-    0x0075: 0x0075,	#  LATIN SMALL LETTER U
-    0x0076: 0x0076,	#  LATIN SMALL LETTER V
-    0x0077: 0x0077,	#  LATIN SMALL LETTER W
-    0x0078: 0x0078,	#  LATIN SMALL LETTER X
-    0x0079: 0x0079,	#  LATIN SMALL LETTER Y
-    0x007a: 0x007a,	#  LATIN SMALL LETTER Z
-    0x007b: 0x007b,	#  LEFT CURLY BRACKET
-    0x007c: 0x007c,	#  VERTICAL LINE
-    0x007d: 0x007d,	#  RIGHT CURLY BRACKET
-    0x007e: 0x007e,	#  TILDE
-    0x007f: 0x007f,	#  DELETE
-    0x00a0: 0x00a0,	#  NO-BREAK SPACE
-    0x00a2: 0x00a2,	#  CENT SIGN
-    0x00a3: 0x00a3,	#  POUND SIGN
-    0x00a4: 0x00a4,	#  CURRENCY SIGN
-    0x00a5: 0x00a5,	#  YEN SIGN
-    0x00a6: 0x00a6,	#  BROKEN BAR
-    0x00a7: 0x00a7,	#  SECTION SIGN
-    0x00a8: 0x00a8,	#  DIAERESIS
-    0x00a9: 0x00a9,	#  COPYRIGHT SIGN
-    0x00ab: 0x00ab,	#  LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
-    0x00ac: 0x00ac,	#  NOT SIGN
-    0x00ad: 0x00ad,	#  SOFT HYPHEN
-    0x00ae: 0x00ae,	#  REGISTERED SIGN
-    0x00af: 0x00af,	#  MACRON
-    0x00b0: 0x00b0,	#  DEGREE SIGN
-    0x00b1: 0x00b1,	#  PLUS-MINUS SIGN
-    0x00b2: 0x00b2,	#  SUPERSCRIPT TWO
-    0x00b3: 0x00b3,	#  SUPERSCRIPT THREE
-    0x00b4: 0x00b4,	#  ACUTE ACCENT
-    0x00b5: 0x00b5,	#  MICRO SIGN
-    0x00b6: 0x00b6,	#  PILCROW SIGN
-    0x00b7: 0x00b7,	#  MIDDLE DOT
-    0x00b8: 0x00b8,	#  CEDILLA
-    0x00b9: 0x00b9,	#  SUPERSCRIPT ONE
-    0x00bb: 0x00bb,	#  RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
-    0x00bc: 0x00bc,	#  VULGAR FRACTION ONE QUARTER
-    0x00bd: 0x00bd,	#  VULGAR FRACTION ONE HALF
-    0x00be: 0x00be,	#  VULGAR FRACTION THREE QUARTERS
-    0x00d7: 0x00d7,	#  MULTIPLICATION SIGN
-    0x00e0: 0x00e0,	#  LATIN SMALL LETTER A WITH GRAVE
-    0x00e2: 0x00e2,	#  LATIN SMALL LETTER A WITH CIRCUMFLEX
-    0x00e7: 0x00e7,	#  LATIN SMALL LETTER C WITH CEDILLA
-    0x00e8: 0x00e8,	#  LATIN SMALL LETTER E WITH GRAVE
-    0x00e9: 0x00e9,	#  LATIN SMALL LETTER E WITH ACUTE
-    0x00ea: 0x00ea,	#  LATIN SMALL LETTER E WITH CIRCUMFLEX
-    0x00eb: 0x00eb,	#  LATIN SMALL LETTER E WITH DIAERESIS
-    0x00ee: 0x00ee,	#  LATIN SMALL LETTER I WITH CIRCUMFLEX
-    0x00ef: 0x00ef,	#  LATIN SMALL LETTER I WITH DIAERESIS
-    0x00f4: 0x00f4,	#  LATIN SMALL LETTER O WITH CIRCUMFLEX
-    0x00f7: 0x00f7,	#  DIVISION SIGN
-    0x00f9: 0x00f9,	#  LATIN SMALL LETTER U WITH GRAVE
-    0x00fb: 0x00fb,	#  LATIN SMALL LETTER U WITH CIRCUMFLEX
-    0x00fc: 0x00fc,	#  LATIN SMALL LETTER U WITH DIAERESIS
-    0x0152: 0x008c,	#  LATIN CAPITAL LIGATURE OE
-    0x0153: 0x009c,	#  LATIN SMALL LIGATURE OE
-    0x0192: 0x0083,	#  LATIN SMALL LETTER F WITH HOOK
-    0x02c6: 0x0088,	#  MODIFIER LETTER CIRCUMFLEX ACCENT
-    0x060c: 0x00a1,	#  ARABIC COMMA
-    0x061b: 0x00ba,	#  ARABIC SEMICOLON
-    0x061f: 0x00bf,	#  ARABIC QUESTION MARK
-    0x0621: 0x00c1,	#  ARABIC LETTER HAMZA
-    0x0622: 0x00c2,	#  ARABIC LETTER ALEF WITH MADDA ABOVE
-    0x0623: 0x00c3,	#  ARABIC LETTER ALEF WITH HAMZA ABOVE
-    0x0624: 0x00c4,	#  ARABIC LETTER WAW WITH HAMZA ABOVE
-    0x0625: 0x00c5,	#  ARABIC LETTER ALEF WITH HAMZA BELOW
-    0x0626: 0x00c6,	#  ARABIC LETTER YEH WITH HAMZA ABOVE
-    0x0627: 0x00c7,	#  ARABIC LETTER ALEF
-    0x0628: 0x00c8,	#  ARABIC LETTER BEH
-    0x0629: 0x00c9,	#  ARABIC LETTER TEH MARBUTA
-    0x062a: 0x00ca,	#  ARABIC LETTER TEH
-    0x062b: 0x00cb,	#  ARABIC LETTER THEH
-    0x062c: 0x00cc,	#  ARABIC LETTER JEEM
-    0x062d: 0x00cd,	#  ARABIC LETTER HAH
-    0x062e: 0x00ce,	#  ARABIC LETTER KHAH
-    0x062f: 0x00cf,	#  ARABIC LETTER DAL
-    0x0630: 0x00d0,	#  ARABIC LETTER THAL
-    0x0631: 0x00d1,	#  ARABIC LETTER REH
-    0x0632: 0x00d2,	#  ARABIC LETTER ZAIN
-    0x0633: 0x00d3,	#  ARABIC LETTER SEEN
-    0x0634: 0x00d4,	#  ARABIC LETTER SHEEN
-    0x0635: 0x00d5,	#  ARABIC LETTER SAD
-    0x0636: 0x00d6,	#  ARABIC LETTER DAD
-    0x0637: 0x00d8,	#  ARABIC LETTER TAH
-    0x0638: 0x00d9,	#  ARABIC LETTER ZAH
-    0x0639: 0x00da,	#  ARABIC LETTER AIN
-    0x063a: 0x00db,	#  ARABIC LETTER GHAIN
-    0x0640: 0x00dc,	#  ARABIC TATWEEL
-    0x0641: 0x00dd,	#  ARABIC LETTER FEH
-    0x0642: 0x00de,	#  ARABIC LETTER QAF
-    0x0643: 0x00df,	#  ARABIC LETTER KAF
-    0x0644: 0x00e1,	#  ARABIC LETTER LAM
-    0x0645: 0x00e3,	#  ARABIC LETTER MEEM
-    0x0646: 0x00e4,	#  ARABIC LETTER NOON
-    0x0647: 0x00e5,	#  ARABIC LETTER HEH
-    0x0648: 0x00e6,	#  ARABIC LETTER WAW
-    0x0649: 0x00ec,	#  ARABIC LETTER ALEF MAKSURA
-    0x064a: 0x00ed,	#  ARABIC LETTER YEH
-    0x064b: 0x00f0,	#  ARABIC FATHATAN
-    0x064c: 0x00f1,	#  ARABIC DAMMATAN
-    0x064d: 0x00f2,	#  ARABIC KASRATAN
-    0x064e: 0x00f3,	#  ARABIC FATHA
-    0x064f: 0x00f5,	#  ARABIC DAMMA
-    0x0650: 0x00f6,	#  ARABIC KASRA
-    0x0651: 0x00f8,	#  ARABIC SHADDA
-    0x0652: 0x00fa,	#  ARABIC SUKUN
-    0x0679: 0x008a,	#  ARABIC LETTER TTEH
-    0x067e: 0x0081,	#  ARABIC LETTER PEH
-    0x0686: 0x008d,	#  ARABIC LETTER TCHEH
-    0x0688: 0x008f,	#  ARABIC LETTER DDAL
-    0x0691: 0x009a,	#  ARABIC LETTER RREH
-    0x0698: 0x008e,	#  ARABIC LETTER JEH
-    0x06a9: 0x0098,	#  ARABIC LETTER KEHEH
-    0x06af: 0x0090,	#  ARABIC LETTER GAF
-    0x06ba: 0x009f,	#  ARABIC LETTER NOON GHUNNA
-    0x06be: 0x00aa,	#  ARABIC LETTER HEH DOACHASHMEE
-    0x06c1: 0x00c0,	#  ARABIC LETTER HEH GOAL
-    0x06d2: 0x00ff,	#  ARABIC LETTER YEH BARREE
-    0x200c: 0x009d,	#  ZERO WIDTH NON-JOINER
-    0x200d: 0x009e,	#  ZERO WIDTH JOINER
-    0x200e: 0x00fd,	#  LEFT-TO-RIGHT MARK
-    0x200f: 0x00fe,	#  RIGHT-TO-LEFT MARK
-    0x2013: 0x0096,	#  EN DASH
-    0x2014: 0x0097,	#  EM DASH
-    0x2018: 0x0091,	#  LEFT SINGLE QUOTATION MARK
-    0x2019: 0x0092,	#  RIGHT SINGLE QUOTATION MARK
-    0x201a: 0x0082,	#  SINGLE LOW-9 QUOTATION MARK
-    0x201c: 0x0093,	#  LEFT DOUBLE QUOTATION MARK
-    0x201d: 0x0094,	#  RIGHT DOUBLE QUOTATION MARK
-    0x201e: 0x0084,	#  DOUBLE LOW-9 QUOTATION MARK
-    0x2020: 0x0086,	#  DAGGER
-    0x2021: 0x0087,	#  DOUBLE DAGGER
-    0x2022: 0x0095,	#  BULLET
-    0x2026: 0x0085,	#  HORIZONTAL ELLIPSIS
-    0x2030: 0x0089,	#  PER MILLE SIGN
-    0x2039: 0x008b,	#  SINGLE LEFT-POINTING ANGLE QUOTATION MARK
-    0x203a: 0x009b,	#  SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
-    0x20ac: 0x0080,	#  EURO SIGN
-    0x2122: 0x0099,	#  TRADE MARK SIGN
+    0x0000: 0x00,	#  NULL
+    0x0001: 0x01,	#  START OF HEADING
+    0x0002: 0x02,	#  START OF TEXT
+    0x0003: 0x03,	#  END OF TEXT
+    0x0004: 0x04,	#  END OF TRANSMISSION
+    0x0005: 0x05,	#  ENQUIRY
+    0x0006: 0x06,	#  ACKNOWLEDGE
+    0x0007: 0x07,	#  BELL
+    0x0008: 0x08,	#  BACKSPACE
+    0x0009: 0x09,	#  HORIZONTAL TABULATION
+    0x000a: 0x0a,	#  LINE FEED
+    0x000b: 0x0b,	#  VERTICAL TABULATION
+    0x000c: 0x0c,	#  FORM FEED
+    0x000d: 0x0d,	#  CARRIAGE RETURN
+    0x000e: 0x0e,	#  SHIFT OUT
+    0x000f: 0x0f,	#  SHIFT IN
+    0x0010: 0x10,	#  DATA LINK ESCAPE
+    0x0011: 0x11,	#  DEVICE CONTROL ONE
+    0x0012: 0x12,	#  DEVICE CONTROL TWO
+    0x0013: 0x13,	#  DEVICE CONTROL THREE
+    0x0014: 0x14,	#  DEVICE CONTROL FOUR
+    0x0015: 0x15,	#  NEGATIVE ACKNOWLEDGE
+    0x0016: 0x16,	#  SYNCHRONOUS IDLE
+    0x0017: 0x17,	#  END OF TRANSMISSION BLOCK
+    0x0018: 0x18,	#  CANCEL
+    0x0019: 0x19,	#  END OF MEDIUM
+    0x001a: 0x1a,	#  SUBSTITUTE
+    0x001b: 0x1b,	#  ESCAPE
+    0x001c: 0x1c,	#  FILE SEPARATOR
+    0x001d: 0x1d,	#  GROUP SEPARATOR
+    0x001e: 0x1e,	#  RECORD SEPARATOR
+    0x001f: 0x1f,	#  UNIT SEPARATOR
+    0x0020: 0x20,	#  SPACE
+    0x0021: 0x21,	#  EXCLAMATION MARK
+    0x0022: 0x22,	#  QUOTATION MARK
+    0x0023: 0x23,	#  NUMBER SIGN
+    0x0024: 0x24,	#  DOLLAR SIGN
+    0x0025: 0x25,	#  PERCENT SIGN
+    0x0026: 0x26,	#  AMPERSAND
+    0x0027: 0x27,	#  APOSTROPHE
+    0x0028: 0x28,	#  LEFT PARENTHESIS
+    0x0029: 0x29,	#  RIGHT PARENTHESIS
+    0x002a: 0x2a,	#  ASTERISK
+    0x002b: 0x2b,	#  PLUS SIGN
+    0x002c: 0x2c,	#  COMMA
+    0x002d: 0x2d,	#  HYPHEN-MINUS
+    0x002e: 0x2e,	#  FULL STOP
+    0x002f: 0x2f,	#  SOLIDUS
+    0x0030: 0x30,	#  DIGIT ZERO
+    0x0031: 0x31,	#  DIGIT ONE
+    0x0032: 0x32,	#  DIGIT TWO
+    0x0033: 0x33,	#  DIGIT THREE
+    0x0034: 0x34,	#  DIGIT FOUR
+    0x0035: 0x35,	#  DIGIT FIVE
+    0x0036: 0x36,	#  DIGIT SIX
+    0x0037: 0x37,	#  DIGIT SEVEN
+    0x0038: 0x38,	#  DIGIT EIGHT
+    0x0039: 0x39,	#  DIGIT NINE
+    0x003a: 0x3a,	#  COLON
+    0x003b: 0x3b,	#  SEMICOLON
+    0x003c: 0x3c,	#  LESS-THAN SIGN
+    0x003d: 0x3d,	#  EQUALS SIGN
+    0x003e: 0x3e,	#  GREATER-THAN SIGN
+    0x003f: 0x3f,	#  QUESTION MARK
+    0x0040: 0x40,	#  COMMERCIAL AT
+    0x0041: 0x41,	#  LATIN CAPITAL LETTER A
+    0x0042: 0x42,	#  LATIN CAPITAL LETTER B
+    0x0043: 0x43,	#  LATIN CAPITAL LETTER C
+    0x0044: 0x44,	#  LATIN CAPITAL LETTER D
+    0x0045: 0x45,	#  LATIN CAPITAL LETTER E
+    0x0046: 0x46,	#  LATIN CAPITAL LETTER F
+    0x0047: 0x47,	#  LATIN CAPITAL LETTER G
+    0x0048: 0x48,	#  LATIN CAPITAL LETTER H
+    0x0049: 0x49,	#  LATIN CAPITAL LETTER I
+    0x004a: 0x4a,	#  LATIN CAPITAL LETTER J
+    0x004b: 0x4b,	#  LATIN CAPITAL LETTER K
+    0x004c: 0x4c,	#  LATIN CAPITAL LETTER L
+    0x004d: 0x4d,	#  LATIN CAPITAL LETTER M
+    0x004e: 0x4e,	#  LATIN CAPITAL LETTER N
+    0x004f: 0x4f,	#  LATIN CAPITAL LETTER O
+    0x0050: 0x50,	#  LATIN CAPITAL LETTER P
+    0x0051: 0x51,	#  LATIN CAPITAL LETTER Q
+    0x0052: 0x52,	#  LATIN CAPITAL LETTER R
+    0x0053: 0x53,	#  LATIN CAPITAL LETTER S
+    0x0054: 0x54,	#  LATIN CAPITAL LETTER T
+    0x0055: 0x55,	#  LATIN CAPITAL LETTER U
+    0x0056: 0x56,	#  LATIN CAPITAL LETTER V
+    0x0057: 0x57,	#  LATIN CAPITAL LETTER W
+    0x0058: 0x58,	#  LATIN CAPITAL LETTER X
+    0x0059: 0x59,	#  LATIN CAPITAL LETTER Y
+    0x005a: 0x5a,	#  LATIN CAPITAL LETTER Z
+    0x005b: 0x5b,	#  LEFT SQUARE BRACKET
+    0x005c: 0x5c,	#  REVERSE SOLIDUS
+    0x005d: 0x5d,	#  RIGHT SQUARE BRACKET
+    0x005e: 0x5e,	#  CIRCUMFLEX ACCENT
+    0x005f: 0x5f,	#  LOW LINE
+    0x0060: 0x60,	#  GRAVE ACCENT
+    0x0061: 0x61,	#  LATIN SMALL LETTER A
+    0x0062: 0x62,	#  LATIN SMALL LETTER B
+    0x0063: 0x63,	#  LATIN SMALL LETTER C
+    0x0064: 0x64,	#  LATIN SMALL LETTER D
+    0x0065: 0x65,	#  LATIN SMALL LETTER E
+    0x0066: 0x66,	#  LATIN SMALL LETTER F
+    0x0067: 0x67,	#  LATIN SMALL LETTER G
+    0x0068: 0x68,	#  LATIN SMALL LETTER H
+    0x0069: 0x69,	#  LATIN SMALL LETTER I
+    0x006a: 0x6a,	#  LATIN SMALL LETTER J
+    0x006b: 0x6b,	#  LATIN SMALL LETTER K
+    0x006c: 0x6c,	#  LATIN SMALL LETTER L
+    0x006d: 0x6d,	#  LATIN SMALL LETTER M
+    0x006e: 0x6e,	#  LATIN SMALL LETTER N
+    0x006f: 0x6f,	#  LATIN SMALL LETTER O
+    0x0070: 0x70,	#  LATIN SMALL LETTER P
+    0x0071: 0x71,	#  LATIN SMALL LETTER Q
+    0x0072: 0x72,	#  LATIN SMALL LETTER R
+    0x0073: 0x73,	#  LATIN SMALL LETTER S
+    0x0074: 0x74,	#  LATIN SMALL LETTER T
+    0x0075: 0x75,	#  LATIN SMALL LETTER U
+    0x0076: 0x76,	#  LATIN SMALL LETTER V
+    0x0077: 0x77,	#  LATIN SMALL LETTER W
+    0x0078: 0x78,	#  LATIN SMALL LETTER X
+    0x0079: 0x79,	#  LATIN SMALL LETTER Y
+    0x007a: 0x7a,	#  LATIN SMALL LETTER Z
+    0x007b: 0x7b,	#  LEFT CURLY BRACKET
+    0x007c: 0x7c,	#  VERTICAL LINE
+    0x007d: 0x7d,	#  RIGHT CURLY BRACKET
+    0x007e: 0x7e,	#  TILDE
+    0x007f: 0x7f,	#  DELETE
+    0x00a0: 0xa0,	#  NO-BREAK SPACE
+    0x00a2: 0xa2,	#  CENT SIGN
+    0x00a3: 0xa3,	#  POUND SIGN
+    0x00a4: 0xa4,	#  CURRENCY SIGN
+    0x00a5: 0xa5,	#  YEN SIGN
+    0x00a6: 0xa6,	#  BROKEN BAR
+    0x00a7: 0xa7,	#  SECTION SIGN
+    0x00a8: 0xa8,	#  DIAERESIS
+    0x00a9: 0xa9,	#  COPYRIGHT SIGN
+    0x00ab: 0xab,	#  LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
+    0x00ac: 0xac,	#  NOT SIGN
+    0x00ad: 0xad,	#  SOFT HYPHEN
+    0x00ae: 0xae,	#  REGISTERED SIGN
+    0x00af: 0xaf,	#  MACRON
+    0x00b0: 0xb0,	#  DEGREE SIGN
+    0x00b1: 0xb1,	#  PLUS-MINUS SIGN
+    0x00b2: 0xb2,	#  SUPERSCRIPT TWO
+    0x00b3: 0xb3,	#  SUPERSCRIPT THREE
+    0x00b4: 0xb4,	#  ACUTE ACCENT
+    0x00b5: 0xb5,	#  MICRO SIGN
+    0x00b6: 0xb6,	#  PILCROW SIGN
+    0x00b7: 0xb7,	#  MIDDLE DOT
+    0x00b8: 0xb8,	#  CEDILLA
+    0x00b9: 0xb9,	#  SUPERSCRIPT ONE
+    0x00bb: 0xbb,	#  RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
+    0x00bc: 0xbc,	#  VULGAR FRACTION ONE QUARTER
+    0x00bd: 0xbd,	#  VULGAR FRACTION ONE HALF
+    0x00be: 0xbe,	#  VULGAR FRACTION THREE QUARTERS
+    0x00d7: 0xd7,	#  MULTIPLICATION SIGN
+    0x00e0: 0xe0,	#  LATIN SMALL LETTER A WITH GRAVE
+    0x00e2: 0xe2,	#  LATIN SMALL LETTER A WITH CIRCUMFLEX
+    0x00e7: 0xe7,	#  LATIN SMALL LETTER C WITH CEDILLA
+    0x00e8: 0xe8,	#  LATIN SMALL LETTER E WITH GRAVE
+    0x00e9: 0xe9,	#  LATIN SMALL LETTER E WITH ACUTE
+    0x00ea: 0xea,	#  LATIN SMALL LETTER E WITH CIRCUMFLEX
+    0x00eb: 0xeb,	#  LATIN SMALL LETTER E WITH DIAERESIS
+    0x00ee: 0xee,	#  LATIN SMALL LETTER I WITH CIRCUMFLEX
+    0x00ef: 0xef,	#  LATIN SMALL LETTER I WITH DIAERESIS
+    0x00f4: 0xf4,	#  LATIN SMALL LETTER O WITH CIRCUMFLEX
+    0x00f7: 0xf7,	#  DIVISION SIGN
+    0x00f9: 0xf9,	#  LATIN SMALL LETTER U WITH GRAVE
+    0x00fb: 0xfb,	#  LATIN SMALL LETTER U WITH CIRCUMFLEX
+    0x00fc: 0xfc,	#  LATIN SMALL LETTER U WITH DIAERESIS
+    0x0152: 0x8c,	#  LATIN CAPITAL LIGATURE OE
+    0x0153: 0x9c,	#  LATIN SMALL LIGATURE OE
+    0x0192: 0x83,	#  LATIN SMALL LETTER F WITH HOOK
+    0x02c6: 0x88,	#  MODIFIER LETTER CIRCUMFLEX ACCENT
+    0x060c: 0xa1,	#  ARABIC COMMA
+    0x061b: 0xba,	#  ARABIC SEMICOLON
+    0x061f: 0xbf,	#  ARABIC QUESTION MARK
+    0x0621: 0xc1,	#  ARABIC LETTER HAMZA
+    0x0622: 0xc2,	#  ARABIC LETTER ALEF WITH MADDA ABOVE
+    0x0623: 0xc3,	#  ARABIC LETTER ALEF WITH HAMZA ABOVE
+    0x0624: 0xc4,	#  ARABIC LETTER WAW WITH HAMZA ABOVE
+    0x0625: 0xc5,	#  ARABIC LETTER ALEF WITH HAMZA BELOW
+    0x0626: 0xc6,	#  ARABIC LETTER YEH WITH HAMZA ABOVE
+    0x0627: 0xc7,	#  ARABIC LETTER ALEF
+    0x0628: 0xc8,	#  ARABIC LETTER BEH
+    0x0629: 0xc9,	#  ARABIC LETTER TEH MARBUTA
+    0x062a: 0xca,	#  ARABIC LETTER TEH
+    0x062b: 0xcb,	#  ARABIC LETTER THEH
+    0x062c: 0xcc,	#  ARABIC LETTER JEEM
+    0x062d: 0xcd,	#  ARABIC LETTER HAH
+    0x062e: 0xce,	#  ARABIC LETTER KHAH
+    0x062f: 0xcf,	#  ARABIC LETTER DAL
+    0x0630: 0xd0,	#  ARABIC LETTER THAL
+    0x0631: 0xd1,	#  ARABIC LETTER REH
+    0x0632: 0xd2,	#  ARABIC LETTER ZAIN
+    0x0633: 0xd3,	#  ARABIC LETTER SEEN
+    0x0634: 0xd4,	#  ARABIC LETTER SHEEN
+    0x0635: 0xd5,	#  ARABIC LETTER SAD
+    0x0636: 0xd6,	#  ARABIC LETTER DAD
+    0x0637: 0xd8,	#  ARABIC LETTER TAH
+    0x0638: 0xd9,	#  ARABIC LETTER ZAH
+    0x0639: 0xda,	#  ARABIC LETTER AIN
+    0x063a: 0xdb,	#  ARABIC LETTER GHAIN
+    0x0640: 0xdc,	#  ARABIC TATWEEL
+    0x0641: 0xdd,	#  ARABIC LETTER FEH
+    0x0642: 0xde,	#  ARABIC LETTER QAF
+    0x0643: 0xdf,	#  ARABIC LETTER KAF
+    0x0644: 0xe1,	#  ARABIC LETTER LAM
+    0x0645: 0xe3,	#  ARABIC LETTER MEEM
+    0x0646: 0xe4,	#  ARABIC LETTER NOON
+    0x0647: 0xe5,	#  ARABIC LETTER HEH
+    0x0648: 0xe6,	#  ARABIC LETTER WAW
+    0x0649: 0xec,	#  ARABIC LETTER ALEF MAKSURA
+    0x064a: 0xed,	#  ARABIC LETTER YEH
+    0x064b: 0xf0,	#  ARABIC FATHATAN
+    0x064c: 0xf1,	#  ARABIC DAMMATAN
+    0x064d: 0xf2,	#  ARABIC KASRATAN
+    0x064e: 0xf3,	#  ARABIC FATHA
+    0x064f: 0xf5,	#  ARABIC DAMMA
+    0x0650: 0xf6,	#  ARABIC KASRA
+    0x0651: 0xf8,	#  ARABIC SHADDA
+    0x0652: 0xfa,	#  ARABIC SUKUN
+    0x0679: 0x8a,	#  ARABIC LETTER TTEH
+    0x067e: 0x81,	#  ARABIC LETTER PEH
+    0x0686: 0x8d,	#  ARABIC LETTER TCHEH
+    0x0688: 0x8f,	#  ARABIC LETTER DDAL
+    0x0691: 0x9a,	#  ARABIC LETTER RREH
+    0x0698: 0x8e,	#  ARABIC LETTER JEH
+    0x06a9: 0x98,	#  ARABIC LETTER KEHEH
+    0x06af: 0x90,	#  ARABIC LETTER GAF
+    0x06ba: 0x9f,	#  ARABIC LETTER NOON GHUNNA
+    0x06be: 0xaa,	#  ARABIC LETTER HEH DOACHASHMEE
+    0x06c1: 0xc0,	#  ARABIC LETTER HEH GOAL
+    0x06d2: 0xff,	#  ARABIC LETTER YEH BARREE
+    0x200c: 0x9d,	#  ZERO WIDTH NON-JOINER
+    0x200d: 0x9e,	#  ZERO WIDTH JOINER
+    0x200e: 0xfd,	#  LEFT-TO-RIGHT MARK
+    0x200f: 0xfe,	#  RIGHT-TO-LEFT MARK
+    0x2013: 0x96,	#  EN DASH
+    0x2014: 0x97,	#  EM DASH
+    0x2018: 0x91,	#  LEFT SINGLE QUOTATION MARK
+    0x2019: 0x92,	#  RIGHT SINGLE QUOTATION MARK
+    0x201a: 0x82,	#  SINGLE LOW-9 QUOTATION MARK
+    0x201c: 0x93,	#  LEFT DOUBLE QUOTATION MARK
+    0x201d: 0x94,	#  RIGHT DOUBLE QUOTATION MARK
+    0x201e: 0x84,	#  DOUBLE LOW-9 QUOTATION MARK
+    0x2020: 0x86,	#  DAGGER
+    0x2021: 0x87,	#  DOUBLE DAGGER
+    0x2022: 0x95,	#  BULLET
+    0x2026: 0x85,	#  HORIZONTAL ELLIPSIS
+    0x2030: 0x89,	#  PER MILLE SIGN
+    0x2039: 0x8b,	#  SINGLE LEFT-POINTING ANGLE QUOTATION MARK
+    0x203a: 0x9b,	#  SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
+    0x20ac: 0x80,	#  EURO SIGN
+    0x2122: 0x99,	#  TRADE MARK SIGN
 }
\ No newline at end of file