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, 644 deletions

diff --git a/Lib/encodings/mac_greek.py b/Lib/encodings/mac_greek.py
index c4fe120..d0757c1 100644
--- a/Lib/encodings/mac_greek.py
+++ b/Lib/encodings/mac_greek.py
@@ -1,4 +1,4 @@
-""" Python Character Mapping Codec generated from 'VENDORS/APPLE/GREEK.TXT' with gencodec.py.
+""" Python Character Mapping Codec generated from 'MAPPINGS/VENDORS/APPLE/GREEK.TXT' with gencodec.py.
 
 """#"
 
@@ -28,656 +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: 0x00c4,	#  LATIN CAPITAL LETTER A WITH DIAERESIS
-    0x0081: 0x00b9,	#  SUPERSCRIPT ONE
-    0x0082: 0x00b2,	#  SUPERSCRIPT TWO
-    0x0083: 0x00c9,	#  LATIN CAPITAL LETTER E WITH ACUTE
-    0x0084: 0x00b3,	#  SUPERSCRIPT THREE
-    0x0085: 0x00d6,	#  LATIN CAPITAL LETTER O WITH DIAERESIS
-    0x0086: 0x00dc,	#  LATIN CAPITAL LETTER U WITH DIAERESIS
-    0x0087: 0x0385,	#  GREEK DIALYTIKA TONOS
-    0x0088: 0x00e0,	#  LATIN SMALL LETTER A WITH GRAVE
-    0x0089: 0x00e2,	#  LATIN SMALL LETTER A WITH CIRCUMFLEX
-    0x008a: 0x00e4,	#  LATIN SMALL LETTER A WITH DIAERESIS
-    0x008b: 0x0384,	#  GREEK TONOS
-    0x008c: 0x00a8,	#  DIAERESIS
-    0x008d: 0x00e7,	#  LATIN SMALL LETTER C WITH CEDILLA
-    0x008e: 0x00e9,	#  LATIN SMALL LETTER E WITH ACUTE
-    0x008f: 0x00e8,	#  LATIN SMALL LETTER E WITH GRAVE
-    0x0090: 0x00ea,	#  LATIN SMALL LETTER E WITH CIRCUMFLEX
-    0x0091: 0x00eb,	#  LATIN SMALL LETTER E WITH DIAERESIS
-    0x0092: 0x00a3,	#  POUND SIGN
-    0x0093: 0x2122,	#  TRADE MARK SIGN
-    0x0094: 0x00ee,	#  LATIN SMALL LETTER I WITH CIRCUMFLEX
-    0x0095: 0x00ef,	#  LATIN SMALL LETTER I WITH DIAERESIS
-    0x0096: 0x2022,	#  BULLET
-    0x0097: 0x00bd,	#  VULGAR FRACTION ONE HALF
-    0x0098: 0x2030,	#  PER MILLE SIGN
-    0x0099: 0x00f4,	#  LATIN SMALL LETTER O WITH CIRCUMFLEX
-    0x009a: 0x00f6,	#  LATIN SMALL LETTER O WITH DIAERESIS
-    0x009b: 0x00a6,	#  BROKEN BAR
-    0x009c: 0x20ac,	#  EURO SIGN # before Mac OS 9.2.2, was SOFT HYPHEN
-    0x009d: 0x00f9,	#  LATIN SMALL LETTER U WITH GRAVE
-    0x009e: 0x00fb,	#  LATIN SMALL LETTER U WITH CIRCUMFLEX
-    0x009f: 0x00fc,	#  LATIN SMALL LETTER U WITH DIAERESIS
-    0x00a0: 0x2020,	#  DAGGER
-    0x00a1: 0x0393,	#  GREEK CAPITAL LETTER GAMMA
-    0x00a2: 0x0394,	#  GREEK CAPITAL LETTER DELTA
-    0x00a3: 0x0398,	#  GREEK CAPITAL LETTER THETA
-    0x00a4: 0x039b,	#  GREEK CAPITAL LETTER LAMDA
-    0x00a5: 0x039e,	#  GREEK CAPITAL LETTER XI
-    0x00a6: 0x03a0,	#  GREEK CAPITAL LETTER PI
-    0x00a7: 0x00df,	#  LATIN SMALL LETTER SHARP S
-    0x00a8: 0x00ae,	#  REGISTERED SIGN
-    0x00aa: 0x03a3,	#  GREEK CAPITAL LETTER SIGMA
-    0x00ab: 0x03aa,	#  GREEK CAPITAL LETTER IOTA WITH DIALYTIKA
-    0x00ac: 0x00a7,	#  SECTION SIGN
-    0x00ad: 0x2260,	#  NOT EQUAL TO
-    0x00ae: 0x00b0,	#  DEGREE SIGN
-    0x00af: 0x00b7,	#  MIDDLE DOT
-    0x00b0: 0x0391,	#  GREEK CAPITAL LETTER ALPHA
-    0x00b2: 0x2264,	#  LESS-THAN OR EQUAL TO
-    0x00b3: 0x2265,	#  GREATER-THAN OR EQUAL TO
-    0x00b4: 0x00a5,	#  YEN SIGN
-    0x00b5: 0x0392,	#  GREEK CAPITAL LETTER BETA
-    0x00b6: 0x0395,	#  GREEK CAPITAL LETTER EPSILON
-    0x00b7: 0x0396,	#  GREEK CAPITAL LETTER ZETA
-    0x00b8: 0x0397,	#  GREEK CAPITAL LETTER ETA
-    0x00b9: 0x0399,	#  GREEK CAPITAL LETTER IOTA
-    0x00ba: 0x039a,	#  GREEK CAPITAL LETTER KAPPA
-    0x00bb: 0x039c,	#  GREEK CAPITAL LETTER MU
-    0x00bc: 0x03a6,	#  GREEK CAPITAL LETTER PHI
-    0x00bd: 0x03ab,	#  GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA
-    0x00be: 0x03a8,	#  GREEK CAPITAL LETTER PSI
-    0x00bf: 0x03a9,	#  GREEK CAPITAL LETTER OMEGA
-    0x00c0: 0x03ac,	#  GREEK SMALL LETTER ALPHA WITH TONOS
-    0x00c1: 0x039d,	#  GREEK CAPITAL LETTER NU
-    0x00c2: 0x00ac,	#  NOT SIGN
-    0x00c3: 0x039f,	#  GREEK CAPITAL LETTER OMICRON
-    0x00c4: 0x03a1,	#  GREEK CAPITAL LETTER RHO
-    0x00c5: 0x2248,	#  ALMOST EQUAL TO
-    0x00c6: 0x03a4,	#  GREEK CAPITAL LETTER TAU
-    0x00c7: 0x00ab,	#  LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
-    0x00c8: 0x00bb,	#  RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
-    0x00c9: 0x2026,	#  HORIZONTAL ELLIPSIS
-    0x00ca: 0x00a0,	#  NO-BREAK SPACE
-    0x00cb: 0x03a5,	#  GREEK CAPITAL LETTER UPSILON
-    0x00cc: 0x03a7,	#  GREEK CAPITAL LETTER CHI
-    0x00cd: 0x0386,	#  GREEK CAPITAL LETTER ALPHA WITH TONOS
-    0x00ce: 0x0388,	#  GREEK CAPITAL LETTER EPSILON WITH TONOS
-    0x00cf: 0x0153,	#  LATIN SMALL LIGATURE OE
-    0x00d0: 0x2013,	#  EN DASH
-    0x00d1: 0x2015,	#  HORIZONTAL BAR
-    0x00d2: 0x201c,	#  LEFT DOUBLE QUOTATION MARK
-    0x00d3: 0x201d,	#  RIGHT DOUBLE QUOTATION MARK
-    0x00d4: 0x2018,	#  LEFT SINGLE QUOTATION MARK
-    0x00d5: 0x2019,	#  RIGHT SINGLE QUOTATION MARK
-    0x00d6: 0x00f7,	#  DIVISION SIGN
-    0x00d7: 0x0389,	#  GREEK CAPITAL LETTER ETA WITH TONOS
-    0x00d8: 0x038a,	#  GREEK CAPITAL LETTER IOTA WITH TONOS
-    0x00d9: 0x038c,	#  GREEK CAPITAL LETTER OMICRON WITH TONOS
-    0x00da: 0x038e,	#  GREEK CAPITAL LETTER UPSILON WITH TONOS
-    0x00db: 0x03ad,	#  GREEK SMALL LETTER EPSILON WITH TONOS
-    0x00dc: 0x03ae,	#  GREEK SMALL LETTER ETA WITH TONOS
-    0x00dd: 0x03af,	#  GREEK SMALL LETTER IOTA WITH TONOS
-    0x00de: 0x03cc,	#  GREEK SMALL LETTER OMICRON WITH TONOS
-    0x00df: 0x038f,	#  GREEK CAPITAL LETTER OMEGA WITH TONOS
-    0x00e0: 0x03cd,	#  GREEK SMALL LETTER UPSILON WITH TONOS
-    0x00e1: 0x03b1,	#  GREEK SMALL LETTER ALPHA
-    0x00e2: 0x03b2,	#  GREEK SMALL LETTER BETA
-    0x00e3: 0x03c8,	#  GREEK SMALL LETTER PSI
-    0x00e4: 0x03b4,	#  GREEK SMALL LETTER DELTA
-    0x00e5: 0x03b5,	#  GREEK SMALL LETTER EPSILON
-    0x00e6: 0x03c6,	#  GREEK SMALL LETTER PHI
-    0x00e7: 0x03b3,	#  GREEK SMALL LETTER GAMMA
-    0x00e8: 0x03b7,	#  GREEK SMALL LETTER ETA
-    0x00e9: 0x03b9,	#  GREEK SMALL LETTER IOTA
-    0x00ea: 0x03be,	#  GREEK SMALL LETTER XI
-    0x00eb: 0x03ba,	#  GREEK SMALL LETTER KAPPA
-    0x00ec: 0x03bb,	#  GREEK SMALL LETTER LAMDA
-    0x00ed: 0x03bc,	#  GREEK SMALL LETTER MU
-    0x00ee: 0x03bd,	#  GREEK SMALL LETTER NU
-    0x00ef: 0x03bf,	#  GREEK SMALL LETTER OMICRON
-    0x00f0: 0x03c0,	#  GREEK SMALL LETTER PI
-    0x00f1: 0x03ce,	#  GREEK SMALL LETTER OMEGA WITH TONOS
-    0x00f2: 0x03c1,	#  GREEK SMALL LETTER RHO
-    0x00f3: 0x03c3,	#  GREEK SMALL LETTER SIGMA
-    0x00f4: 0x03c4,	#  GREEK SMALL LETTER TAU
-    0x00f5: 0x03b8,	#  GREEK SMALL LETTER THETA
-    0x00f6: 0x03c9,	#  GREEK SMALL LETTER OMEGA
-    0x00f7: 0x03c2,	#  GREEK SMALL LETTER FINAL SIGMA
-    0x00f8: 0x03c7,	#  GREEK SMALL LETTER CHI
-    0x00f9: 0x03c5,	#  GREEK SMALL LETTER UPSILON
-    0x00fa: 0x03b6,	#  GREEK SMALL LETTER ZETA
-    0x00fb: 0x03ca,	#  GREEK SMALL LETTER IOTA WITH DIALYTIKA
-    0x00fc: 0x03cb,	#  GREEK SMALL LETTER UPSILON WITH DIALYTIKA
-    0x00fd: 0x0390,	#  GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
-    0x00fe: 0x03b0,	#  GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS
-    0x00ff: 0x00ad,	#  SOFT HYPHEN # before Mac OS 9.2.2, was undefined
-})
 
 ### Decoding Table
 
 decoding_table = (
-    u'\x00'	#  0x0000 -> CONTROL CHARACTER
-    u'\x01'	#  0x0001 -> CONTROL CHARACTER
-    u'\x02'	#  0x0002 -> CONTROL CHARACTER
-    u'\x03'	#  0x0003 -> CONTROL CHARACTER
-    u'\x04'	#  0x0004 -> CONTROL CHARACTER
-    u'\x05'	#  0x0005 -> CONTROL CHARACTER
-    u'\x06'	#  0x0006 -> CONTROL CHARACTER
-    u'\x07'	#  0x0007 -> CONTROL CHARACTER
-    u'\x08'	#  0x0008 -> CONTROL CHARACTER
-    u'\t'	#  0x0009 -> CONTROL CHARACTER
-    u'\n'	#  0x000a -> CONTROL CHARACTER
-    u'\x0b'	#  0x000b -> CONTROL CHARACTER
-    u'\x0c'	#  0x000c -> CONTROL CHARACTER
-    u'\r'	#  0x000d -> CONTROL CHARACTER
-    u'\x0e'	#  0x000e -> CONTROL CHARACTER
-    u'\x0f'	#  0x000f -> CONTROL CHARACTER
-    u'\x10'	#  0x0010 -> CONTROL CHARACTER
-    u'\x11'	#  0x0011 -> CONTROL CHARACTER
-    u'\x12'	#  0x0012 -> CONTROL CHARACTER
-    u'\x13'	#  0x0013 -> CONTROL CHARACTER
-    u'\x14'	#  0x0014 -> CONTROL CHARACTER
-    u'\x15'	#  0x0015 -> CONTROL CHARACTER
-    u'\x16'	#  0x0016 -> CONTROL CHARACTER
-    u'\x17'	#  0x0017 -> CONTROL CHARACTER
-    u'\x18'	#  0x0018 -> CONTROL CHARACTER
-    u'\x19'	#  0x0019 -> CONTROL CHARACTER
-    u'\x1a'	#  0x001a -> CONTROL CHARACTER
-    u'\x1b'	#  0x001b -> CONTROL CHARACTER
-    u'\x1c'	#  0x001c -> CONTROL CHARACTER
-    u'\x1d'	#  0x001d -> CONTROL CHARACTER
-    u'\x1e'	#  0x001e -> CONTROL CHARACTER
-    u'\x1f'	#  0x001f -> CONTROL CHARACTER
-    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 -> CONTROL CHARACTER
-    u'\xc4'	#  0x0080 -> LATIN CAPITAL LETTER A WITH DIAERESIS
-    u'\xb9'	#  0x0081 -> SUPERSCRIPT ONE
-    u'\xb2'	#  0x0082 -> SUPERSCRIPT TWO
-    u'\xc9'	#  0x0083 -> LATIN CAPITAL LETTER E WITH ACUTE
-    u'\xb3'	#  0x0084 -> SUPERSCRIPT THREE
-    u'\xd6'	#  0x0085 -> LATIN CAPITAL LETTER O WITH DIAERESIS
-    u'\xdc'	#  0x0086 -> LATIN CAPITAL LETTER U WITH DIAERESIS
-    u'\u0385'	#  0x0087 -> GREEK DIALYTIKA TONOS
-    u'\xe0'	#  0x0088 -> LATIN SMALL LETTER A WITH GRAVE
-    u'\xe2'	#  0x0089 -> LATIN SMALL LETTER A WITH CIRCUMFLEX
-    u'\xe4'	#  0x008a -> LATIN SMALL LETTER A WITH DIAERESIS
-    u'\u0384'	#  0x008b -> GREEK TONOS
-    u'\xa8'	#  0x008c -> DIAERESIS
-    u'\xe7'	#  0x008d -> LATIN SMALL LETTER C WITH CEDILLA
-    u'\xe9'	#  0x008e -> LATIN SMALL LETTER E WITH ACUTE
-    u'\xe8'	#  0x008f -> LATIN SMALL LETTER E WITH GRAVE
-    u'\xea'	#  0x0090 -> LATIN SMALL LETTER E WITH CIRCUMFLEX
-    u'\xeb'	#  0x0091 -> LATIN SMALL LETTER E WITH DIAERESIS
-    u'\xa3'	#  0x0092 -> POUND SIGN
-    u'\u2122'	#  0x0093 -> TRADE MARK SIGN
-    u'\xee'	#  0x0094 -> LATIN SMALL LETTER I WITH CIRCUMFLEX
-    u'\xef'	#  0x0095 -> LATIN SMALL LETTER I WITH DIAERESIS
-    u'\u2022'	#  0x0096 -> BULLET
-    u'\xbd'	#  0x0097 -> VULGAR FRACTION ONE HALF
-    u'\u2030'	#  0x0098 -> PER MILLE SIGN
-    u'\xf4'	#  0x0099 -> LATIN SMALL LETTER O WITH CIRCUMFLEX
-    u'\xf6'	#  0x009a -> LATIN SMALL LETTER O WITH DIAERESIS
-    u'\xa6'	#  0x009b -> BROKEN BAR
-    u'\u20ac'	#  0x009c -> EURO SIGN # before Mac OS 9.2.2, was SOFT HYPHEN
-    u'\xf9'	#  0x009d -> LATIN SMALL LETTER U WITH GRAVE
-    u'\xfb'	#  0x009e -> LATIN SMALL LETTER U WITH CIRCUMFLEX
-    u'\xfc'	#  0x009f -> LATIN SMALL LETTER U WITH DIAERESIS
-    u'\u2020'	#  0x00a0 -> DAGGER
-    u'\u0393'	#  0x00a1 -> GREEK CAPITAL LETTER GAMMA
-    u'\u0394'	#  0x00a2 -> GREEK CAPITAL LETTER DELTA
-    u'\u0398'	#  0x00a3 -> GREEK CAPITAL LETTER THETA
-    u'\u039b'	#  0x00a4 -> GREEK CAPITAL LETTER LAMDA
-    u'\u039e'	#  0x00a5 -> GREEK CAPITAL LETTER XI
-    u'\u03a0'	#  0x00a6 -> GREEK CAPITAL LETTER PI
-    u'\xdf'	#  0x00a7 -> LATIN SMALL LETTER SHARP S
-    u'\xae'	#  0x00a8 -> REGISTERED SIGN
-    u'\xa9'	#  0x00a9 -> COPYRIGHT SIGN
-    u'\u03a3'	#  0x00aa -> GREEK CAPITAL LETTER SIGMA
-    u'\u03aa'	#  0x00ab -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA
-    u'\xa7'	#  0x00ac -> SECTION SIGN
-    u'\u2260'	#  0x00ad -> NOT EQUAL TO
-    u'\xb0'	#  0x00ae -> DEGREE SIGN
-    u'\xb7'	#  0x00af -> MIDDLE DOT
-    u'\u0391'	#  0x00b0 -> GREEK CAPITAL LETTER ALPHA
-    u'\xb1'	#  0x00b1 -> PLUS-MINUS SIGN
-    u'\u2264'	#  0x00b2 -> LESS-THAN OR EQUAL TO
-    u'\u2265'	#  0x00b3 -> GREATER-THAN OR EQUAL TO
-    u'\xa5'	#  0x00b4 -> YEN SIGN
-    u'\u0392'	#  0x00b5 -> GREEK CAPITAL LETTER BETA
-    u'\u0395'	#  0x00b6 -> GREEK CAPITAL LETTER EPSILON
-    u'\u0396'	#  0x00b7 -> GREEK CAPITAL LETTER ZETA
-    u'\u0397'	#  0x00b8 -> GREEK CAPITAL LETTER ETA
-    u'\u0399'	#  0x00b9 -> GREEK CAPITAL LETTER IOTA
-    u'\u039a'	#  0x00ba -> GREEK CAPITAL LETTER KAPPA
-    u'\u039c'	#  0x00bb -> GREEK CAPITAL LETTER MU
-    u'\u03a6'	#  0x00bc -> GREEK CAPITAL LETTER PHI
-    u'\u03ab'	#  0x00bd -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA
-    u'\u03a8'	#  0x00be -> GREEK CAPITAL LETTER PSI
-    u'\u03a9'	#  0x00bf -> GREEK CAPITAL LETTER OMEGA
-    u'\u03ac'	#  0x00c0 -> GREEK SMALL LETTER ALPHA WITH TONOS
-    u'\u039d'	#  0x00c1 -> GREEK CAPITAL LETTER NU
-    u'\xac'	#  0x00c2 -> NOT SIGN
-    u'\u039f'	#  0x00c3 -> GREEK CAPITAL LETTER OMICRON
-    u'\u03a1'	#  0x00c4 -> GREEK CAPITAL LETTER RHO
-    u'\u2248'	#  0x00c5 -> ALMOST EQUAL TO
-    u'\u03a4'	#  0x00c6 -> GREEK CAPITAL LETTER TAU
-    u'\xab'	#  0x00c7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
-    u'\xbb'	#  0x00c8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
-    u'\u2026'	#  0x00c9 -> HORIZONTAL ELLIPSIS
-    u'\xa0'	#  0x00ca -> NO-BREAK SPACE
-    u'\u03a5'	#  0x00cb -> GREEK CAPITAL LETTER UPSILON
-    u'\u03a7'	#  0x00cc -> GREEK CAPITAL LETTER CHI
-    u'\u0386'	#  0x00cd -> GREEK CAPITAL LETTER ALPHA WITH TONOS
-    u'\u0388'	#  0x00ce -> GREEK CAPITAL LETTER EPSILON WITH TONOS
-    u'\u0153'	#  0x00cf -> LATIN SMALL LIGATURE OE
-    u'\u2013'	#  0x00d0 -> EN DASH
-    u'\u2015'	#  0x00d1 -> HORIZONTAL BAR
-    u'\u201c'	#  0x00d2 -> LEFT DOUBLE QUOTATION MARK
-    u'\u201d'	#  0x00d3 -> RIGHT DOUBLE QUOTATION MARK
-    u'\u2018'	#  0x00d4 -> LEFT SINGLE QUOTATION MARK
-    u'\u2019'	#  0x00d5 -> RIGHT SINGLE QUOTATION MARK
-    u'\xf7'	#  0x00d6 -> DIVISION SIGN
-    u'\u0389'	#  0x00d7 -> GREEK CAPITAL LETTER ETA WITH TONOS
-    u'\u038a'	#  0x00d8 -> GREEK CAPITAL LETTER IOTA WITH TONOS
-    u'\u038c'	#  0x00d9 -> GREEK CAPITAL LETTER OMICRON WITH TONOS
-    u'\u038e'	#  0x00da -> GREEK CAPITAL LETTER UPSILON WITH TONOS
-    u'\u03ad'	#  0x00db -> GREEK SMALL LETTER EPSILON WITH TONOS
-    u'\u03ae'	#  0x00dc -> GREEK SMALL LETTER ETA WITH TONOS
-    u'\u03af'	#  0x00dd -> GREEK SMALL LETTER IOTA WITH TONOS
-    u'\u03cc'	#  0x00de -> GREEK SMALL LETTER OMICRON WITH TONOS
-    u'\u038f'	#  0x00df -> GREEK CAPITAL LETTER OMEGA WITH TONOS
-    u'\u03cd'	#  0x00e0 -> GREEK SMALL LETTER UPSILON WITH TONOS
-    u'\u03b1'	#  0x00e1 -> GREEK SMALL LETTER ALPHA
-    u'\u03b2'	#  0x00e2 -> GREEK SMALL LETTER BETA
-    u'\u03c8'	#  0x00e3 -> GREEK SMALL LETTER PSI
-    u'\u03b4'	#  0x00e4 -> GREEK SMALL LETTER DELTA
-    u'\u03b5'	#  0x00e5 -> GREEK SMALL LETTER EPSILON
-    u'\u03c6'	#  0x00e6 -> GREEK SMALL LETTER PHI
-    u'\u03b3'	#  0x00e7 -> GREEK SMALL LETTER GAMMA
-    u'\u03b7'	#  0x00e8 -> GREEK SMALL LETTER ETA
-    u'\u03b9'	#  0x00e9 -> GREEK SMALL LETTER IOTA
-    u'\u03be'	#  0x00ea -> GREEK SMALL LETTER XI
-    u'\u03ba'	#  0x00eb -> GREEK SMALL LETTER KAPPA
-    u'\u03bb'	#  0x00ec -> GREEK SMALL LETTER LAMDA
-    u'\u03bc'	#  0x00ed -> GREEK SMALL LETTER MU
-    u'\u03bd'	#  0x00ee -> GREEK SMALL LETTER NU
-    u'\u03bf'	#  0x00ef -> GREEK SMALL LETTER OMICRON
-    u'\u03c0'	#  0x00f0 -> GREEK SMALL LETTER PI
-    u'\u03ce'	#  0x00f1 -> GREEK SMALL LETTER OMEGA WITH TONOS
-    u'\u03c1'	#  0x00f2 -> GREEK SMALL LETTER RHO
-    u'\u03c3'	#  0x00f3 -> GREEK SMALL LETTER SIGMA
-    u'\u03c4'	#  0x00f4 -> GREEK SMALL LETTER TAU
-    u'\u03b8'	#  0x00f5 -> GREEK SMALL LETTER THETA
-    u'\u03c9'	#  0x00f6 -> GREEK SMALL LETTER OMEGA
-    u'\u03c2'	#  0x00f7 -> GREEK SMALL LETTER FINAL SIGMA
-    u'\u03c7'	#  0x00f8 -> GREEK SMALL LETTER CHI
-    u'\u03c5'	#  0x00f9 -> GREEK SMALL LETTER UPSILON
-    u'\u03b6'	#  0x00fa -> GREEK SMALL LETTER ZETA
-    u'\u03ca'	#  0x00fb -> GREEK SMALL LETTER IOTA WITH DIALYTIKA
-    u'\u03cb'	#  0x00fc -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA
-    u'\u0390'	#  0x00fd -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
-    u'\u03b0'	#  0x00fe -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS
-    u'\xad'	#  0x00ff -> SOFT HYPHEN # before Mac OS 9.2.2, was undefined
+    u'\x00'	#  0x00 -> CONTROL CHARACTER
+    u'\x01'	#  0x01 -> CONTROL CHARACTER
+    u'\x02'	#  0x02 -> CONTROL CHARACTER
+    u'\x03'	#  0x03 -> CONTROL CHARACTER
+    u'\x04'	#  0x04 -> CONTROL CHARACTER
+    u'\x05'	#  0x05 -> CONTROL CHARACTER
+    u'\x06'	#  0x06 -> CONTROL CHARACTER
+    u'\x07'	#  0x07 -> CONTROL CHARACTER
+    u'\x08'	#  0x08 -> CONTROL CHARACTER
+    u'\t'	#  0x09 -> CONTROL CHARACTER
+    u'\n'	#  0x0a -> CONTROL CHARACTER
+    u'\x0b'	#  0x0b -> CONTROL CHARACTER
+    u'\x0c'	#  0x0c -> CONTROL CHARACTER
+    u'\r'	#  0x0d -> CONTROL CHARACTER
+    u'\x0e'	#  0x0e -> CONTROL CHARACTER
+    u'\x0f'	#  0x0f -> CONTROL CHARACTER
+    u'\x10'	#  0x10 -> CONTROL CHARACTER
+    u'\x11'	#  0x11 -> CONTROL CHARACTER
+    u'\x12'	#  0x12 -> CONTROL CHARACTER
+    u'\x13'	#  0x13 -> CONTROL CHARACTER
+    u'\x14'	#  0x14 -> CONTROL CHARACTER
+    u'\x15'	#  0x15 -> CONTROL CHARACTER
+    u'\x16'	#  0x16 -> CONTROL CHARACTER
+    u'\x17'	#  0x17 -> CONTROL CHARACTER
+    u'\x18'	#  0x18 -> CONTROL CHARACTER
+    u'\x19'	#  0x19 -> CONTROL CHARACTER
+    u'\x1a'	#  0x1a -> CONTROL CHARACTER
+    u'\x1b'	#  0x1b -> CONTROL CHARACTER
+    u'\x1c'	#  0x1c -> CONTROL CHARACTER
+    u'\x1d'	#  0x1d -> CONTROL CHARACTER
+    u'\x1e'	#  0x1e -> CONTROL CHARACTER
+    u'\x1f'	#  0x1f -> CONTROL CHARACTER
+    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 -> CONTROL CHARACTER
+    u'\xc4'	#  0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS
+    u'\xb9'	#  0x81 -> SUPERSCRIPT ONE
+    u'\xb2'	#  0x82 -> SUPERSCRIPT TWO
+    u'\xc9'	#  0x83 -> LATIN CAPITAL LETTER E WITH ACUTE
+    u'\xb3'	#  0x84 -> SUPERSCRIPT THREE
+    u'\xd6'	#  0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS
+    u'\xdc'	#  0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS
+    u'\u0385'	#  0x87 -> GREEK DIALYTIKA TONOS
+    u'\xe0'	#  0x88 -> LATIN SMALL LETTER A WITH GRAVE
+    u'\xe2'	#  0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX
+    u'\xe4'	#  0x8a -> LATIN SMALL LETTER A WITH DIAERESIS
+    u'\u0384'	#  0x8b -> GREEK TONOS
+    u'\xa8'	#  0x8c -> DIAERESIS
+    u'\xe7'	#  0x8d -> LATIN SMALL LETTER C WITH CEDILLA
+    u'\xe9'	#  0x8e -> LATIN SMALL LETTER E WITH ACUTE
+    u'\xe8'	#  0x8f -> LATIN SMALL LETTER E WITH GRAVE
+    u'\xea'	#  0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX
+    u'\xeb'	#  0x91 -> LATIN SMALL LETTER E WITH DIAERESIS
+    u'\xa3'	#  0x92 -> POUND SIGN
+    u'\u2122'	#  0x93 -> TRADE MARK SIGN
+    u'\xee'	#  0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX
+    u'\xef'	#  0x95 -> LATIN SMALL LETTER I WITH DIAERESIS
+    u'\u2022'	#  0x96 -> BULLET
+    u'\xbd'	#  0x97 -> VULGAR FRACTION ONE HALF
+    u'\u2030'	#  0x98 -> PER MILLE SIGN
+    u'\xf4'	#  0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX
+    u'\xf6'	#  0x9a -> LATIN SMALL LETTER O WITH DIAERESIS
+    u'\xa6'	#  0x9b -> BROKEN BAR
+    u'\u20ac'	#  0x9c -> EURO SIGN # before Mac OS 9.2.2, was SOFT HYPHEN
+    u'\xf9'	#  0x9d -> LATIN SMALL LETTER U WITH GRAVE
+    u'\xfb'	#  0x9e -> LATIN SMALL LETTER U WITH CIRCUMFLEX
+    u'\xfc'	#  0x9f -> LATIN SMALL LETTER U WITH DIAERESIS
+    u'\u2020'	#  0xa0 -> DAGGER
+    u'\u0393'	#  0xa1 -> GREEK CAPITAL LETTER GAMMA
+    u'\u0394'	#  0xa2 -> GREEK CAPITAL LETTER DELTA
+    u'\u0398'	#  0xa3 -> GREEK CAPITAL LETTER THETA
+    u'\u039b'	#  0xa4 -> GREEK CAPITAL LETTER LAMDA
+    u'\u039e'	#  0xa5 -> GREEK CAPITAL LETTER XI
+    u'\u03a0'	#  0xa6 -> GREEK CAPITAL LETTER PI
+    u'\xdf'	#  0xa7 -> LATIN SMALL LETTER SHARP S
+    u'\xae'	#  0xa8 -> REGISTERED SIGN
+    u'\xa9'	#  0xa9 -> COPYRIGHT SIGN
+    u'\u03a3'	#  0xaa -> GREEK CAPITAL LETTER SIGMA
+    u'\u03aa'	#  0xab -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA
+    u'\xa7'	#  0xac -> SECTION SIGN
+    u'\u2260'	#  0xad -> NOT EQUAL TO
+    u'\xb0'	#  0xae -> DEGREE SIGN
+    u'\xb7'	#  0xaf -> MIDDLE DOT
+    u'\u0391'	#  0xb0 -> GREEK CAPITAL LETTER ALPHA
+    u'\xb1'	#  0xb1 -> PLUS-MINUS SIGN
+    u'\u2264'	#  0xb2 -> LESS-THAN OR EQUAL TO
+    u'\u2265'	#  0xb3 -> GREATER-THAN OR EQUAL TO
+    u'\xa5'	#  0xb4 -> YEN SIGN
+    u'\u0392'	#  0xb5 -> GREEK CAPITAL LETTER BETA
+    u'\u0395'	#  0xb6 -> GREEK CAPITAL LETTER EPSILON
+    u'\u0396'	#  0xb7 -> GREEK CAPITAL LETTER ZETA
+    u'\u0397'	#  0xb8 -> GREEK CAPITAL LETTER ETA
+    u'\u0399'	#  0xb9 -> GREEK CAPITAL LETTER IOTA
+    u'\u039a'	#  0xba -> GREEK CAPITAL LETTER KAPPA
+    u'\u039c'	#  0xbb -> GREEK CAPITAL LETTER MU
+    u'\u03a6'	#  0xbc -> GREEK CAPITAL LETTER PHI
+    u'\u03ab'	#  0xbd -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA
+    u'\u03a8'	#  0xbe -> GREEK CAPITAL LETTER PSI
+    u'\u03a9'	#  0xbf -> GREEK CAPITAL LETTER OMEGA
+    u'\u03ac'	#  0xc0 -> GREEK SMALL LETTER ALPHA WITH TONOS
+    u'\u039d'	#  0xc1 -> GREEK CAPITAL LETTER NU
+    u'\xac'	#  0xc2 -> NOT SIGN
+    u'\u039f'	#  0xc3 -> GREEK CAPITAL LETTER OMICRON
+    u'\u03a1'	#  0xc4 -> GREEK CAPITAL LETTER RHO
+    u'\u2248'	#  0xc5 -> ALMOST EQUAL TO
+    u'\u03a4'	#  0xc6 -> GREEK CAPITAL LETTER TAU
+    u'\xab'	#  0xc7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
+    u'\xbb'	#  0xc8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
+    u'\u2026'	#  0xc9 -> HORIZONTAL ELLIPSIS
+    u'\xa0'	#  0xca -> NO-BREAK SPACE
+    u'\u03a5'	#  0xcb -> GREEK CAPITAL LETTER UPSILON
+    u'\u03a7'	#  0xcc -> GREEK CAPITAL LETTER CHI
+    u'\u0386'	#  0xcd -> GREEK CAPITAL LETTER ALPHA WITH TONOS
+    u'\u0388'	#  0xce -> GREEK CAPITAL LETTER EPSILON WITH TONOS
+    u'\u0153'	#  0xcf -> LATIN SMALL LIGATURE OE
+    u'\u2013'	#  0xd0 -> EN DASH
+    u'\u2015'	#  0xd1 -> HORIZONTAL BAR
+    u'\u201c'	#  0xd2 -> LEFT DOUBLE QUOTATION MARK
+    u'\u201d'	#  0xd3 -> RIGHT DOUBLE QUOTATION MARK
+    u'\u2018'	#  0xd4 -> LEFT SINGLE QUOTATION MARK
+    u'\u2019'	#  0xd5 -> RIGHT SINGLE QUOTATION MARK
+    u'\xf7'	#  0xd6 -> DIVISION SIGN
+    u'\u0389'	#  0xd7 -> GREEK CAPITAL LETTER ETA WITH TONOS
+    u'\u038a'	#  0xd8 -> GREEK CAPITAL LETTER IOTA WITH TONOS
+    u'\u038c'	#  0xd9 -> GREEK CAPITAL LETTER OMICRON WITH TONOS
+    u'\u038e'	#  0xda -> GREEK CAPITAL LETTER UPSILON WITH TONOS
+    u'\u03ad'	#  0xdb -> GREEK SMALL LETTER EPSILON WITH TONOS
+    u'\u03ae'	#  0xdc -> GREEK SMALL LETTER ETA WITH TONOS
+    u'\u03af'	#  0xdd -> GREEK SMALL LETTER IOTA WITH TONOS
+    u'\u03cc'	#  0xde -> GREEK SMALL LETTER OMICRON WITH TONOS
+    u'\u038f'	#  0xdf -> GREEK CAPITAL LETTER OMEGA WITH TONOS
+    u'\u03cd'	#  0xe0 -> GREEK SMALL LETTER UPSILON WITH TONOS
+    u'\u03b1'	#  0xe1 -> GREEK SMALL LETTER ALPHA
+    u'\u03b2'	#  0xe2 -> GREEK SMALL LETTER BETA
+    u'\u03c8'	#  0xe3 -> GREEK SMALL LETTER PSI
+    u'\u03b4'	#  0xe4 -> GREEK SMALL LETTER DELTA
+    u'\u03b5'	#  0xe5 -> GREEK SMALL LETTER EPSILON
+    u'\u03c6'	#  0xe6 -> GREEK SMALL LETTER PHI
+    u'\u03b3'	#  0xe7 -> GREEK SMALL LETTER GAMMA
+    u'\u03b7'	#  0xe8 -> GREEK SMALL LETTER ETA
+    u'\u03b9'	#  0xe9 -> GREEK SMALL LETTER IOTA
+    u'\u03be'	#  0xea -> GREEK SMALL LETTER XI
+    u'\u03ba'	#  0xeb -> GREEK SMALL LETTER KAPPA
+    u'\u03bb'	#  0xec -> GREEK SMALL LETTER LAMDA
+    u'\u03bc'	#  0xed -> GREEK SMALL LETTER MU
+    u'\u03bd'	#  0xee -> GREEK SMALL LETTER NU
+    u'\u03bf'	#  0xef -> GREEK SMALL LETTER OMICRON
+    u'\u03c0'	#  0xf0 -> GREEK SMALL LETTER PI
+    u'\u03ce'	#  0xf1 -> GREEK SMALL LETTER OMEGA WITH TONOS
+    u'\u03c1'	#  0xf2 -> GREEK SMALL LETTER RHO
+    u'\u03c3'	#  0xf3 -> GREEK SMALL LETTER SIGMA
+    u'\u03c4'	#  0xf4 -> GREEK SMALL LETTER TAU
+    u'\u03b8'	#  0xf5 -> GREEK SMALL LETTER THETA
+    u'\u03c9'	#  0xf6 -> GREEK SMALL LETTER OMEGA
+    u'\u03c2'	#  0xf7 -> GREEK SMALL LETTER FINAL SIGMA
+    u'\u03c7'	#  0xf8 -> GREEK SMALL LETTER CHI
+    u'\u03c5'	#  0xf9 -> GREEK SMALL LETTER UPSILON
+    u'\u03b6'	#  0xfa -> GREEK SMALL LETTER ZETA
+    u'\u03ca'	#  0xfb -> GREEK SMALL LETTER IOTA WITH DIALYTIKA
+    u'\u03cb'	#  0xfc -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA
+    u'\u0390'	#  0xfd -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
+    u'\u03b0'	#  0xfe -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS
+    u'\xad'	#  0xff -> SOFT HYPHEN # before Mac OS 9.2.2, was undefined
 )
 
 ### Encoding Map
 
 encoding_map = {
-    0x0000: 0x0000,	#  CONTROL CHARACTER
-    0x0001: 0x0001,	#  CONTROL CHARACTER
-    0x0002: 0x0002,	#  CONTROL CHARACTER
-    0x0003: 0x0003,	#  CONTROL CHARACTER
-    0x0004: 0x0004,	#  CONTROL CHARACTER
-    0x0005: 0x0005,	#  CONTROL CHARACTER
-    0x0006: 0x0006,	#  CONTROL CHARACTER
-    0x0007: 0x0007,	#  CONTROL CHARACTER
-    0x0008: 0x0008,	#  CONTROL CHARACTER
-    0x0009: 0x0009,	#  CONTROL CHARACTER
-    0x000a: 0x000a,	#  CONTROL CHARACTER
-    0x000b: 0x000b,	#  CONTROL CHARACTER
-    0x000c: 0x000c,	#  CONTROL CHARACTER
-    0x000d: 0x000d,	#  CONTROL CHARACTER
-    0x000e: 0x000e,	#  CONTROL CHARACTER
-    0x000f: 0x000f,	#  CONTROL CHARACTER
-    0x0010: 0x0010,	#  CONTROL CHARACTER
-    0x0011: 0x0011,	#  CONTROL CHARACTER
-    0x0012: 0x0012,	#  CONTROL CHARACTER
-    0x0013: 0x0013,	#  CONTROL CHARACTER
-    0x0014: 0x0014,	#  CONTROL CHARACTER
-    0x0015: 0x0015,	#  CONTROL CHARACTER
-    0x0016: 0x0016,	#  CONTROL CHARACTER
-    0x0017: 0x0017,	#  CONTROL CHARACTER
-    0x0018: 0x0018,	#  CONTROL CHARACTER
-    0x0019: 0x0019,	#  CONTROL CHARACTER
-    0x001a: 0x001a,	#  CONTROL CHARACTER
-    0x001b: 0x001b,	#  CONTROL CHARACTER
-    0x001c: 0x001c,	#  CONTROL CHARACTER
-    0x001d: 0x001d,	#  CONTROL CHARACTER
-    0x001e: 0x001e,	#  CONTROL CHARACTER
-    0x001f: 0x001f,	#  CONTROL CHARACTER
-    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,	#  CONTROL CHARACTER
-    0x00a0: 0x00ca,	#  NO-BREAK SPACE
-    0x00a3: 0x0092,	#  POUND SIGN
-    0x00a5: 0x00b4,	#  YEN SIGN
-    0x00a6: 0x009b,	#  BROKEN BAR
-    0x00a7: 0x00ac,	#  SECTION SIGN
-    0x00a8: 0x008c,	#  DIAERESIS
-    0x00a9: 0x00a9,	#  COPYRIGHT SIGN
-    0x00ab: 0x00c7,	#  LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
-    0x00ac: 0x00c2,	#  NOT SIGN
-    0x00ad: 0x00ff,	#  SOFT HYPHEN # before Mac OS 9.2.2, was undefined
-    0x00ae: 0x00a8,	#  REGISTERED SIGN
-    0x00b0: 0x00ae,	#  DEGREE SIGN
-    0x00b1: 0x00b1,	#  PLUS-MINUS SIGN
-    0x00b2: 0x0082,	#  SUPERSCRIPT TWO
-    0x00b3: 0x0084,	#  SUPERSCRIPT THREE
-    0x00b7: 0x00af,	#  MIDDLE DOT
-    0x00b9: 0x0081,	#  SUPERSCRIPT ONE
-    0x00bb: 0x00c8,	#  RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
-    0x00bd: 0x0097,	#  VULGAR FRACTION ONE HALF
-    0x00c4: 0x0080,	#  LATIN CAPITAL LETTER A WITH DIAERESIS
-    0x00c9: 0x0083,	#  LATIN CAPITAL LETTER E WITH ACUTE
-    0x00d6: 0x0085,	#  LATIN CAPITAL LETTER O WITH DIAERESIS
-    0x00dc: 0x0086,	#  LATIN CAPITAL LETTER U WITH DIAERESIS
-    0x00df: 0x00a7,	#  LATIN SMALL LETTER SHARP S
-    0x00e0: 0x0088,	#  LATIN SMALL LETTER A WITH GRAVE
-    0x00e2: 0x0089,	#  LATIN SMALL LETTER A WITH CIRCUMFLEX
-    0x00e4: 0x008a,	#  LATIN SMALL LETTER A WITH DIAERESIS
-    0x00e7: 0x008d,	#  LATIN SMALL LETTER C WITH CEDILLA
-    0x00e8: 0x008f,	#  LATIN SMALL LETTER E WITH GRAVE
-    0x00e9: 0x008e,	#  LATIN SMALL LETTER E WITH ACUTE
-    0x00ea: 0x0090,	#  LATIN SMALL LETTER E WITH CIRCUMFLEX
-    0x00eb: 0x0091,	#  LATIN SMALL LETTER E WITH DIAERESIS
-    0x00ee: 0x0094,	#  LATIN SMALL LETTER I WITH CIRCUMFLEX
-    0x00ef: 0x0095,	#  LATIN SMALL LETTER I WITH DIAERESIS
-    0x00f4: 0x0099,	#  LATIN SMALL LETTER O WITH CIRCUMFLEX
-    0x00f6: 0x009a,	#  LATIN SMALL LETTER O WITH DIAERESIS
-    0x00f7: 0x00d6,	#  DIVISION SIGN
-    0x00f9: 0x009d,	#  LATIN SMALL LETTER U WITH GRAVE
-    0x00fb: 0x009e,	#  LATIN SMALL LETTER U WITH CIRCUMFLEX
-    0x00fc: 0x009f,	#  LATIN SMALL LETTER U WITH DIAERESIS
-    0x0153: 0x00cf,	#  LATIN SMALL LIGATURE OE
-    0x0384: 0x008b,	#  GREEK TONOS
-    0x0385: 0x0087,	#  GREEK DIALYTIKA TONOS
-    0x0386: 0x00cd,	#  GREEK CAPITAL LETTER ALPHA WITH TONOS
-    0x0388: 0x00ce,	#  GREEK CAPITAL LETTER EPSILON WITH TONOS
-    0x0389: 0x00d7,	#  GREEK CAPITAL LETTER ETA WITH TONOS
-    0x038a: 0x00d8,	#  GREEK CAPITAL LETTER IOTA WITH TONOS
-    0x038c: 0x00d9,	#  GREEK CAPITAL LETTER OMICRON WITH TONOS
-    0x038e: 0x00da,	#  GREEK CAPITAL LETTER UPSILON WITH TONOS
-    0x038f: 0x00df,	#  GREEK CAPITAL LETTER OMEGA WITH TONOS
-    0x0390: 0x00fd,	#  GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
-    0x0391: 0x00b0,	#  GREEK CAPITAL LETTER ALPHA
-    0x0392: 0x00b5,	#  GREEK CAPITAL LETTER BETA
-    0x0393: 0x00a1,	#  GREEK CAPITAL LETTER GAMMA
-    0x0394: 0x00a2,	#  GREEK CAPITAL LETTER DELTA
-    0x0395: 0x00b6,	#  GREEK CAPITAL LETTER EPSILON
-    0x0396: 0x00b7,	#  GREEK CAPITAL LETTER ZETA
-    0x0397: 0x00b8,	#  GREEK CAPITAL LETTER ETA
-    0x0398: 0x00a3,	#  GREEK CAPITAL LETTER THETA
-    0x0399: 0x00b9,	#  GREEK CAPITAL LETTER IOTA
-    0x039a: 0x00ba,	#  GREEK CAPITAL LETTER KAPPA
-    0x039b: 0x00a4,	#  GREEK CAPITAL LETTER LAMDA
-    0x039c: 0x00bb,	#  GREEK CAPITAL LETTER MU
-    0x039d: 0x00c1,	#  GREEK CAPITAL LETTER NU
-    0x039e: 0x00a5,	#  GREEK CAPITAL LETTER XI
-    0x039f: 0x00c3,	#  GREEK CAPITAL LETTER OMICRON
-    0x03a0: 0x00a6,	#  GREEK CAPITAL LETTER PI
-    0x03a1: 0x00c4,	#  GREEK CAPITAL LETTER RHO
-    0x03a3: 0x00aa,	#  GREEK CAPITAL LETTER SIGMA
-    0x03a4: 0x00c6,	#  GREEK CAPITAL LETTER TAU
-    0x03a5: 0x00cb,	#  GREEK CAPITAL LETTER UPSILON
-    0x03a6: 0x00bc,	#  GREEK CAPITAL LETTER PHI
-    0x03a7: 0x00cc,	#  GREEK CAPITAL LETTER CHI
-    0x03a8: 0x00be,	#  GREEK CAPITAL LETTER PSI
-    0x03a9: 0x00bf,	#  GREEK CAPITAL LETTER OMEGA
-    0x03aa: 0x00ab,	#  GREEK CAPITAL LETTER IOTA WITH DIALYTIKA
-    0x03ab: 0x00bd,	#  GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA
-    0x03ac: 0x00c0,	#  GREEK SMALL LETTER ALPHA WITH TONOS
-    0x03ad: 0x00db,	#  GREEK SMALL LETTER EPSILON WITH TONOS
-    0x03ae: 0x00dc,	#  GREEK SMALL LETTER ETA WITH TONOS
-    0x03af: 0x00dd,	#  GREEK SMALL LETTER IOTA WITH TONOS
-    0x03b0: 0x00fe,	#  GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS
-    0x03b1: 0x00e1,	#  GREEK SMALL LETTER ALPHA
-    0x03b2: 0x00e2,	#  GREEK SMALL LETTER BETA
-    0x03b3: 0x00e7,	#  GREEK SMALL LETTER GAMMA
-    0x03b4: 0x00e4,	#  GREEK SMALL LETTER DELTA
-    0x03b5: 0x00e5,	#  GREEK SMALL LETTER EPSILON
-    0x03b6: 0x00fa,	#  GREEK SMALL LETTER ZETA
-    0x03b7: 0x00e8,	#  GREEK SMALL LETTER ETA
-    0x03b8: 0x00f5,	#  GREEK SMALL LETTER THETA
-    0x03b9: 0x00e9,	#  GREEK SMALL LETTER IOTA
-    0x03ba: 0x00eb,	#  GREEK SMALL LETTER KAPPA
-    0x03bb: 0x00ec,	#  GREEK SMALL LETTER LAMDA
-    0x03bc: 0x00ed,	#  GREEK SMALL LETTER MU
-    0x03bd: 0x00ee,	#  GREEK SMALL LETTER NU
-    0x03be: 0x00ea,	#  GREEK SMALL LETTER XI
-    0x03bf: 0x00ef,	#  GREEK SMALL LETTER OMICRON
-    0x03c0: 0x00f0,	#  GREEK SMALL LETTER PI
-    0x03c1: 0x00f2,	#  GREEK SMALL LETTER RHO
-    0x03c2: 0x00f7,	#  GREEK SMALL LETTER FINAL SIGMA
-    0x03c3: 0x00f3,	#  GREEK SMALL LETTER SIGMA
-    0x03c4: 0x00f4,	#  GREEK SMALL LETTER TAU
-    0x03c5: 0x00f9,	#  GREEK SMALL LETTER UPSILON
-    0x03c6: 0x00e6,	#  GREEK SMALL LETTER PHI
-    0x03c7: 0x00f8,	#  GREEK SMALL LETTER CHI
-    0x03c8: 0x00e3,	#  GREEK SMALL LETTER PSI
-    0x03c9: 0x00f6,	#  GREEK SMALL LETTER OMEGA
-    0x03ca: 0x00fb,	#  GREEK SMALL LETTER IOTA WITH DIALYTIKA
-    0x03cb: 0x00fc,	#  GREEK SMALL LETTER UPSILON WITH DIALYTIKA
-    0x03cc: 0x00de,	#  GREEK SMALL LETTER OMICRON WITH TONOS
-    0x03cd: 0x00e0,	#  GREEK SMALL LETTER UPSILON WITH TONOS
-    0x03ce: 0x00f1,	#  GREEK SMALL LETTER OMEGA WITH TONOS
-    0x2013: 0x00d0,	#  EN DASH
-    0x2015: 0x00d1,	#  HORIZONTAL BAR
-    0x2018: 0x00d4,	#  LEFT SINGLE QUOTATION MARK
-    0x2019: 0x00d5,	#  RIGHT SINGLE QUOTATION MARK
-    0x201c: 0x00d2,	#  LEFT DOUBLE QUOTATION MARK
-    0x201d: 0x00d3,	#  RIGHT DOUBLE QUOTATION MARK
-    0x2020: 0x00a0,	#  DAGGER
-    0x2022: 0x0096,	#  BULLET
-    0x2026: 0x00c9,	#  HORIZONTAL ELLIPSIS
-    0x2030: 0x0098,	#  PER MILLE SIGN
-    0x20ac: 0x009c,	#  EURO SIGN # before Mac OS 9.2.2, was SOFT HYPHEN
-    0x2122: 0x0093,	#  TRADE MARK SIGN
-    0x2248: 0x00c5,	#  ALMOST EQUAL TO
-    0x2260: 0x00ad,	#  NOT EQUAL TO
-    0x2264: 0x00b2,	#  LESS-THAN OR EQUAL TO
-    0x2265: 0x00b3,	#  GREATER-THAN OR EQUAL TO
+    0x0000: 0x00,	#  CONTROL CHARACTER
+    0x0001: 0x01,	#  CONTROL CHARACTER
+    0x0002: 0x02,	#  CONTROL CHARACTER
+    0x0003: 0x03,	#  CONTROL CHARACTER
+    0x0004: 0x04,	#  CONTROL CHARACTER
+    0x0005: 0x05,	#  CONTROL CHARACTER
+    0x0006: 0x06,	#  CONTROL CHARACTER
+    0x0007: 0x07,	#  CONTROL CHARACTER
+    0x0008: 0x08,	#  CONTROL CHARACTER
+    0x0009: 0x09,	#  CONTROL CHARACTER
+    0x000a: 0x0a,	#  CONTROL CHARACTER
+    0x000b: 0x0b,	#  CONTROL CHARACTER
+    0x000c: 0x0c,	#  CONTROL CHARACTER
+    0x000d: 0x0d,	#  CONTROL CHARACTER
+    0x000e: 0x0e,	#  CONTROL CHARACTER
+    0x000f: 0x0f,	#  CONTROL CHARACTER
+    0x0010: 0x10,	#  CONTROL CHARACTER
+    0x0011: 0x11,	#  CONTROL CHARACTER
+    0x0012: 0x12,	#  CONTROL CHARACTER
+    0x0013: 0x13,	#  CONTROL CHARACTER
+    0x0014: 0x14,	#  CONTROL CHARACTER
+    0x0015: 0x15,	#  CONTROL CHARACTER
+    0x0016: 0x16,	#  CONTROL CHARACTER
+    0x0017: 0x17,	#  CONTROL CHARACTER
+    0x0018: 0x18,	#  CONTROL CHARACTER
+    0x0019: 0x19,	#  CONTROL CHARACTER
+    0x001a: 0x1a,	#  CONTROL CHARACTER
+    0x001b: 0x1b,	#  CONTROL CHARACTER
+    0x001c: 0x1c,	#  CONTROL CHARACTER
+    0x001d: 0x1d,	#  CONTROL CHARACTER
+    0x001e: 0x1e,	#  CONTROL CHARACTER
+    0x001f: 0x1f,	#  CONTROL CHARACTER
+    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,	#  CONTROL CHARACTER
+    0x00a0: 0xca,	#  NO-BREAK SPACE
+    0x00a3: 0x92,	#  POUND SIGN
+    0x00a5: 0xb4,	#  YEN SIGN
+    0x00a6: 0x9b,	#  BROKEN BAR
+    0x00a7: 0xac,	#  SECTION SIGN
+    0x00a8: 0x8c,	#  DIAERESIS
+    0x00a9: 0xa9,	#  COPYRIGHT SIGN
+    0x00ab: 0xc7,	#  LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
+    0x00ac: 0xc2,	#  NOT SIGN
+    0x00ad: 0xff,	#  SOFT HYPHEN # before Mac OS 9.2.2, was undefined
+    0x00ae: 0xa8,	#  REGISTERED SIGN
+    0x00b0: 0xae,	#  DEGREE SIGN
+    0x00b1: 0xb1,	#  PLUS-MINUS SIGN
+    0x00b2: 0x82,	#  SUPERSCRIPT TWO
+    0x00b3: 0x84,	#  SUPERSCRIPT THREE
+    0x00b7: 0xaf,	#  MIDDLE DOT
+    0x00b9: 0x81,	#  SUPERSCRIPT ONE
+    0x00bb: 0xc8,	#  RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
+    0x00bd: 0x97,	#  VULGAR FRACTION ONE HALF
+    0x00c4: 0x80,	#  LATIN CAPITAL LETTER A WITH DIAERESIS
+    0x00c9: 0x83,	#  LATIN CAPITAL LETTER E WITH ACUTE
+    0x00d6: 0x85,	#  LATIN CAPITAL LETTER O WITH DIAERESIS
+    0x00dc: 0x86,	#  LATIN CAPITAL LETTER U WITH DIAERESIS
+    0x00df: 0xa7,	#  LATIN SMALL LETTER SHARP S
+    0x00e0: 0x88,	#  LATIN SMALL LETTER A WITH GRAVE
+    0x00e2: 0x89,	#  LATIN SMALL LETTER A WITH CIRCUMFLEX
+    0x00e4: 0x8a,	#  LATIN SMALL LETTER A WITH DIAERESIS
+    0x00e7: 0x8d,	#  LATIN SMALL LETTER C WITH CEDILLA
+    0x00e8: 0x8f,	#  LATIN SMALL LETTER E WITH GRAVE
+    0x00e9: 0x8e,	#  LATIN SMALL LETTER E WITH ACUTE
+    0x00ea: 0x90,	#  LATIN SMALL LETTER E WITH CIRCUMFLEX
+    0x00eb: 0x91,	#  LATIN SMALL LETTER E WITH DIAERESIS
+    0x00ee: 0x94,	#  LATIN SMALL LETTER I WITH CIRCUMFLEX
+    0x00ef: 0x95,	#  LATIN SMALL LETTER I WITH DIAERESIS
+    0x00f4: 0x99,	#  LATIN SMALL LETTER O WITH CIRCUMFLEX
+    0x00f6: 0x9a,	#  LATIN SMALL LETTER O WITH DIAERESIS
+    0x00f7: 0xd6,	#  DIVISION SIGN
+    0x00f9: 0x9d,	#  LATIN SMALL LETTER U WITH GRAVE
+    0x00fb: 0x9e,	#  LATIN SMALL LETTER U WITH CIRCUMFLEX
+    0x00fc: 0x9f,	#  LATIN SMALL LETTER U WITH DIAERESIS
+    0x0153: 0xcf,	#  LATIN SMALL LIGATURE OE
+    0x0384: 0x8b,	#  GREEK TONOS
+    0x0385: 0x87,	#  GREEK DIALYTIKA TONOS
+    0x0386: 0xcd,	#  GREEK CAPITAL LETTER ALPHA WITH TONOS
+    0x0388: 0xce,	#  GREEK CAPITAL LETTER EPSILON WITH TONOS
+    0x0389: 0xd7,	#  GREEK CAPITAL LETTER ETA WITH TONOS
+    0x038a: 0xd8,	#  GREEK CAPITAL LETTER IOTA WITH TONOS
+    0x038c: 0xd9,	#  GREEK CAPITAL LETTER OMICRON WITH TONOS
+    0x038e: 0xda,	#  GREEK CAPITAL LETTER UPSILON WITH TONOS
+    0x038f: 0xdf,	#  GREEK CAPITAL LETTER OMEGA WITH TONOS
+    0x0390: 0xfd,	#  GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
+    0x0391: 0xb0,	#  GREEK CAPITAL LETTER ALPHA
+    0x0392: 0xb5,	#  GREEK CAPITAL LETTER BETA
+    0x0393: 0xa1,	#  GREEK CAPITAL LETTER GAMMA
+    0x0394: 0xa2,	#  GREEK CAPITAL LETTER DELTA
+    0x0395: 0xb6,	#  GREEK CAPITAL LETTER EPSILON
+    0x0396: 0xb7,	#  GREEK CAPITAL LETTER ZETA
+    0x0397: 0xb8,	#  GREEK CAPITAL LETTER ETA
+    0x0398: 0xa3,	#  GREEK CAPITAL LETTER THETA
+    0x0399: 0xb9,	#  GREEK CAPITAL LETTER IOTA
+    0x039a: 0xba,	#  GREEK CAPITAL LETTER KAPPA
+    0x039b: 0xa4,	#  GREEK CAPITAL LETTER LAMDA
+    0x039c: 0xbb,	#  GREEK CAPITAL LETTER MU
+    0x039d: 0xc1,	#  GREEK CAPITAL LETTER NU
+    0x039e: 0xa5,	#  GREEK CAPITAL LETTER XI
+    0x039f: 0xc3,	#  GREEK CAPITAL LETTER OMICRON
+    0x03a0: 0xa6,	#  GREEK CAPITAL LETTER PI
+    0x03a1: 0xc4,	#  GREEK CAPITAL LETTER RHO
+    0x03a3: 0xaa,	#  GREEK CAPITAL LETTER SIGMA
+    0x03a4: 0xc6,	#  GREEK CAPITAL LETTER TAU
+    0x03a5: 0xcb,	#  GREEK CAPITAL LETTER UPSILON
+    0x03a6: 0xbc,	#  GREEK CAPITAL LETTER PHI
+    0x03a7: 0xcc,	#  GREEK CAPITAL LETTER CHI
+    0x03a8: 0xbe,	#  GREEK CAPITAL LETTER PSI
+    0x03a9: 0xbf,	#  GREEK CAPITAL LETTER OMEGA
+    0x03aa: 0xab,	#  GREEK CAPITAL LETTER IOTA WITH DIALYTIKA
+    0x03ab: 0xbd,	#  GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA
+    0x03ac: 0xc0,	#  GREEK SMALL LETTER ALPHA WITH TONOS
+    0x03ad: 0xdb,	#  GREEK SMALL LETTER EPSILON WITH TONOS
+    0x03ae: 0xdc,	#  GREEK SMALL LETTER ETA WITH TONOS
+    0x03af: 0xdd,	#  GREEK SMALL LETTER IOTA WITH TONOS
+    0x03b0: 0xfe,	#  GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS
+    0x03b1: 0xe1,	#  GREEK SMALL LETTER ALPHA
+    0x03b2: 0xe2,	#  GREEK SMALL LETTER BETA
+    0x03b3: 0xe7,	#  GREEK SMALL LETTER GAMMA
+    0x03b4: 0xe4,	#  GREEK SMALL LETTER DELTA
+    0x03b5: 0xe5,	#  GREEK SMALL LETTER EPSILON
+    0x03b6: 0xfa,	#  GREEK SMALL LETTER ZETA
+    0x03b7: 0xe8,	#  GREEK SMALL LETTER ETA
+    0x03b8: 0xf5,	#  GREEK SMALL LETTER THETA
+    0x03b9: 0xe9,	#  GREEK SMALL LETTER IOTA
+    0x03ba: 0xeb,	#  GREEK SMALL LETTER KAPPA
+    0x03bb: 0xec,	#  GREEK SMALL LETTER LAMDA
+    0x03bc: 0xed,	#  GREEK SMALL LETTER MU
+    0x03bd: 0xee,	#  GREEK SMALL LETTER NU
+    0x03be: 0xea,	#  GREEK SMALL LETTER XI
+    0x03bf: 0xef,	#  GREEK SMALL LETTER OMICRON
+    0x03c0: 0xf0,	#  GREEK SMALL LETTER PI
+    0x03c1: 0xf2,	#  GREEK SMALL LETTER RHO
+    0x03c2: 0xf7,	#  GREEK SMALL LETTER FINAL SIGMA
+    0x03c3: 0xf3,	#  GREEK SMALL LETTER SIGMA
+    0x03c4: 0xf4,	#  GREEK SMALL LETTER TAU
+    0x03c5: 0xf9,	#  GREEK SMALL LETTER UPSILON
+    0x03c6: 0xe6,	#  GREEK SMALL LETTER PHI
+    0x03c7: 0xf8,	#  GREEK SMALL LETTER CHI
+    0x03c8: 0xe3,	#  GREEK SMALL LETTER PSI
+    0x03c9: 0xf6,	#  GREEK SMALL LETTER OMEGA
+    0x03ca: 0xfb,	#  GREEK SMALL LETTER IOTA WITH DIALYTIKA
+    0x03cb: 0xfc,	#  GREEK SMALL LETTER UPSILON WITH DIALYTIKA
+    0x03cc: 0xde,	#  GREEK SMALL LETTER OMICRON WITH TONOS
+    0x03cd: 0xe0,	#  GREEK SMALL LETTER UPSILON WITH TONOS
+    0x03ce: 0xf1,	#  GREEK SMALL LETTER OMEGA WITH TONOS
+    0x2013: 0xd0,	#  EN DASH
+    0x2015: 0xd1,	#  HORIZONTAL BAR
+    0x2018: 0xd4,	#  LEFT SINGLE QUOTATION MARK
+    0x2019: 0xd5,	#  RIGHT SINGLE QUOTATION MARK
+    0x201c: 0xd2,	#  LEFT DOUBLE QUOTATION MARK
+    0x201d: 0xd3,	#  RIGHT DOUBLE QUOTATION MARK
+    0x2020: 0xa0,	#  DAGGER
+    0x2022: 0x96,	#  BULLET
+    0x2026: 0xc9,	#  HORIZONTAL ELLIPSIS
+    0x2030: 0x98,	#  PER MILLE SIGN
+    0x20ac: 0x9c,	#  EURO SIGN # before Mac OS 9.2.2, was SOFT HYPHEN
+    0x2122: 0x93,	#  TRADE MARK SIGN
+    0x2248: 0xc5,	#  ALMOST EQUAL TO
+    0x2260: 0xad,	#  NOT EQUAL TO
+    0x2264: 0xb2,	#  LESS-THAN OR EQUAL TO
+    0x2265: 0xb3,	#  GREATER-THAN OR EQUAL TO
 }
\ No newline at end of file