2005-10-24 09:07:49 -03:00
|
|
|
""" Python Character Mapping Codec generated from 'MAPPINGS/VENDORS/MICSFT/WINDOWS/CP1257.TXT' with gencodec.py.
|
2000-03-10 19:17:24 -04:00
|
|
|
|
|
|
|
|
"""#"
|
|
|
|
|
|
|
|
|
|
import codecs
|
|
|
|
|
|
|
|
|
|
### Codec APIs
|
|
|
|
|
|
|
|
|
|
class Codec(codecs.Codec):
|
|
|
|
|
|
|
|
|
|
def encode(self,input,errors='strict'):
|
|
|
|
|
|
|
|
|
|
return codecs.charmap_encode(input,errors,encoding_map)
|
2002-08-08 17:19:19 -03:00
|
|
|
|
2000-03-10 19:17:24 -04:00
|
|
|
def decode(self,input,errors='strict'):
|
|
|
|
|
|
2005-10-21 10:49:12 -03:00
|
|
|
return codecs.charmap_decode(input,errors,decoding_table)
|
|
|
|
|
|
2000-03-10 19:17:24 -04:00
|
|
|
class StreamWriter(Codec,codecs.StreamWriter):
|
|
|
|
|
pass
|
2002-08-08 17:19:19 -03:00
|
|
|
|
2000-03-10 19:17:24 -04:00
|
|
|
class StreamReader(Codec,codecs.StreamReader):
|
|
|
|
|
pass
|
|
|
|
|
|
|
|
|
|
### encodings module API
|
|
|
|
|
|
|
|
|
|
def getregentry():
|
|
|
|
|
|
|
|
|
|
return (Codec().encode,Codec().decode,StreamReader,StreamWriter)
|
|
|
|
|
|
|
|
|
|
|
2005-10-21 10:49:12 -03:00
|
|
|
### Decoding Table
|
|
|
|
|
|
|
|
|
|
decoding_table = (
|
2005-10-24 09:07:49 -03:00
|
|
|
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'\ufffe' # 0x81 -> UNDEFINED
|
|
|
|
|
u'\u201a' # 0x82 -> SINGLE LOW-9 QUOTATION MARK
|
|
|
|
|
u'\ufffe' # 0x83 -> UNDEFINED
|
|
|
|
|
u'\u201e' # 0x84 -> DOUBLE LOW-9 QUOTATION MARK
|
|
|
|
|
u'\u2026' # 0x85 -> HORIZONTAL ELLIPSIS
|
|
|
|
|
u'\u2020' # 0x86 -> DAGGER
|
|
|
|
|
u'\u2021' # 0x87 -> DOUBLE DAGGER
|
|
|
|
|
u'\ufffe' # 0x88 -> UNDEFINED
|
|
|
|
|
u'\u2030' # 0x89 -> PER MILLE SIGN
|
|
|
|
|
u'\ufffe' # 0x8a -> UNDEFINED
|
|
|
|
|
u'\u2039' # 0x8b -> SINGLE LEFT-POINTING ANGLE QUOTATION MARK
|
|
|
|
|
u'\ufffe' # 0x8c -> UNDEFINED
|
|
|
|
|
u'\xa8' # 0x8d -> DIAERESIS
|
|
|
|
|
u'\u02c7' # 0x8e -> CARON
|
|
|
|
|
u'\xb8' # 0x8f -> CEDILLA
|
|
|
|
|
u'\ufffe' # 0x90 -> UNDEFINED
|
|
|
|
|
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'\ufffe' # 0x98 -> UNDEFINED
|
|
|
|
|
u'\u2122' # 0x99 -> TRADE MARK SIGN
|
|
|
|
|
u'\ufffe' # 0x9a -> UNDEFINED
|
|
|
|
|
u'\u203a' # 0x9b -> SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
|
|
|
|
|
u'\ufffe' # 0x9c -> UNDEFINED
|
|
|
|
|
u'\xaf' # 0x9d -> MACRON
|
|
|
|
|
u'\u02db' # 0x9e -> OGONEK
|
|
|
|
|
u'\ufffe' # 0x9f -> UNDEFINED
|
|
|
|
|
u'\xa0' # 0xa0 -> NO-BREAK SPACE
|
|
|
|
|
u'\ufffe' # 0xa1 -> UNDEFINED
|
|
|
|
|
u'\xa2' # 0xa2 -> CENT SIGN
|
|
|
|
|
u'\xa3' # 0xa3 -> POUND SIGN
|
|
|
|
|
u'\xa4' # 0xa4 -> CURRENCY SIGN
|
|
|
|
|
u'\ufffe' # 0xa5 -> UNDEFINED
|
|
|
|
|
u'\xa6' # 0xa6 -> BROKEN BAR
|
|
|
|
|
u'\xa7' # 0xa7 -> SECTION SIGN
|
|
|
|
|
u'\xd8' # 0xa8 -> LATIN CAPITAL LETTER O WITH STROKE
|
|
|
|
|
u'\xa9' # 0xa9 -> COPYRIGHT SIGN
|
|
|
|
|
u'\u0156' # 0xaa -> LATIN CAPITAL LETTER R WITH CEDILLA
|
|
|
|
|
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'\xc6' # 0xaf -> LATIN CAPITAL LETTER AE
|
|
|
|
|
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'\xf8' # 0xb8 -> LATIN SMALL LETTER O WITH STROKE
|
|
|
|
|
u'\xb9' # 0xb9 -> SUPERSCRIPT ONE
|
|
|
|
|
u'\u0157' # 0xba -> LATIN SMALL LETTER R WITH CEDILLA
|
|
|
|
|
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'\xe6' # 0xbf -> LATIN SMALL LETTER AE
|
|
|
|
|
u'\u0104' # 0xc0 -> LATIN CAPITAL LETTER A WITH OGONEK
|
|
|
|
|
u'\u012e' # 0xc1 -> LATIN CAPITAL LETTER I WITH OGONEK
|
|
|
|
|
u'\u0100' # 0xc2 -> LATIN CAPITAL LETTER A WITH MACRON
|
|
|
|
|
u'\u0106' # 0xc3 -> LATIN CAPITAL LETTER C WITH ACUTE
|
|
|
|
|
u'\xc4' # 0xc4 -> LATIN CAPITAL LETTER A WITH DIAERESIS
|
|
|
|
|
u'\xc5' # 0xc5 -> LATIN CAPITAL LETTER A WITH RING ABOVE
|
|
|
|
|
u'\u0118' # 0xc6 -> LATIN CAPITAL LETTER E WITH OGONEK
|
|
|
|
|
u'\u0112' # 0xc7 -> LATIN CAPITAL LETTER E WITH MACRON
|
|
|
|
|
u'\u010c' # 0xc8 -> LATIN CAPITAL LETTER C WITH CARON
|
|
|
|
|
u'\xc9' # 0xc9 -> LATIN CAPITAL LETTER E WITH ACUTE
|
|
|
|
|
u'\u0179' # 0xca -> LATIN CAPITAL LETTER Z WITH ACUTE
|
|
|
|
|
u'\u0116' # 0xcb -> LATIN CAPITAL LETTER E WITH DOT ABOVE
|
|
|
|
|
u'\u0122' # 0xcc -> LATIN CAPITAL LETTER G WITH CEDILLA
|
|
|
|
|
u'\u0136' # 0xcd -> LATIN CAPITAL LETTER K WITH CEDILLA
|
|
|
|
|
u'\u012a' # 0xce -> LATIN CAPITAL LETTER I WITH MACRON
|
|
|
|
|
u'\u013b' # 0xcf -> LATIN CAPITAL LETTER L WITH CEDILLA
|
|
|
|
|
u'\u0160' # 0xd0 -> LATIN CAPITAL LETTER S WITH CARON
|
|
|
|
|
u'\u0143' # 0xd1 -> LATIN CAPITAL LETTER N WITH ACUTE
|
|
|
|
|
u'\u0145' # 0xd2 -> LATIN CAPITAL LETTER N WITH CEDILLA
|
|
|
|
|
u'\xd3' # 0xd3 -> LATIN CAPITAL LETTER O WITH ACUTE
|
|
|
|
|
u'\u014c' # 0xd4 -> LATIN CAPITAL LETTER O WITH MACRON
|
|
|
|
|
u'\xd5' # 0xd5 -> LATIN CAPITAL LETTER O WITH TILDE
|
|
|
|
|
u'\xd6' # 0xd6 -> LATIN CAPITAL LETTER O WITH DIAERESIS
|
|
|
|
|
u'\xd7' # 0xd7 -> MULTIPLICATION SIGN
|
|
|
|
|
u'\u0172' # 0xd8 -> LATIN CAPITAL LETTER U WITH OGONEK
|
|
|
|
|
u'\u0141' # 0xd9 -> LATIN CAPITAL LETTER L WITH STROKE
|
|
|
|
|
u'\u015a' # 0xda -> LATIN CAPITAL LETTER S WITH ACUTE
|
|
|
|
|
u'\u016a' # 0xdb -> LATIN CAPITAL LETTER U WITH MACRON
|
|
|
|
|
u'\xdc' # 0xdc -> LATIN CAPITAL LETTER U WITH DIAERESIS
|
|
|
|
|
u'\u017b' # 0xdd -> LATIN CAPITAL LETTER Z WITH DOT ABOVE
|
|
|
|
|
u'\u017d' # 0xde -> LATIN CAPITAL LETTER Z WITH CARON
|
|
|
|
|
u'\xdf' # 0xdf -> LATIN SMALL LETTER SHARP S
|
|
|
|
|
u'\u0105' # 0xe0 -> LATIN SMALL LETTER A WITH OGONEK
|
|
|
|
|
u'\u012f' # 0xe1 -> LATIN SMALL LETTER I WITH OGONEK
|
|
|
|
|
u'\u0101' # 0xe2 -> LATIN SMALL LETTER A WITH MACRON
|
|
|
|
|
u'\u0107' # 0xe3 -> LATIN SMALL LETTER C WITH ACUTE
|
|
|
|
|
u'\xe4' # 0xe4 -> LATIN SMALL LETTER A WITH DIAERESIS
|
|
|
|
|
u'\xe5' # 0xe5 -> LATIN SMALL LETTER A WITH RING ABOVE
|
|
|
|
|
u'\u0119' # 0xe6 -> LATIN SMALL LETTER E WITH OGONEK
|
|
|
|
|
u'\u0113' # 0xe7 -> LATIN SMALL LETTER E WITH MACRON
|
|
|
|
|
u'\u010d' # 0xe8 -> LATIN SMALL LETTER C WITH CARON
|
|
|
|
|
u'\xe9' # 0xe9 -> LATIN SMALL LETTER E WITH ACUTE
|
|
|
|
|
u'\u017a' # 0xea -> LATIN SMALL LETTER Z WITH ACUTE
|
|
|
|
|
u'\u0117' # 0xeb -> LATIN SMALL LETTER E WITH DOT ABOVE
|
|
|
|
|
u'\u0123' # 0xec -> LATIN SMALL LETTER G WITH CEDILLA
|
|
|
|
|
u'\u0137' # 0xed -> LATIN SMALL LETTER K WITH CEDILLA
|
|
|
|
|
u'\u012b' # 0xee -> LATIN SMALL LETTER I WITH MACRON
|
|
|
|
|
u'\u013c' # 0xef -> LATIN SMALL LETTER L WITH CEDILLA
|
|
|
|
|
u'\u0161' # 0xf0 -> LATIN SMALL LETTER S WITH CARON
|
|
|
|
|
u'\u0144' # 0xf1 -> LATIN SMALL LETTER N WITH ACUTE
|
|
|
|
|
u'\u0146' # 0xf2 -> LATIN SMALL LETTER N WITH CEDILLA
|
|
|
|
|
u'\xf3' # 0xf3 -> LATIN SMALL LETTER O WITH ACUTE
|
|
|
|
|
u'\u014d' # 0xf4 -> LATIN SMALL LETTER O WITH MACRON
|
|
|
|
|
u'\xf5' # 0xf5 -> LATIN SMALL LETTER O WITH TILDE
|
|
|
|
|
u'\xf6' # 0xf6 -> LATIN SMALL LETTER O WITH DIAERESIS
|
|
|
|
|
u'\xf7' # 0xf7 -> DIVISION SIGN
|
|
|
|
|
u'\u0173' # 0xf8 -> LATIN SMALL LETTER U WITH OGONEK
|
|
|
|
|
u'\u0142' # 0xf9 -> LATIN SMALL LETTER L WITH STROKE
|
|
|
|
|
u'\u015b' # 0xfa -> LATIN SMALL LETTER S WITH ACUTE
|
|
|
|
|
u'\u016b' # 0xfb -> LATIN SMALL LETTER U WITH MACRON
|
|
|
|
|
u'\xfc' # 0xfc -> LATIN SMALL LETTER U WITH DIAERESIS
|
|
|
|
|
u'\u017c' # 0xfd -> LATIN SMALL LETTER Z WITH DOT ABOVE
|
|
|
|
|
u'\u017e' # 0xfe -> LATIN SMALL LETTER Z WITH CARON
|
|
|
|
|
u'\u02d9' # 0xff -> DOT ABOVE
|
2005-10-21 10:49:12 -03:00
|
|
|
)
|
|
|
|
|
|
2000-03-10 19:17:24 -04:00
|
|
|
### Encoding Map
|
|
|
|
|
|
2005-10-21 10:49:12 -03:00
|
|
|
encoding_map = {
|
2005-10-24 09:07:49 -03:00
|
|
|
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
|
|
|
|
|
0x00a6: 0xa6, # BROKEN BAR
|
|
|
|
|
0x00a7: 0xa7, # SECTION SIGN
|
|
|
|
|
0x00a8: 0x8d, # 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: 0x9d, # 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: 0x8f, # 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
|
|
|
|
|
0x00c4: 0xc4, # LATIN CAPITAL LETTER A WITH DIAERESIS
|
|
|
|
|
0x00c5: 0xc5, # LATIN CAPITAL LETTER A WITH RING ABOVE
|
|
|
|
|
0x00c6: 0xaf, # LATIN CAPITAL LETTER AE
|
|
|
|
|
0x00c9: 0xc9, # LATIN CAPITAL LETTER E WITH ACUTE
|
|
|
|
|
0x00d3: 0xd3, # LATIN CAPITAL LETTER O WITH ACUTE
|
|
|
|
|
0x00d5: 0xd5, # LATIN CAPITAL LETTER O WITH TILDE
|
|
|
|
|
0x00d6: 0xd6, # LATIN CAPITAL LETTER O WITH DIAERESIS
|
|
|
|
|
0x00d7: 0xd7, # MULTIPLICATION SIGN
|
|
|
|
|
0x00d8: 0xa8, # LATIN CAPITAL LETTER O WITH STROKE
|
|
|
|
|
0x00dc: 0xdc, # LATIN CAPITAL LETTER U WITH DIAERESIS
|
|
|
|
|
0x00df: 0xdf, # LATIN SMALL LETTER SHARP S
|
|
|
|
|
0x00e4: 0xe4, # LATIN SMALL LETTER A WITH DIAERESIS
|
|
|
|
|
0x00e5: 0xe5, # LATIN SMALL LETTER A WITH RING ABOVE
|
|
|
|
|
0x00e6: 0xbf, # LATIN SMALL LETTER AE
|
|
|
|
|
0x00e9: 0xe9, # LATIN SMALL LETTER E WITH ACUTE
|
|
|
|
|
0x00f3: 0xf3, # LATIN SMALL LETTER O WITH ACUTE
|
|
|
|
|
0x00f5: 0xf5, # LATIN SMALL LETTER O WITH TILDE
|
|
|
|
|
0x00f6: 0xf6, # LATIN SMALL LETTER O WITH DIAERESIS
|
|
|
|
|
0x00f7: 0xf7, # DIVISION SIGN
|
|
|
|
|
0x00f8: 0xb8, # LATIN SMALL LETTER O WITH STROKE
|
|
|
|
|
0x00fc: 0xfc, # LATIN SMALL LETTER U WITH DIAERESIS
|
|
|
|
|
0x0100: 0xc2, # LATIN CAPITAL LETTER A WITH MACRON
|
|
|
|
|
0x0101: 0xe2, # LATIN SMALL LETTER A WITH MACRON
|
|
|
|
|
0x0104: 0xc0, # LATIN CAPITAL LETTER A WITH OGONEK
|
|
|
|
|
0x0105: 0xe0, # LATIN SMALL LETTER A WITH OGONEK
|
|
|
|
|
0x0106: 0xc3, # LATIN CAPITAL LETTER C WITH ACUTE
|
|
|
|
|
0x0107: 0xe3, # LATIN SMALL LETTER C WITH ACUTE
|
|
|
|
|
0x010c: 0xc8, # LATIN CAPITAL LETTER C WITH CARON
|
|
|
|
|
0x010d: 0xe8, # LATIN SMALL LETTER C WITH CARON
|
|
|
|
|
0x0112: 0xc7, # LATIN CAPITAL LETTER E WITH MACRON
|
|
|
|
|
0x0113: 0xe7, # LATIN SMALL LETTER E WITH MACRON
|
|
|
|
|
0x0116: 0xcb, # LATIN CAPITAL LETTER E WITH DOT ABOVE
|
|
|
|
|
0x0117: 0xeb, # LATIN SMALL LETTER E WITH DOT ABOVE
|
|
|
|
|
0x0118: 0xc6, # LATIN CAPITAL LETTER E WITH OGONEK
|
|
|
|
|
0x0119: 0xe6, # LATIN SMALL LETTER E WITH OGONEK
|
|
|
|
|
0x0122: 0xcc, # LATIN CAPITAL LETTER G WITH CEDILLA
|
|
|
|
|
0x0123: 0xec, # LATIN SMALL LETTER G WITH CEDILLA
|
|
|
|
|
0x012a: 0xce, # LATIN CAPITAL LETTER I WITH MACRON
|
|
|
|
|
0x012b: 0xee, # LATIN SMALL LETTER I WITH MACRON
|
|
|
|
|
0x012e: 0xc1, # LATIN CAPITAL LETTER I WITH OGONEK
|
|
|
|
|
0x012f: 0xe1, # LATIN SMALL LETTER I WITH OGONEK
|
|
|
|
|
0x0136: 0xcd, # LATIN CAPITAL LETTER K WITH CEDILLA
|
|
|
|
|
0x0137: 0xed, # LATIN SMALL LETTER K WITH CEDILLA
|
|
|
|
|
0x013b: 0xcf, # LATIN CAPITAL LETTER L WITH CEDILLA
|
|
|
|
|
0x013c: 0xef, # LATIN SMALL LETTER L WITH CEDILLA
|
|
|
|
|
0x0141: 0xd9, # LATIN CAPITAL LETTER L WITH STROKE
|
|
|
|
|
0x0142: 0xf9, # LATIN SMALL LETTER L WITH STROKE
|
|
|
|
|
0x0143: 0xd1, # LATIN CAPITAL LETTER N WITH ACUTE
|
|
|
|
|
0x0144: 0xf1, # LATIN SMALL LETTER N WITH ACUTE
|
|
|
|
|
0x0145: 0xd2, # LATIN CAPITAL LETTER N WITH CEDILLA
|
|
|
|
|
0x0146: 0xf2, # LATIN SMALL LETTER N WITH CEDILLA
|
|
|
|
|
0x014c: 0xd4, # LATIN CAPITAL LETTER O WITH MACRON
|
|
|
|
|
0x014d: 0xf4, # LATIN SMALL LETTER O WITH MACRON
|
|
|
|
|
0x0156: 0xaa, # LATIN CAPITAL LETTER R WITH CEDILLA
|
|
|
|
|
0x0157: 0xba, # LATIN SMALL LETTER R WITH CEDILLA
|
|
|
|
|
0x015a: 0xda, # LATIN CAPITAL LETTER S WITH ACUTE
|
|
|
|
|
0x015b: 0xfa, # LATIN SMALL LETTER S WITH ACUTE
|
|
|
|
|
0x0160: 0xd0, # LATIN CAPITAL LETTER S WITH CARON
|
|
|
|
|
0x0161: 0xf0, # LATIN SMALL LETTER S WITH CARON
|
|
|
|
|
0x016a: 0xdb, # LATIN CAPITAL LETTER U WITH MACRON
|
|
|
|
|
0x016b: 0xfb, # LATIN SMALL LETTER U WITH MACRON
|
|
|
|
|
0x0172: 0xd8, # LATIN CAPITAL LETTER U WITH OGONEK
|
|
|
|
|
0x0173: 0xf8, # LATIN SMALL LETTER U WITH OGONEK
|
|
|
|
|
0x0179: 0xca, # LATIN CAPITAL LETTER Z WITH ACUTE
|
|
|
|
|
0x017a: 0xea, # LATIN SMALL LETTER Z WITH ACUTE
|
|
|
|
|
0x017b: 0xdd, # LATIN CAPITAL LETTER Z WITH DOT ABOVE
|
|
|
|
|
0x017c: 0xfd, # LATIN SMALL LETTER Z WITH DOT ABOVE
|
|
|
|
|
0x017d: 0xde, # LATIN CAPITAL LETTER Z WITH CARON
|
|
|
|
|
0x017e: 0xfe, # LATIN SMALL LETTER Z WITH CARON
|
|
|
|
|
0x02c7: 0x8e, # CARON
|
|
|
|
|
0x02d9: 0xff, # DOT ABOVE
|
|
|
|
|
0x02db: 0x9e, # OGONEK
|
|
|
|
|
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
|
2005-10-21 10:49:12 -03:00
|
|
|
}
|
