Skip to content

C
cpython
Commit 469cdad8 authored by Tim Peters's avatar Tim Peters
Browse files
Options

Whitespace normalization.

parent d7e8a0dd
Expand all Hide whitespace changes
Inline Side-by-side
Showing with 5817 additions and 5828 deletions
Lib/_strptime.py
View file @ 469cdad8
......@@ -2,7 +2,7 @@
CLASSES:
LocaleTime -- Discovers and/or stores locale-specific time information
TimeRE -- Creates regexes for pattern matching string of text containing
TimeRE -- Creates regexes for pattern matching string of text containing
time information as is returned by time.strftime()
FUNCTIONS:
......@@ -10,7 +10,7 @@ FUNCTIONS:
year
gregorian -- Calculates the Gregorian date based on the Julian day and
year
julianday -- Calculates the Julian day since the first of the year based
julianday -- Calculates the Julian day since the first of the year based
on the Gregorian date
dayofweek -- Calculates the day of the week from the Gregorian date.
strptime -- Calculates the time struct represented by the passed-in string
......@@ -40,23 +40,23 @@ class LocaleTime(object):
store the values have mangled names):
f_weekday -- full weekday names (7-item list)
a_weekday -- abbreviated weekday names (7-item list)
f_month -- full weekday names (14-item list; dummy value in [0], which
f_month -- full weekday names (14-item list; dummy value in [0], which
is added by code)
a_month -- abbreviated weekday names (13-item list, dummy value in
a_month -- abbreviated weekday names (13-item list, dummy value in
[0], which is added by code)
am_pm -- AM/PM representation (2-item list)
LC_date_time -- format string for date/time representation (string)
LC_date -- format string for date representation (string)
LC_time -- format string for time representation (string)
timezone -- daylight- and non-daylight-savings timezone representation
(3-item list; code tacks on blank item at end for
timezone -- daylight- and non-daylight-savings timezone representation
(3-item list; code tacks on blank item at end for
possible lack of timezone such as UTC)
lang -- Language used by instance (string)
"""
def __init__(self, f_weekday=None, a_weekday=None, f_month=None,
a_month=None, am_pm=None, LC_date_time=None, LC_time=None, LC_date=None,
def __init__(self, f_weekday=None, a_weekday=None, f_month=None,
a_month=None, am_pm=None, LC_date_time=None, LC_time=None, LC_date=None,
timezone=None, lang=None):
"""Optionally set attributes with passed-in values."""
if f_weekday is None: self.__f_weekday = None
......@@ -117,9 +117,9 @@ class LocaleTime(object):
if not self.__a_weekday: self.__calc_weekday()
return self.__a_weekday
f_weekday = property(__get_f_weekday, __set_nothing,
f_weekday = property(__get_f_weekday, __set_nothing,
doc="Full weekday names")
a_weekday = property(__get_a_weekday, __set_nothing,
a_weekday = property(__get_a_weekday, __set_nothing,
doc="Abbreviated weekday names")
def __get_f_month(self):
......@@ -187,7 +187,7 @@ class LocaleTime(object):
f_weekday = [calendar.day_name[i] for i in range(7)]
if not self.__a_weekday: self.__a_weekday = a_weekday
if not self.__f_weekday: self.__f_weekday = f_weekday
def __calc_month(self):
"""Set self.__f_month and self.__a_month using the calendar module."""
a_month = [calendar.month_abbr[i] for i in range(13)]
......@@ -197,11 +197,11 @@ class LocaleTime(object):
def __calc_am_pm(self):
"""Set self.__am_pm by using time.strftime().
The magic date (2002, 3, 17, hour, 44, 44, 2, 76, 0) is not really
that magical; just happened to have used it everywhere else where a
The magic date (2002, 3, 17, hour, 44, 44, 2, 76, 0) is not really
that magical; just happened to have used it everywhere else where a
static date was needed.
"""
am_pm = []
for hour in (01,22):
......@@ -211,10 +211,10 @@ class LocaleTime(object):
def __calc_date_time(self):
"""Set self.__date_time, self.__date, & self.__time by using time.strftime().
Use (1999,3,17,22,44,55,2,76,0) for magic date because the amount of
overloaded numbers is minimized. The order in which searches for
values within the format string is very important; it eliminates
Use (1999,3,17,22,44,55,2,76,0) for magic date because the amount of
overloaded numbers is minimized. The order in which searches for
values within the format string is very important; it eliminates
possible ambiguity for what something represents.
"""
......@@ -255,17 +255,17 @@ class LocaleTime(object):
def __calc_timezone(self):
"""Set self.__timezone by using time.tzname.
Empty string used for matching when timezone is not used/needed such
Empty string used for matching when timezone is not used/needed such
as with UTC.
"""
self.__timezone = self.__pad(time.tzname, 0)
def __calc_lang(self):
"""Set self.lang by using locale.getlocale() or
"""Set self.lang by using locale.getlocale() or
locale.getdefaultlocale().
"""
current_lang = locale.getlocale(locale.LC_TIME)[0]
if current_lang: self.__lang = current_lang
......@@ -277,7 +277,7 @@ class TimeRE(dict):
def __init__(self, locale_time=LocaleTime()):
"""Initialize instance with non-locale regexes and store LocaleTime object."""
super(TimeRE,self).__init__({
'd': r"(?P<d>3[0-1]|[0-2]\d|\d| \d)", #The " \d" option is
'd': r"(?P<d>3[0-1]|[0-2]\d|\d| \d)", #The " \d" option is
#to make %c from ANSI
#C work
'H': r"(?P<H>2[0-3]|[0-1]\d|\d)",
......@@ -299,16 +299,16 @@ class TimeRE(dict):
return super(TimeRE,self).__getitem__(fetch)
except KeyError:
if fetch == 'A':
self[fetch] = self.__seqToRE(self.locale_time.f_weekday,
self[fetch] = self.__seqToRE(self.locale_time.f_weekday,
fetch)
elif fetch == 'a':
self[fetch] = self.__seqToRE(self.locale_time.a_weekday,
self[fetch] = self.__seqToRE(self.locale_time.a_weekday,
fetch)
elif fetch == 'B':
self[fetch] = self.__seqToRE(self.locale_time.f_month[1:],
self[fetch] = self.__seqToRE(self.locale_time.f_month[1:],
fetch)
elif fetch == 'b':
self[fetch] = self.__seqToRE(self.locale_time.a_month[1:],
self[fetch] = self.__seqToRE(self.locale_time.a_month[1:],
fetch)
elif fetch == 'c':
self[fetch] = self.pattern(self.locale_time.LC_date_time)
......@@ -319,28 +319,28 @@ class TimeRE(dict):
elif fetch == 'X':
self[fetch] = self.pattern(self.locale_time.LC_time)
elif fetch == 'Z':
self[fetch] = self.__seqToRE(self.locale_time.timezone,
self[fetch] = self.__seqToRE(self.locale_time.timezone,
fetch)
elif fetch == '%':
return '%'
return super(TimeRE,self).__getitem__(fetch)
def __seqToRE(self, to_convert, directive):
"""Convert a list to a regex string for matching directive."""
def sorter(a, b):
"""Sort based on length.
Done in case for some strange reason that names in the locale only
differ by a suffix and thus want the name with the suffix to match
first.
"""
try: a_length = len(a)
except TypeError: a_length = 0
try: b_length = len(b)
except TypeError: b_length = 0
return cmp(b_length, a_length)
to_convert = to_convert[:] #Don't want to change value in-place.
to_convert.sort(sorter)
regex = '(?P<%s>' % directive
......@@ -357,7 +357,7 @@ class TimeRE(dict):
format = format.replace(whitespace, r'\s*')
while format.find('%') != -1:
directive_index = format.index('%')+1
processed_format = "%s%s%s" % (processed_format,
processed_format = "%s%s%s" % (processed_format,
format[:directive_index-1],
self[format[directive_index]])
format = format[directive_index+1:]
......@@ -371,12 +371,12 @@ class TimeRE(dict):
def strptime(data_string, format="%a %b %d %H:%M:%S %Y"):
"""Convert data_string to a time struct based on the format string or re object; will return an re object for format if data_string is False.
The object passed in for format may either be a re object compiled by
strptime() or a format string. If False is passed in for data_string
then an re object for format will be returned. The re object
The object passed in for format may either be a re object compiled by
strptime() or a format string. If False is passed in for data_string
then an re object for format will be returned. The re object
must be used with the same language as used to compile the re object.
"""
locale_time = LocaleTime()
if isinstance(format, type(re_compile(''))):
......
Lib/encodings/__init__.py
View file @ 469cdad8
......@@ -39,7 +39,7 @@ class CodecRegistryError(exceptions.LookupError,
pass
def search_function(encoding):
# Cache lookup
entry = _cache.get(encoding, _unknown)
if entry is not _unknown:
......@@ -72,8 +72,8 @@ def search_function(encoding):
if mod is None:
# Cache misses
_cache[encoding] = None
return None
return None
# Now ask the module for the registry entry
entry = tuple(getregentry())
if len(entry) != 4:
......
Lib/encodings/ascii.py
View file @ 469cdad8
......@@ -19,7 +19,7 @@ class Codec(codecs.Codec):
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......
Lib/encodings/base64_codec.py
View file @ 469cdad8
......@@ -51,7 +51,7 @@ class Codec(codecs.Codec):
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......
Lib/encodings/charmap.py
View file @ 469cdad8
......@@ -2,7 +2,7 @@
Use this codec directly rather than through the automatic
conversion mechanisms supplied by unicode() and .encode().
Written by Marc-Andre Lemburg (mal@lemburg.com).
......@@ -31,7 +31,7 @@ class StreamWriter(Codec,codecs.StreamWriter):
def encode(self,input,errors='strict'):
return Codec.encode(input,errors,self.mapping)
class StreamReader(Codec,codecs.StreamReader):
def __init__(self,stream,errors='strict',mapping=None):
......@@ -48,4 +48,3 @@ class StreamReader(Codec,codecs.StreamReader):
def getregentry():
return (Codec.encode,Codec.decode,StreamReader,StreamWriter)
Lib/encodings/cp037.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp1006.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp1026.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp1140.py
View file @ 469cdad8
......@@ -14,14 +14,14 @@ class Codec(codecs.Codec):
def encode(self,input,errors='strict'):
return codecs.charmap_encode(input,errors,encoding_map)
def decode(self,input,errors='strict'):
return codecs.charmap_decode(input,errors,decoding_map)
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......@@ -36,10 +36,9 @@ def getregentry():
decoding_map = copy.copy(cp037.decoding_map)
decoding_map.update({
0x009f: 0x20ac # EURO SIGN
0x009f: 0x20ac # EURO SIGN
})
### Encoding Map
encoding_map = codecs.make_encoding_map(decoding_map)
Lib/encodings/cp1250.py
View file @ 469cdad8
......@@ -16,14 +16,14 @@ class Codec(codecs.Codec):
def encode(self,input,errors='strict'):
return codecs.charmap_encode(input,errors,encoding_map)
def decode(self,input,errors='strict'):
return codecs.charmap_decode(input,errors,decoding_map)
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......@@ -37,85 +37,85 @@ def getregentry():
decoding_map = codecs.make_identity_dict(range(256))
decoding_map.update({
0x0080: 0x20ac, # EURO SIGN
0x0081: None, # UNDEFINED
0x0082: 0x201a, # SINGLE LOW-9 QUOTATION MARK
0x0083: None, # UNDEFINED
0x0084: 0x201e, # DOUBLE LOW-9 QUOTATION MARK
0x0085: 0x2026, # HORIZONTAL ELLIPSIS
0x0086: 0x2020, # DAGGER
0x0087: 0x2021, # DOUBLE DAGGER
0x0088: None, # UNDEFINED
0x0089: 0x2030, # PER MILLE SIGN
0x008a: 0x0160, # LATIN CAPITAL LETTER S WITH CARON
0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
0x008c: 0x015a, # LATIN CAPITAL LETTER S WITH ACUTE
0x008d: 0x0164, # LATIN CAPITAL LETTER T WITH CARON
0x008e: 0x017d, # LATIN CAPITAL LETTER Z WITH CARON
0x008f: 0x0179, # LATIN CAPITAL LETTER Z WITH ACUTE
0x0090: None, # UNDEFINED
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: None, # UNDEFINED
0x0099: 0x2122, # TRADE MARK SIGN
0x009a: 0x0161, # LATIN SMALL LETTER S WITH CARON
0x009b: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
0x009c: 0x015b, # LATIN SMALL LETTER S WITH ACUTE
0x009d: 0x0165, # LATIN SMALL LETTER T WITH CARON
0x009e: 0x017e, # LATIN SMALL LETTER Z WITH CARON
0x009f: 0x017a, # LATIN SMALL LETTER Z WITH ACUTE
0x00a1: 0x02c7, # CARON
0x00a2: 0x02d8, # BREVE
0x00a3: 0x0141, # LATIN CAPITAL LETTER L WITH STROKE
0x00a5: 0x0104, # LATIN CAPITAL LETTER A WITH OGONEK
0x00aa: 0x015e, # LATIN CAPITAL LETTER S WITH CEDILLA
0x00af: 0x017b, # LATIN CAPITAL LETTER Z WITH DOT ABOVE
0x00b2: 0x02db, # OGONEK
0x00b3: 0x0142, # LATIN SMALL LETTER L WITH STROKE
0x00b9: 0x0105, # LATIN SMALL LETTER A WITH OGONEK
0x00ba: 0x015f, # LATIN SMALL LETTER S WITH CEDILLA
0x00bc: 0x013d, # LATIN CAPITAL LETTER L WITH CARON
0x00bd: 0x02dd, # DOUBLE ACUTE ACCENT
0x00be: 0x013e, # LATIN SMALL LETTER L WITH CARON
0x00bf: 0x017c, # LATIN SMALL LETTER Z WITH DOT ABOVE
0x00c0: 0x0154, # LATIN CAPITAL LETTER R WITH ACUTE
0x00c3: 0x0102, # LATIN CAPITAL LETTER A WITH BREVE
0x00c5: 0x0139, # LATIN CAPITAL LETTER L WITH ACUTE
0x00c6: 0x0106, # LATIN CAPITAL LETTER C WITH ACUTE
0x00c8: 0x010c, # LATIN CAPITAL LETTER C WITH CARON
0x00ca: 0x0118, # LATIN CAPITAL LETTER E WITH OGONEK
0x00cc: 0x011a, # LATIN CAPITAL LETTER E WITH CARON
0x00cf: 0x010e, # LATIN CAPITAL LETTER D WITH CARON
0x00d0: 0x0110, # LATIN CAPITAL LETTER D WITH STROKE
0x00d1: 0x0143, # LATIN CAPITAL LETTER N WITH ACUTE
0x00d2: 0x0147, # LATIN CAPITAL LETTER N WITH CARON
0x00d5: 0x0150, # LATIN CAPITAL LETTER O WITH DOUBLE ACUTE
0x00d8: 0x0158, # LATIN CAPITAL LETTER R WITH CARON
0x00d9: 0x016e, # LATIN CAPITAL LETTER U WITH RING ABOVE
0x00db: 0x0170, # LATIN CAPITAL LETTER U WITH DOUBLE ACUTE
0x00de: 0x0162, # LATIN CAPITAL LETTER T WITH CEDILLA
0x00e0: 0x0155, # LATIN SMALL LETTER R WITH ACUTE
0x00e3: 0x0103, # LATIN SMALL LETTER A WITH BREVE
0x00e5: 0x013a, # LATIN SMALL LETTER L WITH ACUTE
0x00e6: 0x0107, # LATIN SMALL LETTER C WITH ACUTE
0x00e8: 0x010d, # LATIN SMALL LETTER C WITH CARON
0x00ea: 0x0119, # LATIN SMALL LETTER E WITH OGONEK
0x00ec: 0x011b, # LATIN SMALL LETTER E WITH CARON
0x00ef: 0x010f, # LATIN SMALL LETTER D WITH CARON
0x00f0: 0x0111, # LATIN SMALL LETTER D WITH STROKE
0x00f1: 0x0144, # LATIN SMALL LETTER N WITH ACUTE
0x00f2: 0x0148, # LATIN SMALL LETTER N WITH CARON
0x00f5: 0x0151, # LATIN SMALL LETTER O WITH DOUBLE ACUTE
0x00f8: 0x0159, # LATIN SMALL LETTER R WITH CARON
0x00f9: 0x016f, # LATIN SMALL LETTER U WITH RING ABOVE
0x00fb: 0x0171, # LATIN SMALL LETTER U WITH DOUBLE ACUTE
0x00fe: 0x0163, # LATIN SMALL LETTER T WITH CEDILLA
0x00ff: 0x02d9, # DOT ABOVE
0x0080: 0x20ac, # EURO SIGN
0x0081: None, # UNDEFINED
0x0082: 0x201a, # SINGLE LOW-9 QUOTATION MARK
0x0083: None, # UNDEFINED
0x0084: 0x201e, # DOUBLE LOW-9 QUOTATION MARK
0x0085: 0x2026, # HORIZONTAL ELLIPSIS
0x0086: 0x2020, # DAGGER
0x0087: 0x2021, # DOUBLE DAGGER
0x0088: None, # UNDEFINED
0x0089: 0x2030, # PER MILLE SIGN
0x008a: 0x0160, # LATIN CAPITAL LETTER S WITH CARON
0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
0x008c: 0x015a, # LATIN CAPITAL LETTER S WITH ACUTE
0x008d: 0x0164, # LATIN CAPITAL LETTER T WITH CARON
0x008e: 0x017d, # LATIN CAPITAL LETTER Z WITH CARON
0x008f: 0x0179, # LATIN CAPITAL LETTER Z WITH ACUTE
0x0090: None, # UNDEFINED
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: None, # UNDEFINED
0x0099: 0x2122, # TRADE MARK SIGN
0x009a: 0x0161, # LATIN SMALL LETTER S WITH CARON
0x009b: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
0x009c: 0x015b, # LATIN SMALL LETTER S WITH ACUTE
0x009d: 0x0165, # LATIN SMALL LETTER T WITH CARON
0x009e: 0x017e, # LATIN SMALL LETTER Z WITH CARON
0x009f: 0x017a, # LATIN SMALL LETTER Z WITH ACUTE
0x00a1: 0x02c7, # CARON
0x00a2: 0x02d8, # BREVE
0x00a3: 0x0141, # LATIN CAPITAL LETTER L WITH STROKE
0x00a5: 0x0104, # LATIN CAPITAL LETTER A WITH OGONEK
0x00aa: 0x015e, # LATIN CAPITAL LETTER S WITH CEDILLA
0x00af: 0x017b, # LATIN CAPITAL LETTER Z WITH DOT ABOVE
0x00b2: 0x02db, # OGONEK
0x00b3: 0x0142, # LATIN SMALL LETTER L WITH STROKE
0x00b9: 0x0105, # LATIN SMALL LETTER A WITH OGONEK
0x00ba: 0x015f, # LATIN SMALL LETTER S WITH CEDILLA
0x00bc: 0x013d, # LATIN CAPITAL LETTER L WITH CARON
0x00bd: 0x02dd, # DOUBLE ACUTE ACCENT
0x00be: 0x013e, # LATIN SMALL LETTER L WITH CARON
0x00bf: 0x017c, # LATIN SMALL LETTER Z WITH DOT ABOVE
0x00c0: 0x0154, # LATIN CAPITAL LETTER R WITH ACUTE
0x00c3: 0x0102, # LATIN CAPITAL LETTER A WITH BREVE
0x00c5: 0x0139, # LATIN CAPITAL LETTER L WITH ACUTE
0x00c6: 0x0106, # LATIN CAPITAL LETTER C WITH ACUTE
0x00c8: 0x010c, # LATIN CAPITAL LETTER C WITH CARON
0x00ca: 0x0118, # LATIN CAPITAL LETTER E WITH OGONEK
0x00cc: 0x011a, # LATIN CAPITAL LETTER E WITH CARON
0x00cf: 0x010e, # LATIN CAPITAL LETTER D WITH CARON
0x00d0: 0x0110, # LATIN CAPITAL LETTER D WITH STROKE
0x00d1: 0x0143, # LATIN CAPITAL LETTER N WITH ACUTE
0x00d2: 0x0147, # LATIN CAPITAL LETTER N WITH CARON
0x00d5: 0x0150, # LATIN CAPITAL LETTER O WITH DOUBLE ACUTE
0x00d8: 0x0158, # LATIN CAPITAL LETTER R WITH CARON
0x00d9: 0x016e, # LATIN CAPITAL LETTER U WITH RING ABOVE
0x00db: 0x0170, # LATIN CAPITAL LETTER U WITH DOUBLE ACUTE
0x00de: 0x0162, # LATIN CAPITAL LETTER T WITH CEDILLA
0x00e0: 0x0155, # LATIN SMALL LETTER R WITH ACUTE
0x00e3: 0x0103, # LATIN SMALL LETTER A WITH BREVE
0x00e5: 0x013a, # LATIN SMALL LETTER L WITH ACUTE
0x00e6: 0x0107, # LATIN SMALL LETTER C WITH ACUTE
0x00e8: 0x010d, # LATIN SMALL LETTER C WITH CARON
0x00ea: 0x0119, # LATIN SMALL LETTER E WITH OGONEK
0x00ec: 0x011b, # LATIN SMALL LETTER E WITH CARON
0x00ef: 0x010f, # LATIN SMALL LETTER D WITH CARON
0x00f0: 0x0111, # LATIN SMALL LETTER D WITH STROKE
0x00f1: 0x0144, # LATIN SMALL LETTER N WITH ACUTE
0x00f2: 0x0148, # LATIN SMALL LETTER N WITH CARON
0x00f5: 0x0151, # LATIN SMALL LETTER O WITH DOUBLE ACUTE
0x00f8: 0x0159, # LATIN SMALL LETTER R WITH CARON
0x00f9: 0x016f, # LATIN SMALL LETTER U WITH RING ABOVE
0x00fb: 0x0171, # LATIN SMALL LETTER U WITH DOUBLE ACUTE
0x00fe: 0x0163, # LATIN SMALL LETTER T WITH CEDILLA
0x00ff: 0x02d9, # DOT ABOVE
})
### Encoding Map
......
Lib/encodings/cp1251.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp1252.py
View file @ 469cdad8
......@@ -16,14 +16,14 @@ class Codec(codecs.Codec):
def encode(self,input,errors='strict'):
return codecs.charmap_encode(input,errors,encoding_map)
def decode(self,input,errors='strict'):
return codecs.charmap_decode(input,errors,decoding_map)
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......@@ -37,38 +37,38 @@ def getregentry():
decoding_map = codecs.make_identity_dict(range(256))
decoding_map.update({
0x0080: 0x20ac, # EURO SIGN
0x0081: None, # UNDEFINED
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: 0x0160, # LATIN CAPITAL LETTER S WITH CARON
0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
0x008c: 0x0152, # LATIN CAPITAL LIGATURE OE
0x008d: None, # UNDEFINED
0x008e: 0x017d, # LATIN CAPITAL LETTER Z WITH CARON
0x008f: None, # UNDEFINED
0x0090: None, # UNDEFINED
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: 0x02dc, # SMALL TILDE
0x0099: 0x2122, # TRADE MARK SIGN
0x009a: 0x0161, # LATIN SMALL LETTER S WITH CARON
0x009b: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
0x009c: 0x0153, # LATIN SMALL LIGATURE OE
0x009d: None, # UNDEFINED
0x009e: 0x017e, # LATIN SMALL LETTER Z WITH CARON
0x009f: 0x0178, # LATIN CAPITAL LETTER Y WITH DIAERESIS
0x0080: 0x20ac, # EURO SIGN
0x0081: None, # UNDEFINED
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: 0x0160, # LATIN CAPITAL LETTER S WITH CARON
0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
0x008c: 0x0152, # LATIN CAPITAL LIGATURE OE
0x008d: None, # UNDEFINED
0x008e: 0x017d, # LATIN CAPITAL LETTER Z WITH CARON
0x008f: None, # UNDEFINED
0x0090: None, # UNDEFINED
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: 0x02dc, # SMALL TILDE
0x0099: 0x2122, # TRADE MARK SIGN
0x009a: 0x0161, # LATIN SMALL LETTER S WITH CARON
0x009b: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
0x009c: 0x0153, # LATIN SMALL LIGATURE OE
0x009d: None, # UNDEFINED
0x009e: 0x017e, # LATIN SMALL LETTER Z WITH CARON
0x009f: 0x0178, # LATIN CAPITAL LETTER Y WITH DIAERESIS
})
### Encoding Map
......
Lib/encodings/cp1253.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp1254.py
View file @ 469cdad8
......@@ -16,14 +16,14 @@ class Codec(codecs.Codec):
def encode(self,input,errors='strict'):
return codecs.charmap_encode(input,errors,encoding_map)
def decode(self,input,errors='strict'):
return codecs.charmap_decode(input,errors,decoding_map)
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......@@ -37,44 +37,44 @@ def getregentry():
decoding_map = codecs.make_identity_dict(range(256))
decoding_map.update({
0x0080: 0x20ac, # EURO SIGN
0x0081: None, # UNDEFINED
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: 0x0160, # LATIN CAPITAL LETTER S WITH CARON
0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
0x008c: 0x0152, # LATIN CAPITAL LIGATURE OE
0x008d: None, # UNDEFINED
0x008e: None, # UNDEFINED
0x008f: None, # UNDEFINED
0x0090: None, # UNDEFINED
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: 0x02dc, # SMALL TILDE
0x0099: 0x2122, # TRADE MARK SIGN
0x009a: 0x0161, # LATIN SMALL LETTER S WITH CARON
0x009b: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
0x009c: 0x0153, # LATIN SMALL LIGATURE OE
0x009d: None, # UNDEFINED
0x009e: None, # UNDEFINED
0x009f: 0x0178, # LATIN CAPITAL LETTER Y WITH DIAERESIS
0x00d0: 0x011e, # LATIN CAPITAL LETTER G WITH BREVE
0x00dd: 0x0130, # LATIN CAPITAL LETTER I WITH DOT ABOVE
0x00de: 0x015e, # LATIN CAPITAL LETTER S WITH CEDILLA
0x00f0: 0x011f, # LATIN SMALL LETTER G WITH BREVE
0x00fd: 0x0131, # LATIN SMALL LETTER DOTLESS I
0x00fe: 0x015f, # LATIN SMALL LETTER S WITH CEDILLA
0x0080: 0x20ac, # EURO SIGN
0x0081: None, # UNDEFINED
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: 0x0160, # LATIN CAPITAL LETTER S WITH CARON
0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
0x008c: 0x0152, # LATIN CAPITAL LIGATURE OE
0x008d: None, # UNDEFINED
0x008e: None, # UNDEFINED
0x008f: None, # UNDEFINED
0x0090: None, # UNDEFINED
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: 0x02dc, # SMALL TILDE
0x0099: 0x2122, # TRADE MARK SIGN
0x009a: 0x0161, # LATIN SMALL LETTER S WITH CARON
0x009b: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
0x009c: 0x0153, # LATIN SMALL LIGATURE OE
0x009d: None, # UNDEFINED
0x009e: None, # UNDEFINED
0x009f: 0x0178, # LATIN CAPITAL LETTER Y WITH DIAERESIS
0x00d0: 0x011e, # LATIN CAPITAL LETTER G WITH BREVE
0x00dd: 0x0130, # LATIN CAPITAL LETTER I WITH DOT ABOVE
0x00de: 0x015e, # LATIN CAPITAL LETTER S WITH CEDILLA
0x00f0: 0x011f, # LATIN SMALL LETTER G WITH BREVE
0x00fd: 0x0131, # LATIN SMALL LETTER DOTLESS I
0x00fe: 0x015f, # LATIN SMALL LETTER S WITH CEDILLA
})
### Encoding Map
......
Lib/encodings/cp1255.py
View file @ 469cdad8
......@@ -16,14 +16,14 @@ class Codec(codecs.Codec):
def encode(self,input,errors='strict'):
return codecs.charmap_encode(input,errors,encoding_map)
def decode(self,input,errors='strict'):
return codecs.charmap_decode(input,errors,decoding_map)
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......@@ -37,105 +37,105 @@ def getregentry():
decoding_map = codecs.make_identity_dict(range(256))
decoding_map.update({
0x0080: 0x20ac, # EURO SIGN
0x0081: None, # UNDEFINED
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: None, # UNDEFINED
0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
0x008c: None, # UNDEFINED
0x008d: None, # UNDEFINED
0x008e: None, # UNDEFINED
0x008f: None, # UNDEFINED
0x0090: None, # UNDEFINED
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: 0x02dc, # SMALL TILDE
0x0099: 0x2122, # TRADE MARK SIGN
0x009a: None, # UNDEFINED
0x009b: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
0x009c: None, # UNDEFINED
0x009d: None, # UNDEFINED
0x009e: None, # UNDEFINED
0x009f: None, # UNDEFINED
0x00a4: 0x20aa, # NEW SHEQEL SIGN
0x00aa: 0x00d7, # MULTIPLICATION SIGN
0x00ba: 0x00f7, # DIVISION SIGN
0x00c0: 0x05b0, # HEBREW POINT SHEVA
0x00c1: 0x05b1, # HEBREW POINT HATAF SEGOL
0x00c2: 0x05b2, # HEBREW POINT HATAF PATAH
0x00c3: 0x05b3, # HEBREW POINT HATAF QAMATS
0x00c4: 0x05b4, # HEBREW POINT HIRIQ
0x00c5: 0x05b5, # HEBREW POINT TSERE
0x00c6: 0x05b6, # HEBREW POINT SEGOL
0x00c7: 0x05b7, # HEBREW POINT PATAH
0x00c8: 0x05b8, # HEBREW POINT QAMATS
0x00c9: 0x05b9, # HEBREW POINT HOLAM
0x00ca: None, # UNDEFINED
0x00cb: 0x05bb, # HEBREW POINT QUBUTS
0x00cc: 0x05bc, # HEBREW POINT DAGESH OR MAPIQ
0x00cd: 0x05bd, # HEBREW POINT METEG
0x00ce: 0x05be, # HEBREW PUNCTUATION MAQAF
0x00cf: 0x05bf, # HEBREW POINT RAFE
0x00d0: 0x05c0, # HEBREW PUNCTUATION PASEQ
0x00d1: 0x05c1, # HEBREW POINT SHIN DOT
0x00d2: 0x05c2, # HEBREW POINT SIN DOT
0x00d3: 0x05c3, # HEBREW PUNCTUATION SOF PASUQ
0x00d4: 0x05f0, # HEBREW LIGATURE YIDDISH DOUBLE VAV
0x00d5: 0x05f1, # HEBREW LIGATURE YIDDISH VAV YOD
0x00d6: 0x05f2, # HEBREW LIGATURE YIDDISH DOUBLE YOD
0x00d7: 0x05f3, # HEBREW PUNCTUATION GERESH
0x00d8: 0x05f4, # HEBREW PUNCTUATION GERSHAYIM
0x00d9: None, # UNDEFINED
0x00da: None, # UNDEFINED
0x00db: None, # UNDEFINED
0x00dc: None, # UNDEFINED
0x00dd: None, # UNDEFINED
0x00de: None, # UNDEFINED
0x00df: None, # UNDEFINED
0x00e0: 0x05d0, # HEBREW LETTER ALEF
0x00e1: 0x05d1, # HEBREW LETTER BET
0x00e2: 0x05d2, # HEBREW LETTER GIMEL
0x00e3: 0x05d3, # HEBREW LETTER DALET
0x00e4: 0x05d4, # HEBREW LETTER HE
0x00e5: 0x05d5, # HEBREW LETTER VAV
0x00e6: 0x05d6, # HEBREW LETTER ZAYIN
0x00e7: 0x05d7, # HEBREW LETTER HET
0x00e8: 0x05d8, # HEBREW LETTER TET
0x00e9: 0x05d9, # HEBREW LETTER YOD
0x00ea: 0x05da, # HEBREW LETTER FINAL KAF
0x00eb: 0x05db, # HEBREW LETTER KAF
0x00ec: 0x05dc, # HEBREW LETTER LAMED
0x00ed: 0x05dd, # HEBREW LETTER FINAL MEM
0x00ee: 0x05de, # HEBREW LETTER MEM
0x00ef: 0x05df, # HEBREW LETTER FINAL NUN
0x00f0: 0x05e0, # HEBREW LETTER NUN
0x00f1: 0x05e1, # HEBREW LETTER SAMEKH
0x00f2: 0x05e2, # HEBREW LETTER AYIN
0x00f3: 0x05e3, # HEBREW LETTER FINAL PE
0x00f4: 0x05e4, # HEBREW LETTER PE
0x00f5: 0x05e5, # HEBREW LETTER FINAL TSADI
0x00f6: 0x05e6, # HEBREW LETTER TSADI
0x00f7: 0x05e7, # HEBREW LETTER QOF
0x00f8: 0x05e8, # HEBREW LETTER RESH
0x00f9: 0x05e9, # HEBREW LETTER SHIN
0x00fa: 0x05ea, # HEBREW LETTER TAV
0x00fb: None, # UNDEFINED
0x00fc: None, # UNDEFINED
0x00fd: 0x200e, # LEFT-TO-RIGHT MARK
0x00fe: 0x200f, # RIGHT-TO-LEFT MARK
0x00ff: None, # UNDEFINED
0x0080: 0x20ac, # EURO SIGN
0x0081: None, # UNDEFINED
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: None, # UNDEFINED
0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
0x008c: None, # UNDEFINED
0x008d: None, # UNDEFINED
0x008e: None, # UNDEFINED
0x008f: None, # UNDEFINED
0x0090: None, # UNDEFINED
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: 0x02dc, # SMALL TILDE
0x0099: 0x2122, # TRADE MARK SIGN
0x009a: None, # UNDEFINED
0x009b: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
0x009c: None, # UNDEFINED
0x009d: None, # UNDEFINED
0x009e: None, # UNDEFINED
0x009f: None, # UNDEFINED
0x00a4: 0x20aa, # NEW SHEQEL SIGN
0x00aa: 0x00d7, # MULTIPLICATION SIGN
0x00ba: 0x00f7, # DIVISION SIGN
0x00c0: 0x05b0, # HEBREW POINT SHEVA
0x00c1: 0x05b1, # HEBREW POINT HATAF SEGOL
0x00c2: 0x05b2, # HEBREW POINT HATAF PATAH
0x00c3: 0x05b3, # HEBREW POINT HATAF QAMATS
0x00c4: 0x05b4, # HEBREW POINT HIRIQ
0x00c5: 0x05b5, # HEBREW POINT TSERE
0x00c6: 0x05b6, # HEBREW POINT SEGOL
0x00c7: 0x05b7, # HEBREW POINT PATAH
0x00c8: 0x05b8, # HEBREW POINT QAMATS
0x00c9: 0x05b9, # HEBREW POINT HOLAM
0x00ca: None, # UNDEFINED
0x00cb: 0x05bb, # HEBREW POINT QUBUTS
0x00cc: 0x05bc, # HEBREW POINT DAGESH OR MAPIQ
0x00cd: 0x05bd, # HEBREW POINT METEG
0x00ce: 0x05be, # HEBREW PUNCTUATION MAQAF
0x00cf: 0x05bf, # HEBREW POINT RAFE
0x00d0: 0x05c0, # HEBREW PUNCTUATION PASEQ
0x00d1: 0x05c1, # HEBREW POINT SHIN DOT
0x00d2: 0x05c2, # HEBREW POINT SIN DOT
0x00d3: 0x05c3, # HEBREW PUNCTUATION SOF PASUQ
0x00d4: 0x05f0, # HEBREW LIGATURE YIDDISH DOUBLE VAV
0x00d5: 0x05f1, # HEBREW LIGATURE YIDDISH VAV YOD
0x00d6: 0x05f2, # HEBREW LIGATURE YIDDISH DOUBLE YOD
0x00d7: 0x05f3, # HEBREW PUNCTUATION GERESH
0x00d8: 0x05f4, # HEBREW PUNCTUATION GERSHAYIM
0x00d9: None, # UNDEFINED
0x00da: None, # UNDEFINED
0x00db: None, # UNDEFINED
0x00dc: None, # UNDEFINED
0x00dd: None, # UNDEFINED
0x00de: None, # UNDEFINED
0x00df: None, # UNDEFINED
0x00e0: 0x05d0, # HEBREW LETTER ALEF
0x00e1: 0x05d1, # HEBREW LETTER BET
0x00e2: 0x05d2, # HEBREW LETTER GIMEL
0x00e3: 0x05d3, # HEBREW LETTER DALET
0x00e4: 0x05d4, # HEBREW LETTER HE
0x00e5: 0x05d5, # HEBREW LETTER VAV
0x00e6: 0x05d6, # HEBREW LETTER ZAYIN
0x00e7: 0x05d7, # HEBREW LETTER HET
0x00e8: 0x05d8, # HEBREW LETTER TET
0x00e9: 0x05d9, # HEBREW LETTER YOD
0x00ea: 0x05da, # HEBREW LETTER FINAL KAF
0x00eb: 0x05db, # HEBREW LETTER KAF
0x00ec: 0x05dc, # HEBREW LETTER LAMED
0x00ed: 0x05dd, # HEBREW LETTER FINAL MEM
0x00ee: 0x05de, # HEBREW LETTER MEM
0x00ef: 0x05df, # HEBREW LETTER FINAL NUN
0x00f0: 0x05e0, # HEBREW LETTER NUN
0x00f1: 0x05e1, # HEBREW LETTER SAMEKH
0x00f2: 0x05e2, # HEBREW LETTER AYIN
0x00f3: 0x05e3, # HEBREW LETTER FINAL PE
0x00f4: 0x05e4, # HEBREW LETTER PE
0x00f5: 0x05e5, # HEBREW LETTER FINAL TSADI
0x00f6: 0x05e6, # HEBREW LETTER TSADI
0x00f7: 0x05e7, # HEBREW LETTER QOF
0x00f8: 0x05e8, # HEBREW LETTER RESH
0x00f9: 0x05e9, # HEBREW LETTER SHIN
0x00fa: 0x05ea, # HEBREW LETTER TAV
0x00fb: None, # UNDEFINED
0x00fc: None, # UNDEFINED
0x00fd: 0x200e, # LEFT-TO-RIGHT MARK
0x00fe: 0x200f, # RIGHT-TO-LEFT MARK
0x00ff: None, # UNDEFINED
})
### Encoding Map
......
Lib/encodings/cp1256.py
View file @ 469cdad8
......@@ -16,14 +16,14 @@ class Codec(codecs.Codec):
def encode(self,input,errors='strict'):
return codecs.charmap_encode(input,errors,encoding_map)
def decode(self,input,errors='strict'):
return codecs.charmap_decode(input,errors,decoding_map)
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......@@ -37,91 +37,91 @@ def getregentry():
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
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
})
### Encoding Map
......
Lib/encodings/cp1257.py
View file @ 469cdad8
......@@ -16,14 +16,14 @@ class Codec(codecs.Codec):
def encode(self,input,errors='strict'):
return codecs.charmap_encode(input,errors,encoding_map)
def decode(self,input,errors='strict'):
return codecs.charmap_decode(input,errors,decoding_map)
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......@@ -37,93 +37,93 @@ def getregentry():
decoding_map = codecs.make_identity_dict(range(256))
decoding_map.update({
0x0080: 0x20ac, # EURO SIGN
0x0081: None, # UNDEFINED
0x0082: 0x201a, # SINGLE LOW-9 QUOTATION MARK
0x0083: None, # UNDEFINED
0x0084: 0x201e, # DOUBLE LOW-9 QUOTATION MARK
0x0085: 0x2026, # HORIZONTAL ELLIPSIS
0x0086: 0x2020, # DAGGER
0x0087: 0x2021, # DOUBLE DAGGER
0x0088: None, # UNDEFINED
0x0089: 0x2030, # PER MILLE SIGN
0x008a: None, # UNDEFINED
0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
0x008c: None, # UNDEFINED
0x008d: 0x00a8, # DIAERESIS
0x008e: 0x02c7, # CARON
0x008f: 0x00b8, # CEDILLA
0x0090: None, # UNDEFINED
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: None, # UNDEFINED
0x0099: 0x2122, # TRADE MARK SIGN
0x009a: None, # UNDEFINED
0x009b: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
0x009c: None, # UNDEFINED
0x009d: 0x00af, # MACRON
0x009e: 0x02db, # OGONEK
0x009f: None, # UNDEFINED
0x00a1: None, # UNDEFINED
0x00a5: None, # UNDEFINED
0x00a8: 0x00d8, # LATIN CAPITAL LETTER O WITH STROKE
0x00aa: 0x0156, # LATIN CAPITAL LETTER R WITH CEDILLA
0x00af: 0x00c6, # LATIN CAPITAL LETTER AE
0x00b8: 0x00f8, # LATIN SMALL LETTER O WITH STROKE
0x00ba: 0x0157, # LATIN SMALL LETTER R WITH CEDILLA
0x00bf: 0x00e6, # LATIN SMALL LETTER AE
0x00c0: 0x0104, # LATIN CAPITAL LETTER A WITH OGONEK
0x00c1: 0x012e, # LATIN CAPITAL LETTER I WITH OGONEK
0x00c2: 0x0100, # LATIN CAPITAL LETTER A WITH MACRON
0x00c3: 0x0106, # LATIN CAPITAL LETTER C WITH ACUTE
0x00c6: 0x0118, # LATIN CAPITAL LETTER E WITH OGONEK
0x00c7: 0x0112, # LATIN CAPITAL LETTER E WITH MACRON
0x00c8: 0x010c, # LATIN CAPITAL LETTER C WITH CARON
0x00ca: 0x0179, # LATIN CAPITAL LETTER Z WITH ACUTE
0x00cb: 0x0116, # LATIN CAPITAL LETTER E WITH DOT ABOVE
0x00cc: 0x0122, # LATIN CAPITAL LETTER G WITH CEDILLA
0x00cd: 0x0136, # LATIN CAPITAL LETTER K WITH CEDILLA
0x00ce: 0x012a, # LATIN CAPITAL LETTER I WITH MACRON
0x00cf: 0x013b, # LATIN CAPITAL LETTER L WITH CEDILLA
0x00d0: 0x0160, # LATIN CAPITAL LETTER S WITH CARON
0x00d1: 0x0143, # LATIN CAPITAL LETTER N WITH ACUTE
0x00d2: 0x0145, # LATIN CAPITAL LETTER N WITH CEDILLA
0x00d4: 0x014c, # LATIN CAPITAL LETTER O WITH MACRON
0x00d8: 0x0172, # LATIN CAPITAL LETTER U WITH OGONEK
0x00d9: 0x0141, # LATIN CAPITAL LETTER L WITH STROKE
0x00da: 0x015a, # LATIN CAPITAL LETTER S WITH ACUTE
0x00db: 0x016a, # LATIN CAPITAL LETTER U WITH MACRON
0x00dd: 0x017b, # LATIN CAPITAL LETTER Z WITH DOT ABOVE
0x00de: 0x017d, # LATIN CAPITAL LETTER Z WITH CARON
0x00e0: 0x0105, # LATIN SMALL LETTER A WITH OGONEK
0x00e1: 0x012f, # LATIN SMALL LETTER I WITH OGONEK
0x00e2: 0x0101, # LATIN SMALL LETTER A WITH MACRON
0x00e3: 0x0107, # LATIN SMALL LETTER C WITH ACUTE
0x00e6: 0x0119, # LATIN SMALL LETTER E WITH OGONEK
0x00e7: 0x0113, # LATIN SMALL LETTER E WITH MACRON
0x00e8: 0x010d, # LATIN SMALL LETTER C WITH CARON
0x00ea: 0x017a, # LATIN SMALL LETTER Z WITH ACUTE
0x00eb: 0x0117, # LATIN SMALL LETTER E WITH DOT ABOVE
0x00ec: 0x0123, # LATIN SMALL LETTER G WITH CEDILLA
0x00ed: 0x0137, # LATIN SMALL LETTER K WITH CEDILLA
0x00ee: 0x012b, # LATIN SMALL LETTER I WITH MACRON
0x00ef: 0x013c, # LATIN SMALL LETTER L WITH CEDILLA
0x00f0: 0x0161, # LATIN SMALL LETTER S WITH CARON
0x00f1: 0x0144, # LATIN SMALL LETTER N WITH ACUTE
0x00f2: 0x0146, # LATIN SMALL LETTER N WITH CEDILLA
0x00f4: 0x014d, # LATIN SMALL LETTER O WITH MACRON
0x00f8: 0x0173, # LATIN SMALL LETTER U WITH OGONEK
0x00f9: 0x0142, # LATIN SMALL LETTER L WITH STROKE
0x00fa: 0x015b, # LATIN SMALL LETTER S WITH ACUTE
0x00fb: 0x016b, # LATIN SMALL LETTER U WITH MACRON
0x00fd: 0x017c, # LATIN SMALL LETTER Z WITH DOT ABOVE
0x00fe: 0x017e, # LATIN SMALL LETTER Z WITH CARON
0x00ff: 0x02d9, # DOT ABOVE
0x0080: 0x20ac, # EURO SIGN
0x0081: None, # UNDEFINED
0x0082: 0x201a, # SINGLE LOW-9 QUOTATION MARK
0x0083: None, # UNDEFINED
0x0084: 0x201e, # DOUBLE LOW-9 QUOTATION MARK
0x0085: 0x2026, # HORIZONTAL ELLIPSIS
0x0086: 0x2020, # DAGGER
0x0087: 0x2021, # DOUBLE DAGGER
0x0088: None, # UNDEFINED
0x0089: 0x2030, # PER MILLE SIGN
0x008a: None, # UNDEFINED
0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
0x008c: None, # UNDEFINED
0x008d: 0x00a8, # DIAERESIS
0x008e: 0x02c7, # CARON
0x008f: 0x00b8, # CEDILLA
0x0090: None, # UNDEFINED
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: None, # UNDEFINED
0x0099: 0x2122, # TRADE MARK SIGN
0x009a: None, # UNDEFINED
0x009b: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
0x009c: None, # UNDEFINED
0x009d: 0x00af, # MACRON
0x009e: 0x02db, # OGONEK
0x009f: None, # UNDEFINED
0x00a1: None, # UNDEFINED
0x00a5: None, # UNDEFINED
0x00a8: 0x00d8, # LATIN CAPITAL LETTER O WITH STROKE
0x00aa: 0x0156, # LATIN CAPITAL LETTER R WITH CEDILLA
0x00af: 0x00c6, # LATIN CAPITAL LETTER AE
0x00b8: 0x00f8, # LATIN SMALL LETTER O WITH STROKE
0x00ba: 0x0157, # LATIN SMALL LETTER R WITH CEDILLA
0x00bf: 0x00e6, # LATIN SMALL LETTER AE
0x00c0: 0x0104, # LATIN CAPITAL LETTER A WITH OGONEK
0x00c1: 0x012e, # LATIN CAPITAL LETTER I WITH OGONEK
0x00c2: 0x0100, # LATIN CAPITAL LETTER A WITH MACRON
0x00c3: 0x0106, # LATIN CAPITAL LETTER C WITH ACUTE
0x00c6: 0x0118, # LATIN CAPITAL LETTER E WITH OGONEK
0x00c7: 0x0112, # LATIN CAPITAL LETTER E WITH MACRON
0x00c8: 0x010c, # LATIN CAPITAL LETTER C WITH CARON
0x00ca: 0x0179, # LATIN CAPITAL LETTER Z WITH ACUTE
0x00cb: 0x0116, # LATIN CAPITAL LETTER E WITH DOT ABOVE
0x00cc: 0x0122, # LATIN CAPITAL LETTER G WITH CEDILLA
0x00cd: 0x0136, # LATIN CAPITAL LETTER K WITH CEDILLA
0x00ce: 0x012a, # LATIN CAPITAL LETTER I WITH MACRON
0x00cf: 0x013b, # LATIN CAPITAL LETTER L WITH CEDILLA
0x00d0: 0x0160, # LATIN CAPITAL LETTER S WITH CARON
0x00d1: 0x0143, # LATIN CAPITAL LETTER N WITH ACUTE
0x00d2: 0x0145, # LATIN CAPITAL LETTER N WITH CEDILLA
0x00d4: 0x014c, # LATIN CAPITAL LETTER O WITH MACRON
0x00d8: 0x0172, # LATIN CAPITAL LETTER U WITH OGONEK
0x00d9: 0x0141, # LATIN CAPITAL LETTER L WITH STROKE
0x00da: 0x015a, # LATIN CAPITAL LETTER S WITH ACUTE
0x00db: 0x016a, # LATIN CAPITAL LETTER U WITH MACRON
0x00dd: 0x017b, # LATIN CAPITAL LETTER Z WITH DOT ABOVE
0x00de: 0x017d, # LATIN CAPITAL LETTER Z WITH CARON
0x00e0: 0x0105, # LATIN SMALL LETTER A WITH OGONEK
0x00e1: 0x012f, # LATIN SMALL LETTER I WITH OGONEK
0x00e2: 0x0101, # LATIN SMALL LETTER A WITH MACRON
0x00e3: 0x0107, # LATIN SMALL LETTER C WITH ACUTE
0x00e6: 0x0119, # LATIN SMALL LETTER E WITH OGONEK
0x00e7: 0x0113, # LATIN SMALL LETTER E WITH MACRON
0x00e8: 0x010d, # LATIN SMALL LETTER C WITH CARON
0x00ea: 0x017a, # LATIN SMALL LETTER Z WITH ACUTE
0x00eb: 0x0117, # LATIN SMALL LETTER E WITH DOT ABOVE
0x00ec: 0x0123, # LATIN SMALL LETTER G WITH CEDILLA
0x00ed: 0x0137, # LATIN SMALL LETTER K WITH CEDILLA
0x00ee: 0x012b, # LATIN SMALL LETTER I WITH MACRON
0x00ef: 0x013c, # LATIN SMALL LETTER L WITH CEDILLA
0x00f0: 0x0161, # LATIN SMALL LETTER S WITH CARON
0x00f1: 0x0144, # LATIN SMALL LETTER N WITH ACUTE
0x00f2: 0x0146, # LATIN SMALL LETTER N WITH CEDILLA
0x00f4: 0x014d, # LATIN SMALL LETTER O WITH MACRON
0x00f8: 0x0173, # LATIN SMALL LETTER U WITH OGONEK
0x00f9: 0x0142, # LATIN SMALL LETTER L WITH STROKE
0x00fa: 0x015b, # LATIN SMALL LETTER S WITH ACUTE
0x00fb: 0x016b, # LATIN SMALL LETTER U WITH MACRON
0x00fd: 0x017c, # LATIN SMALL LETTER Z WITH DOT ABOVE
0x00fe: 0x017e, # LATIN SMALL LETTER Z WITH CARON
0x00ff: 0x02d9, # DOT ABOVE
})
### Encoding Map
......
Lib/encodings/cp1258.py
View file @ 469cdad8
......@@ -16,14 +16,14 @@ class Codec(codecs.Codec):
def encode(self,input,errors='strict'):
return codecs.charmap_encode(input,errors,encoding_map)
def decode(self,input,errors='strict'):
return codecs.charmap_decode(input,errors,decoding_map)
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......@@ -37,52 +37,52 @@ def getregentry():
decoding_map = codecs.make_identity_dict(range(256))
decoding_map.update({
0x0080: 0x20ac, # EURO SIGN
0x0081: None, # UNDEFINED
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: None, # UNDEFINED
0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
0x008c: 0x0152, # LATIN CAPITAL LIGATURE OE
0x008d: None, # UNDEFINED
0x008e: None, # UNDEFINED
0x008f: None, # UNDEFINED
0x0090: None, # UNDEFINED
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: 0x02dc, # SMALL TILDE
0x0099: 0x2122, # TRADE MARK SIGN
0x009a: None, # UNDEFINED
0x009b: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
0x009c: 0x0153, # LATIN SMALL LIGATURE OE
0x009d: None, # UNDEFINED
0x009e: None, # UNDEFINED
0x009f: 0x0178, # LATIN CAPITAL LETTER Y WITH DIAERESIS
0x00c3: 0x0102, # LATIN CAPITAL LETTER A WITH BREVE
0x00cc: 0x0300, # COMBINING GRAVE ACCENT
0x00d0: 0x0110, # LATIN CAPITAL LETTER D WITH STROKE
0x00d2: 0x0309, # COMBINING HOOK ABOVE
0x00d5: 0x01a0, # LATIN CAPITAL LETTER O WITH HORN
0x00dd: 0x01af, # LATIN CAPITAL LETTER U WITH HORN
0x00de: 0x0303, # COMBINING TILDE
0x00e3: 0x0103, # LATIN SMALL LETTER A WITH BREVE
0x00ec: 0x0301, # COMBINING ACUTE ACCENT
0x00f0: 0x0111, # LATIN SMALL LETTER D WITH STROKE
0x00f2: 0x0323, # COMBINING DOT BELOW
0x00f5: 0x01a1, # LATIN SMALL LETTER O WITH HORN
0x00fd: 0x01b0, # LATIN SMALL LETTER U WITH HORN
0x00fe: 0x20ab, # DONG SIGN
0x0080: 0x20ac, # EURO SIGN
0x0081: None, # UNDEFINED
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: None, # UNDEFINED
0x008b: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
0x008c: 0x0152, # LATIN CAPITAL LIGATURE OE
0x008d: None, # UNDEFINED
0x008e: None, # UNDEFINED
0x008f: None, # UNDEFINED
0x0090: None, # UNDEFINED
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: 0x02dc, # SMALL TILDE
0x0099: 0x2122, # TRADE MARK SIGN
0x009a: None, # UNDEFINED
0x009b: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
0x009c: 0x0153, # LATIN SMALL LIGATURE OE
0x009d: None, # UNDEFINED
0x009e: None, # UNDEFINED
0x009f: 0x0178, # LATIN CAPITAL LETTER Y WITH DIAERESIS
0x00c3: 0x0102, # LATIN CAPITAL LETTER A WITH BREVE
0x00cc: 0x0300, # COMBINING GRAVE ACCENT
0x00d0: 0x0110, # LATIN CAPITAL LETTER D WITH STROKE
0x00d2: 0x0309, # COMBINING HOOK ABOVE
0x00d5: 0x01a0, # LATIN CAPITAL LETTER O WITH HORN
0x00dd: 0x01af, # LATIN CAPITAL LETTER U WITH HORN
0x00de: 0x0303, # COMBINING TILDE
0x00e3: 0x0103, # LATIN SMALL LETTER A WITH BREVE
0x00ec: 0x0301, # COMBINING ACUTE ACCENT
0x00f0: 0x0111, # LATIN SMALL LETTER D WITH STROKE
0x00f2: 0x0323, # COMBINING DOT BELOW
0x00f5: 0x01a1, # LATIN SMALL LETTER O WITH HORN
0x00fd: 0x01b0, # LATIN SMALL LETTER U WITH HORN
0x00fe: 0x20ab, # DONG SIGN
})
### Encoding Map
......
Lib/encodings/cp424.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp437.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp500.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp737.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp775.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp850.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp852.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp855.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp856.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp857.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp860.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp861.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp862.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp863.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp864.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp865.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp866.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp869.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp874.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/cp875.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/hex_codec.py
View file @ 469cdad8
......@@ -51,7 +51,7 @@ class Codec(codecs.Codec):
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......
Lib/encodings/iso8859_1.py
View file @ 469cdad8
......@@ -16,14 +16,14 @@ class Codec(codecs.Codec):
def encode(self,input,errors='strict'):
return codecs.charmap_encode(input,errors,encoding_map)
def decode(self,input,errors='strict'):
return codecs.charmap_decode(input,errors,decoding_map)
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......
Lib/encodings/iso8859_10.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/iso8859_13.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/iso8859_14.py
View file @ 469cdad8
......@@ -16,14 +16,14 @@ class Codec(codecs.Codec):
def encode(self,input,errors='strict'):
return codecs.charmap_encode(input,errors,encoding_map)
def decode(self,input,errors='strict'):
return codecs.charmap_decode(input,errors,decoding_map)
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......@@ -37,37 +37,37 @@ def getregentry():
decoding_map = codecs.make_identity_dict(range(256))
decoding_map.update({
0x00a1: 0x1e02, # LATIN CAPITAL LETTER B WITH DOT ABOVE
0x00a2: 0x1e03, # LATIN SMALL LETTER B WITH DOT ABOVE
0x00a4: 0x010a, # LATIN CAPITAL LETTER C WITH DOT ABOVE
0x00a5: 0x010b, # LATIN SMALL LETTER C WITH DOT ABOVE
0x00a6: 0x1e0a, # LATIN CAPITAL LETTER D WITH DOT ABOVE
0x00a8: 0x1e80, # LATIN CAPITAL LETTER W WITH GRAVE
0x00aa: 0x1e82, # LATIN CAPITAL LETTER W WITH ACUTE
0x00ab: 0x1e0b, # LATIN SMALL LETTER D WITH DOT ABOVE
0x00ac: 0x1ef2, # LATIN CAPITAL LETTER Y WITH GRAVE
0x00af: 0x0178, # LATIN CAPITAL LETTER Y WITH DIAERESIS
0x00b0: 0x1e1e, # LATIN CAPITAL LETTER F WITH DOT ABOVE
0x00b1: 0x1e1f, # LATIN SMALL LETTER F WITH DOT ABOVE
0x00b2: 0x0120, # LATIN CAPITAL LETTER G WITH DOT ABOVE
0x00b3: 0x0121, # LATIN SMALL LETTER G WITH DOT ABOVE
0x00b4: 0x1e40, # LATIN CAPITAL LETTER M WITH DOT ABOVE
0x00b5: 0x1e41, # LATIN SMALL LETTER M WITH DOT ABOVE
0x00b7: 0x1e56, # LATIN CAPITAL LETTER P WITH DOT ABOVE
0x00b8: 0x1e81, # LATIN SMALL LETTER W WITH GRAVE
0x00b9: 0x1e57, # LATIN SMALL LETTER P WITH DOT ABOVE
0x00ba: 0x1e83, # LATIN SMALL LETTER W WITH ACUTE
0x00bb: 0x1e60, # LATIN CAPITAL LETTER S WITH DOT ABOVE
0x00bc: 0x1ef3, # LATIN SMALL LETTER Y WITH GRAVE
0x00bd: 0x1e84, # LATIN CAPITAL LETTER W WITH DIAERESIS
0x00be: 0x1e85, # LATIN SMALL LETTER W WITH DIAERESIS
0x00bf: 0x1e61, # LATIN SMALL LETTER S WITH DOT ABOVE
0x00d0: 0x0174, # LATIN CAPITAL LETTER W WITH CIRCUMFLEX
0x00d7: 0x1e6a, # LATIN CAPITAL LETTER T WITH DOT ABOVE
0x00de: 0x0176, # LATIN CAPITAL LETTER Y WITH CIRCUMFLEX
0x00f0: 0x0175, # LATIN SMALL LETTER W WITH CIRCUMFLEX
0x00f7: 0x1e6b, # LATIN SMALL LETTER T WITH DOT ABOVE
0x00fe: 0x0177, # LATIN SMALL LETTER Y WITH CIRCUMFLEX
0x00a1: 0x1e02, # LATIN CAPITAL LETTER B WITH DOT ABOVE
0x00a2: 0x1e03, # LATIN SMALL LETTER B WITH DOT ABOVE
0x00a4: 0x010a, # LATIN CAPITAL LETTER C WITH DOT ABOVE
0x00a5: 0x010b, # LATIN SMALL LETTER C WITH DOT ABOVE
0x00a6: 0x1e0a, # LATIN CAPITAL LETTER D WITH DOT ABOVE
0x00a8: 0x1e80, # LATIN CAPITAL LETTER W WITH GRAVE
0x00aa: 0x1e82, # LATIN CAPITAL LETTER W WITH ACUTE
0x00ab: 0x1e0b, # LATIN SMALL LETTER D WITH DOT ABOVE
0x00ac: 0x1ef2, # LATIN CAPITAL LETTER Y WITH GRAVE
0x00af: 0x0178, # LATIN CAPITAL LETTER Y WITH DIAERESIS
0x00b0: 0x1e1e, # LATIN CAPITAL LETTER F WITH DOT ABOVE
0x00b1: 0x1e1f, # LATIN SMALL LETTER F WITH DOT ABOVE
0x00b2: 0x0120, # LATIN CAPITAL LETTER G WITH DOT ABOVE
0x00b3: 0x0121, # LATIN SMALL LETTER G WITH DOT ABOVE
0x00b4: 0x1e40, # LATIN CAPITAL LETTER M WITH DOT ABOVE
0x00b5: 0x1e41, # LATIN SMALL LETTER M WITH DOT ABOVE
0x00b7: 0x1e56, # LATIN CAPITAL LETTER P WITH DOT ABOVE
0x00b8: 0x1e81, # LATIN SMALL LETTER W WITH GRAVE
0x00b9: 0x1e57, # LATIN SMALL LETTER P WITH DOT ABOVE
0x00ba: 0x1e83, # LATIN SMALL LETTER W WITH ACUTE
0x00bb: 0x1e60, # LATIN CAPITAL LETTER S WITH DOT ABOVE
0x00bc: 0x1ef3, # LATIN SMALL LETTER Y WITH GRAVE
0x00bd: 0x1e84, # LATIN CAPITAL LETTER W WITH DIAERESIS
0x00be: 0x1e85, # LATIN SMALL LETTER W WITH DIAERESIS
0x00bf: 0x1e61, # LATIN SMALL LETTER S WITH DOT ABOVE
0x00d0: 0x0174, # LATIN CAPITAL LETTER W WITH CIRCUMFLEX
0x00d7: 0x1e6a, # LATIN CAPITAL LETTER T WITH DOT ABOVE
0x00de: 0x0176, # LATIN CAPITAL LETTER Y WITH CIRCUMFLEX
0x00f0: 0x0175, # LATIN SMALL LETTER W WITH CIRCUMFLEX
0x00f7: 0x1e6b, # LATIN SMALL LETTER T WITH DOT ABOVE
0x00fe: 0x0177, # LATIN SMALL LETTER Y WITH CIRCUMFLEX
})
### Encoding Map
......
Lib/encodings/iso8859_15.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/iso8859_2.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/iso8859_3.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/iso8859_4.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/iso8859_5.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/iso8859_6.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/iso8859_7.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/iso8859_8.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/iso8859_9.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/koi8_r.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/latin_1.py
View file @ 469cdad8
......@@ -19,7 +19,7 @@ class Codec(codecs.Codec):
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......
Lib/encodings/mac_cyrillic.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/mac_greek.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/mac_iceland.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/mac_latin2.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/mac_roman.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/mac_turkish.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/mbcs.py
View file @ 469cdad8
......@@ -20,7 +20,7 @@ class Codec(codecs.Codec):
class StreamWriter(Codec,codecs.StreamWriter):
pass
class StreamReader(Codec,codecs.StreamReader):
pass
......
Lib/encodings/palmos.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/raw_unicode_escape.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/rot_13.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/undefined.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/unicode_escape.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/unicode_internal.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/utf_16.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/utf_16_be.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/utf_16_le.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/utf_7.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/utf_8.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/uu_codec.py
View file @ 469cdad8
This diff is collapsed.
Lib/encodings/zlib_codec.py
View file @ 469cdad8
This diff is collapsed.
Lib/heapq.py
View file @ 469cdad8
This diff is collapsed.
Lib/httplib.py
View file @ 469cdad8
This diff is collapsed.
Lib/smtplib.py
View file @ 469cdad8
This diff is collapsed.
Lib/test/string_tests.py
View file @ 469cdad8
......@@ -314,4 +314,3 @@ def run_contains_tests(test):
vereq('asdf' in 'asdf', True)
vereq('asdf' in 'asd', False)
vereq('asdf' in '', False)
Lib/test/test_format.py
View file @ 469cdad8
This diff is collapsed.
Lib/test/test_slice.py
View file @ 469cdad8
This diff is collapsed.
Lib/test/test_socket.py
View file @ 469cdad8
This diff is collapsed.
Lib/test/test_strptime.py
View file @ 469cdad8
This diff is collapsed.
Lib/test/test_whichdb.py
View file @ 469cdad8
This diff is collapsed.
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7