def sortingvalue(self):
"""
DStringANG.sortingvalue
Return a SortingValue object
"""
res = SortingValue()
# Pylint can't know that <self> has an 'options' member
# created when <self> has been initialized by new_dstring() :
# pylint: disable=E1101
# -> "Instance of 'DStringANG' has no 'options' member"
if self.options["sorting method"] == "default":
# base character :
data = []
for char in self:
sorting_order = -1
if char.base_char in SORTING_ORDER:
sorting_order = SORTING_ORDER[char.base_char]
data.append( ({False:0,
True:1}[char.unknown_char],
sorting_order ))
res.append( data )
# makron :
data = []
for char in self:
data.append( { False : 0,
True : 1,}[char.makron] )
res.append(data)
else:
# Pylint can't know that <self> has an 'options' member
# created when <self> has been initialized by new_dstring() :
# pylint: disable=E1101
# -> "Instance of 'DStringANG' has no 'options' member"
err_msg = "unknown sorting method '{0}'."
raise DCharsError( context = "DStringANG.sortingvalue",
message = err_msg.format(self.options["sorting method"]) )
return res
def sortingvalue(self):
"""
DStringLAT.sortingvalue
Return a SortingValue object
"""
res = SortingValue()
# Pylint can't know that <self> has an 'options' member
# created when <self> has been initialized by new_dstring() :
# pylint: disable=E1101
# -> "Instance of 'DStringLAT' has no 'options' member"
if self.options["sorting method"] == "default":
# base character :
data = []
for char in self:
data.append( ({False:0,
True:1}[char.unknown_char],
ord(char.base_char) ))
res.append(data)
# length :
data = []
for char in self:
data.append( { None : 0,
"short" : 1,
"long" : 2, }[char.length])
res.append(data)
else:
# Pylint can't know that <self> has an 'options' member
# created when <self> has been initialized by new_dstring() :
# pylint: disable=E1101
# -> "Instance of 'DStringLAT' has no 'options' member"
err_msg = "unknown sorting method '{0}'."
raise DCharsError( context = "DStringLAT.sortingvalue",
message = err_msg.format(self.options["sorting method"]) )
return res
def sortingvalue(self):
"""
DCharacterFRO.sortingvalue
Return an SortingValue object
NB : this function has almost no interest; you must use DStringFRO.sortingvalue()
to compare two strings. Use this function if you just want to compare
two characters.
"""
res = SortingValue()
if self.dstring_object.options["sorting method"] == "default":
if self.unknown_char:
# unknown char :
res.append(1)
return res
# known char :
res.append(0)
# # capital_letter :
# if res.capital_letter:
# res.append( 0 )
# else:
# res.append( 1 )
# base_char :
if self.base_char not in SORTING_ORDER:
res.append(-1)
else:
res.append(SORTING_ORDER[self.base_char])
else:
raise DCharsError(
context="DCharacterFRO.sortingvalue",
message="unknown sorting method =" + str(self.dstring_object.options["sorting method"]),
)
return res
def sortingvalue(self):
"""
DCharacterLAT.sortingvalue
Return an SortingValue object
NB : this function has almost no interest; you must use DStringLAT.sortingvalue()
to compare two strings. Use this function if you just want to compare
two characters.
"""
res = SortingValue()
if self.dstring_object.options["sorting method"] == "default":
if self.unknown_char:
# unknown char :
res.append(1)
return res
# known char :
res.append(0)
# # capital_letter :
# if res.capital_letter:
# res.append( 0 )
# else:
# res.append( 1 )
# base_char :
res.append( ord(self.base_char) )
# length :
if self.length is None:
res.append( 0 )
elif self.length == "short":
res.append( 0 )
elif self.length == "long":
res.append( 1 )
else:
raise DCharsError( context = "DCharacterLAT.sortingvalue",
message = "unknown value for length ="+\
str(self.length) )
# # stress :
# if not res.stress:
# self.append( 0 )
# else:
# self.append( 1 )
# # diaeresis :
# if not res.diaeresis:
# self.append( 0 )
# else:
# self.append( 1 )
else:
raise DCharsError( context = "DCharacterLAT.sortingvalue",
message = "unknown sorting method ="+\
str(self.dstring_object.options["sorting method"]) )
return res
def sortingvalue(self):
"""
DStringJPN.sortingvalue
sorting methods :
o "default" : cf Kanji & Kana, Hadamitzky and Spahn, p. 22
Return a SortingValue object
"""
res = SortingValue()
# Pylint can't know that <self> has an 'options' member
# created when <self> has been initialized by new_dstring() :
# pylint: disable=E1101
# -> "Instance of 'DStringJPN' has no 'options' member"
if self.options["sorting method"] == "default":
# base character :
data = []
previous_char = None
for index, char in enumerate(self):
if char.unknown_char:
data.append( (1, 999) )
elif char.chartype == 'choonpu':
# we treat the choonpu symbol as if it was the last vowel
# cf Kanji & Kana, Hadamitzky and Spahn, p. 22
if index == 0:
# problem : no preceding vowel
data.append( (0, 0 ))
else:
vowel = VOWEL_IN_HIRAGANA[previous_char.base_char]
if vowel is not None:
# normal case : if char == 'か', vowel = 'あ'
data.append( (0, HIRAGANA_ORDER[vowel] ))
else:
# abnormal case : if char == 'ん', there's no vowel...
# so ... we take the order of ん.
data.append( (0, HIRAGANA_ORDER['ん'] ))
elif char.chartype in ('hiragana', 'katakana'):
data.append( (0, HIRAGANA_ORDER[char.base_char] ))
else:
# other cases : kanji or unknown symbol.
data.append( (0, ord(char.base_char) ))
previous_char = char
res.append(data)
# small size :
data = []
for char in self:
data.append( { False:0,
True:1, }[char.smallsize])
res.append(data)
# diacritic :
data = []
for char in self:
data.append( { None:0,
"dakuten":1,
"handakuten":2 }[char.diacritic] )
res.append(data)
# hiragana < katakana:
data = []
for char in self:
if char.chartype == 'hiragana':
data.append( 0 )
elif char.chartype == 'choonpu':
# we treat the choonpu symbol as a katakana
# cf Kanji & Kana, Hadamitzky and Spahn, p. 22
data.append( 1 )
elif char.chartype == 'katakana':
data.append( 1 )
else:
data.append( 2 )
res.append(data)
else:
# Pylint can't know that <self> has an 'options' member
# created when <self> has been initialized by new_dstring() :
# pylint: disable=E1101
# -> "Instance of 'DStringJPN' has no 'options' member"
err_msg = "unknown sorting method '{0}'."
raise DCharsError( context = "DStringJPN.sortingvalue",
message = err_msg.format(self.options["sorting method"]) )
return res
def sortingvalue(self):
"""
DCharacterSAN.sortingvalue
Return an SortingValue object
Output : a list of number :
[(un)known_char,
base_char,
nukta,
virama/dependent vowel,
anusvara/candrabindu,
anudatta,
accent]
!!! BEWARE : the number #4 (anusvara/candrabindu) will be equal to 999 if
!!! this character is followed by a visarga : see DStringSAN.sortingvalue()
!!! BEWARE : if you modify this function, don't forget to change the
!!! attribute DCharacterSAN.sortingvalue_for_visarga
"""
res = SortingValue()
if self.dstring_object.options["sorting method"] == "default":
if self.unknown_char:
# unknown char :
res.append(1)
return res
# known char :
res.append(0)
# base_char :
if self.base_char == "SHORT A":
res.append(0)
elif self.base_char == "A":
res.append(1)
elif self.base_char == "AA":
res.append(2)
elif self.base_char == "I":
res.append(3)
elif self.base_char == "II":
res.append(4)
elif self.base_char == "U":
res.append(5)
elif self.base_char == "UU":
res.append(6)
elif self.base_char == "VOCALIC R":
res.append(7)
elif self.base_char == "VOCALIC RR":
res.append(8)
elif self.base_char == "VOCALIC L":
res.append(9)
elif self.base_char == "VOCALIC LL":
res.append(10)
elif self.base_char == "SHORT E":
res.append(11)
elif self.base_char == "E":
res.append(12)
elif self.base_char == "AI":
res.append(13)
elif self.base_char == "SHORT O":
res.append(14)
elif self.base_char == "O":
res.append(15)
elif self.base_char == "AU":
res.append(16)
elif self.base_char == "DEVANAGARI SIGN VISARGA":
res.append(100)
elif self.base_char == "KA":
res.append(101)
elif self.base_char == "KHA":
res.append(102)
elif self.base_char == "GA":
res.append(103)
elif self.base_char == "GHA":
res.append(104)
elif self.base_char == "NGA":
res.append(105)
elif self.base_char == "CA":
res.append(106)
elif self.base_char == "CHA":
res.append(107)
elif self.base_char == "JA":
res.append(108)
elif self.base_char == "JHA":
res.append(109)
elif self.base_char == "NYA":
res.append(110)
elif self.base_char == "TTA":
res.append(111)
elif self.base_char == "TTHA":
res.append(112)
elif self.base_char == "DDA":
res.append(113)
elif self.base_char == "DDHA":
res.append(114)
elif self.base_char == "NNA":
res.append(115)
elif self.base_char == "TA":
res.append(116)
elif self.base_char == "THA":
res.append(117)
elif self.base_char == "DA":
res.append(118)
elif self.base_char == "DHA":
res.append(119)
elif self.base_char == "NA":
res.append(120)
elif self.base_char == "PA":
res.append(121)
elif self.base_char == "PHA":
res.append(122)
elif self.base_char == "BA":
res.append(123)
elif self.base_char == "BHA":
res.append(124)
elif self.base_char == "MA":
res.append(125)
elif self.base_char == "YA":
res.append(126)
elif self.base_char == "RA":
res.append(127)
elif self.base_char == "LA":
res.append(128)
elif self.base_char == "LLA":
res.append(129)
elif self.base_char == "VA":
res.append(130)
elif self.base_char == "SHA":
res.append(131)
elif self.base_char == "SSA":
res.append(132)
elif self.base_char == "SA":
res.append(133)
elif self.base_char == "HA":
res.append(134)
else:
# E.g. "1", ...
base_char_num = 0
for index_char, char in enumerate(self.base_char):
base_char_num += ord(char) << index_char
res.append(1000 + base_char_num)
# nukta :
if not self.nukta:
res.append(0)
else:
res.append(1)
# dependent vowel :
# virama :
if self.virama:
res.append(0)
else:
if self.dependentvowel is None:
res.append(1)
elif self.dependentvowel == "A":
res.append(2)
elif self.dependentvowel == "AA":
res.append(3)
elif self.dependentvowel == "I":
res.append(4)
elif self.dependentvowel == "II":
res.append(5)
elif self.dependentvowel == "U":
res.append(6)
elif self.dependentvowel == "UU":
res.append(7)
elif self.dependentvowel == "VOCALIC R":
res.append(8)
elif self.dependentvowel == "VOCALIC RR":
res.append(9)
elif self.dependentvowel == "VOCALIC L":
res.append(10)
elif self.dependentvowel == "VOCALIC LL":
res.append(11)
elif self.dependentvowel == "CANDRA E":
res.append(12)
elif self.dependentvowel == "SHORT E":
res.append(13)
elif self.dependentvowel == "E":
res.append(14)
elif self.dependentvowel == "AI":
res.append(15)
elif self.dependentvowel == "CANDRA O":
res.append(16)
elif self.dependentvowel == "SHORT O":
res.append(17)
elif self.dependentvowel == "O":
res.append(18)
elif self.dependentvowel == "AU":
res.append(19)
else:
raise DCharsError(
context="DCharacterSAN.sortingvalue",
message="unknown value for dependentvowel =" + str(self.dependentvowel),
)
# anusvara / candrabindu :
if self.anusvara_candrabindu is None:
res.append(0)
elif self.anusvara_candrabindu == "DEVANAGARI SIGN ANUSVARA":
res.append(1)
elif self.anusvara_candrabindu == "DEVANAGARI SIGN CANDRABINDU":
res.append(2)
elif self.anusvara_candrabindu == "DEVANAGARI SIGN INVERTED CANDRABINDU":
res.append(3)
else:
raise DCharsError(
context="DCharacterSAN.sortingvalue",
message="unknown value for anusvara_candrabindu =" + str(self.anusvara_candrabindu),
)
# anudatta :
if not self.anudatta:
res.append(0)
else:
res.append(1)
# accent :
if self.accent is None:
res.append(0)
elif self.accent == "DEVANAGARI STRESS SIGN UDATTA":
res.append(1)
elif self.accent == "DEVANAGARI GRAVE ACCENT":
res.append(2)
elif self.accent == "DEVANAGARI ACUTE ACCENT":
res.append(3)
else:
raise DCharsError(
context="DCharacterSAN.sortingvalue", message="unknown value for accent =" + str(self.accent)
)
else:
raise DCharsError(
context="DCharacterSAN.sortingvalue",
message="unknown sorting method =" + str(self.dstring_object.options["sorting method"]),
)
return res
def sortingvalue(self):
"""
DCharacterJPN.sortingvalue
Return a SortingValue object
NB : this function has almost no interest; you must use DStringJPN.sortingvalue()
to compare two strings. Use this function if you just want to compare
two characters.
"""
res = SortingValue()
if self.dstring_object.options["sorting method"] == "default":
if self.unknown_char:
# unknown char :
res.append(1)
res.append( ord(self.base_char) )
return res
# known char :
res.append(0)
# base_char :
res.append( HIRAGANA_ORDER[self.base_char] )
# small size ?
res.append( {True:0,
False:1}[self.smallsize] )
# hiragana < katakana :
if self.chartype == 'hiragana':
res.append(0)
elif self.chartype == 'katakana':
res.append(1)
else:
res.append(2)
else:
raise DCharsError( context = "DCharacterJPN.sortingvalue",
message = "unknown sorting method ="+\
str(self.dstring_object.options["sorting method"]) )
return res
def sortingvalue(self):
"""
DStringGRC.sortingvalue
Return a SortingValue object
"""
res = SortingValue()
# Pylint can't know that <self> has an 'options' member
# created when <self> has been initialized by new_dstring() :
# pylint: disable=E1101
# -> "Instance of 'DStringGRC' has no 'options' member"
if self.options["sorting method"] == "default":
# base character :
data = []
for char in self:
data.append( ({False:0,
True:1}[char.unknown_char],
char.base_char ))
res.append(data)
# pneuma :
data = []
for char in self:
data.append( { None : 0,
"ψιλὸν" : 1,
"δασὺ" : 2, }[char.pneuma])
res.append(data)
# tonos :
data = []
for char in self:
data.append( { None : 0,
"ὀξεῖα" : 1,
"βαρεῖα" : 2,
"περισπωμένη" : 3, }[char.tonos])
res.append(data)
# hypogegrammene :
data = []
for char in self:
data.append( { False : 0,
True : 1, }[char.hypogegrammene])
res.append(data)
# mekos :
data = []
for char in self:
data.append( { None : 0,
"βραχύ" : 1,
"μακρόν" : 2, }[char.mekos])
res.append(data)
else:
# Pylint can't know that <self> has an 'options' member
# created when <self> has been initialized by new_dstring() :
# pylint: disable=E1101
# -> "Instance of 'DStringGRC' has no 'options' member"
err_msg = "unknown sorting method '{0}'."
raise DCharsError( context = "DStringGRC.sortingvalue",
message = err_msg.format(self.options["sorting method"]) )
return res
def sortingvalue(self):
"""
DCharacterGRC.sortingvalue
Return a SortingValue object
NB : this function has almost no interest; you must use DStringGRC.sortingvalue()
to compare two strings. Use this function if you just want to compare
two characters.
"""
res = SortingValue()
if self.dstring_object.options["sorting method"] == "default":
if self.unknown_char:
# unknown char :
res.append(1)
# Some base_char may contain more than one character, like "β2".
base_char_num = 0
for index_char, char in enumerate(self.base_char):
base_char_num += ord(char) << index_char
res.append( base_char_num )
return res
# known char :
res.append(0)
# base_char :
#
# Some base_char may contain more than one character, like "β2".
#
base_char_num = 0
for index_char, char in enumerate(self.base_char):
base_char_num += ord(char) << index_char
res.append( base_char_num )
# pneuma :
if self.pneuma is None:
res.append(0)
elif self.pneuma == "ψιλὸν":
res.append(1)
elif self.pneuma == "δασὺ":
res.append(2)
else:
raise DCharsError( context = "DCharacterGRC.sortingvalue",
message = "unknown value for pneuma ="+\
str(self.pneuma) )
# tonos :
if self.tonos is None:
res.append(0)
elif self.tonos == "ὀξεῖα":
res.append(1)
elif self.tonos == "βαρεῖα":
res.append(2)
elif self.tonos == "περισπωμένη":
res.append(3)
else:
raise DCharsError( context = "DCharacterGRC.sortingvalue",
message = "unknown value for tonos ="+\
str(self.tonos) )
# hypogegrammene :
if not self.hypogegrammene:
res.append(0)
else:
res.append(1)
# mekos :
if self.mekos is None:
res.append( 0 )
elif self.mekos == "βραχύ":
res.append( 0 )
elif self.mekos == "μακρόν":
res.append( 1 )
else:
raise DCharsError( context = "DCharacterGRC.sortingvalue",
message = "unknown value for mekos ="+\
str(self.mekos) )
else:
raise DCharsError( context = "DCharacterGRC.sortingvalue",
message = "unknown sorting method ="+\
str(self.dstring_object.options["sorting method"]) )
return res
def sortingvalue(self):
"""
DStringHBO.sortingvalue
Return a SortingValue object
"""
res = SortingValue()
# Pylint can't know that <self> has an 'options' member
# created when <self> has been initialized by new_dstring() :
# pylint: disable=E1101
# -> "Instance of 'DStringHBO' has no 'options' member"
if self.options["sorting method"] == "default":
# base character :
data = []
for char in self:
data.append( ({False:0,
True:1}[char.unknown_char],
char.base_char ))
res.append(data)
# shin_sin_dot :
data = []
for char in self:
data.append( ({None : 0,
"HEBREW POINT SHIN DOT" : 1,
"HEBREW POINT SIN DOT" : 2,}[char.shin_sin_dot]
))
res.append(data)
# vowel :
data = []
for char in self:
data.append( ({None : 0,
"HEBREW POINT PATAH" : 1,
"HEBREW POINT SEGOL" : 2,
"HEBREW POINT HIRIQ" : 3,
"HEBREW POINT QUBUTS" : 4,
"HEBREW POINT QAMATS" : 5,
"HEBREW POINT QAMATS QATAN" : 6,
"HEBREW POINT TSERE" : 7,
"HEBREW POINT HOLAM" : 8,
"HEBREW POINT HOLAM HASER FOR VAV": 9,
"HEBREW POINT HATAF SEGOL" : 10,
"HEBREW POINT HATAF PATAH" : 11,
"HEBREW POINT HATAF QAMATS" : 12,
"HEBREW POINT SHEVA" : 13,}[char.vowel]
))
res.append(data)
# daghesh_mapiq :
data = []
for char in self:
data.append( ({False:0,
True:1}[char.daghesh_mapiq] ))
res.append(data)
# methegh :
data = []
for char in self:
data.append( ({False:0,
True:1}[char.methegh] ))
res.append(data)
# raphe :
data = []
for char in self:
data.append( ({False:0,
True:1}[char.raphe] ))
res.append(data)
# special point :
data = []
for char in self:
data.append( ({None : 0,
"HEBREW MARK UPPER DOT" : 1,
"HEBREW MARK LOWER DOT" : 2,}[char.specialpoint]
))
res.append(data)
else:
# Pylint can't know that <self> has an 'options' member
# created when <self> has been initialized by new_dstring() :
# pylint: disable=E1101
# -> "Instance of 'DStringHBO' has no 'options' member"
err_msg = "unknown sorting method '{0}'."
raise DCharsError( context = "DStringHBO.sortingvalue",
message = err_msg.format(self.options["sorting method"]) )
return res
def sortingvalue(self):
"""
DCharacterHBO.sortingvalue
Return a SortingValue object
NB : this function has almost no interest; you must use DStringHBO.sortingvalue()
to compare two strings. Use this function if you just want to compare
two characters.
"""
res = SortingValue()
if self.dstring_object.options["sorting method"] == "default":
if self.unknown_char:
# unknown char :
res.append(1)
return res
# known char :
res.append(0)
# base_char :
res.append( ord(self.base_char) )
# shin_sin_dot :
if self.shin_sin_dot is None:
res.append(0)
elif self.shin_sin_dot == "HEBREW POINT SHIN DOT":
res.append(1)
elif self.shin_sin_dot == "HEBREW POINT SIN DOT":
res.append(2)
else:
raise DCharsError( context = "DCharacterHBO.sortingvalue",
message = "unknown value for shin_sin_dot ="+\
str(self.shin_sin_dot) )
# vowel :
if self.vowel is None:
res.append(0)
elif self.vowel == "HEBREW POINT PATAH":
res.append(1)
elif self.vowel == "HEBREW POINT SEGOL":
res.append(2)
elif self.vowel == "HEBREW POINT HIRIQ":
res.append(3)
elif self.vowel == "HEBREW POINT QUBUTS":
res.append(4)
elif self.vowel == "HEBREW POINT QAMATS":
res.append(5)
elif self.vowel == "HEBREW POINT QAMATS QATAN":
res.append(6)
elif self.vowel == "HEBREW POINT TSERE":
res.append(7)
elif self.vowel == "HEBREW POINT HOLAM":
res.append(8)
elif self.vowel == "HEBREW POINT HOLAM HASER FOR VAV":
res.append(9)
elif self.vowel == "HEBREW POINT HATAF SEGOL":
res.append(10)
elif self.vowel == "HEBREW POINT HATAF PATAH":
res.append(11)
elif self.vowel == "HEBREW POINT HATAF QAMATS":
res.append(12)
elif self.vowel == "HEBREW POINT SHEVA":
res.append(13)
else:
raise DCharsError( context = "DCharacterHBO.sortingvalue",
message = "unknown vowel ="+\
str(self.vowel) )
# daghesh_mapiq :
if not self.daghesh_mapiq:
res.append(0)
else:
res.append(1)
# methegh :
if not self.methegh:
res.append(0)
else:
res.append(1)
# raphe :
if not self.raphe:
res.append(0)
else:
res.append(1)
# special point :
if self.specialpoint is None:
res.append(0)
elif self.specialpoint == "HEBREW MARK UPPER DOT":
res.append(1)
elif self.specialpoint == "HEBREW MARK LOWER DOT":
res.append(2)
else:
raise DCharsError( context = "DCharacterHBO.sortingvalue",
message = "unknown value for special_point ="+\
str(self.specialpoint) )
else:
raise DCharsError( context = "DCharacterHBO.sortingvalue",
message = "unknown sorting method ="+\
str(self.dstring_object.options["sorting method"]) )
return res
