def get_sourcestr_representation(self):
"""
DCharacterLAT.get_sourcestr_representation
Return a string.
"""
#.......................................................................
# unknown char ? Nothing to do :
#.......................................................................
if self.unknown_char:
if self.dstring_object.options["anonymize the unknown characters"] == 'yes':
return UNKNOWN_CHAR_SYMBOL
else:
return self.base_char
#.......................................................................
# ok, the function can analyse <self> :
#.......................................................................
res = []
if self.base_char is not None:
if self.punctuation:
# punctuation symbol :
res.append( self.base_char )
elif not self.capital_letter:
# lower case :
res.append( SYMB_LOWER_CASE.get_default_symbol(self.base_char) )
else:
# upper case :
res.append( SYMB_UPPER_CASE.get_default_symbol(self.base_char) )
if self.stress:
res.append( DEFAULTSYMB__STRESS )
if self.length == 'short' or self.length == 'long':
res.append( SYMB_DIACRITICS.get_default_symbol(self.length) )
if self.diaeresis:
res.append( DEFAULTSYMB__DIAERESIS )
res = "".join(res)
# (1/2) composition with unicodedata.normalize :
res = unicodedata.normalize('NFC', res)
# (2/2) composition with COMPLETE_NORMALIZE_NFC :
for src, dest in COMPLETE_NORMALIZE_NFC:
res = res.replace(src, dest)
return res
