def get_sourcestr_representation(self):
"""
DCharacterHBO.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:
base_char = self.base_char
if base_char == "כ" and self.contextual_form == "final":
base_char = "ך"
if base_char == "מ" and self.contextual_form == "final":
base_char = "ם"
if base_char == "נ" and self.contextual_form == "final":
base_char = "ן"
if base_char == "פ" and self.contextual_form == "final":
base_char = "ף"
if base_char == "צ" and self.contextual_form == "final":
base_char = "ץ"
res.append( base_char )
if self.shin_sin_dot is not None:
res.append( SYMB_POINTS.get_default_symbol(self.shin_sin_dot) )
if self.daghesh_mapiq:
res.append( DEFAULTSYMB__DAGHESHMAPIQ )
if self.vowel is not None:
res.append( SYMB_VOWELS.get_default_symbol(self.vowel) )
if self.methegh:
res.append( DEFAULTSYMB__METEG )
if self.raphe:
res.append( DEFAULTSYMB__RAFE )
if self.specialpoint is not None:
res.append( SYMB_SPECIALPOINTS.get_default_symbol(self.specialpoint) )
if self.cantillation_mark is not None:
for cmark in self.cantillation_mark:
res.append( SYMB_CANTILLATION_MARKS.get_default_symbol(cmark) )
res = "".join(res)
# composition with COMPLETE_NORMALIZE_NFC
# we don't use NFC which "shuffle" the elements in an order incompatible
# with our code.
for src, dest in COMPLETE_NORMALIZE_NFC:
res = res.replace(src, dest)
return res
