def remove(self, *strings):
"""
Remove the lower case and no-period version of the string arguments from the set.
Returns ``self`` for chaining.
"""
[self.elements.remove(lc(s)) for s in strings if lc(s) in self.elements]
return self
def is_suffix(self, piece):
"""
Is in the suffixes set and not :py:func:`is_an_initial()`.
Some suffixes may be acronyms (M.B.A) while some are not (Jr.),
so we remove the periods from `piece` when testing against
`C.suffix_acronyms`.
"""
# suffixes may have periods inside them like "M.D."
return ((lc(piece).replace('.','') in self.C.suffix_acronyms) \
or (lc(piece) in self.C.suffix_not_acronyms)) \
and not self.is_an_initial(piece)
def cap_word(self, word):
if self.is_prefix(word) or self.is_conjunction(word):
return word.lower()
exceptions = self.C.capitalization_exceptions
if lc(word) in exceptions:
return exceptions[lc(word)]
mac_match = self.C.regexes.mac.match(word)
if mac_match:
def cap_after_mac(m):
return m.group(1).capitalize() + m.group(2).capitalize()
return self.C.regexes.mac.sub(cap_after_mac, word)
else:
return word.capitalize()
def add(self, *strings):
"""
Add the lower case and no-period version of the string arguments to the set.
Returns ``self`` for chaining.
"""
[self.elements.add(lc(s)) for s in strings]
return self
def add(self, *strings):
"""
Add the lower case and no-period version of the string arguments to the set.
Returns ``self`` for chaining.
"""
for s in strings:
if type(s) == binary_type:
s = s.decode(sys.stdin.encoding)
self.elements.add(lc(s))
return self
def handle_firstnames(self):
"""
If there are only two parts and one is a title, assume it's a last name
instead of a first name. e.g. Mr. Johnson. Unless it's a special title
like "Sir", then when it's followed by a single name that name is always
a first name.
"""
if self.title \
and len(self) == 2 \
and not lc(self.title) in self.C.first_name_titles:
self.last, self.first = self.first, self.last
def handle_firstnames(self):
"""
If there are only two parts and one is a title, assume it's a last name
instead of a first name. e.g. Mr. Johnson. Unless it's a special title
like "Sir", then when it's followed by a single name that name is always
a first name.
"""
if self.title \
and len(self) == 2 \
and not lc(self.title) in self.C.first_name_titles:
self.last, self.first = self.first, self.last
def add_with_encoding(self, s, encoding=None):
"""
Add the lower case and no-period version of the string to the set. Pass an
explicit `encoding` parameter to specify the encoding of binary strings that
are not DEFAULT_ENCODING (UTF-8).
"""
stdin_encoding = None
if sys.stdin:
stdin_encoding = sys.stdin.encoding
encoding = encoding or stdin_encoding or DEFAULT_ENCODING
if type(s) == binary_type:
s = s.decode(encoding)
self.elements.add(lc(s))
def add_with_encoding(self, s, encoding=None):
"""
Add the lower case and no-period version of the string to the set. Pass an
explicit `encoding` parameter to specify the encoding of binary strings that
are not DEFAULT_ENCODING (UTF-8).
"""
stdin_encoding = None
if sys.stdin:
stdin_encoding = sys.stdin.encoding
encoding = encoding or stdin_encoding or DEFAULT_ENCODING
if type(s) == binary_type:
s = s.decode(encoding)
self.elements.add(lc(s))
def cap_word(self, word):
if self.is_prefix(word) or self.is_conjunction(word):
return lc(word)
exceptions = self.C.capitalization_exceptions
if word in exceptions:
return exceptions[word]
mac_match = self.C.regexes.mac.match(word)
if mac_match:
def cap_after_mac(m):
return m.group(1).capitalize() + m.group(2).capitalize()
return self.C.regexes.mac.sub(cap_after_mac, word)
else:
return word.capitalize()
def is_prefix(self, piece):
"""Is in the prefixes set and not :py:func:`is_an_initial()`."""
return lc(piece) in self.C.prefixes and not self.is_an_initial(piece)
def is_conjunction(self, piece):
"""Is in the conjuctions set or :py:func:`is_an_initial()`."""
return lc(piece) in self.C.conjunctions and not self.is_an_initial(piece)
def is_rootname(self, piece):
"""
Is not a known title, suffix or prefix. Just first, middle, last names.
"""
return lc(piece) not in self.C.suffixes_prefixes_titles \
and not self.is_an_initial(piece)
def is_suffix(self, piece):
"""Is in the suffixes set or :py:func:`is_an_initial()`."""
return lc(piece) in self.C.suffixes and not self.is_an_initial(piece)
def is_conjunction(self, piece):
"""Is in the conjuctions set or :py:func:`is_an_initial()`."""
return lc(
piece) in self.C.conjunctions and not self.is_an_initial(piece)
def is_suffix(self, piece):
"""Is in the suffixes set or :py:func:`is_an_initial()`."""
# suffixes may have periods inside them like "M.D."
return lc(piece).replace(
'.', '') in self.C.suffixes and not self.is_an_initial(piece)
def is_rootname(self, piece):
'''Is not a known title, suffix or prefix. Just first, middle, last names.'''
return lc(piece) not in self.C.suffixes_prefixes_titles \
and not self.is_an_initial(piece)
def is_prefix(self, piece):
"""
Lowercase and no periods version of piece is in the
`~nameparser.config.titles.PREFIXES` set.
"""
return lc(piece) in self.C.prefixes
def is_suffix(self, piece):
"""Is in the suffixes set or :py:func:`is_an_initial()`."""
return lc(piece) in self.C.suffixes and not self.is_an_initial(piece)
def is_title(self, value):
"""Is in the :py:data:`~nameparser.config.titles.TITLES` set."""
return lc(value) in self.C.titles
def is_title(self, value):
"""Is in the :py:data:`~nameparser.config.titles.TITLES` set."""
return lc(value) in self.C.titles
def is_prefix(self, piece):
"""
Lowercase and no periods version of piece is in the
:py:data:`~nameparser.config.prefixes.PREFIXES` set.
"""
return lc(piece) in self.C.prefixes
def is_suffix(self, piece):
"""Is in the suffixes set or :py:func:`is_an_initial()`."""
# suffixes may have periods inside them like "M.D."
return lc(piece).replace('.','') in self.C.suffixes and not self.is_an_initial(piece)
