def test_characters_do_not_break(raw):
_ = Name(raw).raw
def t_strip_margin_hyphens(self, raw, cooked):
assert Name.clean(raw) == cooked
def issue_7_allow_initials_written_properly():
n = Name("J.R.R. Tolkien")
assert n.first == "j"
assert n.middle == "r r"
assert n.last == "tolkien"
def test_unit_do_not_grab_junior(clusters):
blank = Name()
_ = blank._grab_junior(clusters)
assert "jr" not in blank.detail["suffix"]
def t_drop_spacing(self, raw):
assert Name.clean(raw) == "mr eric praline"
def t_ignore_most_symbols(self, raw):
assert Name.clean(raw) == "mr eric praline"
def test_conjunction_avoid(static):
assert Name._combine_conjunctions(static) == static
def t_drop_emoji(self, raw):
name = Name()
assert name.clean(name._pre_clean(raw)) == "spiny norman"
def test_prefix_avoid(static):
raw = " ".join(static)
assert set(static) <= set(Name(raw).values())
def test_conjunction_combine(incoming, outgoing):
assert Name._combine_conjunctions(incoming) == outgoing
def test_prefix_combining(incoming, outgoing):
assert Name._combine_rightmost_prefixes(incoming) == outgoing
def issue_40_smart_quotes():
name = Name("Bergholt “BSJ” Johnson")
assert name.nickname
def test_commas_do_not_break(raw1, raw2, raw3):
_ = Name(f"{raw1},{raw2}").raw
_ = Name(f"{raw1},{raw2},{raw3}").raw
def issue_4_clean_nicknames():
name = Name("ang%ua (ba%%rky) uberwald")
assert name.first == "angua"
assert name.nickname == "barky"
def only_weird_prefixes():
assert Name("van de von der von")
def issue_52_hyphenated_number():
name = Name("¼,A,A")
assert name.first == "a"
assert name.last == "a"
def issue_14_mc_and_mac_as_prefix(first, second):
assert Name(first) == Name(second)
def t_convert_unicode(self, raw):
"""This is handled by unidecode and should not be extensively tested"""
assert Name._pre_clean(raw) == "mr eric praline"
def issue_22_consistently_handle_of_prefixes_in_first_name():
name1 = Name("de ook, van ook yolo moomoo")
name2 = Name("van ook yolo moomoo de ook")
assert name1 == name2
def t_colons_and_commas(self, raw):
assert Name.clean(raw) == "praline, mr eric"
def cli_report(raw_name: str, details: bool = True) -> int:
"""Parse into Name, output (core or report) attributes."""
name = Name(raw_name)
output = name.report() if details else dict(name)
prettier_print(output)
return 0
def t_handle_certain_symbols(self, raw, cooked):
assert Name.clean(raw) == cooked
def parse_name(s: str) -> T.Dict[str, T.Any]:
"""Parse into Name, return core name attributes as a dict.
This is the simplest function interface to *nominally*.
"""
return dict(Name(s))
def test_sweep_suffixes(incoming, outs, outwork):
bob = Name()
observed_string = bob._sweep_suffixes(incoming)
assert observed_string.strip() == outs
assert bob.detail["suffix"] == outwork
def t_force_lower(self, raw):
assert Name._pre_clean(raw) == "dinsdale"
def test_way_too_many_name_parts_post_suffix():
name = Name("johnson, bergholt, stuttley, joonyor")
assert name.first == "bergholt"
assert name.middle == "stuttley joonyor"
assert name.last == "johnson"
def test_alphabetical_text_does_not_break(raw):
_ = Name(raw).raw