def testSNameStandardiser(self):
"""Test name standardiser routines (surname first)"""
# return
ns = standardisation.NameStandardiser(descript = 'Test name standardiser',
input_fields = ['in_sname'],
output_fiel = ['title',
'gender_guess',
'given_name',
'alt_given_name',
'surname',
'alt_surname'],
female_t = self.name_female_titles,
male_t = self.name_male_titles,
tag_t=self.name_tag_table,
corr_l=self.name_corr_list,
first_name_c = 'sname',
hmm_train_fi = 'test-hmm-train.txt')
rs = standardisation.RecordStandardiser(descr = 'Test record standardiser',
input_dataset = self.in_ds,
output_dataset = self.out_ds,
comp_stand_list =[ns])
for (name_str, name_res) in self.names_snames:
clean_name_str = ns.clean_component(name_str)
test_name_res = ns.standardise(name_str, clean_name_str)
# assert name_res == test_name_res, \
# 'Wrong surname first standardisation: %s, should be: %s' % \
# (str(test_name_res), str(name_res))
print 'Count dict:', ns.count_dict
def testGNameStandardiser(
self): # -----------------------------------------
"""Test name standardiser routines (given name first)"""
# return
ns = standardisation.NameStandardiser(
descript="Test name standardiser",
input_fields=["in_gname"],
output_fiel=[
"title",
"gender_guess",
"given_name",
"alt_given_name",
"surname",
"alt_surname",
],
female_t=self.name_female_titles,
male_t=self.name_male_titles,
tag_t=self.name_tag_table,
corr_l=self.name_corr_list,
hmm_train_fil="test-hmm-train.txt",
)
rs = standardisation.RecordStandardiser(
descr="Test record standardiser",
input_dataset=self.in_ds,
output_dataset=self.out_ds,
comp_stand_list=[ns],
)
for (name_str, name_res) in self.names_gnames:
clean_name_str = ns.clean_component(name_str)
test_name_res = ns.standardise(name_str, clean_name_str)
# assert name_res == test_name_res, \
# 'Wrong given name first standardisation: %s, should be: %s' % \
# (str(test_name_res), str(name_res))
# rs.standardise() # Use record standardiser and write output file
print("Count dict:", ns.count_dict)