def compare_va_to_ra(va_id, ra_id):
'''
Compares the origin names of a virtual author against the name list of
a real author
@param va_id: ID of the virtual author
@type va_id: int
@param ra_id: ID of the real author
@type ra_id: int
@return: The probability resulting from the name comparison.
@rtype: float
'''
bconfig.LOGGER.info("|-> Start of name comparison (va %s : ra %s)"
% (va_id, ra_id))
ra_names = get_realauthor_names_from_set(ra_id)
va_nameid_recs = get_virtualauthor_records(va_id, tag='orig_authorname_id')
# print "RA Names: ", ra_names
# print "VA Name: ", va_name
authorname_id = -1
if va_nameid_recs:
authorname_id = va_nameid_recs[0]['value']
authorname_strings = get_name_and_db_name_strings(authorname_id)
if not authorname_strings["name"]:
return 0.0
comparisons = []
for ra_name in ra_names:
comparison = compare_names(authorname_strings["name"], ra_name)
bconfig.LOGGER.info("|-> %s & %s -> %s"
% (authorname_strings["name"],
ra_name, comparison))
comparisons.append(comparison)
#print "checking ",name_1," against ", name_2
bconfig.LOGGER.debug("|--> Name comparisons: %s" % (comparisons))
bconfig.LOGGER.info("|-> End of name comparison")
# ret = average(comparisons)
ret = float(sum(comparisons)) / len(comparisons)
if ret < .1:
ret = 0 #.1
bconfig.LOGGER.info("|--> Resulting name probability: %s" % (ret))
return ret
def compare_va_to_ra(va_id, ra_id):
"""
Compares the origin names of a virtual author against the name list of
a real author
@param va_id: ID of the virtual author
@type va_id: int
@param ra_id: ID of the real author
@type ra_id: int
@return: The probability resulting from the name comparison.
@rtype: float
"""
bconfig.LOGGER.info("|-> Start of name comparison (va %s : ra %s)" % (va_id, ra_id))
ra_names = get_realauthor_names_from_set(ra_id)
va_nameid_recs = get_virtualauthor_records(va_id, tag="orig_authorname_id")
# print "RA Names: ", ra_names
# print "VA Name: ", va_name
authorname_id = -1
if va_nameid_recs:
authorname_id = va_nameid_recs[0]["value"]
authorname_strings = get_name_and_db_name_strings(authorname_id)
if not authorname_strings["name"]:
return 0.0
comparisons = []
for ra_name in ra_names:
comparison = compare_names(authorname_strings["name"], ra_name)
bconfig.LOGGER.info("|-> %s & %s -> %s" % (authorname_strings["name"], ra_name, comparison))
comparisons.append(comparison)
# print "checking ",name_1," against ", name_2
bconfig.LOGGER.debug("|--> Name comparisons: %s" % (comparisons))
bconfig.LOGGER.info("|-> End of name comparison")
# ret = average(comparisons)
ret = float(sum(comparisons)) / len(comparisons)
if ret < 0.1:
ret = 0 # .1
bconfig.LOGGER.info("|--> Resulting name probability: %s" % (ret))
return ret
def test_compare_names(self):
"""bibauthorid - test names comparison funcions"""
self.assertEqual(0.94999999999999996,
bau.compare_names('Ellis, j.', 'Ellis, j.'))
self.assertEqual(1.0, bau.compare_names('Ellis, john', 'Ellis, john'))
self.assertEqual(1.0, bau.compare_names('John Ellis', 'John Ellis'))
# self.assertEqual(0.94999999999999996,
# bau.compare_names('J. Ellis','J. Ellis'))
self.assertEqual(0.0, bau.compare_names('John Ellis', 'John Mark'))
self.assertEqual(0.0, bau.compare_names('Ellis, John', 'Mark, John'))
self.assertEqual(0.0, bau.compare_names('', ''))
def test_compare_names(self):
"""bibauthorid - test names comparison funcions"""
self.assertEqual(0.94999999999999996,
bau.compare_names('Ellis, j.', 'Ellis, j.'))
self.assertEqual(1.0,
bau.compare_names('Ellis, john', 'Ellis, john'))
self.assertEqual(1.0,
bau.compare_names('John Ellis', 'John Ellis'))
# self.assertEqual(0.94999999999999996,
# bau.compare_names('J. Ellis','J. Ellis'))
self.assertEqual(0.0,
bau.compare_names('John Ellis', 'John Mark'))
self.assertEqual(0.0,
bau.compare_names('Ellis, John', 'Mark, John'))
self.assertEqual(0.0,
bau.compare_names('', ''))
