def generate_person_long_names(engine):
log.info("Generating person fingerprints and slugs...")
from offenesparlament.transform.namematch import match_speaker
nkp = nk_persons()
Person = sl.get_table(engine, 'person')
for person in sl.find(engine, Person):
long_name = make_long_name(person)
try:
long_name = match_speaker(long_name)
except NKNoMatch:
pass
log.info(" -> %s" % long_name.strip())
slug = url_slug(long_name)
sl.upsert(engine, Person, {
'fingerprint': long_name,
'slug': slug,
'id': person['id']},
unique=['id'])
tries = 0
while True:
try:
nkp.ensure_value(long_name, data=person)
except ValueError, E:
log.warn('Exception: %s' % str(E))
tries = tries + 1
if tries > 5:
raise
else:
break
def generate_person_long_names(db):
log.info("Generating person fingerprints and slugs...")
Person = db['person']
for person in Person:
long_name = make_long_name(person)
log.debug(" -> %s" % long_name.strip())
slug = url_slug(long_name)
Person.writerow({'fingerprint': long_name,
'slug': slug,
'__id__': person['__id__']},
unique_columns=['__id__'])
for fp in Person.distinct('fingerprint'):
if fp['_count'] > 1:
raise ValueError("Partial fingerprint: %s" % fp['fingerprint'])
log.info("Updating 'rollen' to have fingerprints...")
Rolle = db['rolle']
for person in Person:
if person['mdb_id']:
Rolle.writerow({
'mdb_id': person['mdb_id'],
'fingerprint': person['fingerprint']
}, unique_columns=['mdb_id'])
elif person['source_url']:
Rolle.writerow({
'person_source_url': person['person_source_url'],
'fingerprint': person['fingerprint']
}, unique_columns=['person_source_url'])