def test_resolve(self):
resolver = ResolvePopoloName(
date = datetime.date(month=11, year=2010, day=1) )
popolo_person = resolver.get_person('J Q Smith')
self.assertEqual(self.john_q, popolo_person)
def test_resolve_with_filter(self):
resolver = ResolvePopoloName(
date=datetime.date(month=11, year=2010, day=1),
person_filter=ExcludePeopleWithNoMiddleNames(),
)
popolo_person = resolver.get_person('John Smith')
self.assertEqual(popolo_person.name, 'John Quentin Smith')
def set_resolver_for_date(self, date_string='', date=None):
self.resolver = ResolvePopoloName(
date=date,
date_string=date_string,
person_filter=AllowedPersonFilter(self.pombola_id_blacklist),
)
class ImportZAMixin(object):
def __init__(self, instance=None, commit=True, pombola_id_blacklist=None, **kwargs):
super(ImportZAMixin, self).__init__(
instance=instance,
commit=commit,
**kwargs
)
self.person_cache = {}
self.pombola_id_blacklist = pombola_id_blacklist
def set_resolver_for_date(self, date_string='', date=None):
self.resolver = ResolvePopoloName(
date=date,
date_string=date_string,
person_filter=AllowedPersonFilter(self.pombola_id_blacklist),
)
def debug_output_csv_row(self, pombola_person_slug, from_cache, our_speaker, speaker_from_slug):
'''A method to help produce data for analyzing name resolution results'''
import csv
with open('name-resolution.log', 'ab') as f:
writer = csv.writer(f)
writer.writerow([
str(self.resolver.date),
'' if pombola_person_slug is None else pombola_person_slug,
our_speaker.id if our_speaker else '',
our_speaker.name.encode('utf-8') if our_speaker else '',
speaker_from_slug.id if speaker_from_slug else '',
speaker_from_slug.name.encode('utf-8') if speaker_from_slug else '',
])
def get_person(self, name, party, pombola_person_slug=None):
# If we can directly find the person from the
# pombola_person_slug, use that - the Code4SA / PMG
# identification of speakers seems to be better than that from
# popolo_name_resolver.
speaker_from_slug = None
if pombola_person_slug is not None:
speaker_from_slug = Speaker.objects.filter(
identifiers__scheme='pombola_person_slug',
identifiers__identifier=pombola_person_slug).first()
if speaker_from_slug:
return speaker_from_slug
cached = self.person_cache.get(name, None)
if cached:
return cached
display_name = name or '(narrative)'
speaker = None
person = None
if name:
person = self.resolver.get_person(display_name, party)
if person:
speaker = person.speaker
if not speaker:
try:
speaker = Speaker.objects.get(instance=self.instance, name=display_name)
except Speaker.DoesNotExist:
speaker = Speaker(instance=self.instance, name=display_name)
if self.commit:
speaker.save()
self.person_cache[name] = speaker
return speaker
def set_resolver_for_date(self, date_string='', date=None):
self.resolver = ResolvePopoloName(
date=date,
date_string=date_string,
person_filter=AllowedPersonFilter(self.pombola_id_blacklist),
)
class ImportZAMixin(object):
def __init__(self, instance=None, commit=True, pombola_id_blacklist=None, **kwargs):
super(ImportZAMixin, self).__init__(
instance=instance,
commit=commit,
**kwargs
)
self.person_cache = {}
self.pombola_id_blacklist = pombola_id_blacklist
def set_resolver_for_date(self, date_string='', date=None):
self.resolver = ResolvePopoloName(
date=date,
date_string=date_string,
person_filter=AllowedPersonFilter(self.pombola_id_blacklist),
)
def debug_output_csv_row(self, pombola_person_slug, from_cache, our_speaker, speaker_from_slug):
'''A method to help produce data for analyzing name resolution results'''
import csv
with open('name-resolution.log', 'ab') as f:
writer = csv.writer(f)
writer.writerow([
str(self.resolver.date),
'' if pombola_person_slug is None else pombola_person_slug,
our_speaker.id if our_speaker else '',
our_speaker.name.encode('utf-8') if our_speaker else '',
speaker_from_slug.id if speaker_from_slug else '',
speaker_from_slug.name.encode(
'utf-8') if speaker_from_slug else '',
])
def get_person(self, name, party, pombola_person_slug=None):
# If we can directly find the person from the
# pombola_person_slug, use that - the Code4SA / PMG
# identification of speakers seems to be better than that from
# popolo_name_resolver.
speaker_from_slug = None
if pombola_person_slug is not None:
speaker_from_slug = Speaker.objects.filter(
identifiers__scheme='pombola_person_slug',
identifiers__identifier=pombola_person_slug).first()
if speaker_from_slug:
return speaker_from_slug
cached = self.person_cache.get(name, None)
if cached:
return cached
display_name = name or '(narrative)'
speaker = None
person = None
if name:
person = self.resolver.get_person(display_name, party)
if person:
speaker = person.speaker
if not speaker:
try:
speaker = Speaker.objects.get(
instance=self.instance, name=display_name)
except Speaker.DoesNotExist:
speaker = Speaker(instance=self.instance, name=display_name)
if self.commit:
speaker.save()
self.person_cache[name] = speaker
return speaker
def test_resolve(self):
resolver = ResolvePopoloName(date=date(2013, 06, 21))
person = resolver.get_person('Mrs A Steyn', None)
self.assertTrue(person)
