def get_existing_popit_person(vi_person_id):
from candidates.models import PopItPerson
from candidates.popit import get_search_url
# See if this person already exists by searching for the
# ID they were imported with:
query_format = \
'identifiers.identifier:"{id}" AND ' + \
'identifiers.scheme:"{scheme}"'
search_url = get_search_url(
'persons',
query_format.format(
id=vi_person_id, scheme='import-id'
),
embed='membership.organization'
)
results = requests.get(search_url).json()
total = results['total']
if total > 1:
message = "Multiple matches for CI ID {0}"
raise Exception(message.format(vi_person_id))
if total == 0:
return None
# Otherwise there was exactly one result:
return PopItPerson.create_from_dict(results['result'][0])
def get_party_data(party_name):
from candidates.popit import get_search_url
# See if this person already exists by searching for the
# ID they were imported with:
party_name = party_name.replace('/', '')
party_name = party_name.decode('utf-8')
query_format = \
'name:"{name}"'
search_url = get_search_url(
'organizations',
query_format.format(
name=party_name
)
)
print(party_name)
results = requests.get(search_url).json()
print(results)
total = results['total']
if total > 1:
message = "Multiple matches for party {0}"
raise Exception(message.format(party_name))
if total == 0:
return None
# Otherwise there was exactly one result:
return results['result'][0]
def _objects_from_popit_search(self):
self.page = int(self.request.GET.get(self.page_kwarg) or 1)
url = get_search_url(
'persons',
"_missing_:%s AND _exists_:standing_in.2015.post_id" %
self.get_field(),
page=self.page,
per_page=20
)
page_result = requests.get(url).json()
return page_result
def get_existing_popit_person(vi_person_id):
# See if this person already exists by searching for the
# ID they were imported with:
query_format = \
'identifiers.identifier:"{id}" AND ' + \
'identifiers.scheme:"{scheme}"'
search_url = get_search_url('persons',
query_format.format(id=vi_person_id,
scheme='import-id'),
embed='membership.organization')
results = requests.get(search_url).json()
total = results['total']
if total > 1:
message = "Multiple matches for CI ID {0}"
raise Exception(message.format(vi_person_id))
if total == 0:
return None
# Otherwise there was exactly one result:
return PopItPerson.create_from_dict(results['result'][0])
def get_party_data(party_name):
from candidates.popit import get_search_url
# See if this person already exists by searching for the
# ID they were imported with:
party_name = party_name.replace('/', '')
party_name = party_name.decode('utf-8')
query_format = \
'name:"{name}"'
search_url = get_search_url('organizations',
query_format.format(name=party_name))
print(party_name)
results = requests.get(search_url).json()
print(results)
total = results['total']
if total > 1:
message = "Multiple matches for party {0}"
raise Exception(message.format(party_name))
if total == 0:
return None
# Otherwise there was exactly one result:
return results['result'][0]
def handle_person(self, ppc_data, image_filename):
print u"PPC ({party_slug}): {name}".format(**ppc_data).encode('utf-8')
# Search PopIt for anyone with the same name. (FIXME: we
# should make this a bit fuzzier when the PopIt API
# supports that.)
person_search_url = get_search_url(
'persons', '"' + ppc_data['name'] + '"'
)
r = requests.get(
person_search_url
)
add_new_person = True
for result in r.json()['result']:
# FIXME: With the current code, updating Elasticsearch
# fails because it rejects the empty string as a null date
# value. To just get this working, GET the person directly
# so the information's defintely up to date.
result = self.api.persons(result['id']).get()['result']
# We don't need past versions of the person or their
# memberships, and it's unnecessary verbose to record them
# with decisions.
del result['versions']
del result['memberships']
popit_person_id = result['id']
message = " Considering PopIt person search result {0}"
print message.format(popit_person_id)
added_for_2015 = self.already_added_for_2015(ppc_data, result)
if added_for_2015 is None:
add_new_person = False
print " already_added_for_2015 returned None, so ignoring"
print " !!! CHECK THIS MANUALLY !!!"
break
elif added_for_2015:
# Then just update that person with the possibly
# updated scraped data:
add_new_person = False
print " matched a 2015 candidate, so updating that"
self.update_popit_person(popit_person_id, ppc_data, image_filename)
break
else:
# Do we already have a decision about whether this PPC is
# the same as another from 2010?
decision = self.already_matched_to_a_person(ppc_data, popit_person_id)
if decision is None:
print " no previous decision found, so asking"
# We have to ask the user for a decision:
if self.decision_from_user(ppc_data, result):
print " updating, and recording the decision that they're a match"
add_new_person = False
self.update_popit_person(popit_person_id, ppc_data, image_filename)
record_human_decision(ppc_data, popit_person_id, True)
break
else:
print " recording the decision that they're not a match"
record_human_decision(ppc_data, popit_person_id, False)
else:
# Otherwise we'd previously decided that they're
# the same person, so just update that person.
print " there was a previous decision found"
if decision:
print " the previous decision was that they're a match, so updating"
add_new_person = False
self.update_popit_person(popit_person_id, ppc_data, image_filename)
break
if add_new_person:
new_person_id = self.add_popit_person(ppc_data, image_filename)
print " Added them as a new person ({0})".format(new_person_id)
