def link_sections_to_a_new_person(self, sections, section_distances,
person_id):
assert len(section_distances) == len(sections)
if person_id is None:
person_id = self.permalinks_db._get_new_id()
self.logger.debug("create new person.id: {}".format(person_id))
else:
self.logger.debug("use old person.id: {}".format(person_id))
try:
person = models.Person.objects.get(id=person_id)
#reuse person record from declarator
if person.declarator_person_id is None:
self.logger.error(
"Warning! Reuse existing person_id = {} for different sections (cluster), it could happen"
"if this person_id was used for different person created by copy_person_id.py "
"but should not happen if declarator_person_id is None. ".
format(person_id))
except models.Person.DoesNotExist as exp:
#create new person record
person = models.Person(id=person_id)
person.save()
for (section, distance) in zip(sections, section_distances):
self.link_section_to_person(section, person, distance)
def test(self):
src_doc = create_default_source_document()
person = models.Person(id=2, declarator_person_id=1111, person_name="Иванов Иван Иванович")
person.save()
models.Section(id=1, source_document=src_doc, person_name="Иванов Иван Иванович", person=person).save()
models.Section(id=2, source_document=src_doc, person_name="Иванов И. И.").save()
person.refresh_from_db()
permalinks_path = os.path.join(os.path.dirname(__file__), "permalinks.dbm")
p = TPermaLinksDB(permalinks_path)
p.create_db()
p.save_next_primary_key_value(models.Person, 3)
p.create_sql_sequences()
p.close()
run_dedupe = RunDedupe(None, None)
run_dedupe.handle(None,
permanent_links_db=permalinks_path,
write_to_db=True,
fake_dedupe=True,
surname_bounds=',',
rebuild=True)
sec1 = models.Section.objects.get(id=1)
self.assertEqual(sec1.person_id, person.id)
sec2 = models.Section.objects.get(id=2)
self.assertEqual(sec2.person_id, person.id)
def test(self):
self.create_test_db()
person_id = 1
person = models.Person(id=person_id, person_name=self.fio)
self.assertIsNone(person.declarator_person_id)
person.save()
section1 = models.Section.objects.get(id=self.section_id1)
section1.person = person
section1.save()
TPermaLinksPerson(CopyPersonIdTestCaseBase.permalinks_folder
).create_and_save_empty_db()
CreatePermalinksStorageCommand(None, None).handle(
None, directory=CopyPersonIdTestCaseBase.permalinks_folder)
permalinks_db = TPermaLinksPerson(
CopyPersonIdTestCaseBase.permalinks_folder)
permalinks_db.open_db_read_only()
permalinks_db.recreate_auto_increment_table()
self.run_copy_person_id(False, False)
self.assertEqual(models.Person.objects.count(), 1)
section1 = models.Section.objects.get(id=self.section_id1)
self.assertEqual(section1.person.declarator_person_id,
self.declarator_person_id)
self.assertEqual(section1.person.id, person_id)
def test(self):
src_doc = create_default_source_document()
models.Section(id=1, source_document=src_doc, person_name="Иванов Иван Иванович").save()
models.Section(id=2, source_document=src_doc, person_name="Иванов И. И.").save()
permalinks_path = os.path.join(os.path.dirname(__file__), "permalinks.dbm")
db = TPermaLinksDB(permalinks_path)
db.create_db()
person = models.Person(id=99)
person.tmp_section_set = {str(1), str(2)}
db.put_record_id(person)
db.save_next_primary_key_value(models.Person, 100)
db.create_sql_sequences()
db.close()
run_dedupe = RunDedupe(None, None)
run_dedupe.handle(None,
permanent_links_db=permalinks_path,
write_to_db=True,
fake_dedupe=True,
surname_bounds=',',
rebuild=True)
sec1 = models.Section.objects.get(id=1)
self.assertEqual(sec1.person_id, person.id)
sec2 = models.Section.objects.get(id=2)
self.assertEqual(sec2.person_id, person.id)
def link_sections_to_a_new_person(self, section_ids):
person = models.Person()
person.tmp_section_set = set(str(id) for (id, score) in section_ids)
person.id = self.primary_keys_builder.get_record_id(person)
person.save()
for (section_id, score) in section_ids:
section = models.Section.objects.get(id=section_id)
self.link_section_to_person(section, person, score)
def copy_human_merge(self, section, declarator_person_id):
# we think that person ids in declarator db are stable
person = models.Person(declarator_person_id=declarator_person_id)
person.id = self.primary_keys_builder.get_record_id(person)
self.logger.debug("connect section {} to person {}, declarator_person_id={}".format(
section.id, person.id, declarator_person_id))
if person.person_name is None or len(person.person_name) < len(section.person_name):
person.person_name = section.person_name
person.save()
section.person = person
section.dedupe_score = None
section.save()
def test_rating(self):
models.Person.objects.all().delete()
models.Section.objects.all().delete()
models.Source_Document.objects.all().delete()
src_doc = models.Source_Document(id=1)
src_doc.save()
person_id = 99
person = models.Person(id=person_id)
person.save()
models.Section(id=1,
source_document=src_doc,
person_name="i1",
income_year=2019,
office_id=1,
person=person).save()
models.Section(id=2,
source_document=src_doc,
person_name="i2",
income_year=2019,
office_id=1,
person=person).save()
models.Section(id=3,
source_document=src_doc,
person_name="i3",
income_year=2019,
office_id=1,
person=person).save()
models.Income(section_id=1, size=1, relative=models.Relative.main_declarant_code).save()
models.Income(section_id=2, size=2, relative=models.Relative.main_declarant_code).save()
models.Income(section_id=3, size=3, relative=models.Relative.main_declarant_code).save()
builder = BuildRatingCommand(None, None)
builder.handle(None, min_members_count=3)
self.assertEqual(models.Person_Rating_Items.objects.count(), 3)
rating = list(models.Person_Rating_Items.objects.all())
self.assertEqual(rating[0].rating_value, 3)
self.assertEqual(rating[0].person_place, 1)
self.assertEqual(rating[1].rating_value, 2)
self.assertEqual(rating[1].person_place, 2)
self.assertEqual(rating[2].rating_value, 1)
self.assertEqual(rating[2].person_place, 3)
def test(self):
self.initialize()
permalinks_folder = os.path.dirname(__file__)
person_id = 99
person = models.Person(id=person_id)
person.save()
section1 = self.create_section(1,
"Иванов Иван Иванович",
person=person)
section2 = self.create_section(2,
"Иванов Иван Иванович",
person=person)
CreatePermalinksStorageCommand(None, None).handle(
None, directory=permalinks_folder)
TPermaLinksPerson(permalinks_folder).open_db_read_only(
).recreate_auto_increment_table()
section1.person = None
section1.save()
section2.person = None
section2.save()
person.delete()
run_dedupe = RunDedupe(None, None)
run_dedupe.handle(None,
permalinks_folder=permalinks_folder,
write_to_db=True,
fake_dedupe=True,
separate_sections=True,
surname_bounds=',',
take_sections_with_empty_income=True,
rebuild=True)
self.assertEqual(2, models.Person.objects.count())
#"person_id" is inherited by the minimal section_id, if there is no other grounds
sec1 = models.Section.objects.get(id=1)
self.assertEqual(sec1.person_id, person_id)
sec2 = models.Section.objects.get(id=2)
self.assertEqual(sec2.person_id, person_id + 1) # a new person_id
def read_dumped_objects(self, file_name):
if self.options.get('recreate_db'):
assert models.Section.objects.count() == 0
with open(file_name) as inp:
for line in inp:
js = json.loads(line)
o = TDeduplicationObject().from_json(js)
if self.options.get('recreate_db'):
if o.record_id.source_table == TDeduplicationObject.SECTION:
assert len(o.offices) == 1
s = models.Section(id=o.record_id.id,
office_id=list(o.offices)[0])
self.section_cache[o.record_id.id] = s
s.save()
else:
models.Person(id=o.record_id.id).save()
self.cluster_by_minimal_fio[
o.fio.build_fio_with_initials()].append(o)
def test(self):
self.initialize()
person_id = 99
person = models.Person(id=person_id)
person.save()
section1 = self.create_section(1,
"Иванов Иван Иванович",
person=person)
section2 = self.create_section(2, "Иванов И. И.", person=person)
section3 = self.create_section(3, "Петров И. И.")
permalinks_folder = os.path.dirname(__file__)
db = TPermaLinksPerson(permalinks_folder)
db.create_db()
db.save_dataset(setup_logging())
db.recreate_auto_increment_table()
db.close_db()
section1.person = None
section1.save()
section2.person = None
section2.save()
person.delete()
run_dedupe = RunDedupe(None, None)
run_dedupe.handle(None,
permalinks_folder=permalinks_folder,
write_to_db=True,
fake_dedupe=True,
surname_bounds=',',
take_sections_with_empty_income=True,
rebuild=True)
self.assertEqual(models.Person.objects.count(), 1)
sec1 = models.Section.objects.get(id=1)
self.assertEqual(sec1.person_id, person_id)
sec2 = models.Section.objects.get(id=2)
self.assertEqual(sec2.person_id, person_id)
sec3 = models.Section.objects.get(id=3)
self.assertEqual(sec3.person_id, person_id)
def create_records(self, records):
models.Section.objects.all().delete()
models.Source_Document.objects.all().delete()
models.Person.objects.all().delete()
models.PersonRedirect.objects.all().delete()
assert models.Office.objects.all().count() > 0
for d in records.get('source_documents', []):
d = models.Source_Document(**d)
d.save()
for d in records.get('persons', []):
models.Person(**d).save()
for d in records.get('sections', []):
if len(models.Office.objects.filter(id=d['office_id'])) == 0:
o = models.Office(id=d['office_id'], name="aaa")
o.save()
models.Section(**d).save()
for d in records.get('redirects', []):
models.PersonRedirect(**d).save()
def test(self):
self.initialize()
person_id = 2
declarator_person_id = 1111
person = models.Person(id=person_id,
declarator_person_id=declarator_person_id,
person_name="Иванов Иван Иванович")
person.save()
self.create_section(1, "Иванов Иван Иванович", person)
self.create_section(2, "Иванов И. И.")
permalinks_folder = os.path.dirname(__file__)
db = TPermaLinksPerson(permalinks_folder)
db.create_db()
db.save_dataset(setup_logging())
#db.save_max_plus_one_primary_key(3)
db.recreate_auto_increment_table()
db.close_db()
run_dedupe = RunDedupe(None, None)
run_dedupe.handle(None,
permalinks_folder=permalinks_folder,
write_to_db=True,
fake_dedupe=True,
surname_bounds=',',
take_sections_with_empty_income=True,
rebuild=True)
self.assertEqual(models.Person.objects.count(), 1)
person = models.Person.objects.get(id=person_id)
self.assertIsNotNone(person)
self.assertEqual(declarator_person_id, person.declarator_person_id)
sec1 = models.Section.objects.get(id=1)
self.assertEqual(sec1.person_id, person.id)
sec2 = models.Section.objects.get(id=2)
self.assertEqual(sec2.person_id, person.id)
def copy_human_merge(self, section, declarator_person_id):
person = self.declarator_person_id_to_disclosures_person.get(
declarator_person_id)
if person is None:
person_id = self.permalinks_db.get_person_id_by_declarator_id(
declarator_person_id, section.id)
if person_id in self.disclosures_person_id_to_disclosures_person:
person = self.disclosures_person_id_to_disclosures_person.get(
person_id)
if declarator_person_id != person.declarator_person_id:
self.logger.error(
"Person id={} has conflict declarator_person_id ({} != {}), use the first person id {}"
.format(person_id, declarator_person_id,
person.declarator_person_id,
person.declarator_person_id))
else:
person = models.Person(
id=person_id,
declarator_person_id=declarator_person_id,
person_name=section.person_name)
person.save()
self.declarator_person_id_to_disclosures_person[
declarator_person_id] = person
self.disclosures_person_id_to_disclosures_person[
declarator_person_id] = person
elif person.person_name is None or len(person.person_name) < len(
section.person_name):
person.person_name = section.person_name
person.save()
assert person.declarator_person_id is not None
self.logger.debug(
"connect section {} to person {}, declarator_person_id={}".format(
section.id, person.id, person.declarator_person_id))
section.person = person
section.dedupe_score = None
section.save()
def test_corrected_person(self):
models.Section.objects.all().delete()
models.Source_Document.objects.all().delete()
models.Person.objects.all().delete()
src_doc = models.Source_Document(id=1)
src_doc.save()
assert SECTION_CORRECTIONS.get_corrected_section_id(8048661) == 9798543
models.Person(id=1, person_name="Иванов Иван Ильич").save()
models.Section(id=8048661,
income_year=2016,
person_name="Иванов Иван Ильич",
source_document=src_doc,
office_id=1,
person_id=1).save()
models.Section(id=9798543,
income_year=2016,
person_name="Иванов Иван Ильич",
source_document=src_doc,
office_id=1,
person_id=1).save()
person = models.Person.objects.get(id=1)
sections = person.sections_ordered_by_year
self.assertEqual(1, len(sections))
