def create_test_db(self):
models.Section.objects.all().delete()
models.Declarator_File_Reference.objects.all().delete()
models.Source_Document.objects.all().delete()
models.Person.objects.all().delete()
self.assertGreater(models.Office.objects.count(), 0)
src_doc = models.Source_Document(id=1)
src_doc.save()
models.Declarator_File_Reference(
source_document=src_doc,
declarator_document_id=self.declarator_document_id).save()
section1 = models.Section(id=self.section_id1,
source_document=src_doc,
person_name=self.fio,
office_id=1)
section1.save()
section2 = models.Section(id=self.section_id2,
source_document=src_doc,
person_name=self.fio,
office_id=1)
section2.save()
models.Income(section=section1,
size=self.income_main1,
relative=models.Relative.main_declarant_code).save()
models.Income(section=section2,
size=self.income_main2,
relative=models.Relative.main_declarant_code).save()
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):
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 test_search_section_by_partial_person_name(self):
models.Section.objects.all().delete()
models.Source_Document.objects.all().delete()
src_doc = models.Source_Document(id=1)
src_doc.save()
models.Section(id=1, person_name="Один Иван Ильич", source_document=src_doc, office_id=1).save()
models.Section(id=2, person_name="Два Иван Ильич", source_document=src_doc, office_id=1).save()
models.Section(id=3, person_name="Иван Ильич", source_document=src_doc, office_id=1).save()
BuildElasticIndex(None, None).handle(None, model="section")
res = self.search_sections_by_fio("Один Иван")
self.assertEqual(len(res), 1)
self.assertEqual(res[0].id, 1)
res = self.search_sections_by_fio("Два Иван")
self.assertEqual(len(res), 1)
self.assertEqual(res[0].id, 2)
res = self.search_sections_by_fio("Один Иван Ильич")
self.assertEqual(len(res), 1)
self.assertEqual(res[0].id, 1)
res = self.search_sections_by_fio("Иван Ильич")
self.assertEqual(len(res), 3)
res = self.search_sections_by_fio("Ильич")
self.assertEqual(len(res), 3)
res = self.search_sections_by_fio("Один")
self.assertEqual(len(res), 1)
def test_elastic(self):
self.assertGreater(models.Office.objects.count(), 0)
#delete all documents
index = Index(settings.ELASTICSEARCH_INDEX_NAMES['section_index_name'],
Elasticsearch())
index.delete()
index.create()
time.sleep(2)
#search to get no results
people = list(ElasticSectionDocument.search().query(
'match', person_name='Иванов'))
self.assertEqual(len(people), 0)
models.Section.objects.all().delete()
models.Source_Document.objects.all().delete()
ofc = models.Office.objects.get(id=1)
src_doc = models.Source_Document()
src_doc.id = 1
src_doc.save()
models.Section(id=1,
person_name="Иванов Иван",
source_document=src_doc,
office=ofc).save()
models.Section(id=2,
person_name="Петров Петр",
source_document=src_doc,
office=ofc).save()
models.Section(id=3,
person_name="Сидоров Федор",
source_document=src_doc,
office=ofc).save()
#reindex section index
TSectionElasticIndexator.chunk_size = 2
BuildElasticIndex(None, None).handle(None, model="section")
time.sleep(2)
people = list(ElasticSectionDocument.search().query(
'match', person_name='Иванов'))
self.assertEqual(len(people), 1)
people = list(ElasticSectionDocument.search().query(
'match', person_name='Петров'))
self.assertEqual(len(people), 1)
people = list(ElasticSectionDocument.search().query(
'match', person_name='Сидоров'))
self.assertEqual(len(people), 1)
people = list(ElasticSectionDocument.search().query(
'match', person_name='Сокирко'))
self.assertEqual(len(people), 0)
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_elastic(self):
ElasticSectionDocument.init()
ElasticSectionDocument._index._name.endswith("_test")
ElasticSectionDocument.search().query().delete()
time.sleep(2)
people = list(ElasticSectionDocument.search().query(
'match', person_name='Иванов'))
self.assertEqual(len(people), 0)
people = list(ElasticSectionDocument.search().query(
'match', person_name='Иванов'))
models.Section.objects.all().delete()
models.Source_Document.objects.all().delete()
ofc = models.Office()
ofc.name = "some name"
ofc.save()
src_doc = models.Source_Document()
src_doc.id = 1
src_doc.office = ofc
src_doc.save()
section = models.Section()
section.id = 1
section.person_name = "Иванов Иван"
section.source_document = src_doc
section.save()
print("sleep 2 seconds till elastic processes records")
time.sleep(2)
people = list(ElasticSectionDocument.search().query(
'match', person_name='Иванов'))
print(len(people))
self.assertEqual(len(people), 1)
def check_case(self, use_only_surname, check_ambiguity):
models.Section.objects.all().delete()
self.assertGreater(models.Office.objects.count(), 0)
fio = "Иванов Иван Иванович"
document_id = 1784
income_main = 12534
declarator_person_id = 178
person_ids_path = os.path.join(os.path.dirname(__file__), "person_ids.json")
with open(person_ids_path, "w") as outp:
fio_key = fio
if use_only_surname:
fio_key = fio.split()[0]
value = declarator_person_id
if check_ambiguity:
value = "AMBIGUOUS_KEY"
record = {
build_section_passport(document_id, fio_key, income_main): value
}
json.dump(record, outp, ensure_ascii=False, indent=4)
src_doc = models.Source_Document()
src_doc.office_id = 1
src_doc.id = 1
src_doc.save()
decl_info = models.Declarator_File_Reference(source_document=src_doc,
declarator_document_id=document_id)
decl_info.save()
section = models.Section(source_document=src_doc,
person_name=fio)
section.id = 1
section.save()
income = models.Income(section=section,
size=income_main,
relative=models.Relative.main_declarant_code
)
income.save()
permalinks_path = os.path.join(os.path.dirname(__file__), "permalinks.dbm")
p = TPermaLinksDB(permalinks_path)
p.create_db()
p.close_db()
copier = CopyPersonIdCommand(None, None)
copier.handle(None, read_person_from_json=person_ids_path, permanent_links_db=permalinks_path)
section.refresh_from_db()
if check_ambiguity:
self.assertEqual(section.person, None)
else:
self.assertEqual(section.person.declarator_person_id, declarator_person_id)
def test_search_section_by_person_name(self):
self.assertGreater(models.Office.objects.count(), 0)
models.Section.objects.all().delete()
models.Source_Document.objects.all().delete()
src_doc = models.Source_Document(id=1)
src_doc.save()
models.Section(id=1, person_name="Иванов Иван Иванович", source_document=src_doc, office_id=1).save()
BuildElasticIndex(None, None).handle(None, model="section")
self.assertEqual(self.search_sections_by_fio("Иванов И.И.")[0].id, 1)
self.assertEqual(self.search_sections_by_fio("Иванов Иван Иванович")[0].id, 1)
self.assertEqual(self.search_sections_by_fio("Иванов Иван")[0].id, 1)
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))
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 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 create_section(self, section_id, person_name, person=None):
section = models.Section(id=section_id, office_id=TEST_OFFICE_ID,
source_document=self.src_doc, person_name=person_name, person=person)
section.save()
return section