def test_can_add_subs_to_duplicate_set(self):
dup_set = models.DuplicateSubmissionSet()
dup_set.save()
subs = [factories.FormSubmissionFactory.create() for i in range(2)]
dup_set.submissions.add(*subs)
for sub in subs:
self.assertEqual(sub.duplicate_set, dup_set)
self.assertTrue(sub.duplicate_set.id)
def handle(self, *args, **options):
dup_sets = SubmissionsService.find_duplicates(
models.FormSubmission.objects.all())
self.stdout.write("Found {} duplicate sets".format(len(dup_sets)))
existing_dup_sets = models.DuplicateSubmissionSet.objects.all()
self.stdout.write("{} duplicate sets already exist".format(
existing_dup_sets.count()))
count_already_existed = 0
dup_set_extensions = {}
new_dup_sets = []
existing_dup_set_lookups = {
frozenset(dup_set.submissions.all()): dup_set
for dup_set in existing_dup_sets
}
for dup_set in dup_sets:
found_existing = False
for lookup_set, existing in existing_dup_set_lookups.items():
if dup_set == lookup_set:
found_existing = True
count_already_existed += 1
break
elif dup_set & lookup_set:
dup_set_extensions[existing] = dup_set
found_existing = True
break
if not found_existing:
new_dup_sets.append(dup_set)
for existing, new_dups in dup_set_extensions:
existing.submissions.add(*new_dups)
self.stdout.write("{} found duplicate sets were existing".format(
count_already_existed))
self.stdout.write("Extended {} existing duplicate sets".format(
len(dup_set_extensions)))
for new_set in new_dup_sets:
new_dup_set_object = models.DuplicateSubmissionSet()
new_dup_set_object.save()
new_dup_set_object.submissions.add(*new_set)
self.stdout.write("Created {} new duplicate sets".format(
len(new_dup_sets)))
def link_with_any_duplicates(submission, applicant_id):
"""Links submission with any duplicates from the same applicant
If duplicates are found, returns the DuplicateSubmissionSet id
If no duplicates are found, returns False
"""
duplicates = check_for_existing_duplicates(submission, applicant_id)
if duplicates:
dup_set_id = None
unadded_duplicates = []
for dup in duplicates:
if dup.duplicate_set_id:
dup_set_id = dup.duplicate_set_id
else:
unadded_duplicates.append(dup)
if not dup_set_id:
new_dup_set = models.DuplicateSubmissionSet()
new_dup_set.save()
new_dup_set.submissions.add(*unadded_duplicates)
dup_set_id = new_dup_set.id
submission.duplicate_set_id = dup_set_id
submission.save()
return dup_set_id
return False