def test_new_record(self):
cit = TempCitizenFactory()
stats = mirror_database(from_model=TempCitizen, to_model=Citizen)
self.assertEqual(0, stats.unchanged_count)
self.assertEqual(0, stats.modified_record_count)
self.assertEqual(1, stats.new_record_count)
self.assertEqual(0, stats.not_there_anymore_count)
self.assertTrue(Citizen.objects.filter(pk=cit.pk).exists())
def test_unchanged_record(self):
cit = CitizenFactory()
TempCitizenFactory(**model_to_dict(cit))
stats = mirror_database(from_model=TempCitizen, to_model=Citizen)
self.assertEqual(1, stats.unchanged_count)
self.assertEqual(0, stats.modified_record_count)
self.assertEqual(0, stats.new_record_count)
self.assertEqual(0, stats.not_there_anymore_count)
def test_empty_dbs(self):
stats = mirror_database(
from_model=TempCitizen,
to_model=Citizen
)
self.assertEqual(0, stats.unchanged_count)
self.assertEqual(0, stats.modified_record_count)
self.assertEqual(0, stats.new_record_count)
self.assertEqual(0, stats.not_there_anymore_count)
def test_changed_record(self):
cit = CitizenFactory()
temp_cit = TempCitizenFactory(pk=cit.pk,
first_name=cit.first_name + '_CHANGED')
stats = mirror_database(from_model=TempCitizen, to_model=Citizen)
self.assertEqual(0, stats.unchanged_count)
self.assertEqual(1, stats.modified_record_count)
self.assertEqual(0, stats.new_record_count)
self.assertEqual(0, stats.not_there_anymore_count)
c = Citizen.objects.get(pk=cit.pk)
self.assertEqual(temp_cit.first_name, c.first_name)
def test_new_record(self):
cit = TempCitizenFactory()
stats = mirror_database(
from_model=TempCitizen,
to_model=Citizen
)
self.assertEqual(0, stats.unchanged_count)
self.assertEqual(0, stats.modified_record_count)
self.assertEqual(1, stats.new_record_count)
self.assertEqual(0, stats.not_there_anymore_count)
self.assertTrue(Citizen.objects.filter(pk=cit.pk).exists())
def test_unchanged_record(self):
cit = CitizenFactory()
TempCitizenFactory(**model_to_dict(cit))
stats = mirror_database(
from_model=TempCitizen,
to_model=Citizen
)
self.assertEqual(1, stats.unchanged_count)
self.assertEqual(0, stats.modified_record_count)
self.assertEqual(0, stats.new_record_count)
self.assertEqual(0, stats.not_there_anymore_count)
def test_record_gone_do_not_delete_with_existing(self):
existing = CitizenFactory(pk=1)
TempCitizenFactory(pk=1, **model_to_dict(existing))
cit = CitizenFactory(pk=2)
existing2 = CitizenFactory(pk=3)
TempCitizenFactory(pk=3, **model_to_dict(existing2))
stats = mirror_database(from_model=TempCitizen, to_model=Citizen)
self.assertEqual(2, stats.unchanged_count)
self.assertEqual(0, stats.modified_record_count)
self.assertEqual(0, stats.new_record_count)
self.assertEqual(1, stats.not_there_anymore_count)
# NOT deleted
self.assertTrue(Citizen.objects.filter(pk=cit.pk).exists())
def test_changed_record(self):
cit = CitizenFactory()
temp_cit = TempCitizenFactory(pk=cit.pk, first_name=cit.first_name + '_CHANGED')
stats = mirror_database(
from_model=TempCitizen,
to_model=Citizen
)
self.assertEqual(0, stats.unchanged_count)
self.assertEqual(1, stats.modified_record_count)
self.assertEqual(0, stats.new_record_count)
self.assertEqual(0, stats.not_there_anymore_count)
c = Citizen.objects.get(pk=cit.pk)
self.assertEqual(temp_cit.first_name, c.first_name)
def test_new_record_with_existing(self):
nid = 100000000000
existing = CitizenFactory(pk=1, national_id=nid)
TempCitizenFactory(pk=1, **model_to_dict(existing))
nid += 1
cit = TempCitizenFactory(pk=2, national_id=nid)
nid += 1
existing2 = CitizenFactory(pk=3, national_id=nid)
TempCitizenFactory(pk=3, **model_to_dict(existing2))
stats = mirror_database(from_model=TempCitizen, to_model=Citizen)
self.assertEqual(2, stats.unchanged_count)
self.assertEqual(0, stats.modified_record_count)
self.assertEqual(1, stats.new_record_count)
self.assertEqual(0, stats.not_there_anymore_count)
self.assertTrue(Citizen.objects.filter(pk=cit.pk).exists())
def test_record_gone_do_not_delete_with_existing(self):
existing = CitizenFactory(pk=1)
TempCitizenFactory(pk=1, **model_to_dict(existing))
cit = CitizenFactory(pk=2)
existing2 = CitizenFactory(pk=3)
TempCitizenFactory(pk=3, **model_to_dict(existing2))
stats = mirror_database(
from_model=TempCitizen,
to_model=Citizen
)
self.assertEqual(2, stats.unchanged_count)
self.assertEqual(0, stats.modified_record_count)
self.assertEqual(0, stats.new_record_count)
self.assertEqual(1, stats.not_there_anymore_count)
# NOT deleted
self.assertTrue(Citizen.objects.filter(pk=cit.pk).exists())
def test_new_record_with_existing(self):
nid = 100000000000
existing = CitizenFactory(pk=1, national_id=nid)
TempCitizenFactory(pk=1, **model_to_dict(existing))
nid += 1
cit = TempCitizenFactory(pk=2, national_id=nid)
nid += 1
existing2 = CitizenFactory(pk=3, national_id=nid)
TempCitizenFactory(pk=3, **model_to_dict(existing2))
stats = mirror_database(
from_model=TempCitizen,
to_model=Citizen
)
self.assertEqual(2, stats.unchanged_count)
self.assertEqual(0, stats.modified_record_count)
self.assertEqual(1, stats.new_record_count)
self.assertEqual(0, stats.not_there_anymore_count)
self.assertTrue(Citizen.objects.filter(pk=cit.pk).exists())
def test_changed_record_with_existing(self):
nid = 100000000000
existing = CitizenFactory(pk=1, national_id=nid)
nid += 1
TempCitizenFactory(pk=existing.pk, **model_to_dict(existing))
nid += 1
cit = CitizenFactory(pk=2, national_id=nid)
temp_cit = TempCitizenFactory(pk=cit.pk,
national_id=nid,
first_name=cit.first_name + '_CHANGED')
existing2 = CitizenFactory(pk=3)
TempCitizenFactory(pk=existing2.pk, **model_to_dict(existing2))
stats = mirror_database(from_model=TempCitizen, to_model=Citizen)
self.assertEqual(2, stats.unchanged_count)
self.assertEqual(1, stats.modified_record_count)
self.assertEqual(0, stats.new_record_count)
self.assertEqual(0, stats.not_there_anymore_count)
c = Citizen.objects.get(pk=cit.pk)
self.assertEqual(temp_cit.first_name, c.first_name)
def test_changed_record_with_existing(self):
nid = 100000000000
existing = CitizenFactory(pk=1, national_id=nid)
nid += 1
TempCitizenFactory(pk=existing.pk, **model_to_dict(existing))
nid += 1
cit = CitizenFactory(pk=2, national_id=nid)
temp_cit = TempCitizenFactory(pk=cit.pk, national_id=nid,
first_name=cit.first_name + '_CHANGED')
existing2 = CitizenFactory(pk=3)
TempCitizenFactory(pk=existing2.pk, **model_to_dict(existing2))
stats = mirror_database(
from_model=TempCitizen,
to_model=Citizen
)
self.assertEqual(2, stats.unchanged_count)
self.assertEqual(1, stats.modified_record_count)
self.assertEqual(0, stats.new_record_count)
self.assertEqual(0, stats.not_there_anymore_count)
c = Citizen.objects.get(pk=cit.pk)
self.assertEqual(temp_cit.first_name, c.first_name)
def test_empty_dbs(self):
stats = mirror_database(from_model=TempCitizen, to_model=Citizen)
self.assertEqual(0, stats.unchanged_count)
self.assertEqual(0, stats.modified_record_count)
self.assertEqual(0, stats.new_record_count)
self.assertEqual(0, stats.not_there_anymore_count)
def import_citizen_dump(input_filename,
max_change_percent=DEFAULT_MAX_CHANGE_PERCENT,
encoding='UTF-8'):
with transaction.atomic():
# Clear out TempCitizen table. (We clear it at the end too, but this makes
# extra sure that we start with it empty.)
delete_all('default', [TempCitizen])
num_records_at_start = Citizen.objects.count()
#
# 1. Fill our temp table with the data from the latest dump
#
logger.info("Loading data from dump")
input_file = codecs.open(input_filename, encoding=encoding)
logger.info("Reading %s" % input_filename)
batch = BatchOperations(TempCitizen)
records_read = 0
for record in get_records(input_file):
records_read += 1
batch.add(record)
batch.flush()
#
# 2. Sync data from temp table to our real table
#
logger.info("Updating our own database")
stats = mirror_database(from_model=TempCitizen,
to_model=Citizen)
# See what % of the records we're changing
if num_records_at_start > 0:
num_changes = (stats.modified_record_count + stats.new_record_count
+ stats.not_there_anymore_count)
percent_changed = 100 * (num_changes / num_records_at_start)
if percent_changed > max_change_percent:
raise TooManyChanges(
"Too many changes, aborting Citizen data import. Max change is %f%% but "
"the import would have changed %f%% records (%d/%d). Use "
"--max-change-percent=NN to override this limit if necessary."
% (max_change_percent, percent_changed, num_changes, num_records_at_start))
# Add our data
stats.records_read = records_read
# Make a note of when we did it
timestamp = now()
CitizenMetadata.objects.update_or_create(defaults=dict(dump_time=timestamp))
# Flag any records that turned up missing
if stats.missing_pks:
Citizen.objects.filter(pk__in=stats.missing_pks, missing=None).update(missing=timestamp)
# And we're done!
# Clear out our temp table (no point in taking up disk space)
delete_all('default', [TempCitizen])
return stats