class Command(BaseCommand):
def add_arguments(self, parser):
parser.add_argument(
'domain',
help="File path for log file",
)
parser.add_argument(
'log_file',
help="File path for log file",
)
def handle(self, domain, log_file, **options):
total_cases = CaseES().domain(domain).case_type('household').is_closed().count()
self.case_accessor = CaseAccessors(domain)
failed_updates = []
with open(log_file, "w", encoding='utf-8') as fh:
fh.write('--------Successful Form Ids----------\n')
for cases in chunked(with_progress_bar(self._get_cases_to_process(domain), total_cases), 100):
related_cases = self._get_related_cases(cases)
case_tupes = [(case_id, {}, True) for case_id in related_cases]
try:
xform, cases = bulk_update_cases(
domain, case_tupes, self.__module__)
fh.write(xform.form_id + '\n')
except LocalSubmissionError as e:
print('submission error')
print(six.text_type(e))
failed_updates.extend(related_cases)
except Exception as e:
print('unexpected error')
print(six.text_type(e))
failed_updates.extend(related_cases)
fh.write('--------Failed Cases--------------\n')
for case_id in failed_updates:
fh.write(case_id)
print('-------------COMPLETE--------------')
def _get_cases_to_process(self, domain):
from corehq.sql_db.util import get_db_aliases_for_partitioned_query
dbs = get_db_aliases_for_partitioned_query()
for db in dbs:
cases = CommCareCaseSQL.objects.using(db).filter(domain=domain, type='household', closed=True)
for case in cases:
yield case.case_id
def _get_related_cases(self, cases):
related_cases = {case.case_id for case in self.case_accessor.get_all_reverse_indices_info(list(cases))
if case.relationship == CommCareCaseIndexSQL.CHILD}
related_cases |= {case.case_id for case in
self.case_accessor.get_all_reverse_indices_info(list(related_cases))
if case.relationship == CommCareCaseIndexSQL.CHILD}
return related_cases
def get_cleanliness_flag_from_scratch(domain, owner_id):
casedb = CaseAccessors(domain)
footprint_info = get_case_footprint_info(domain, owner_id)
owned_cases = footprint_info.base_ids
cases_to_check = footprint_info.all_ids - owned_cases
if cases_to_check:
closed_owned_case_ids = set(casedb.get_closed_case_ids_for_owner(owner_id))
cases_to_check = cases_to_check - closed_owned_case_ids - footprint_info.extension_ids
# check extension cases that are unowned or owned by others
extension_cases_to_check = footprint_info.extension_ids - closed_owned_case_ids - owned_cases
while extension_cases_to_check:
extension_case = extension_cases_to_check.pop()
dependent_cases = set(get_dependent_case_info(domain, [extension_case]).all_ids)
unowned_dependent_cases = dependent_cases - owned_cases
extension_cases_to_check = extension_cases_to_check - dependent_cases
dependent_cases_owned_by_other_owners = {
dependent_case.case_id
for dependent_case in casedb.get_cases(list(unowned_dependent_cases))
if dependent_case.owner_id != UNOWNED_EXTENSION_OWNER_ID
}
if dependent_cases_owned_by_other_owners:
hint_id = dependent_cases & owned_cases
# can't get back from extension case to owned case e.g. host is a child of owned case
if hint_id:
return CleanlinessFlag(False, hint_id.pop())
if cases_to_check:
# it wasn't in any of the open or closed IDs - it must be dirty
reverse_index_infos = casedb.get_all_reverse_indices_info(list(cases_to_check))
reverse_index_ids = set([r.case_id for r in reverse_index_infos])
indexed_with_right_owner = (reverse_index_ids & (owned_cases | closed_owned_case_ids))
found_deleted_cases = False
while indexed_with_right_owner:
hint_id = indexed_with_right_owner.pop()
infos_for_this_owner = _get_info_by_case_id(reverse_index_infos, hint_id)
for info in infos_for_this_owner:
try:
case = CaseAccessors(domain).get_case(info.referenced_id)
if not case.is_deleted:
return CleanlinessFlag(False, hint_id)
else:
found_deleted_cases = True
except ResourceNotFound:
# the case doesn't exist - don't use it as a dirty flag
found_deleted_cases = True
if found_deleted_cases:
# if we made it all the way to the end of the loop without returning anything
# then the owner was only flagged as dirty due to missing cases,
# This implies the owner is still clean.
return CleanlinessFlag(True, None)
else:
# I don't believe code can ever be hit, but if it is we should fail hard
# until we can better understand it.
raise IllegalCaseId('Owner {} in domain {} has an invalid index reference chain!!'.format(
owner_id, domain
))
return CleanlinessFlag(True, None)
def get_cleanliness_flag_from_scratch(domain, owner_id):
casedb = CaseAccessors(domain)
footprint_info = get_case_footprint_info(domain, owner_id)
owned_cases = footprint_info.base_ids
cases_to_check = footprint_info.all_ids - owned_cases
if cases_to_check:
closed_owned_case_ids = set(casedb.get_closed_case_ids_for_owner(owner_id))
cases_to_check = cases_to_check - closed_owned_case_ids - footprint_info.extension_ids
# check extension cases that are unowned or owned by others
extension_cases_to_check = footprint_info.extension_ids - closed_owned_case_ids - owned_cases
while extension_cases_to_check:
extension_case = extension_cases_to_check.pop()
dependent_cases = set(get_dependent_case_info(domain, [extension_case]).all_ids)
unowned_dependent_cases = dependent_cases - owned_cases
extension_cases_to_check = extension_cases_to_check - dependent_cases
dependent_cases_owned_by_other_owners = {
dependent_case.case_id
for dependent_case in casedb.get_cases(list(unowned_dependent_cases))
if dependent_case.owner_id != UNOWNED_EXTENSION_OWNER_ID
}
if dependent_cases_owned_by_other_owners:
hint_id = dependent_cases & owned_cases
# can't get back from extension case to owned case e.g. host is a child of owned case
if hint_id:
return CleanlinessFlag(False, hint_id.pop())
if cases_to_check:
# it wasn't in any of the open or closed IDs - it must be dirty
reverse_index_infos = casedb.get_all_reverse_indices_info(list(cases_to_check))
reverse_index_ids = set([r.case_id for r in reverse_index_infos])
indexed_with_right_owner = (reverse_index_ids & (owned_cases | closed_owned_case_ids))
found_deleted_cases = False
while indexed_with_right_owner:
hint_id = indexed_with_right_owner.pop()
infos_for_this_owner = _get_info_by_case_id(reverse_index_infos, hint_id)
for info in infos_for_this_owner:
try:
case = CaseAccessors(domain).get_case(info.referenced_id)
if not case.is_deleted:
return CleanlinessFlag(False, hint_id)
else:
found_deleted_cases = True
except ResourceNotFound:
# the case doesn't exist - don't use it as a dirty flag
found_deleted_cases = True
if found_deleted_cases:
# if we made it all the way to the end of the loop without returning anything
# then the owner was only flagged as dirty due to missing cases,
# This implies the owner is still clean.
return CleanlinessFlag(True, None)
else:
# I don't believe code can ever be hit, but if it is we should fail hard
# until we can better understand it.
raise IllegalCaseId('Owner {} in domain {} has an invalid index reference chain!!'.format(
owner_id, domain
))
class Command(BaseCommand):
def add_arguments(self, parser):
parser.add_argument(
'domain',
help="File path for log file",
)
parser.add_argument(
'log_file',
help="File path for log file",
)
def handle(self, domain, log_file, **options):
total_cases = CaseES().domain(domain).case_type(
'household').is_closed().count()
self.case_accessor = CaseAccessors(domain)
failed_updates = []
with open(log_file, "w", encoding='utf-8') as fh:
fh.write('--------Successful Form Ids----------\n')
for cases in chunked(
with_progress_bar(self._get_cases_to_process(domain),
total_cases), 100):
related_cases = self._get_related_cases(cases)
case_tupes = [(case_id, {}, True) for case_id in related_cases]
try:
xform, cases = bulk_update_cases(domain, case_tupes,
self.__module__)
fh.write(xform.form_id + '\n')
except LocalSubmissionError as e:
print('submission error')
print(six.text_type(e))
failed_updates.extend(related_cases)
except Exception as e:
print('unexpected error')
print(six.text_type(e))
failed_updates.extend(related_cases)
fh.write('--------Failed Cases--------------\n')
for case_id in failed_updates:
fh.write(case_id)
print('-------------COMPLETE--------------')
def _get_cases_to_process(self, domain):
from corehq.sql_db.util import get_db_aliases_for_partitioned_query
dbs = get_db_aliases_for_partitioned_query()
for db in dbs:
cases = CommCareCaseSQL.objects.using(db).filter(domain=domain,
type='household',
closed=True)
for case in cases:
yield case.case_id
def _get_related_cases(self, cases):
related_cases = {
case.case_id
for case in self.case_accessor.get_all_reverse_indices_info(
list(cases)) if case.relationship == CommCareCaseIndexSQL.CHILD
}
