def setUpClass(cls, *args, **kwargs):
super().setUpClass(*args, **kwargs)
cls.domain = 'case-dedupe-test'
cls.case_type = 'adult'
cls.factory = CaseFactory(cls.domain)
cls.rule = AutomaticUpdateRule.objects.create(
domain=cls.domain,
name='test',
case_type=cls.case_type,
active=True,
deleted=False,
filter_on_server_modified=False,
server_modified_boundary=None,
workflow=AutomaticUpdateRule.WORKFLOW_DEDUPLICATE,
)
_, cls.action = cls.rule.add_action(
CaseDeduplicationActionDefinition,
match_type=CaseDeduplicationMatchTypeChoices.ALL,
case_properties=["name", "age"],
)
cls.action.set_properties_to_update([
CaseDeduplicationActionDefinition.PropertyDefinition(
name='is_potential_duplicate',
value_type=CaseDeduplicationActionDefinition.VALUE_TYPE_EXACT,
value='yes',
)
])
cls.action.save()
def reset_deduplicate_rule(rule):
"""Deletes all case duplicates for this rule
"""
from corehq.apps.data_interfaces.models import (
CaseDeduplicationActionDefinition,
CaseDuplicate,
)
deduplicate_action = CaseDeduplicationActionDefinition.from_rule(rule)
CaseDuplicate.objects.filter(action=deduplicate_action).delete()
def _get_duplicates_count(self, rule):
if rule.locked_for_editing:
progress_helper = MessagingRuleProgressHelper(rule.id)
return _(
"Processing - {progress_percent}% ({cases_processed}/{total_cases} cases) complete"
).format(
progress_percent=progress_helper.get_progress_pct(),
cases_processed=progress_helper.get_cases_processed(),
total_cases=progress_helper.get_total_cases_to_process(),
)
action = CaseDeduplicationActionDefinition.from_rule(rule)
return CaseDuplicate.objects.filter(action=action).count()
def setUpClass(cls):
super().setUpClass()
cls.domain = 'naboo'
cls.case_type = 'people'
cls.factory = CaseFactory(cls.domain)
cls.pillow = get_xform_pillow(skip_ucr=True)
cls.rule = AutomaticUpdateRule.objects.create(
domain=cls.domain,
name='test',
case_type=cls.case_type,
active=True,
deleted=False,
filter_on_server_modified=False,
server_modified_boundary=None,
workflow=AutomaticUpdateRule.WORKFLOW_DEDUPLICATE,
)
_, cls.action = cls.rule.add_action(
CaseDeduplicationActionDefinition,
match_type=CaseDeduplicationMatchTypeChoices.ALL,
case_properties=["case_name", "age"],
)
cls.action.set_properties_to_update([
CaseDeduplicationActionDefinition.PropertyDefinition(
name='age',
value_type=CaseDeduplicationActionDefinition.VALUE_TYPE_EXACT,
value='5',
),
CaseDeduplicationActionDefinition.PropertyDefinition(
name='case_name',
value_type=CaseDeduplicationActionDefinition.VALUE_TYPE_EXACT,
value='Herman Miller',
)
])
cls.action.save()
AutomaticUpdateRule.clear_caches(
cls.domain, AutomaticUpdateRule.WORKFLOW_DEDUPLICATE)
def _format_rule(self, rule):
ret = super()._format_rule(rule)
rule_properties = (set(
CaseDeduplicationActionDefinition.from_rule(rule).case_properties)
- set(CaseListExplorerColumns.DEFAULT_COLUMNS))
ret['duplicates_count'] = self._get_duplicates_count(rule)
ret['explore_url'] = reverse_with_params(
'project_report_dispatcher',
args=(self.domain, 'duplicate_cases'),
params={
"duplicate_case_rule":
rule.id,
"explorer_columns":
json.dumps(CaseListExplorerColumns.DEFAULT_COLUMNS +
list(rule_properties)),
},
)
return ret
def test_update_parent(self):
es = get_es_new()
with trap_extra_setup(ConnectionError):
initialize_index_and_mapping(es, CASE_SEARCH_INDEX_INFO)
self.addCleanup(ensure_index_deleted, CASE_SEARCH_INDEX_INFO.index)
duplicates, uniques = self._create_cases(num_cases=2)
parent = uniques[0]
child = duplicates[0]
set_parent_case(self.domain, child, parent)
parent_case_property_value = parent.get_case_property('name')
new_parent_case_property_value = f'{parent_case_property_value}-someextratext'
self.action.set_properties_to_update([
CaseDeduplicationActionDefinition.PropertyDefinition(
name='parent/name',
value_type=CaseDeduplicationActionDefinition.VALUE_TYPE_EXACT,
value=new_parent_case_property_value,
)
])
self.action.save()
for case in chain(duplicates, uniques):
send_to_elasticsearch(
'case_search',
transform_case_for_elasticsearch(case.to_json()))
es.indices.refresh(CASE_SEARCH_INDEX_INFO.index)
self.rule = AutomaticUpdateRule.objects.get(id=self.rule.id)
self.rule.run_actions_when_case_matches(child)
updated_parent_case = CommCareCase.objects.get_case(
parent.case_id, self.domain)
self.assertEqual(updated_parent_case.get_case_property('name'),
new_parent_case_property_value)
def backfill_deduplicate_rule(domain, rule):
from corehq.apps.data_interfaces.models import (
AutomaticUpdateRule,
CaseDeduplicationActionDefinition,
DomainCaseRuleRun,
)
progress_helper = MessagingRuleProgressHelper(rule.pk)
total_cases_count = CaseSearchES().domain(domain).case_type(
rule.case_type).count()
progress_helper.set_total_cases_to_be_processed(total_cases_count)
now = datetime.utcnow()
try:
run_record = DomainCaseRuleRun.objects.create(
domain=domain,
started_on=now,
status=DomainCaseRuleRun.STATUS_RUNNING,
case_type=rule.case_type,
)
action = CaseDeduplicationActionDefinition.from_rule(rule)
case_iterator = AutomaticUpdateRule.iter_cases(
domain, rule.case_type, include_closed=action.include_closed)
iter_cases_and_run_rules(
domain,
case_iterator,
[rule],
now,
run_record.id,
rule.case_type,
progress_helper=progress_helper,
)
finally:
progress_helper.set_rule_complete()
AutomaticUpdateRule.objects.filter(pk=rule.pk).update(
locked_for_editing=False,
last_run=now,
)
def dedupe_action(self):
return CaseDeduplicationActionDefinition.from_rule(self.rule)