Spaces:
Sleeping
Sleeping
Download src/tests/runtime/test_failure_localization.py from dellweb3/VibeCoding: direct link, hf CLI and curl.
- Browser
- Download file 4.54 kB
-
https://huggingface.co/spaces/dellweb3/VibeCoding/resolve/main/src/tests/runtime/test_failure_localization.py
- Command line
-
hf download hf://spaces/dellweb3/VibeCoding/src/tests/runtime/test_failure_localization.py
-
curl -L -o test_failure_localization.py https://huggingface.co/spaces/dellweb3/VibeCoding/resolve/main/src/tests/runtime/test_failure_localization.py
4.54 kB
| """Explicit synthetic diagnosis contracts; live source/VM tests are separate.""" | |
| import json | |
| import unittest | |
| from src.tests.runtime.fixtures import setup,workspace,USER,AUTHOR,SOURCE | |
| from src.modules.skill.atomic_registry.catalog import routing_operations | |
| from src.modules.skill.domain.models import Draft,Budget | |
| from src.modules.llm_gateway.service import ModelReply,ProviderFailure | |
| from src.modules.governance.local_validation import LocalValidation | |
| class FailureLocalizationTests(unittest.TestCase): | |
| def execute(self, mode='valid'): | |
| runtime,request,model,sandbox=setup(self.enterContext(workspace())) | |
| for op in routing_operations():runtime.skills.add_operation(AUTHOR,op) | |
| doc=runtime.skills.read_draft(AUTHOR,'erp-java-development') | |
| body={k:v for k,v in doc['body'].items() if k not in {'owner_id','last_editor_id'}} | |
| body['project_id']='fixture-project' | |
| body['budget']=Budget(max_model_calls=6,max_tokens=120000,max_repairs=2).model_dump() | |
| for node in body['graph']['nodes']: | |
| if node.get('operation_ref')=='AO05@0.1.0':node['operation_ref']='AO05@0.2.0' | |
| runtime.skills.save_draft(AUTHOR,Draft.model_validate(body),doc['revision']) | |
| release=runtime.skills.stage_local_validation(AUTHOR,'erp-java-development','unit-localization',validation_owner=USER.user_id) | |
| request['release_ref']=release['key'] | |
| model.task_roles=['diagnose'];captured=[];original=model.complete | |
| sandbox.fail='build' | |
| def complete(messages,*args): | |
| data=json.loads(messages[-1]['content']);captured.append(data) | |
| if data['task']=='diagnose': | |
| model.count+=1 | |
| if mode=='unknown':raise ProviderFailure('MODEL_RESULT_UNKNOWN') | |
| if mode=='invalid':return ModelReply({'files':SOURCE},80,80,model.name) | |
| return ModelReply({'summary':'Concrete synthetic compiler cause and minimal correction.'},80,80,model.name) | |
| if data['task']=='repair' and data['data'].get('failure_analysis'):sandbox.fail=None | |
| return original(messages,*args) | |
| model.complete=complete | |
| run=runtime.create(USER,request,'failure-localization-unit') | |
| validator=LocalValidation(runtime,USER,'unit-localization') | |
| for _ in range(30): | |
| run=runtime.get(USER,run['id']) | |
| if run['status'] in {'unknown','failed'} or (run.get('pending_approval') or {}).get('scope')=='acceptance':break | |
| if mode=='one-call-left' and run['repairs']==2: | |
| runtime.store.mutate(run['id'],lambda r:r['budget'].update(max_model_calls=4)) | |
| validator.step(run['id']) | |
| return runtime,runtime.get(USER,run['id']),captured | |
| def test_repeat_failure_is_localized_then_repaired_and_independently_retested(self): | |
| runtime,run,messages=self.execute() | |
| self.assertEqual(run['status'],'waiting_approval') | |
| self.assertEqual(run['pending_approval']['scope'],'acceptance') | |
| self.assertEqual([m['task'] for m in messages],['design','code','repair','diagnose','repair']) | |
| self.assertEqual(run['failure_analysis']['trust'],'model_hypothesis_requires_independent_tests') | |
| self.assertEqual(runtime.validation_evidence(USER,run['id'])['manifest']['failure_analysis'],run['failure_analysis']) | |
| self.assertEqual(len(runtime.store.calls(run['id'])),5) | |
| def test_unknown_diagnosis_stops_without_applying_or_replaying(self): | |
| runtime,run,messages=self.execute('unknown') | |
| self.assertEqual(run['status'],'unknown') | |
| self.assertEqual(messages[-1]['task'],'diagnose') | |
| self.assertEqual(runtime.store.calls(run['id'])[-1]['state'],'unknown') | |
| def test_invalid_optional_diagnosis_is_settled_once_and_does_not_consume_a_retry(self): | |
| runtime,run,messages=self.execute('invalid') | |
| self.assertEqual(run['status'],'failed') | |
| self.assertEqual(sum(m['task']=='diagnose' for m in messages),1) | |
| self.assertEqual(len(runtime.store.calls(run['id'])),5) | |
| self.assertTrue(all(c['state']=='settled' for c in runtime.store.calls(run['id']))) | |
| def test_diagnosis_does_not_steal_the_last_repair_call_or_expand_budget(self): | |
| runtime,run,messages=self.execute('one-call-left') | |
| self.assertEqual(run['status'],'failed') | |
| self.assertFalse(any(m['task']=='diagnose' for m in messages)) | |
| self.assertEqual(len(runtime.store.calls(run['id'])),4) | |