gpt-5.3 codex review: Phase 4

tests/services/test_v45_candidates.py

@@ -121,3 +121,105 @@ async def test_retranscription_job_locks_source_and_frozen_model(default_session
     assert loaded.model == "vendor/alternate"
     assert loaded.user_prompt == "Transcribe verbatim."
     assert [link.source_id for link in loaded.job_sources] == [source.id]
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_promoting_candidate_does_not_mutate_execution_attempt_history(default_session_factory):
+    settings = Settings(openrouter_api_key="test-key", provider_models=None)
+    services = _services(default_session_factory, settings)
+    source = await _seed_source(services)
+
+    first_job = await services.jobs.create_job(Job(document_id=source.document_id))
+    second_job = await services.jobs.create_job(Job(document_id=source.document_id))
+    await services.sources.create_job_source(JobSource(job_id=first_job.id, source_id=source.id))
+    await services.sources.create_job_source(JobSource(job_id=second_job.id, source_id=source.id))
+
+    await services.sources.update_job_source_transcription(
+        job_id=first_job.id,
+        source_id=source.id,
+        text="baseline",
+        provider="fixture",
+        model="model-a",
+    )
+    await services.sources.update_job_source_transcription(
+        job_id=second_job.id,
+        source_id=source.id,
+        text="candidate",
+        provider="fixture",
+        model="model-b",
+    )
+
+    before = [
+        (
+            attempt.id,
+            attempt.job_id,
+            attempt.source_id,
+            attempt.attempt_number,
+            attempt.status.value,
+            attempt.raw_transcription,
+            attempt.error_category,
+            attempt.error_detail,
+            attempt.failure_phase,
+        )
+        for attempt in await services.evidence.list_execution_attempts(source_id=source.id)
+    ]
+
+    candidate = next(
+        attempt
+        for attempt in await services.evidence.list_execution_attempts(source_id=source.id)
+        if attempt.raw_transcription == "candidate"
+    )
+    await services.evidence.promote_machine_attempt(source_id=source.id, execution_attempt_id=candidate.id)
+
+    after = [
+        (
+            attempt.id,
+            attempt.job_id,
+            attempt.source_id,
+            attempt.attempt_number,
+            attempt.status.value,
+            attempt.raw_transcription,
+            attempt.error_category,
+            attempt.error_detail,
+            attempt.failure_phase,
+        )
+        for attempt in await services.evidence.list_execution_attempts(source_id=source.id)
+    ]
+
+    assert after == before
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_retry_appends_new_attempt_instead_of_rewriting_history(default_session_factory):
+    settings = Settings(openrouter_api_key="test-key", provider_models=None)
+    services = _services(default_session_factory, settings)
+    source = await _seed_source(services)
+
+    job = await services.jobs.create_job(Job(document_id=source.document_id))
+    await services.sources.create_job_source(JobSource(job_id=job.id, source_id=source.id))
+
+    await services.sources.update_job_source_transcription(
+        job_id=job.id,
+        source_id=source.id,
+        text=None,
+        error_detail="timeout",
+        error_category="timeout",
+        provider="fixture",
+        model="model-a",
+        failure_phase="provider_call",
+    )
+    await services.sources.update_job_source_transcription(
+        job_id=job.id,
+        source_id=source.id,
+        text="retry-success",
+        provider="fixture",
+        model="model-a",
+    )
+
+    attempts = list(await services.evidence.list_execution_attempts(source_id=source.id))
+    assert len(attempts) == 2
+    assert [attempt.attempt_number for attempt in attempts] == [1, 2]
+    assert attempts[0].raw_transcription is None
+    assert attempts[1].raw_transcription == "retry-success"

tests/services/test_workflows_reliability.py

@@ -181,6 +181,74 @@ class TestWorkflowReliability:
         assert duration_ms >= int(budget_seconds * 1000 * 0.9)
         assert duration_ms < int((budget_seconds + setup_seconds) * 1000 * 0.9)
 
+    @pytest.mark.asyncio
+    async def test_attempt_metadata_persists_provider_and_processing_durations(
+        self,
+        default_session_factory,
+        monkeypatch,
+    ):
+        """Execution metadata records both provider-only and end-to-end durations."""
+        services = ServiceBundle.from_session_factory(default_session_factory)
+        async with services.jobs._session_scope() as session:
+            document = Document(id=uuid4(), name="timing-metadata-doc")
+            session.add(document)
+            await session.flush()
+            job = Job(document_id=document.id, status=JobStatus.QUEUED)
+            session.add(job)
+            await session.flush()
+            source = Source(
+                document_id=document.id,
+                page_number=1,
+                upload_name="timing.jpg",
+                filename="timing.jpg",
+                file_path=str(Path("tests/fixtures/images/real/Book Two - page 02.jpg")),
+                file_hash="f" * 64,
+                file_size_bytes=1,
+            )
+            session.add(source)
+            await session.flush()
+            session.add(JobSource(job_id=job.id, source_id=source.id, status=JobSourceStatus.PENDING))
+            await session.commit()
+            loaded = await services.jobs.read_job(job_id=job.id, session=session)
+
+        async def _returns_text(*args, **kwargs):
+            _ = (args, kwargs)
+            await asyncio.sleep(0.03)
+            return TranscriptionResult(text="timed output", provider="fixture", model="fixture-model")
+
+        monkeypatch.setattr("transcription.services.workflows.transcribe_document_image", _returns_text)
+
+        result = await process_queued_job(
+            job=loaded,
+            services=services,
+            settings=Settings(openrouter_api_key="test-key", worker_provider_timeout_seconds=2.0),
+        )
+        assert result is not None
+        assert result.status == JobStatus.TRANSCRIBED
+
+        async with services.jobs._session_scope() as session:
+            attempts = (
+                (
+                    await session.exec(
+                        select(ExecutionAttempt).where(
+                            col(ExecutionAttempt.job_source_id).in_([js.id for js in result.job_sources])
+                        )
+                    )
+                )
+                .all()
+            )
+        assert len(attempts) == 1
+        attempt = attempts[0]
+        timing = (attempt.normalized_metadata or {}).get("processing_timing")
+        assert isinstance(timing, dict)
+        provider_call_ms = timing.get("provider_call_duration_ms")
+        processing_ms = timing.get("processing_duration_ms")
+        assert isinstance(provider_call_ms, int)
+        assert isinstance(processing_ms, int)
+        assert provider_call_ms >= 0
+        assert processing_ms >= provider_call_ms
+        assert attempt.duration_ms == provider_call_ms
+
     @pytest.mark.asyncio
     async def test_error_after_claim_fails_the_job_instead_of_stranding_it(
         self,
@@ -239,7 +307,7 @@ class TestWorkflowReliability:
             assert final.status == JobStatus.FAILED
 
     @pytest.mark.asyncio
-    async def test_completed_page_is_committed_before_next_provider_call_finishes(
+    async def test_transcribed_page_is_committed_before_next_provider_call_finishes(
         self,
         default_session_factory,
         monkeypatch,