model used being carried thru

This commit is contained in:
John Lancaster
2026-06-29 19:04:16 -05:00
parent 9ada09accf
commit 7df687d6f5
27 changed files with 1442 additions and 210 deletions
+53 -19
View File
@@ -3,12 +3,15 @@
from pathlib import Path
import pytest
from sqlmodel import desc
from sqlmodel import select
from transcription.config import Settings
from transcription.models import Job, JobStatus, Transcript
from transcription.models import Job
from transcription.models import JobStatus
from transcription.models import Transcript
from transcription.providers.base import TranscriptionResult
from transcription.services.upload import create_upload_job
from transcription.services.store import create_upload_job
from transcription.worker import process_next_queued_job
@@ -16,24 +19,42 @@ from transcription.worker import process_next_queued_job
class TestPipelineSuccessFlow:
"""Verify end-to-end success lifecycle behavior."""
def test_upload_then_worker_persists_transcribed_terminal_state(self, session, tmp_path: Path, monkeypatch):
@pytest.mark.asyncio
async def test_upload_then_worker_persists_transcribed_terminal_state(
self,
async_session,
tmp_path: Path,
monkeypatch,
):
"""Upload followed by worker processing persists transcript and transcribed status."""
settings = Settings(openrouter_api_key="test-key", upload_dir=tmp_path)
upload_result = create_upload_job(
upload_result = await create_upload_job(
filename="pipeline.jpg",
file_bytes=b"pipeline-bytes",
session=session,
session=async_session,
settings=settings,
)
def _fake_transcribe(_path: str) -> TranscriptionResult:
return TranscriptionResult(text="Pipeline transcript", provider="openrouter", model="test-model")
async def _fake_transcribe(_path: str) -> TranscriptionResult:
return TranscriptionResult(
text="Pipeline transcript",
provider="openrouter",
prompt_name="transcribe_document.md",
model="test-model",
)
monkeypatch.setattr("transcription.worker.transcribe_document_image", _fake_transcribe)
monkeypatch.setattr("transcription.services.workflows.transcribe_document_image", _fake_transcribe)
processed = process_next_queued_job(session=session)
job = session.get(Job, upload_result.job_id)
transcript = session.exec(select(Transcript).where(Transcript.job_id == upload_result.job_id)).first()
processed = await process_next_queued_job(session=async_session)
job = await async_session.get(Job, upload_result.job_id)
transcript = (
await async_session.exec(
select(Transcript)
.where(Transcript.job_id == upload_result.job_id)
.order_by(desc(Transcript.revision))
.limit(1)
)
).first()
assert processed is True
assert job is not None
@@ -47,24 +68,37 @@ class TestPipelineSuccessFlow:
class TestPipelineFailureFlow:
"""Verify end-to-end failure lifecycle behavior."""
def test_upload_then_worker_persists_failed_terminal_state(self, session, tmp_path: Path, monkeypatch):
@pytest.mark.asyncio
async def test_upload_then_worker_persists_failed_terminal_state(
self,
async_session,
tmp_path: Path,
monkeypatch,
):
"""Upload followed by worker processing persists error detail and failed status."""
settings = Settings(openrouter_api_key="test-key", upload_dir=tmp_path)
upload_result = create_upload_job(
upload_result = await create_upload_job(
filename="pipeline.jpg",
file_bytes=b"pipeline-bytes",
session=session,
session=async_session,
settings=settings,
)
def _fake_transcribe(_path: str) -> TranscriptionResult:
async def _fake_transcribe(_path: str) -> TranscriptionResult:
raise RuntimeError("pipeline provider failure")
monkeypatch.setattr("transcription.worker.transcribe_document_image", _fake_transcribe)
monkeypatch.setattr("transcription.services.workflows.transcribe_document_image", _fake_transcribe)
processed = process_next_queued_job(session=session)
job = session.get(Job, upload_result.job_id)
transcript = session.exec(select(Transcript).where(Transcript.job_id == upload_result.job_id)).first()
processed = await process_next_queued_job(session=async_session)
job = await async_session.get(Job, upload_result.job_id)
transcript = (
await async_session.exec(
select(Transcript)
.where(Transcript.job_id == upload_result.job_id)
.order_by(desc(Transcript.revision))
.limit(1)
)
).first()
assert processed is True
assert job is not None