A reactive notebook for Python — run reproducible experiments, query with SQL, execute as a script, deploy as an app, and version with git. Stored as pure Python. All in a modern, AI-native editor.
Python artificial-intelligencedagdata-sciencedata-visualizationdataflowdeveloper-toolsmachine-learningnotebookspipelinepythonreactivesqlweb-app
import asyncio
import threading
from pathlib import Path
from marimo._export.exporter import AutoExporter
async def test_newer_auto_export_supersedes_an_older_write(
tmp_path: Path, monkeypatch
) -> None:
monkeypatch.setattr(
"marimo._export.exporter.notebook_output_dir",
lambda _path: tmp_path,
)
exporter = AutoExporter()
try:
older_write_started = threading.Event()
release_older_write = threading.Event()
def write_file(filepath: Path, content: str) -> None:
if content == "older":
older_write_started.set()
assert release_older_write.wait(timeout=2)
filepath.write_text(content)
monkeypatch.setattr(exporter, "_write_file_sync", write_file)
older = exporter.reserve_revision("notebook.py", "html")
older_task = asyncio.create_task(
exporter.save_html("notebook.py", "older", revision=older)
)
assert await asyncio.to_thread(older_write_started.wait, 2)
newer = exporter.reserve_revision("notebook.py", "html")
newer_task = asyncio.create_task(
exporter.save_html("notebook.py", "newer", revision=newer)
)
release_older_write.set()
assert not await older_task
assert await newer_task
assert (tmp_path / "notebook.html").read_text() == "newer"
finally:
exporter.cleanup()