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
from __future__ import annotations
import json
import pytest
from marimo._dependencies.dependencies import DependencyManager
from marimo._runtime.runtime import Kernel
from tests.conftest import ExecReqProvider
HAS_ALTAIR = DependencyManager.altair.has()
HAS_VL_CONVERT = DependencyManager.has("vl-convert")
def assert_vegalite_mimetype(mime: str) -> None:
"""Assert that the mime type is a valid vega-lite mime type (v5 or v6)."""
assert mime in (
"application/vnd.vegalite.v5+json",
"application/vnd.vegalite.v6+json",
), f"Expected vega-lite mime type (v5 or v6), got {mime}"
@pytest.mark.skipif(
not HAS_ALTAIR, reason="optional dependencies not installed"
)
async def test_altair(
executing_kernel: Kernel, exec_req: ExecReqProvider
) -> None:
from marimo._output.formatters.formatters import register_formatters
register_formatters()
await executing_kernel.run(
[
exec_req.get(
"""
import altair as alt
import pandas as pd
from marimo._output.formatting import get_formatter
data = pd.DataFrame({"a": ["A", "B"], "b": [28, 55]})
chart = alt.Chart(data).mark_bar().encode(x="a", y="b")
formatter = get_formatter(chart)
result = formatter(chart)
"""
)
]
)
result = executing_kernel.globals["result"]
assert result is not None
assert_vegalite_mimetype(result[0])
# Valid JSON
assert "$schema" in result[1]
assert isinstance(result[1], str)
assert json.loads(result[1])
@pytest.mark.skipif(
not HAS_ALTAIR, reason="optional dependencies not installed"
)
async def test_altair_with_embed_options(
executing_kernel: Kernel, exec_req: ExecReqProvider
) -> None:
from marimo._output.formatters.formatters import register_formatters
register_formatters()
await executing_kernel.run(
[
exec_req.get(
"""
import altair as alt
import pandas as pd
from marimo._output.formatting import get_formatter
data = pd.DataFrame({"a": ["A", "B"], "b": [28, 55]})
chart = alt.Chart(data).mark_bar().encode(x="a", y="b")
formatter = get_formatter(chart)
before_options_result = formatter(chart)
alt.renderers.set_embed_options(actions=False)
after_options_result = formatter(chart)
"""
)
]
)
result = executing_kernel.globals["before_options_result"]
assert result is not None
assert_vegalite_mimetype(result[0])
json_result = json.loads(result[1])
assert json_result["usermeta"] == {
"embedOptions": {},
}
result = executing_kernel.globals["after_options_result"]
assert result is not None
assert_vegalite_mimetype(result[0])
json_result = json.loads(result[1])
assert json_result["usermeta"] == {
"embedOptions": {
"actions": False,
},
}
@pytest.mark.skipif(
not HAS_ALTAIR or not HAS_VL_CONVERT,
reason="optional dependencies not installed and fails on mac",
)
async def test_altair_with_svg(
executing_kernel: Kernel, exec_req: ExecReqProvider
) -> None:
from marimo._output.formatters.formatters import register_formatters
register_formatters()
await executing_kernel.run(
[
exec_req.get(
"""
import altair as alt
import pandas as pd
from marimo._output.formatting import get_formatter
data = pd.DataFrame({"a": ["A", "B"], "b": [28, 55]})
alt.renderers.enable("svg")
chart = alt.Chart(data).mark_bar().encode(x="a", y="b")
formatter = get_formatter(chart)
result = formatter(chart)
"""
)
]
)
result = executing_kernel.globals["result"]
assert result is not None
assert result[0] == "image/svg+xml"
assert result[1].startswith("<svg")
# TODO: Add tests for vegafusion