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
/* Copyright 2026 Marimo. All rights reserved. */
// Fallback spec
import { mint, orange, slate } from "@radix-ui/colors";
import type { TopLevelSpec } from "vega-lite";
import type { Scale } from "vega-lite/types_unstable/scale.js";
// @ts-expect-error vega-typings does not include formats
import { formats } from "vega-loader";
import { asRemoteURL } from "@/core/runtime/config";
import type { TopLevelFacetedUnitSpec } from "@/plugins/impl/data-explorer/queries/types";
import { arrow } from "@/plugins/impl/vega/formats";
import { parseCsvData } from "@/plugins/impl/vega/loader";
import {
base64ToUint8Array,
extractBase64FromDataURL,
isDataURLString,
} from "@/utils/json/base64";
// Height (px) of the fallback/legacy summary charts. The invisible full-height
// hover layer's `y2` must match this so it covers the whole plot; shared with
// `createBase` so the two can't drift apart.
export const DEFAULT_CHART_HEIGHT = 100;
export function getLegacyNumericSpec(
column: string,
format: string,
base: TopLevelFacetedUnitSpec,
): TopLevelFacetedUnitSpec {
return {
...base, // Assuming base contains shared configurations
// Two layers: one with the visible bars, and one with invisible bars
// that provide a larger tooltip area.
// @ts-expect-error 'layer' property not in TopLevelFacetedUnitSpec
layer: [
{
mark: {
type: "bar",
color: mint.mint11,
},
encoding: {
x: {
field: column,
type: "quantitative",
bin: true,
},
y: {
aggregate: "count",
type: "quantitative",
axis: null,
},
},
},
// Full-height invisible hit targets (pixel coords). Avoid
// `y: { aggregate: "max" }` without a field — Vega drops it (#10303).
{
mark: {
type: "bar",
opacity: 0,
},
encoding: {
x: {
field: column,
type: "quantitative",
bin: true,
axis: {
title: null,
labelFontSize: 8.5,
labelOpacity: 0.5,
labelExpr:
"(datum.value >= 10000 || datum.value <= -10000) ? format(datum.value, '.2e') : format(datum.value, '.2~f')",
},
},
y: {
value: 0,
},
y2: {
value: DEFAULT_CHART_HEIGHT,
},
tooltip: [
{
field: column,
type: "quantitative",
bin: true,
title: column,
format: format,
},
{
aggregate: "count",
type: "quantitative",
title: "Count",
format: ",d",
},
],
},
},
],
};
}
export function getLegacyTemporalSpec(
column: string,
type: "date" | "datetime" | "time",
base: TopLevelFacetedUnitSpec,
scale: Scale,
): TopLevelFacetedUnitSpec {
const format =
type === "date"
? "%Y-%m-%d"
: type === "time"
? "%H:%M:%S"
: "%Y-%m-%dT%H:%M:%S";
return {
...base,
// Two layers: one with the visible bars, and one with invisible bars
// that provide a larger tooltip area.
// @ts-expect-error 'layer' property not in TopLevelFacetedUnitSpec
layer: [
{
mark: {
type: "bar",
color: mint.mint11,
},
encoding: {
x: {
field: column,
type: "temporal",
axis: null,
bin: true,
scale: scale,
},
y: { aggregate: "count", type: "quantitative", axis: null },
// Color nulls
color: {
condition: {
test: `datum["bin_maxbins_10_${column}_range"] === "null"`,
value: orange.orange11,
},
value: mint.mint11,
},
},
},
// Full-height invisible hit targets (pixel coords). Avoid
// `y: { aggregate: "max" }` without a field — Vega drops it (#10303).
{
mark: {
type: "bar",
opacity: 0,
},
encoding: {
x: {
field: column,
type: "temporal",
axis: null,
bin: true,
scale: scale,
},
y: { value: 0 },
y2: { value: DEFAULT_CHART_HEIGHT },
tooltip: [
{
// Can also use column, but this is more explicit
field: `bin_maxbins_10_${column}`,
type: "temporal",
format: format,
bin: { binned: true },
title: `${column} (start)`,
},
{
field: `bin_maxbins_10_${column}_end`,
type: "temporal",
format: format,
bin: { binned: true },
title: `${column} (end)`,
},
{
aggregate: "count",
type: "quantitative",
title: "Count",
format: ",d",
},
],
// Color nulls
color: {
condition: {
test: `datum["bin_maxbins_10_${column}_range"] === "null"`,
value: orange.orange11,
},
value: mint.mint11,
},
},
},
],
};
}
export function getLegacyBooleanSpec(
column: string,
base: TopLevelFacetedUnitSpec,
barHeight: number,
): TopLevelFacetedUnitSpec {
return {
...base,
mark: { type: "bar", color: mint.mint11 },
encoding: {
y: {
field: column,
type: "nominal",
axis: {
labelExpr:
"datum.label === 'true' || datum.label === 'True' ? 'True' : 'False'",
tickWidth: 0,
title: null,
labelColor: slate.slate9,
},
},
x: {
aggregate: "count",
type: "quantitative",
axis: null,
scale: { type: "linear" },
},
tooltip: [
{ field: column, type: "nominal", title: "Value" },
{
aggregate: "count",
type: "quantitative",
title: "Count",
format: ",d",
},
],
},
layer: [
{
mark: {
type: "bar",
color: mint.mint11,
height: barHeight,
},
},
{
mark: {
type: "text",
align: "left",
baseline: "middle",
dx: 3,
color: slate.slate9,
},
encoding: {
text: {
aggregate: "count",
type: "quantitative",
},
},
},
],
} as TopLevelFacetedUnitSpec; // "layer" not in TopLevelFacetedUnitSpec
}
// Arrow formats have a magic number at the beginning of the file.
const ARROW_MAGIC_NUMBER = "ARROW1";
// register arrow reader under type 'arrow'
formats("arrow", arrow);
export function getDataSpecAndSourceName<T>(data: string | T[]): {
dataSpec: TopLevelSpec["data"];
sourceName: "data_0" | "source_0";
} {
let dataSpec: TopLevelSpec["data"];
let sourceName: "data_0" | "source_0";
// Data may come in from a few different sources:
// - A URL
// - A CSV data URI (e.g. "data:text/csv;base64,...")
// - A CSV string (e.g. "a,b,c\n1,2,3\n4,5,6")
// - An array of objects
// For each case, we need to set up the data spec and source name appropriately.
// If its a file, the source name will be "source_0", otherwise it will be "data_0".
// We have a few snapshot tests to ensure that the spec is correct for each case.
if (typeof data === "string") {
if (data.startsWith("./@file") || data.startsWith("/@file")) {
dataSpec = { url: asRemoteURL(data).href };
sourceName = "source_0";
} else if (isDataURLString(data)) {
sourceName = "data_0";
const base64 = extractBase64FromDataURL(data);
const decoded = window.atob(base64);
// oxlint-disable-next-line unicorn/prefer-ternary
if (decoded.startsWith(ARROW_MAGIC_NUMBER)) {
dataSpec = {
values: base64ToUint8Array(base64),
// @ts-expect-error vega-typings does not include arrow format
format: { type: "arrow" },
};
} else {
// Assume it's a CSV string
dataSpec = { values: parseCsvData(decoded) };
}
} else {
// Assume it's a CSV string
dataSpec = { values: parseCsvData(data) };
sourceName = "data_0";
}
} else {
dataSpec = { values: data };
sourceName = "source_0";
}
return { dataSpec, sourceName };
}
export function getScale(sourceName: string): Scale {
return {
align: 0,
paddingInner: 0,
paddingOuter: {
expr: `length(data('${sourceName}')) == 2 ? 1 : length(data('${sourceName}')) == 3 ? 0.5 : length(data('${sourceName}')) == 4 ? 0 : 0`,
},
};
}