observable-notebooks（可观察的笔记本）

Observable Notebooks 2.0

Overview

Observable Notebooks 2.0 is an open-source notebook system for creating interactive documents that combine code, data, and visualization. It features an HTML-based file format, vanilla JavaScript (no more Observable-specific dialect), and a CLI for building static sites.

When to Use This Skill

Use this skill when:

• Creating new Observable notebooks
• Editing existing notebook files
• Building and previewing notebooks locally
• Working with Observable's standard library (Plot, Inputs, D3)

Installation

Install Notebook Kit as a local dependency:

npm add @observablehq/notebook-kit

CLI Commands

Preview (Development Server)

notebooks preview notebook.html
notebooks preview --root ./notebooks
notebooks preview --template template.tmpl notebook.html

Starts a development server with live reload for editing notebooks.

Build (Static Site Generation)

notebooks build notebook.html
notebooks build --root ./notebooks -- *.html

Generates static HTML in .observablehq/dist/ by default.

Download (Convert Existing Notebooks)

notebooks download https://observablehq.com/@d3/bar-chart > bar-chart.html

Downloads an existing Observable notebook to the local HTML format.

Query (Database Queries)

notebooks query --database duckdb 'SELECT * FROM data'

Execute and cache database query results.

File Format

Notebooks are HTML files with a <notebook> root element:

<notebook>
  <title>My Notebook</title>

  <script type="text/markdown">
    # Introduction

    This is a markdown cell with interpolation: ${1 + 1}
  </script>

  <script type="module">
    const data = [1, 2, 3, 4, 5];
    display(data);
  </script>

  <script type="module">
    import * as Plot from "npm:@observablehq/plot";
    display(Plot.barY(data).plot());
  </script>

</notebook>

Use four-space indentation inside <script> tags (trimmed during parsing).

Cell Types

Cell Attributes

• pinned - Displays the cell's source code alongside output
• hidden - Suppresses implicit display (cell still runs)
• id="123" - Unique positive integer for stable editing
• database="mydb" - Specifies database for SQL cells
• format="json" - Output format for data loaders
• output="varname" - Exposes values from non-JavaScript cells

Example:

<script type="module" pinned>
  const x = 42;
  display(x);
</script>

<script type="application/sql" database="duckdb" output="results">
  select * from data limit 10
</script>

Reactivity

Cells run automatically when referenced variables change, like a spreadsheet:

<script type="module">
  const a = 10;
</script>

<script type="module">
  const b = 20;
</script>

<script type="module">
  // This cell re-runs whenever a or b changes
  display(a + b);
</script>

Top-level variables declared in one cell are accessible throughout the notebook.

Standard Library

Core Functions

• display(value) - Render values to the page
• view(input) - Display inputs and return value generators
• FileAttachment(path) - Reference local files
• invalidation - Promise for cleanup on cell re-run
• visibility - Wait for cell to become visible
• width - Current page width (reactive)
• now - Current timestamp (reactive)

Preloaded Libraries

All available without explicit imports:

• Inputs - Form controls (Inputs.range(), Inputs.select(), etc.)
• Plot - Charting library (Plot.barY(), Plot.line(), etc.)
• D3 - Data visualization utilities
• htl - Hypertext Literal for safe DOM creation
• tex - LaTeX rendering

Using npm Packages

<script type="module">
  import confetti from "npm:canvas-confetti";
  confetti();
</script>

Observable Inputs

Create interactive controls that automatically update dependent cells:

<script type="module">
  const gain = view(Inputs.range([0, 11], {
    value: 5,
    step: 0.1,
    label: "Gain"
  }));
</script>

<script type="module">
  // This cell re-runs when the slider changes
  display(`Current gain: ${gain}`);
</script>

Available inputs:

• Inputs.button(label) - Click trigger
• Inputs.toggle({label, value}) - Boolean switch
• Inputs.checkbox(options, {label}) - Multi-select
• Inputs.radio(options, {label}) - Single select
• Inputs.range([min, max], {value, step, label}) - Numeric slider
• Inputs.select(options, {label, multiple}) - Dropdown
• Inputs.text({label, placeholder}) - Text input
• Inputs.textarea({label, rows}) - Multi-line text
• Inputs.date({label, value}) - Date picker
• Inputs.color({label, value}) - Color picker
• Inputs.file({label, accept}) - File upload
• Inputs.search(data, {label}) - Search/filter data
• Inputs.table(data, {columns}) - Interactive data table

Observable Plot

Create charts with the grammar of graphics:

<script type="module">
  const data = [
    {year: 2020, value: 10},
    {year: 2021, value: 20},
    {year: 2022, value: 15}
  ];

  display(Plot.plot({
    marks: [
      Plot.barY(data, {x: "year", y: "value"})
    ]
  }));
</script>

Common marks:

• Plot.dot() - Scatter plots
• Plot.line() - Line charts
• Plot.barX(), Plot.barY() - Bar charts
• Plot.areaY() - Area charts
• Plot.rectY() - Histograms
• Plot.text() - Labels
• Plot.ruleX(), Plot.ruleY() - Reference lines

Loading Data

Local Files

<script type="module">
  const data = await FileAttachment("data.csv").csv({typed: true});
  display(Inputs.table(data));
</script>

Supported formats: .csv(), .json(), .tsv(), .text(), .blob(), .arrayBuffer(), .image(), .xlsx(), .zip().

Remote Data

<script type="module">
  const response = await fetch("https://api.example.com/data");
  const data = await response.json();
  display(data);
</script>

Database Queries

<script type="application/sql" database="duckdb" output="results">
  select * from "data.parquet" limit 100
</script>

<script type="module">
  display(Inputs.table(results));
</script>

Themes

Set a theme on the notebook element:

<notebook theme="dark">
  ...
</notebook>

Built-in themes: air, coffee, cotton, deep-space, glacier, ink, midnight, near-midnight, ocean-floor, parchment, slate, stark, sun-faded.

Page Templates

Create custom templates for consistent layouts:

<!-- template.tmpl -->
<!DOCTYPE html>
<html lang="en">
<head>
  <meta charset="utf-8">
  <meta name="viewport" content="width=device-width, initial-scale=1">
  <meta name="generator" content="Observable Notebook Kit">
  <title>My Site</title>
  <link rel="stylesheet" href="styles.css">
</head>
<body>
  <header>My Site Header</header>
  <main></main>
  <footer>My Site Footer</footer>
</body>
</html>

Use with: notebooks build --template template.tmpl -- *.html

The <main> element receives the rendered notebook cells.

Example: Complete Notebook

<notebook theme="slate">
  <title>Sales Dashboard</title>

  <script type="text/markdown">
    # Sales Dashboard

    Interactive visualization of quarterly sales data.
  </script>

  <script type="module">
    const sales = await FileAttachment("sales.csv").csv({typed: true});
  </script>

  <script type="module">
    const quarter = view(Inputs.select(
      ["Q1", "Q2", "Q3", "Q4"],
      {label: "Quarter", value: "Q1"}
    ));
  </script>

  <script type="module">
    const filtered = sales.filter(d => d.quarter === quarter);

    display(Plot.plot({
      title: `Sales for ${quarter}`,
      marks: [
        Plot.barY(filtered, {
          x: "product",
          y: "revenue",
          fill: "category"
        }),
        Plot.ruleY([0])
      ]
    }));
  </script>

  <script type="module" pinned>
    display(Inputs.table(filtered, {
      columns: ["product", "category", "revenue", "units"]
    }));
  </script>

</notebook>

Resources

• Notebook Kit Documentation
• System Guide
• Observable Plot
• GitHub Repository