datasette-dashboards

Datasette plugin to create interactive data dashboards

Try out a live demo at https://datasette-dashboards-demo.vercel.app

WARNING: this plugin is still experimental and not ready for production. Some breaking changes might happen between releases before reaching a stable version. Use it at your own risks!

Installation

Install this plugin in the same environment as Datasette:

$ datasette install datasette-dashboards

Usage

Define dashboards within metadata.yml / metadata.json:

plugins:
  datasette-dashboards:
    my-dashboard:
      title: My Dashboard
      description: Showing some nice metrics
      layout:
        - [analysis-note, events-count]
        - [analysis-note, events-source]
      filters:
        date_start:
          name: Date Start
          type: date
          default: "2021-01-01"
        date_end:
          name: Date End
          type: date
        category:
          name: My Category
          type: select
          options: [Option 1, Option 2, Option 3]
        dynamic_category:
          name: My Dynamic Category
          type: select
          db: jobs
          query: SELECT DISTINCT col1 FROM mytable ORDER BY col1 ASC
        dependent_filter:
          name: My Dependent Filter
          type: select
          db: jobs
          query: SELECT DISTINCT col2 FROM mytable WHERE col2 IS NOT NULL [[ AND col1 = :dynamic_category ]] ORDER BY col2 ASC
      charts:
        analysis-note:
          library: markdown
          settings:
            extensions:
              - attr_list
            extra_attrs:
              h1:
                - name
          display: |-
            # Analysis notes {: name="notes" }
            > A quick rundown of events statistics and KPIs

        events-count:
          title: Total number of events
          db: jobs
          query: SELECT count(*) as count FROM events
          library: metric
          display:
            field: count
            prefix:
            suffix:

        events-source:
          title: Number of events by source
          db: jobs
          query: SELECT source, count(*) as count FROM events WHERE TRUE [[ AND date >= date(:date_start) ]] [[ AND date <= date(:date_end) ]] GROUP BY source ORDER BY count DESC
          library: vega-lite
          display:
            mark: { type: arc, tooltip: true }
            encoding:
              color: { field: source, type: nominal }
              theta: { field: count, type: quantitative }

A new menu entry is now available, pointing at /-/dashboards to access all defined dashboards.

Properties

Dashboard properties:

Dashboard settings:

Dashboard filters:

Common chart properties for all chart types:

To define SQL queries using dashboard filters:

SELECT * FROM mytable [[ WHERE col >= :my_filter ]]

SELECT * FROM mytable WHERE TRUE [[ AND col1 = :my_filter_1 ]] [[ AND col2 = :my_filter_2 ]]

Important notes:

• When a select filter has more than 100 options, the dropdown list will be automatically converted to a text filter with autocompletion

Vega properties

Available configuration for vega charts:

Notes about the display property:

• Requires a valid Vega specification object
• Some fields are pre-defined: $schema, description, autosize, data, signals
• All fields are passed along as-is (overriding pre-defined fields if any)
• Only mark and encoding fields are required as the bare-minimum

Vega-Lite properties

Available configuration for vega-lite charts:

Notes about the display property:

• Requires a valid Vega-Lite specification object
• Some fields are pre-defined: $schema, description, width, view, config, data
• All fields are passed along as-is (overriding pre-defined fields if any)
• Only mark and encoding fields are required as the bare-minimum

Markdown properties

Available configuration for markdown chart:

Note :

• Some common properties do not apply and can be omitted: title, db, query
• Markdown rendering is done by datasette-render-markdown
• To configure Markdown rendering, use the settings property.

Markdown settings

Available configuration for Markdown rendering:

Note :

• These arguments are passed to the render_markdown() function.
• The extra_tags and extra_attrs are passed to Bleach.
  • extra_attrs is indeed required to be given as a dict.

Metric properties

Available configuration for metric chart:

Note:

• The display.field must reference a single-numerical value from the SQL query (e.g. numerical number field in SELECT count(*) as number FROM events)

Table properties

There is no required configured in display, so you can either ignored or leave it empty for table charts.

Some advice for a nice table chart:

• Set proper column names in the SELECT clause
• Limit the number of columns in the SELECT clause
• Limit the number of rows with the LIMIT clause
• Order the rows explicitely with the ORDER BY clause
• Use SQLite string concatenation operator (||) to format column data (for instance to include HTML markup!)

Map properties

Available configuration for map chart:

Warning: do not try to load more than a thousand rows for a map at the risk of slugginess and being unreadable. Make sensible use of the LIMIT clause to reduce the number of items to display on the map.

Dashboard layout

The default dashboard layout will present two charts per row (one per row on mobile). To make use of custom dashboard layout using CSS Grid Layout, define the layout array property as a grid / matrix:

• Each entry represents a row of charts
• Each column is referring a chart by its property name
• An empty slot in the grid can be specified using the . (full stop) placeholder

WARNINGS:

• All rows must specify the same number of columns
• All charts must be placed somewhere on the custom layout

Here is a simple 2x3 grid example with 4 different charts:

layout:
  - [chart1, chart2, chart3]
  - [chart1, chart4, chart4]

Here is a more subtle example involving an empty spot at the end of the second row:

layout:
  - [chart1, chart2, chart3]
  - [chart1, chart4, .]

Embedding dashboards and charts

Dashboards can be embedded within an HTML page using an iframe element:

<iframe
  src="/-/dashboards/my-dashboard/embed?start_date=2023-01-01&end_date=2023-12-31"
  frameborder="0"
  width="100%"
  height="600"
  allowtransparency
>
</iframe>

Same goes for charts:

<iframe
  src="/-/dashboards/my-dashboard/my-chart/embed?start_date=2023-01-01&end_date=2023-12-31"
  frameborder="0"
  width="100%"
  height="600"
  allowtransparency
>
</iframe>

Development

To set up this plugin locally, first checkout the code. Then create a new virtual environment and the required dependencies:

uv sync

To run the QA suite:

uv run ruff check datasette_dashboards tests
uv run mypy --strict datasette_dashboards tests
uv run pytest -v --cov=datasette_dashboards --cov=tests --cov-branch --cov-report=term-missing tests

Updating JS dependencies

External JS dependencies are tracked and bundled using NPM and package.json (package-lock.json is not needed here):

npm install --no-package-lock

Demo

With the developmnent environment setup, you can run the demo locally:

uv run datasette \
  --metadata demo/metadata.yml \
  --template-dir demo/templates \
  demo/jobs.db

License

Licensed under Apache License, Version 2.0

Copyright (c) 2021 - present Romain Clement