Share your code. npm Orgs help your team discover, share, and reuse code. Create a free org »


    The Flourish SDK

    Flourish is a platform for visualising and storytelling with data. It allows users to quickly create high-quality visualisations, stories, presentations and other interactive content by putting their own data and text into configurable templates.

    This SDK allows developers to make templates and upload them to Flourish.


    See all commands and options. There are also a few example templates you can use for reference.

    Template overview

    There is no limit to what a Flourish template can be or do. It may be a simple chart, a 3D visualisation in WebGL, or a richly interactive app with a complex UI. There's also no limit to how you can write your code, what libraries you can use, etc. The only requirement is that the template directory includes certain files, as specified below, and that the template.js creates a window.template object with the following properties:

    • state The template's current state; updated by Flourish when the user changes a setting or switches slide
    • draw() Called by Flourish when the template is loaded
    • update() Called by Flourish when the user changes a setting or data table, or switches slide
    • data Optional; updated by Flourish with data from any user-editable data tables

    Template files

    A Flourish template is a directory containing, at a minimum:

    Templates often also include:

    • thumbnail.jpg or thumbnail.png Template image for use on Flourish: strongly recommended, and may become mandatory in the future
    • index.html The template HTML
    • src/ JavaScript source files
    • data/ Data tables in CSV format that define the sample data shipped with the template
    • static/ Arbitrary static files that can be referenced by the template
    • Documentation for end users of the template in Flourish, in Markdown format
    • Documentation for developers working on the template, in Markdown format

    The following files are not treated specially by Flourish, but many templates include them as part of the way they are packaged and built:


    The template script. Typically generated using a build process, such as rollup.js, from Javascript modules in src/, though in principle you could write it directly as long as it creates a window.template object that has function properties update() and draw(), and object properties state and optionally data.


    The main Flourish configuration file for your template. The top-level properties are:

    • sdk_version Which major version of the Flourish SDK the template is compatible with. Currently this should be 2.
    • id A unique identifier for your template. If you publish a template and the id is already in use by you, the Flourish server will assume you are updating an existing template and overwrite it.
    • name What the template will be called within Flourish
    • author Who wrote the template
    • description A short description of the template
    • credits Optional credits for data sources, map tiles, etc, in Markdown format
    • autoheight Optional configuration for autoheight embedding (see below)
    • image_download Flag to indicate whether image snapshots work for the template (default is true)

    Other properties are settings, data and build, which are described below.


    When a Flourish user embeds a visualisation or story, the default sizes are width 100% and height auto. (The user can override these with any valid css values.) If no autoheight property is specified in the template.yml, height auto means that the embedding iframe will be automatically updated to match the computed height of the template’s body. This works well for most templates, but not for vertically fluid templates that have no natural height (e.g. a slippy map that fills the window). For these templates, you can optionally specify what auto should mean by setting the autoheight property. You can set it to a px-based CSS value – such as 600px – or to an aspect ratio using the syntax 4x3, 16x9, etc. For additional control you can specify an object representing min-width breakpoints. For example:

        600: 500px # Height is 500px up to width of 600px 
        1200: 4x3 # Aspect ratio is 4x3 600–1200px 
        : 1400px # Height is 1400px for all widths above 1200. 

    Infinity can be specified with ∞ or *. If the infinity option is left off, the largest specified breakpoint will apply to infinity.

    All of this explains how the height is set by default when a visualisation or story based on your template is embedded. However, a Flourish user also has the option to embed at a fixed height. You can test for Flourish.fixed_height in your template to see whether that is the case.

    build configuration

    The template.yml file will usually also include a build section defining build rules. For example:

      # Defines the build processes used to build your template from source. 
      # You can have any number of build processes, and changes to the relevant 
      # files will trigger the appropriate build. Here we define build scripts 
      # for JavaScript and CSS. 
        script: npm run build
        # You can specify a whole directory, or individual files, or both. 
        # A change inside the directory or to any of the listed files will 
        # trigger a rebuild and reload. 
        directory: src
          - rollup.config.js
        script: npm run less
        directory: less


    The template.yml file will usually also include a settings section which populates the settings panel in the Flourish visualisation editor (and SDK). Each setting allows the user to change a specific property in the template state. When a setting is changed by the user , state is updated and the template's update() function is called.

    If an entry in the settings array is a string, it is interpreted as a section title. Otherwise it must be an object with the property and type properties. Other properties are optional, but name and description are recommended to help the user understand the role of the setting.

    - Section title # Headings can be used to break up the settings into collapsible sections 
    property: my_state_property # Required; must be a property of the state object 
      name: Example number setting # Optional; appears next to the setting 
      description: A setting for changing a number # Optional; appears on mouseover 
      type: number # Required; see available types below 

    To improve the layout of your settings, you can set the width of any setting to be full, half, three quarters or quarter of the width of the settings panel. You can also add a horizontal separator above a setting using new_section: true, or a subheading using new_section: This is a subheading.

    property: my_number
      name: Neat little number input
      width: quarter # Optional; sets the width of the setting 
      new_line: true # Optional; starts a new line with the current setting and adds a line above 

    The following types of settings are supported:


    Creates a checkbox that sets the state property to true or false. Alternatively can be displayed as two buttons using the choices attribute to specify labels and/or images. For example:

    property: ranked
      type: boolean
      - [ Ranks, true ]
      - [ Scores, false ]

    Creates a colour picker that sets the state property to a string containing a hex RGB colour e.g. "#123456".


    Creates a number input that sets the state property to a number. Optionally add min and max properties to limit the range, step to control the input’s increment buttons. By default number settings always return a number and blanked inputs are set to zero; to allow blanked input, with null returned as the value, add optional: true. Width defaults to half.


    By default, creates a single-line text input that sets the state property to the relevant string text. If you add a valid choices property, the setting instead creates a dropdown (by default) or button group (if you also add style: buttons). The choices property must be an array. Each of its element can be a string (in which case this string is returned to the state) or an array containing a display name, the associated string value and (for button groups) a background image.

    To add a special dropdown that allows the user to specify any text in addition to choosing from the list, add choices_other: true. This is ignored for button groups.

    property: size
      name: Size
      type: string
      choices: # An array of values to fill the dropdown 
        – small_size # A choice can be a string 
        – # Or a choice can be an array of two or three elements … 
          – Absolutely enormous # … in which case the first string is the display name, 
          – large_size # … the second string is the value passed to the template 
          – massive.jpg # … the third is an image file name in your `static` directory 
      choices_other: true # allows the user to input any value they like 

    Creates a user-resizeable multiline text input. Defaults to two lines tall. Can be made taller by adding size: large, which sets the height to half of the viewport height.


    Same as text but with a monospace font and text-wrapping control.

    Conditional settings

    Sometimes you might want to simplify the user experience for Flourish users by hiding some settings depending on whether they are needed or not. You can use the show_if and hide_if properties to control whether or not a setting should be displayed based on another setting’s value.

    In the following example, the x axis label setting will only be displayed if “Show x axis” is selected:

    property: show_x_axis
      name: Show x axis
      type: boolean
    property: x_axis_label
      name: X axis label
      type: string
        show_x_axis: true

    You can use the following shorthand syntax for booleans. This is equivalent to the previous example:

    show_if: show_x_axis

    If you specify an array of conditional values, this setting will be displayed if the referenced setting has any of the specified values. In the following example, the setting is displayed if color_mode is set to either "diverging" or "continuous":

      color_mode: [diverging, continuous]

    You can specify multiple conditions. All of these tests must pass for the setting to be displayed. For example:

      show_x_axis: true
      color_mode: diverging

    The hide_if option works in exactly the same way, except that the setting is hidden if the conditional test passes.

      color_mode: diverging

    You can also control settings display depending on whether or not particular data bindings have been specified. A binding is specified using the syntax data.[dataset].[key]. For example:

    show_if: data.values.filter1 // shorthand syntax
      data.values.filter1: true

    You cannot specify both show_if and hide_if options on the same setting.

    data bindings

    The template.yml file may also include a data section. This section consists of an array of data ‘bindings’ that sets how the template should use and refer to the template’s editable data tables (which are initially populated by the CSV files in data/). Each binding adds one or more columns of data to a dataset under a particular key. You can define as many datasets as you like. They are made available to the template as properties of the data object. Each one consists of an array containing an object for each row of the relevant data table, as shown in the example below.

    Once your template is published, Flourish users can change the data in the Flourish editor, and also change which columns are linked to each binding. But in your code you don’t need to worry about this because you just refer to the key rather than referencing the column header or index.

    There are two types of data binding: column is used when the number of columns is and must always be one; columns supports any number of columns, including none.

    A default value must be supplied for each data binding, unless you have specified optional: true (only supported for single column bindings). The example below shows how this is done.

    The following example sets up a dataset with two keys, one single-column and one multi-column.

    - Locations # Optional; breaks up the bindings into collapsible sections 
    - Description string # Optional; additional description to explain to the user how the data binding works 
    name: Country code # Name shown in UI 
      description: Requires ISO 3166-1 alpha-2 codes # Optional description for the UI 
      dataset: country_scores # Which dataset this binding is part of 
      key: iso_code # The key used to access the data in this binding in the template code 
      type: column # This binding can take only one column 
      column: By Decade::A  # The default values are drawn from column A of `By Decade.csv` 
    name: Values
      dataset: country_scores
      key: values
      type: columns # This binding can take any number of columns 
      columns: By Decade::B-D,F # The default values are arrays drawing from columns B-D and F of `By Decade.csv` 
    name: Flag image 
      dataset: country_scores
      key: flag_pic
      type: column
      optional: true # Default values can be omitted for an optional binding 

    In this example, if By Decade.csv contained the following…


    … then in your template data.country_scores would be:

        { iso_code: "US", values: [ "3122", "3128", "3129", "3119" ]},
        { iso_code: "GB", values: [ "1203", "1205", "1208", "1200" ]},
        { iso_code: "FR", values: [ "1030", "1005", "1010", "1024" ]}

    The column headers are available in any dataset via the column_names property of the data array. E.g. in the above example data.country_scores.column_names is:

        { iso_code: "Country", values: [ "1980s", "1990s", "2000s", "2010s" ]}

    thumbnail.jpg or thumbnail.png

    A thumbnail image for your template in JPEG or PNG format. No particular size is required – the precise size at which the image is displayed depends on the size of the browser window – but we recommend approximately 600px × 400px.


    The base HTML for the template, if required.

    To reference resources in the static directory use relative links, e.g.

    <img src="logo.png">

    These relative links will be replaced by a suitable path when the template is rendered. If you’re creating links to static resources with code, you need to prefix them with the value of Flourish.static_prefix. See below for more about static resources.

    You can add DOM elements, script tags, external stylesheets, etc, to your index.html, as with any other html page. Do not reference assets at non-https addresses, since these will cause problems when the template is embedded in Flourish or any other secure website.

    If the index.html file is missing, the following default HTML is used:

    <!doctype html>
            <meta charset="utf-8">
                htmlbody { height: 100%; margin: 0; overflow: hidden; }

    There is no need to include a <title> element, because Flourish will insert an appropriate one when it renders the template. If you do include a <title> then it will be replaced.


    JavaScript source code. Typically these will be bundled in a build process to create template.js.

    In simple templates src/ may contain just a single index.js which will export the three key template properties – state, draw(), update() – and, optionally, data.

    If you modify a source file while the SDK is running, the SDK will run the appropriate build script specified in template.yml and then reload the preview.

    Source files are not uploaded to the server when you publish a template.


    A directory containing CSV data tables (with header rows) that the user will be able to see and edit in the Flourish visualisation editor's spreadsheet interface. These can be used for sample data that you intend the end user to replace with their own data. They can also be useful for create editable look-up tables for configuration options.

    Columns from these CSV files (or the data tables in the Flourish editor) are made available in the template as one or more datasets, as specified in the data bindings of the template.yml.

    (It's also possible to import data into your template manually from your static directory or elsewhere, using D3 or any other technique. But the end user won't be able to edit the data or update the column selection if you do that.)


    A directory for any static files used in the template, such as images, fonts, stylesheets or code libraries. To reference the static directory in your index.html file, use relative links:

    <script src="leaflet/leaflet.js"></script>

    Or from JavaScript use Flourish.static_prefix:

    var img_url = Flourish.static_prefix + "/my_image.jpg";

    A file documenting the template for end Flourish users, in Markdown format. This documentation will be displayed on the template’s page on the Flourish site, when the template is published and made live.

    A standard README file for developers, available as part of the source like other files.

    Other files

    In general any files not listed above will be ignored by Flourish, and you may use them as part of your template’s build process or for any other purpose. They won’t be uploaded when you publish the template.

    The skeleton template you get when you run flourish new and the Flourish example templates contain a couple of such files:


    A standard npm configuration file, listing any dependencies. If you’re not starting with a skeleton template, you can create this file using npm init. This file is not treated specially by Flourish, and is only needed if your template is packaged as an npm module.


    This is the default filename for a rollup.js configuration file, and specifies how the source code is to be bundled by rollup.js. It might look something like this:

    export default {
      entry: "src/index.js",
      format: "iife",
      moduleName: "template",
      dest: "template.js",
      sourceMap: true,
      plugins: [

    You only need this if you’re using rollup.js to bundle your template.

    Using the SDK


    Install the SDK with npm, as follows. If you don't already have it you'll need to install node.

    npm install -g @flourish/sdk

    The -g means the package will be installed globally, enabling access to the flourish command across your system.

    Create a new template

    flourish new [dir_name]

    Where dir_name is the name of the template directory that will be created. The new directory will be populated with skeleton files and directories that you can edit with your own code and files.


    flourish build [rules]

    Build the template in the current directory, using the scripts specified in template.yml. if you don’t specify any build rules, it will run all of them. You shouldn’t usually need to use this command explicitly, because flourish run will do a full build before it runs the server, and then monitor your files for changes and trigger appropriate rebuilds automatically when something changes.

    Run and view a template

    flourish run [dir_name]

    Builds the template and runs the SDK viewer in your web browser. If dir_name is omitted it uses the current directory.

    While it’s running it watches the template directory for changes:

    • if you edit any static template files (index.html, template.yml, data/*, static/*) the SDK will refresh the page in the browser;
    • if you edit a file that is the source for a build rule, the SDK will run that build rule and then refresh the page.


    • --open or -o Try to open the SDK in your web browser once the server is running. At present this only works on macOS.
    • --port Specify a particular port; defaults to 1685
    • --no-build Skip the build process

    Publish a template

    flourish publish [dir_name]

    You'll first need to be logged in. If dir_name is omitted it uses the current directory.

    Login, logout, register, check status

    New users should flourish register and follow the prompts. Existing users can log in and out with flourish login and flourish logout. To check which account you are currently logged in with, use flourish whoami.

    Download a template

    To download an existing template, such as one of our examples or an open-source template, simply git clone the repo, then change into the new directory and npm install then flourish run. For example:

    git clone
    cd example-template-circle/
    npm install
    flourish run

    Upgrade from an older version of the SDK

    flourish upgrade [dir_name]

    Sometimes changes to the Flourish SDK will require templates to be changed to be compatible with the new version. This command will attempt to convert a template made for an earlier version of Flourish to be compatible with the current version.

    Template API

    Your compiled JavaScript should assign an object with the following properties to window.template:


    Records the current state of template. Default values for each of its properties are set where the object is declared in the JavaScript. These are updated by Flourish when the user changes a setting in the visualisation editor, or programmatically, e.g. when the user interacts with the template output.

    To make sure your template works nicely in the story editor, ensure that all visual aspects of the template – including the UI – are set from the state by the update() function. For example, if the user clicks a menu item and you want to highlight that item, do this by updating the state and calling update(). Do not do it in the click handler for the menu item, as this will only work when the user clicks the button manually, not when the story editor moves from one state to another.


    Called when the template loads. Typically used for initialisation code such as adding elements to the DOM. In some cases, however, it might do nothing except call update().

    In most templates draw() will not be called except once by Flourish after the template is loaded, though in some cases you may want to call it manually – for example if you want to delete the DOM and redraw it when the user resizes the window.


    Called whenever the user changes a data table or setting in the visualisation editor, or when changing slides in the story editor. Typically will also be called by the template in response to user interaction.


    An object into which Flourish will put the data from user-editable data tables. Usually your code will initialise data as an empty object {}, and read from it in the draw() and update() functions.

    Each property is a dataset: an array containing an object for each row in the relevant data table. The structure of each dataset is defined in the data bindings of the template.yml, and the data is loaded from the tables in the data/ directory.




    npm i @flourish/sdk

    Downloadslast 7 days







    last publish


    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar