alekzonder/discovery

Hackable JSON discovery tool

Discovery

Hackable JSON discovery tool

Documentation in progress...

Examples

• CSS syntax reference (source)

Install

> npm install @discoveryjs/discovery

How to launch server

Run the command in folder with discovery installed:

> npx discovery

By default server server starts on port 8123. You can change it by -p (or --port) option:

> npx discovery -p 7777

See more options with -h (or --help) argument:

> npx discovery -h

How to build model

Run the command in folder with discovery installed:

> npx discovery-build

The result will be placed to build folder by default. You can change it with -o (or --output) option:

> npx discovery-build --output some/path

See more options with -h (or --help) argument:

> npx discovery-build -h

Modes

Discovery can work in three modes:

• Model-free (when no any model specified)
• Single model
• Multiple models

Model-free

In this mode you can upload any data by clicking "Load data" button, or drag'n'drop file right into the browser,

Multiple models

In this mode discovery will start with model selection page. Every model will have own route namespace, and you can switch models and reports at any time.

Single model

If you want only one model, you should start discovery with --model %modelName%. In this mode index page will represent your model default page.

Define a model

To configure discovery you should specify one of config files:

• .discoveryrc.js
• .discoveryrc.json
• .discoveryrc (the same as .discoveryrc.json)

Or you can use discovery key in your package.json file.

Config structure

Config should provide JSON or export js-object with following properties:

• name - name of discovery instance (used in title)
• models - object with model configuration, where key used as model slug and value - model config

Example:

module.exports = {
    name: 'discovery',
    models: {
        one: <modelConfig>,
        two: <modelConfig>
    }
};

Define model

Model config consists of the following fields:

• name - name of model (used in title)
• data - function which returns any|Promise<any>. Result of this function must be JSON serializable
• prepare - path to a file with additional initialization logic. (eg. add cyclic links and relations. extensions for query engine)
• ui object with following fields:
  • basedir - directory to resolve relative path in assets
  • assets - path to .js and .css files

  js files has own scope (as modules) which have reference to discovery instance

• extendRouter - function(router, modelConfig, options)

Base concepts

Model goes through data->prepare->render chain. Data can be modified with prepare function and rendered by various views and its combinations.

View is a function(el, config, data, context) where:

• el - DOM-element in which view will be rendered
• config - configuration of view
• data - data to render
• context - contains of model data, metaifo (createdAt, etc), router (optional), any user defined or view defined additional data

Page is a root view function(el, data, context). Similar to view it has all of its arguments except config.

Page

To define page you should call discovery.page.define(pageId, render(el, data, context), options) where:

• pageId - unique page identifier
• render - page render function described above
• options - object with options:
  • reuseEl - do not clear container before render page (skiped for first render or pageId change)
  • init - invokes on first page render or pageId change
  • keepScrollOffset - dont scroll to page top if pageId didn't changed since last page render

Other handy methods for working with page:

• discovery.renderPage() renderId, renderRef, renderParams
• setPage(pageId, pageRef, renderParams) or setPageParams(renderParams) triggers renderPage()
• discovery.getPageContext() gets context of page

Special pages

There are some built-in special pages:

• default
• report
• not-found

you can override this pages with definePage

View

To define new view just call discovery.view.define(viewId, render, options) where:

• viewId - unique view identifier
• render - function(el, config, data, context) or view definition object
• options - object with following fields: tag - container element. if false|null view will return documentFragment

You can render your view with discovery.view.render(el, view, data, context) where:

• view can be string viewId, ViewDefinition or Array<ViewDefinition> viewId will expand to { view: viewId }.

Also you can specify view as viewId:query, that will give you handy shortcut to { view: viewId, data: query }

• viewDefinition object with view definition settings:
  • view: string viewId
  • when: query, will be coerced to Boolean (but empty arrays and objects will coerce to false)
  • data: query (will be described later)
  • Also there can be view specific additional fields

Special views

Also specail built-in sidebar view is available you can override this view with view.define

Queries

As a query you can specify string which will be processed by Jora so your data will be a result of following flow:

jora(value) -> function(data, context) -> data

Or you can use function(data, context) as query as well. Aslo you can use any data as query right away.

License

MIT