Skip to content

Base library for derby-entity CRUD component and derby-entities-visualizations

Notifications You must be signed in to change notification settings

michael-brade/derby-entities-lib

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Derby Entities Library

This library reads the entity schema definition and provides the API to access entities and their items. It is the base library for derby-entity.

Installation

As usual, the library is installed via

npm install derby-entities-lib

However, I don't like to contribute to huge node_modules directories, so the npm package is compiled and minified, and neither carries the original sources, nor the tests. If you want a readable and debuggable version, install it directly from GitHub using

npm install michael-brade/derby-entities-lib

or clone the repository.

npm install
npm test

will then run the tests.

Usage

EntitiesApi = require('derby-entities-lib/api');

// in the app init() function
EntitiesApi.init(model, entities);  // see below for what an entities definition is

// then get the API instance anywhere in the app:
var instance = EntitiesApi.instance(model);

Terminology

  • entity: in case of Derby: a MongoDB collection
  • item: the actual instances of an entity, a MongoDB document, has at least an id attribute
  • attribute: an attribute of an entity

An analogy would be from the OO world: an entity is a class of which an item is an object.

Schema Definition

Derby uses MongoDB in the background and that is currently the only option available. The structure of how the data is store in the DB can be chosen quite freely with derby-entity. Basically, the components under types/ define how to read and write an attribute of an item, i.e., a document in a collection. Even though MongoDB allows for arbitrary strutures, each item of one entity has to have the same structure. That means all MongoDB documents have the same structure under a certain collection.

Entities

An entity definition has the following structure:

entities =
    * id: "people"

      display:
          attribute: "name"
          decorate: ["photo", "employer"]

      attributes:
        * id:   "name"
        * id:   "email"  

        * id:   "photo"
          type: "image"

        * id:   "employer"
          type: "entity"
          multi: false
          reference: true

Display

This property defines how to display an item of that entity if displayed as an attribute of another item or in a select2 dropdown. It is the summary of that item, so to speak.

attribute, type: string, default: name

This sets the main display attribute of that item.

decorate, type: array of strings, default: none

decorate adds further information, like an image, a color, or just some additional text. This works recursively, so the display attribute is rendered, that output is then input to the attribute renderer for the photo attribute (to stay with the example above), which then is passed to the employer renderer. (Actually, entity doesn't support decorations yet, but whatever, you get the idea.)

Attributes

Currently available attribute types:

Type Description
text default type; a simple string, edited using <input type="text">
textarea also a simple string, but probably longer, thus edited with an automatically expanding textarea
number an integer, edited using <input type="number">
color a color, edited using <input type="color">
image an image, which will be converted to base64 and stored like a string
entity a nested structure, copy/reference the item of another entity type

Planned for the future are:

Type Description
boolean enter a true/false value using a checkbox
password enter and encrypt a password securely
object to allow for arbitrary data structures to be created and edited, essentially creating recursive forms
markdown a markdown text field, edited using CodeMirror and a visual editor
svg instead of an image, use svg
image-upload maybe; this would upload the image to the server and create a link to it so that it doesn't have to be stored in the MongoDB
file-upload same, see above

All attributes must define at least one property: id. The following properties are available for all attribute types:

id, type: string, default: none, mandatory

The id of the attribute to be defined. This will be the key used in the json structure for the value this attribute holds.

type, type: string, default: text

The type of the attribute, as available under types/. See the next section for details.

i18n, type: boolean, default: false

If i18n is false, the value for the key is stored directly, like

{
    <id>: value
}

If i18n is true, then the value for the key will be an object with each supported locale as keys:

{
    <id>: {
        en: <value-in-en>
        de: <value-in-de>
        fr: <value-in-fr>
        ...
    }
}

Attribute Types

Each attribute type allows for certain properties to be set.

Color, number, text, textarea

These types don't take any special properties. Except for number, they all support decorations.

Image

A base64 image storage in the model. Supports decorations.

max-size, type: number, unit: kB, default: unlimited

(not implemented yet)

Since this is a base64 representation of the image, it should be possible to restrict the maximum size. The given number is interpreted as kilobytes.

Entity

entity, type: string, default: none, mandatory

The entity type of which this attribute can select its item(s) from.

reference, type: boolean, default: false

Should the whole item be copied into this item's attribute (reference==false), or should this attribute just be a reference to the other item (reference==true). Copying the item makes only sense if you want to keep the information just as it was when this item was edited; then it is irrelevant if the copied item is changed or deleted later on.

If it is a reference on the other hand then all changes will be visible in this item and the referenced item may not be deleted as long as this item exists in the database.

multi, type: boolean, default: false

If this attribute holds just one value, set multi to false, if it holds an array of items, set multi to true.

uniq, type: boolean, default: true

In case of multi==true, this determines if the same item can appear more than once (uniq==false), or if each item has to be unique.

Validation

I haven't yet spent much time thinking about validation. Just a very basic uniqValidator is implemented. Maybe I will add a validation property to an attribute...

Development

Adding new Attribute Type Definitions

Right now, each attribute type consists of a controller (written in JavaScript/LiveScript/ES6) and a Derby view. It is, in effect, a Derby component.

Controller

Each type class has to provide the following methods:

  • attribute: (item, attr, locale) ->

    This returns the item's attribute attr as plain text in the given locale.

  • renderAttribute: (item, attr, locale, parent) ->

    This returns the item's attribute attr as html in the given locale.

If the type extends the class Type, then renderAttributeData, and optionally attributeData, have to be implemented. Instead of the whole item as first argument, only its attribute is provided as data. This is just so that some common boilerplate can be avoided.

View

The component view takes a parameter mode, which can be either text, html, or edit, and that determines if the view should output plain text or html, or if the editor for that attribute should be rendered.

Consequently, each component should define three subviews: <-text:>, <-html:>, and <-edit:>. There is a leading dash to make sure that views don't get confused with components, like entity:text vs text.

A type finally has to be registered in EntitiesApi (api.ls) by adding it to types.ls.

Testing

Testing is done with mocha. There are two types of tests that are used in this project:

  • API tests running directly in Node; the standard, they are executed by the mocha command

  • Derby tests requiring the DOM; they are run in a browser (electron). Written as mocha tests, run in electron, which is started by nightmare. There is not really a server needed, browserify bundles DerbyStandalone in before() and the test.html file includes Derby as well as the components to test, which is then read locally using file:// by electron (through nightmare).

Note to self: this is a list all possible scenarios

  • mocha tests written in standard js files

    • execute in NodeJS: mocha
    • execute in electron:
      • without DOM: electron-mocha
      • with DOM: electron-mocha --renderer, iron-node _mocha
  • DOM mocha tests loaded into a html file (using mocha.js)

    • execute in phantomjs (with DOM): mocha-phantomjs index.html
    • execute in electron: no need to, use the above electron-mocha instead
    • execute manually by using any browser and opening the .html file
  • DOM tests on a webpage without mocha interface:

    • in electron: nightmare (could still be executed using mocha)
    • in phantomjs or slimerjs: casperjs
    • in any browser: webdriverio (requires selenium server)

License

MIT

Copyright (c) 2015-2016 Michael Brade