Socket
Socket
Sign inDemoInstall

@plotdb/block

Package Overview
Dependencies
11
Maintainers
1
Versions
98
Alerts
File Explorer

Advanced tools

Install Socket

Detect and block malicious and high-risk dependencies

Install

    @plotdb/block

block


Version published
Weekly downloads
31
increased by34.78%
Maintainers
1
Created
Weekly downloads
 

Changelog

Source

v5.3.0

  • support ctx option in block.manager.get and block.class constructor to share caller context with classes.

Readme

Source

@plotdb/block

Frontend module manipulation library with following features:

  • HTML-based module definition
  • scoped JS / CSS for vanilla libraries with no bundling required.
  • reuseable, extendable components

Usage

install @plotdb/block along with all necessary js libraries:

npm install @plotdb/block @plotdb/rescope @plotdb/csscope @plotdb/semver proxise

and include them:

<script src="path-to-semver/index.min.js"></script>
<script src="path-to-proxise/index.min.js"></script>
<script src="path-to-csscope/index.min.js"></script>
<script src="path-to-rescope/index.min.js"></script>
<script src="path-to-block/index.min.js"></script>

Load a sample block:

mgr = new block.manager(...);
mgr.from({name: 'block name', version: 'x.y.z', path: "index.html"})
  .then ({instance, interface}) -> ...

A sample block may look like this:

<div>
  <script type="@plotdb/block">
    module.exports = {};
  </script>
  <h1>
    hello world!
  </h1>
</div>

Concept

Similar to web component, @plotdb/block modularizes frontend codes into components called block. A block is defined with a plain HTML file, containing following 3 parts ( all parts are optional ):

  • HTML
  • CSS
  • JavaScript

This is an example of a block file:

<div>
  <h1> Hello World! </h1>
  <style> /* plain CSS ... */ </style>
  <script type="@plotdb/block"> /* plain javascript ... */ </script>
</div>

Since it's just a valid HTML file, User can use different languages ( Sass, TypeScript, Pug, etc ) and transpile when necessary, once the result file is a plain HTML file.

Following is an example with Pug, LiveScript and Stylus with additional Pug filters ( :stylus and :lsc ) which can be transpiled directly with srcbuild-pug command provided in @plotdb/srcbuild:

div
  h1 Hello World!
  style: :stylus
    h1 { color: #543; }
  script(type="@plotdb/block"): :lsc
    module.export = { init: -> console.log \loaded. };

Script can either be an object described as below, or a function returning that object. Styles will be automatically scoped and limited in this block.

Block Identifier and File Accessing

To use a block, we need to know how to identify it. Like npm modules, blocks are defined with name, version and an optional ns, where:

  • ns: Namespace, such as npm or github. How this works depends on how registry is implemented.
  • name: Block name. Use the naming convention as npm. e.g., @loadingio/spinner
  • version: Block version in semver format, or labels such as main, latest.

Additionally, files in a block are identified with path and `type fields:

  • path: path of the block definition file inside the module name@version.
    • if omitted, inferred by type field, or decided by block manager.
  • type: type of the requested file. if omitted, inferred from path, or decided by block manager.

This block identifier can either be an object or a string, such as this object:

{ns: "local", name: "@plotdb/konfig", version: "1.2.3", path: "index.html"}

with a identical string representation:

local:@plotdb/konfig@1.2.3:index.html

@plotdb/block provides two methods to convert between string and object identifier:

  • block.id(obj): return the corresponding string representation of an identifier object obj.
    • see below for more options.
  • block.id2obj(id): return the corresponding object represetnation of an string identifier id.

To access a block file identified by a block identifier, we can use block.manager:

manager = new block.manager({
  /* indicating where we can find the file */
  registry: ({ns,name,version,path}) -> "/block/#name/#version/#path/index.html"
});
mananger.init!
  .then -> manager.get({name: "my-block", version: "0.1.0"})
  .then (blockClass) -> ...

Manager, Class and Instance of Block

The resolved object from manager.get is a instance of block.class, which represents the definition of the given block file. To use it, we have to create an instance of block.instance from it, such as:

manager.get( ... )
  .then (cls) -> cls.create();
  .then (instace) -> ...

A block instance can then be injected into web page:

instance.attach({root: document.body})
  .then -> ...

Below is a simplified flow of relationship between the above concepts:

  • block file (in HTML)
  • block.manager: load, convert and cache block files.
  • block.class: Object representation of a block file.
  • block.instance: a JS Instance created from block.class

Core modules

As described above, @plotdb/block contains following basic elements:

  • block.manager - to access, register, get and cache block.class
  • block.class - representing the definition of a block, and is used to generate block.instance.
  • block.instance - object for manipulating state / DOM of a given block.

Additionally, block itself provides following functions:

  • block.id(obj) - return an ID corresponding to input object with following possible fields:

    • id: if id exists, it will be returned directly.
    • url: if id is not found abut url exists, url will be returned instead.
    • ns, name, version, path: if none of above is found, use these to generate an ID.
      • name is required in this case.
      • version default to main, path default to index.html if not provided.
  • block.id2obj(id) - reversely convert id into block with ns, name, version and path fields.

  • block.i18n

    • use(obj): use obj to replace module.
    • language: current used language.
    • changeLanguage and addResourceBundle: see below in module.
      • TODO we may want to remove these, and rely on module directly.
    • module: default i18n module object, which should support following APIs:
      • t(..): return translated text based on given input.
      • on(name, cb): listen to event name with listener cb. expected events:
        • languageChanged: fired when language changes.
      • off(name, cb): remove listener cb from event name.
      • changeLangauge(ns): set default language to ns.
      • addResourceBundle(...): add resource bundle, with following parameters ( in order ):
        • lng: ns for this resource to add.
        • id: id if any. default undefined.
        • resource: resource to add
        • deep: default true.
        • overwrite: default true. whether overwrite existing resource or not.
  • block.env(win) - set current environment to win.

block.manager

Since a block is just a plain HTML, it can be stored anywhere once a string can be stored. Common places to store a block may be:

  • local in web page: block HTMLs are served directly along the web page.
  • remote in web server: block HTMLs are stored as files and can be accessed via Ajax through specific URL.

either way we have to provide a way to load, register, cache these blocks - that is, to manage them, which can be done with the help of block.manager.

constructor options

Create a block.manager instance with

mgr = new block.manager(opt);

where the constructor options are as below:

  • registry: either function or string, tell block.manager where to find remote blocks.

    • function({ns,url,name,version,path,type}): return URL for given bid of a block.
      • should respect url or use/transofmr it if provided.
    • string: the registry base url. block.manager will look up blocks under this url with this rule:
      • /assets/block/<name>/<version>/<path>
    • object:
      • either an object with url and fetch (optional) field, or lib and block field.
        • url and fetch:
          • url({ns,url,name, ...}) will be used to transform given bid to an URL.
          • fetch({ns,url,...})) will be used to fetch url or bid ir provided.
            • if url() is provided, bid will be transformed first.
        • lib and block:
          • registry set here will be used for both block and libraries. To distinquish them, use:

            registry: {lib: (-> ...), block: (-> ...)}

          • registry.lib will be used for querying block if registry.block is omitted.

  • rescope: optional. should be a @plotdb/rescope object if provided.

    • will replace internal rescope object if provided.
  • csscope: optional. should be a @plotdb/csscope object if provied.

    • will replace internal csscope object if provided.
  • chain: optional. fallback manager for chaining block lookup if requested block is not found in current manager.

APIs

A block.manager instance provides following methods:

  • registry(v): update registry dynamically.
    • v: can be a function, string or an object, similar to the option in constructor.
  • set({name,version,path,block}): register a block with name, versionandpath`.
    • block: a block-class object, explained below.
    • set also accepts Array of {name,version,block} object for batching set.
  • getUrl({ns,name,version,path}): get url for a block corresponding to the given block identifier.
  • get({ns,name,version,path,force,ctx}): return a block-class object corresponding to the given block identifier.
    • force: by default, block.manager caches result. set force to true to force block.manager re-fetch data.
    • ctx: optional context object for providing context for the requested block class.
      • note: class context is initialized when init() is called, which means that once a class is inited, new ctx provided for manager.get won't work as expected. To re-initialize, set force to true.
    • get also accept an array of {ns,name,version,path,force} tuples for batching get.
      • in this case, get returns an array of block.class.
  • from(block-id-obj, attach-opt): shorthand for manager.get + class.create + instance.attach + instace.interface
    • return a Promise which resolves to an object {interface, instance}:
      • instance: created instance
      • interface: created interface
    • block-id-obj: block identifier object. see get() and above description.
    • attach-opt: attach options. see block.instance's attach() function.
  • chain(mgr): set a fallback manager for chaining lookup of requested block.
  • rescope: rescope object, either global one or customized one.
  • csscope: csscope object, either global one or customized one.
  • id: shortcut for block.id
  • id2obj: shortcut for block.id2obj

block.class

block.class is for generating block instances. It parses the code of a block based on the block specification and convert them into clonable code, preparing for generating block.instance objects on demand.

We usually don't have to create a block.class instance manually since block.manager does this for us, however to manually create one:

cls = new block.class( ... );
constructor options
  • manager: default block manager for this class. mandatory
  • name: block name. mandatory.
  • version: block version. mandatory.
  • path: block path. optional. index.html if omitted.
  • code: use to create DOM / style / internal object. it can be one of following:
    • a function. should return either html code or object; returned value will be parsed by corresponding rules.
    • a string, providing HTML code. structure of HTML should follow the definition of a block.
    • an object, containing dom, style and script members.
      • dom: HTML code string, or a function returning HTML code string.
      • style: should be string for CSS.
      • script: function, object or string of code, for interface of the internal object by:
        • function: return the interface.
        • object: as the interface.
        • string: evaled to the interface, or a function which return the interface.
        • for detail of the "interface", see "interface of the internal object" section below.
  • root: optional. root of a DOM tree representing the block HTML code. Overwrite code.
  • ctx: optional context object, providing additional preloaded dependencies for this class.
APIs
  • create(opt): create a block.instance based on this object. options:
    • data: instance data. defined by user and passed directly to block instance javascript.
    • root and before: parameters passed to attach.
      • instance will and only will be attached automatically if root is provided.
  • context(): get library context corresponding to this block.
  • i18n(text): return translated text based on the current context.

Additional, here are the private members:

  • name: name of this block.
  • version: version of this block.
  • path: path of this block.
  • manager: block manager to use when resolving recursive blocks.
  • dom: block DOM tree.
  • scope: unique id randomly generated each time when block.class is created mainly for scoping purpose.
  • opt: raw constructor options.
  • code: source code for constructing this block.
  • script: source code for this block's script definition.
  • style: source code for this block's style definition.
  • link: reserved for future use.
  • styleNode: node storing parsed / scoped style of this block.
  • interface: javascript interface for this block.
    • This will also be used as prototype of the instance object, created by factory method below.
  • factory: constructor for generating the js context for block script. See below.
  • id: unique name for this block.
    • "name@version/path" or randomly generated one if name and version is not available.
  • \_ctx: js context object from rescope.
  • csscopes
    • local: scope list of css for local scope.
    • global: scope list of css scope name for global scope.
  • extend: extended block, as a block.class object.
  • extendDom: to extend dom or not.
  • extends: array of extended blocks. extends[0] is the direct parent class.

To create an instance from a block.class:

instance = aBlockClass.create();

While block.class is used to create instance of block.instance, JavaScript of a block will be executed when a block class is loaded, in order to prepare for upcoming instance creation. No instance context at this time since we only have the block.class object.

To access block.instance context, block JavaScript should be implemented based on the factory interface described in the following section. This will be discussed in following section JS context of block instance.

block.instance

block.instance is an instance of block created from a block.class. It's responsible for maintaining block's state and DOM status.

constructor options
  • block: block.class for this instance.
  • name, ns, version, path: as defined in the object identifier.
  • data: custom data for this instance. usage and spec of this data is defined by the block file.
APIs
  • attach({root, before, data, autoTransform}): attach DOM and initialize this instance.
    • block instance is attahed to root before before if before is provided.
    • if a factory interface is exported by block JS, it will be used to create an internal context and be inited.
      • see Internal JS context of a block below.
    • return a Promise which resolves with a list of internal object based on inheriance hierarchy after inited.
    • when root is omitted, attach block in headless mode ( for pure script )
    • attach DOM by appendChild when before is omitted, and by insertBefore otherwise.
    • autoTransform: default null. set to i18n to enable auto i18n transformation based on i18n module event.
      • note: will be by default i18n in future release. explicitly set to null if that's what you want.
  • detach(): detach DOM. return Promise.
  • i18n(text): return translated text based on the current context.
  • path(p): return url for the given path p
  • dom(): return DOM corresponding to this block. Create a new one if not yet created.
  • run({node,type}): execute type API provided by block implementation with node as root.
  • transform(cfg): (re)transform DOM based on the given cfg option, which is:
    • string: name of the transform (e.g., i18n) to apply.
  • update(ops): (TBD) update datadom based on provided ops ( array of operational transformation ).

Additionally, following are the private members:

  • obj - list of JS internal context objects created from the exported factory interface.
    • see below for the detail of the internal context object.
    • it's a list of all objects from the inheritant chain. base block comes first.
    • each item in this list contains block's data and interface.

Internal JS Context of a block

While block.instance represents the block instance itself, block JavaScript is run in a different context to prevent intervention. The interface of this context is as below:

  • \_class: the object of block.class for this block, filled automatically when creating this context.
  • \_instance: the object of block.instance for this block, filled automatically when creating this context.
    • Note: currently this is not available in base blocks. use it only for dev / debug purpose.
  • pkg: block information
  • init(opt): initialization function of this context.
  • destroy(opt): destroy function of this context.
  • interface(): JS interface for block users to access.
  • parent: JS Context of parent block, if any.
    • use parent.interface() to reach parent interface if needed.

Except \_class and \_instance, functions in above interface should be implemented by block JavaScript and exported via module.exports:

module.exports = {
  init: (opt) ->
  interface: ->
};

This interface is used in the factory constructor of block.class to create the internal JS Context:

context = new aBlockClass.factory(instance);

which is the object stored in obj member of block.instance described in the block.instance section.

The detail of the fields of interface is as below:

  • pkg: block information, described below. optional.

  • init(opt): initializing a block. optional.

    • return a Promise for asynchronous initialization.
    • opt is an object with following fields:
      • root: root element
      • ctx: dependencies in an object.
      • context: deprecated, use ctx instead.
      • parent: object for the direct base block.
      • pubsub: for communication between block in extend chain. pubsub is an object with following methods:
        • on(event, cb(parmas)): handle event with cb callback, params from fire.
          • return value will be passed and resolved to the returned promise of fire.
        • fire(event, params): fire event`. return promise.
      • data: data passing to create. optional and up to user.
      • path(p): path transformer to convert p to a local string based on the identifier of this block.
      • t(text): translation function based on local, base class and global i18n information. shorthand of i18n.t.
      • i18n: i18n related helpers including:
        • getLanguage()`: return current used language.

        • t(text): as described above.

        • addResourceBundles(res): dynamically adding i18n resources. sample res:

          { "zh-TW": {"string", "文字"}, "en-US": {"string": "string"} }

  • destroy({root, context}): destroying a block. optional.

  • interface: for accessing custom object. optional.

    • either a function returning interface object, or the interface object itself.
    • child block always overwrite parents' interface in an inheritance chain, if available
  • mod: reserved for block javascript. future implement update of @plotdb/block should not use it.

  • exports(global): (TBD) for sharing block as a JS library. return objects to export. optional

    • user can use a block as a library by adding it in the dependencies config, such as:
      • [{name: "some-block", version: "some-vesion", path: "path-to-file"}, ...]

All members are optional thus the minimal definition will be an empty object or even undefined:

{}

Use module.exports to explicitly export the desired object:

module.exports = { .... };
Block Information

The pkg field of a block interface is defined as:

  • ns, name, version, path: from this block's identifier. optional

  • author: author name. optional

  • license: License. optional.

  • description: description of this block. optional

  • syncInit: default false.

    • if true, each init in extend chain runs only after the returned Promise of the previous method resolves.
    • otherwise, order of init methods are not guaranteed.
  • extend: optional. block identifier of block to extend.

    • ns, name, version, path: from parent block's identifier. optional
    • dom: default true. can be any of following:
      • true: use parent's DOM if set true.
      • false: completely ignore extended DOM in any ancestor.
      • "overwrite": overwrite parent DOM but extend DOM from grantparent, if any.
    • style: default true. can be any of following:
      • true: use parent's style if set true.
      • false: completely ignore extended style in any ancestor.
      • "overwrite": overwrite parent style but extend style from grantparent, if any.
    • use plug ( for html ), parent and pubsub ( js ) to work with extended block.
      • for more information about plug, see HTML Plugs section below.
  • dependencies: dependencies of this block.

    • list or modules, in case of mutual dependencies: ["some-url", {url: "some-url", async: false, dev: true, global: true, type: "css, js or block"}]
    • for now, block type dependencies are used for hint of bundling.
    • options in object notation:
      • `async: true to load this module asynchronously. true by default.
      • `global: for CSS. true if the CSS should also work in global scope. ( under body ). default false.
      • type: default js. either css or js.
        • (TBD) support block type for preloading block / export block library.
      • url: path of required module.
        • generated from name + version + path if omitted. ( TODO )
      • name: name of required module ( TODO )
      • version: version of required module ( TODO )
      • mode: use to control when this module should be loaded. ( TODO )
    • dependencies will be additive in inheritance chain.
  • i18n: i18next style i18n resource. e.g.,

    { "zh-TW": { "name": "名字" } }

Block Events

(TBD) Following are possible events:

  • before insert
  • init
  • after insert
  • before destroy
  • destroy
  • after destroy
  • update

i18n configuration

use block.i18n.use(...) to switch the core i18n module, which should at least implement following API:

  • addResourceBundle(lng, ns, resource, deep, overwrite)
  • changeLanguage(lng)
  • t(text)

These API are intentionally aligned with i18next. Check i18next documentation for more information about these API.

A sample setup with i18next and @plotdb/block:

i18next.init({supportedLng: ["en", "zh-TW"], fallbackLng: "en"})
  .then(function() { i18next.changeLanguage("zh-TW"); })
  .then(function() { block.i18n.use(i18next); })

HTML Plugs

Base block may provide slots for child block to plug. use <plug> tag with name attribute:

<plug name="layout"/>

To plug elements In child block to given slot, use plug attribute in child block:

<div plug="layout"> ... </div>

Packed Block with Bundled Packages

To ship bundled packages along with a block, simply append the corresponding <template> tag at the end of the block:

<div> ... </div><template rel="block"> ... </template>

Actually, any template tag in this block with rel attribute set to block will be considered a bundle tag for @plotdb/block.

Why block

At first we just want to make web editing easier across expertise, and block design ( see future of web design comes in blocks, Editor.js ) seems to be a trend in web design. It's similar to web components but we will have to do more for making visually editing possible.

While what @plotdb/block ( web component & management ) provides is already available in other popular frameworks, @plotdb/block is actually designed with following criteria thus makes it different with others:

  • version management
    • blocks are managed with proper versioning.
    • blocks should work even using the same lib with different versions without import.
      • popular frameworks use import which will have to bundle js within.
      • even if bundle is not necessary, many libs don't support import and will need wrapper.
  • framework agnostic
    • prevent from abducted by specific framework
    • while we seem to invent yet another framework:
      • @plotdb/block is only for components. no state management, no reactive.
      • thus, any js frameworks are expected to work well with @plotdb/block.
  • As Simple as Possible
    • making a component is extremely easy. ( KISS principle )
      • there is no new syntax to learn in @plotdb/block, only interface.
  • Collaborative
    • @plotdb/block is built along with @plotdb/datadom for DOM serialization.
      • this makes it by default suitable for serialization, thus also for collaboration
      • editing can be described by concepts such as operational transformation
  • DOM manipulating with UI ( cross expertise editing )
    • this is covered in @plotdb/editable.

Resources

  • Related modules
    • editable: cross-expertise editor interface based on a set of predefined attributes over plain HTML.
    • datadom: DOM in JSON, with extension.
    • registry: block module storage and delivery.

License

MIT

FAQs

Last updated on 26 Jan 2024

Did you know?

Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.

Install

Related posts

SocketSocket SOC 2 Logo

Product

  • Package Alerts
  • Integrations
  • Docs
  • Pricing
  • FAQ
  • Roadmap

Stay in touch

Get open source security insights delivered straight into your inbox.


  • Terms
  • Privacy
  • Security

Made with ⚡️ by Socket Inc