MCSS is a build time development extension that provides class selector intellisense completions and obfuscation support for mithril projects bundled with Rollup. The plugin provides fugazi methods on mithril's m export.
This tool is developed for use with mithril.js and it cannot be appropriated into different frameworks. If you don't use mithril, then you should consider it for your next project.
- Install
- How it works
- Plugins
- Rollup Config
- Typings
- Obfuscation
- Usage
- Selector Example
- Unknown Selectors
- Caveats
pnpm add @brixtol/mcss -D
Because pnpm is dope and does dope shit
MCSS leverages Rollup and PostCSS. The module will parse your stylesheets and generate a type declaration file containing all defined selector (class names). Using Rollup's plugin API it walks the generated acorn AST, intercepting the mcss fugazi instance methods in JavaScript/TypeScript files replacing them with valid hyperscript equivalents.
Curried Fugazi`
HTML Tag names are exposed atop of the m export. Selectors (class names) can be provided with curried expression. This is however optional, you can omit selectors and pass mithril attrs, vnodes (children), text value or combination of all 3.
// Text value
m.div('hello world') // => m('div', 'hello world')
// Selector value (completions are not supported)
m.p('.class', 'hello world') // => m('p.class', 'hello world')
// Attributes with vnodes
m.p({ id: '1' }, [ m.span('bar', 'hello') ]) // => m('p', { id: '1' }, [ m('span.bar', 'hello') ])
// Curried with selectors
// => m('.foo.bar.baz', { onclick: e => {} }, m('ul', [ m('li', 'one'), m('li.active', 'two') ]))
m.div(
'foo',
'bar',
'baz'
)(
{
onclick: e => {}
},
m.ul(
[
m.li('one'), // => m('li', 'one')
m.li('active')('two'), // => m('li.active', 'two')
]
)
);Selector Fugazi
Supplies a css method atop of an m export. The transformed value will be a dot . separated hyperscript selector.
m(m.css.div('foo', 'bar'), 'one'); // m('.foo.bar', 'one')
m(m.css.ul('foo', 'bar'), 'two'); // m('ul.foo.bar', 'two')Mithril Fugazi is a glorified rollup and postcss plugin combination. In order for one to leverage it you need to couple it with a rollup plugin that exposes a method for postcss. It does not matter what plugin as long as you can hook into postcss, below are the two most common:
import mcss from '@brixtol/mcss';
import postcss from 'rollup-plugin-postcss'
export default {
input: 'src/index.js',
output: {
dir: 'dist',
format: 'es',
},
plugins: [
mcss(),
postcss({ plugins: [ mcss.postcss() ])
]
};The plugin will generate a fugazi.d.ts declaration file populated with your css class selectors that have been defined within stylesheets. The generated declaration will extend mithril's definitely typed mithril.d.ts Static interface and implement a fugazi methods on the m export. The fugazi.d.ts file will be automatically be updated when you define or delete selectors within stylesheets.
Enable esModuleInterop or allowSyntheticDefaultImports options to import mithril's commonjs export format in your tsconfig.json or jsconfig.json file. Some projects may need to explicitly include the location of your src files and the fugazi.d.ts declaration file. You can customize the location where the declaration is written. The below config should cover all bases:
{
"include": ["src", "types/mcss"],
"compilerOptions": {
"module": "ES2020",
"target": "ESNext",
"lib": ["ES2020", "DOM"],
"moduleResolution": "Node",
"allowSyntheticDefaultImports": true,
"esModuleInterop": true
}
}Refer to the @types/mithril for more information and if you are not getting completions, restart the TypeScript language server.
Class names can be obfuscated. Obfuscation can help reduce bundle size because it renames classes to single letter and/or number combinations. This is very helpful when you are working on a large project as it reduces both the production size of your stylesheet and javascript files. By default, obfuscation is disabled, you will need to pass a boolean true value to enable, do this when bundling for production.
Your css:
.some-class-name {}
.another-class-name {}
.yet-another-class-name {}Your mithril node:
m('.some-class-name.another-class-name.yet-another-class-name');Your css:
.a {} // .some-class-name
.b {} // .another-class-name
.c {} // .yet-another-class-nameYour mithril node:
m('.a.b.c');Below is an example using rollup-plugin-scss which allows us to pass transpiled code to postCSS via its processor hook option. If you are not processing SASS/SCSS files then you can use the rollup-plugin-postcss plugin and add the mcss.postcss() method to its plugins[] field. The mcss() default export should be placed within rollup plugins[] and its the function which you will use to define options. The mcss.postcss() does not accept any options, it's just a method.
The plugin provides detailed JSDoc annotated descriptions of all options.
import mcss from '@brixtol/mcss';
import scss from 'rollup-plugin-scss';
import sass from 'node-sass';
import postcss from 'postcss'
import clean from 'postcss-clean'
import autoprefixer from 'node-autoprefixer';
import { terser } from 'rollup-plugin-terser';
export default {
input: 'src/index.js',
output: {
dir: 'dist',
format: 'es',
sourcemap: false,
plugins: [
terser()
]
},
plugins: [
mcss({
// Files to include
include: [],
// Files to exclude
exclude: [],
// Clear the cache and generated types before new builds
clear: false,
// Prints warnings about unknown selectors.
warnUnknown: true,
// Generate sourcemaps
sourcemap: true,
// Minify styles using clean-css
minify: false,
// When minify is true, pass output options to cleancss
cleancss: {},
// When true, obfuscation is applied (defaults to false)
obfuscate: false,
// Where cached mappings are stored.
cacheDir: 'node_modules/.cache/mcss',
// Where the generated declaration file is written.
declaration: 'types/mcss/selectors.d.ts',
// The alphabet used to generate the new class names
alphabet: 'abcefghijklmnopqrstuvwxyz0123456789',
// A list of class names to ignore.
ignore: []
}),
scss({
sass,
output: 'dist/style.min.css',
watch: 'src',
processor: () => postcss([
autoprefixer(),
mcss.postcss()
])
})
]
};The plugin allows selectors to be expressed as if they apart of the mithril API and are available as a method of the m default export. The generates type declarations which extend the mithril module will give the impression that mithril provides this method natively (FYI: It doesn't, it's a fugazi).
It's important to note that the tag name, which can normally be omitted in hyperScript is required when using mcss
Below is an example using curried fugazi selector type:
import m from 'mithril';
m.div(
'row',
'jc-center'
)(
m.div(
'col',
'col-sm-6',
'col-lg-12'
)(
m.ul('list')([
m.li('list-item')('one'),
m.li('list-item')('two'),
m.li('list-item')({ class: m.class('foo', 'bar') } , 'three'),
m.li('list-item')({
onclick: (e) => e.target.className = 1 > 0 ? m.class(
'new-list-item-class',
'another-class-in-your-stylesheet'
) : m.class(
'list-item',
'foo',
'bar'
)
}, 'four')
])
)
)Rollup will convert all instances of the fugazi m.*()() selector when processing the output. Every selector matching such schema when encountered in chunks will be swapped out and replaced with a valid hyperscript equivalent. This a fast process because we apply these transformation directly on the AST. The above example would output the following:
m(
"div.row.jc-center"
m(
"div.col",
m("ul.list",
m("li.list-item", "one"),
m("li.list-item", "two"),
m("li.list-item", { class: 'foo bar' }, "three"),
m("li.list-item"), {
onclick: (e) => e.target.className = 1 > 0
? "new-list-item-class another-class-in-your-stylesheet"
: "list-item foo bar"
)
}, "four")
)
)
);The m.*()() selector tag we are using is just sugar and it holds not value outside of the development process. It provides us the capabilities of selector completion and a nice way to express selector class names. When obfuscate is enabled all class names defined in stylesheets and all the selectors of m.* will be replaced with a short-name variation that is auto-generated according to the alphabet sequence provided.
The above code would output something like this:
m(
"div.a.b"
m(
"div.c.d.e",
m("ul.f",
m("li.g", "one"),
m("li.g", "two"),
m("li.g", { class: 'h i' }, "three"),
m("li.g"), {
onclick: (e) => {
e.target.className = 1 > 0
? "j k"
: "g h i"
)
}
}, "four")
)
)
);In the above examples you will notice in onclick and class that we are using m.css.class() to define our selectors. The m.css.class() method will transform selectors and separate them with a single whitespace character. All other methods of m.css will be transform using dot . separators.
m.class('foo', 'bar'); // => 'foo bar'
m.css('foo', 'bar'); // => '.foo.bar'When an unknown selector is encountered that has no be defined in your stylesheets a warning will be printed to the console along with a list all the selector class name missing. This indicated thr selector being used in is undefined in your stylesheet. You can suppress this warning by setting the warnUnknown option to false or alternatively you can add those selectors to the ignore[] option.
The module assumes that all selectors expressed on the
m.*method are css class names defined in stylesheets.
There are a couple of very minor caveats to this approach. When you are writing hyperscript selectors using the fugazi approach.
Because the module will extend upon the m export using the generated global declaration it is not possible to generate typings for multiple stylesheets. If you have multiple stylesheets then mcss will replace selectors based on stylesheet.
import './vendor.scss'
import './styles.scss'All selectors contained within vendor.scss and styles.scss are combined. This logic also applies to multiple entries, wherein multiple inputs that output different bundles that import different stylesheets will be combined. In cases where are leveraging rollup to produce multiple bundles, for every new bundle the selectors declaration is overwritten.
Basically, to avoid any of this until this project matures itself only ever import a single stylesheet from a single input entry, eg:
import './styles.scss'
import m from 'mithril'
import Component from '..'
export default m.mount(document.body, Component)The module will create couple of cache files after reading stylesheets. These files contain JSON references of your class name which are used when generating types declarations and obfuscating selectors. Because the plugin is a Rollup integrated solution coupling PostCSS via third parties the cache files which get generated are processed as assets and their output is handled separate from chunk generation.
When operating in watch mode this is not an issue as fugazi will execute a rebuild upon map generation after PostCSS has processed styles. When operating in build mode (ie: generating production bundles) and if obfuscation is enabled then you will need to generate a class name obfuscation mapping. The problem here that that can only be done in the post-build cycle. In situations where an obfuscation cache mapping does not exists, you will need to execute a rebuild of the project a second time. This can easily be avoided by using fugazi in dev mode before bundling for production.
Don't be hero, no one likes hero's.
The module is not intelligent enough to understand dynamic selectors, ie: selectors which are defined as a variable or data reference value. The class names you provide to the m.*() method need to be string values wrapped in quotation characters, for example:
import m from 'mithril';
const selector = 'my-button';
// DO NOT DO THIS
m.div('foo', 'bar', selector)();Similar not supporting dynamic selectors, you cannot perform actions within the m.*() selector. This means that conditional expressions will not work, for example:
import m from 'mithril';
// DO NOT DO THIS
m.div(i > 1 ? 'foo' : 'bar'));
MIT