Purgecss-custom-extractor NPM

A tool to easily create custom extractors for purgecss.

Install

npm i purgecss-custom-extractor

Usage

Accepts regex as a RegExp or a string with it ('\w+', '/\w+/g') First argument can be a regex or an array of regex and match processor or a list with a mix of both.

const purgeCss = new Purgecss({
  content: ['**/*.html'],
  css: ['**/*.css'],
  extractors: [{
      // 'g' flag will be enforced.
      extractor: Extractor.custom(/[a-zA-Z0-9\-_]+/),
      extensions: ['html']
    }]
})

By default purgecss treats every word in text as potential selector. But what if you're using some selectors that contain a non standard characters, like in TailwindCSS framework (w-1/2, hover:bg-blue). Or your code contains a lot of static text, then you can get lots of selectors you don't really use. You can create a simple regular expression to only keep selectors that are mentioned in class attribute of html tags. If you pass an array of [regex, matchProcessor] then each match will go through that passed function.

// getting all the tags with class attribute
// taking first group in each match that contains
// class list and splitting it by ' '
Extractor.custom([ /<[\w]+.*?class="(.*?)".*?>/mg, m => m[1].split(' ') ])

If you want you can get rid of html comments  to ignore class names in commented code. To do this you can define contentProcessor and remove html comments before looking for matches.

// Extractor.regex.comment() == /<!--([\s\S]*?)-->/mg
Extractor.custom({
  regex: [ /<[\w]+.*?class="(.*?)".*?>/mg, m => m[1].split(' ') ],
  contentProcessor: c => c.replace(Extractor.regex.comment(), '')
})

You can also trim text to a content of a specific html tag. This also can be useful when working with VueJS single file components, you can isolate <template> tag and look for matches only there.

Extractor.custom({
  regex: [ /<[\w]+.*?class="(.*?)".*?>/mg, m => m[1].split(' ') ],
  contentProcessor: content => {
    // generate regexp for lazy template tag
    let regex = Extractor.regex.lazyTag('template')
    let match = regex.exec(content)
    // if match found use second group for tags content
    // for reference see api -> regex
    let res = match ? match[2] : content
    return res
  }
})

API

`custom(regex | {regex, matchProcessor, contentProcessor})`

Function to create custom extractor

First argument can be either of these:

regex: a regex, array of regex and match processor or a mixed list of both

custom(regex);
custom([regex, eachMatch]);
custom([
  [regex, eachMatch],
  [regex],
  regex
]);
custom({regex, matchProcessor, contentProcessor});

opts: object with options:
- regex: see above
- matchProcessor: optional will receive result of each match and will return processed value, can return string or array of strings (m => m[1])
- contentProcessor: optional will receive content before looking for matches (c => c.toLowerCase())
- returns: extractor object for purgecss

`matchAll(re, text, matchProcessor)`

Function to get all the matches from given string

re: RegExp or string with it ('\w+', '/\w+/g')
text: text to match in
matchProcessor: optional will receive result of each match and will return processed value, can return string or array of strings (m => m[1])
returns: array of all the matched strings

`regex`

Object containing methods to create predefined regular expressions.

simple(): returns regex for simple css selector (/[a-zA-Z0-9\-_]+/g)
extended(): returns regex for css selector with : and \ characters (/[a-zA-Z0-9\-_:\/]+/g) such css selectors are being used in some frameworks, like TailwindCSS
lazyTag(tagName): returns regex to match multiline html tag with it's content (/<div(.*?)>([\s\S]*?)<\/divs*>/mg) group 1 - attributes, group 2 - content
greedyTag(tagName): same as above but greedy for content group (/<div(.*?)>([\s\S]*)<\/divs*>/mg)
comment(): returns regex to match html comments (//mg)