node package manager


CSSO (CSS Optimizer) is a CSS minifier with structural optimisations

NPM version Build Status Coverage Status NPM Downloads Twitter

CSSO (CSS Optimizer) is a CSS minifier. It performs three sort of transformations: cleaning (removing redundant), compression (replacement for shorter form) and restructuring (merge of declarations, rulesets and so on). As a result your CSS becomes much smaller.

Originated by Yandex Sponsored by Avito


npm install -g csso

Or try out CSSO right in your browser (web interface).


Command line

csso [input] [output] [options]
      --comments <value>    Comments to keep: exclamation (default), first-exclamation or none
      --debug [level]       Output intermediate state of CSS during compression
  -h, --help                Output usage information
  -i, --input <filename>    Input file
      --input-map <source>  Input source map: none, auto (default) or <filename>
  -m, --map <destination>   Generate source map: none (default), inline, file or <filename>
  -o, --output <filename>   Output file (result outputs to stdout if not set)
      --restructure-off     Turns structure minimization off
      --stat                Output statistics in stderr
  -u, --usage <filenane>    Usage data file
  -v, --version             Output version

Some examples:

> csso in.css
...output result in stdout...
> csso in.css --output out.css
> echo '.test { color: #ff0000; }' | csso
> cat source1.css source2.css | csso | gzip -9 -c > production.css.gz

Source maps

Source map doesn't generate by default. To generate map use --map CLI option, that can be:

  • none (default) – don't generate source map
  • inline – add source map into result CSS (via /*# sourceMappingURL=application/json;base64,... */)
  • file – write source map into file with same name as output file, but with .map extension (in this case --output option is required)
  • any other values treat as filename for generated source map


> csso my.css --map inline
> csso my.css --output my.min.css --map file
> csso my.css --output my.min.css --map maps/

Use --input-map option to specify input source map if needed. Possible values for option:

  • auto (default) - attempt to fetch input source map by follow steps:
    • try to fetch inline map from input
    • try to fetch source map filename from input and read its content
    • (when --input is specified) check file with same name as input file but with .map extension exists and read its content
  • none - don't use input source map; actually it's using to disable auto-fetching
  • any other values treat as filename for input source map

Generally you shouldn't care about input source map since defaults behaviour (auto) covers most use cases.

NOTE: Input source map is using only if output source map is generating.

Usage data

CSSO can use data about how CSS is using for better compression. File with this data (JSON format) can be set using --usage option. Usage data may contain follow sections:

  • tags – white list of tags
  • ids – white list of ids
  • classes – white list of classes
  • scopes – groups of classes which never used with classes from other groups on single element

All sections are optional. Value of tags, ids and classes should be array of strings, value of scopes should be an array of arrays of strings. Other values are ignoring.

Selector filtering

tags, ids and classes are using on clean stage to filter selectors that contains something that not in list. Selectors are filtering only by those kind of simple selector which white list is specified. For example, if only tags list is specified then type selectors are checking, and if selector hasn't any type selector (or even any type selector) it isn't filter.

ids and classes names are case sensitive, tags – is not.

Input CSS:

* { color: green; }
ulolli { color: blue; } { color: red; }

Usage data:

    "tags": ["ul", "LI"]

Result CSS:



Scopes is designed for CSS scope isolation solutions such as css-modules. Scopes are similar to namespaces and defines lists of class names that exclusively used on some markup. This information allows the optimizer to move rulesets more agressive. Since it assumes selectors from different scopes can't to be matched on the same element. That leads to better ruleset merging.

Suppose we have a file:

.module1-foo { color: red; }
.module1-bar { font-size: 1.5em; background: yellow; }
.module2-baz { color: red; }
.module2-qux { font-size: 1.5em; background: yellow; width: 50px; }

It can be assumed that first two rules never used with second two on the same markup. But we can't know that for sure without markup. The optimizer doesn't know it eather and will perform safe transformations only. The result will be the same as input but with no spaces and some semicolons:


But with usage data CSSO can get better output. If follow usage data is provided:

    "scopes": [
        ["module1-foo", "module1-bar"],
        ["module2-baz", "module2-qux"]

New result (29 bytes extra saving):


If class name doesn't specified in scopes it belongs to default "scope". scopes doesn't affect classes. If class name presents in scopes but missed in classes (both sections specified) it will be filtered.

Note that class name can't be specified in several scopes. Also selector can't has classes from different scopes. In both cases an exception throws.

Currently the optimizer doesn't care about out-of-bounds selectors order changing safety (i.e. selectors that may be matched to elements with no class name of scope, e.g. .scope div or .scope ~ :last-child) since assumes scoped CSS modules doesn't relay on it's order. It may be fix in future if to be an issue.


var csso = require('csso');
var compressedCss = csso.minify('.test { color: #ff0000; }').css;
// .test{color:red} 

You may minify CSS by yourself step by step:

var ast = csso.parse('.test { color: #ff0000; }');
var compressResult = csso.compress(ast);
var compressedCss = csso.translate(compressResult.ast);
// .test{color:red} 

Working with source maps:

var css = fs.readFileSync('path/to/my.css', 'utf8');
var result = csso.minify(css, {
  filename: 'path/to/my.css', // will be added to source map as reference to source file 
  sourceMap: true             // generate source map 
// { css: '...minified...', map: SourceMapGenerator {} } 
// '{ .. source map content .. }' 

minify(source[, options])

Minify source CSS passed as String.


  • sourceMap Boolean - generate source map if true
  • filename String - filename of input, uses for source map
  • debug Boolean - output debug information to stderr
  • beforeCompress function|array<function> - called right after parse is run. Callbacks arguments are ast, options.
  • afterCompress function|array<function> - called right after compress is run. Callbacks arguments are compressResult, options.
  • other options are the same as for compress()

Returns an object with properties:

  • css String – resulting CSS
  • map Object – instance of SourceMapGenerator or null
var result = csso.minify('.test { color: #ff0000; }', {
    restructure: false,   // don't change CSS structure, i.e. don't merge declarations, rulesets etc 
    debug: true           // show additional debug information: 
                          // true or number from 1 to 3 (greater number - more details) 
// > .test{color:red} 

minifyBlock(source[, options])

The same as minify() but for style block. Usualy it's a style attribute content.

var result = csso.minifyBlock('color: rgba(255, 0, 0, 1); color: #ff0000').css;
// > color:red 

parse(source[, options])

Parse CSS to AST.

NOTE: Currenly parser omit redundant separators, spaces and comments (except exclamation comments, i.e. /*! comment */) on AST build, since those things are removing by compressor anyway.


  • context String – parsing context, useful when some part of CSS is parsing (see below)
  • positions Boolean – should AST contains node position or not, store data in info property of nodes (false by default)
  • filename String – filename of source that adds to info when positions is true, uses for source map generation (<unknown> by default)
  • line Number – initial line number, useful when parse fragment of CSS to compute correct positions
  • column Number – initial column number, useful when parse fragment of CSS to compute correct positions


  • stylesheet (default) – regular stylesheet, should be suitable in most cases
  • atrule – at-rule (e.g. @media screen, print { ... })
  • atruleExpression – at-rule expression (screen, print for example above)
  • ruleset – rule (e.g. .foo, .bar:hover { color: red; border: 1px solid black; })
  • selector – selector group (.foo, .bar:hover for ruleset example)
  • simpleSelector – selector (.foo or .bar:hover for ruleset example)
  • block – block content w/o curly braces (color: red; border: 1px solid black; for ruleset example)
  • declaration – declaration (color: red or border: 1px solid black for ruleset example)
  • value – declaration value (red or 1px solid black for ruleset example)
// simple parsing with no options 
var ast = csso.parse('.example { color: red }');
// parse with options 
var ast = csso.parse('', {
    context: 'simpleSelector',
    positions: true

compress(ast[, options])

Does the main task – compress AST.

NOTE: compress performs AST compression by transforming input AST by default (since AST cloning is expensive and needed in rare cases). Use clone option with truthy value in case you want to keep input AST untouched.


  • restructure Boolean – do the structure optimisations or not (true by default)
  • clone Boolean - transform a copy of input AST if true, useful in case of AST reuse (false by default)
  • comments String or Boolean – specify what comments to left
    • 'exclamation' or true (default) – left all exclamation comments (i.e. /*! .. */)
    • 'first-exclamation' – remove every comments except first one
    • false – remove every comments
  • usage Object - usage data for advanced optimisations (see Usage data for details)
  • logger Function - function to track every step of transformations


Make an AST node deep copy.

var orig = csso.parse('.test { color: red }');
var copy = csso.clone(orig);
csso.walk(copy, function(node) {
    if (node.type === 'Class') { = 'replaced';
// .test{color:red} 
// .replaced{color:red} 


Converts AST to string.

var ast = csso.parse('.test { color: red }');
// > .test{color:red} 


The same as translate() but also generates source map (nodes should contain positions in info property).

var ast = csso.parse('.test { color: red }', {
    filename: 'my.css',
    positions: true
// { css: '.test{color:red}', map: SourceMapGenerator {} } 

walk(ast, handler)

Visit all nodes of AST and call handler for each one. handler receives three arguments:

  • node – current AST node
  • item – node wrapper when node is a list member; this wrapper contains references to prev and next nodes in list
  • list – reference to list when node is a list member; it's useful for operations on list like remove() or insert()

Context for handler an object, that contains references to some parent nodes:

  • root – refers to ast or root node
  • stylesheet – refers to closest StyleSheet node, it may be a top-level or at-rule block stylesheet
  • atruleExpression – refers to AtruleExpression node if current node inside at-rule expression
  • ruleset – refers to Ruleset node if current node inside a ruleset
  • selector – refers to Selector node if current node inside a selector
  • declaration – refers to Declaration node if current node inside a declaration
  • function – refers to closest Function or FunctionalPseudo node if current node inside one of them
// collect all urls in declarations 
var csso = require('./lib/index.js');
var urls = [];
var ast = csso.parse(`
  @import url(import.css);
  .foo { background: url('foo.jpg'); }
  .bar { background-image: url(bar.png); }
csso.walk(ast, function(node) {
    if (this.declaration !== null && node.type === 'Url') {
        var value = node.value;
        if (value.type === 'Raw') {
        } else {
            urls.push(value.value.substr(1, value.value.length - 2));
// [ 'foo.jpg', 'bar.png' ] 

walkRules(ast, handler)

Same as walk() but visits Ruleset and Atrule nodes only.

walkRulesRight(ast, handler)

Same as walkRules() but visits nodes in reverse order (from last to first).

More reading