Extract rulesets that are language specific into their own file so they can be served at runtime in a more optimal way.

The AST is walked only once and language specific rulesets are collected along the way. After processing, the language specific selectors can be written out to their own files or re-combined in various ways according to how you desire to send the locale specific CSS over the wire.

If a single selector has more than one lang or is combined with non-lang-specific selectors, then the ruleset is duplicated and the language specific selectors are removed.

langOptimizer = require("postcss-lang-optimizer");
postcss = require("postcss");

var optimizerOptions = {};
var input = ".foo { font-size: 14px; } .foo:lang(zh) { font-size: 18px; }"
var result = postcss(langOptimizer(optimizerOptions)).process(input);
// language agnostic CSS stripped of all language specific rulesets
result.css; // => ".foo { font-size: 14px; }"
// zh-specific CSS
langOptimizer.extract(result, "zh"); // => " .foo:lang(zh) { font-size: 18px; }"

You can disable and re-enable the language optimizer for portions of the CSS file by using control comments. When these comments are encountered, it will change the enabled state of the optimizer until the end of the file or until another control comment is encountered.

/* lang-optimizer: disable */

/* lang-optimizer: enable */

• subtags - When set to true, then any language selectors will be extracted to their own file. That is de, de-DE and de-DE-1996 would each be extracted to their own file.

• Wildcard language matching E.g. :lang(*-CH) - These will be left in the base file.

• Sourcemaps will not work with language specific output files.

There is an ember addon for this plugin. You can find it in the ember-addon branch.