Awesome
DEPRECATED
This repo is no longer maintained.
If you still need what it provided you should take a look at this fork:
https://github.com/MichaelRoosz/dr-css-inliner
dr-css-inliner
Puppeteer script to inline above-the-fold CSS on a webpage.
Inlining CSS for above-the-fold content (and loading stylesheets in a non-blocking manner) will make pages render instantly. This script will help extract the CSS needed.
As proposed by the Google Pagespeed team: Optimizing the Critical Rendering Path for Instant Mobile Websites - Velocity SC - 2013
How it works
There are two ways of processing a webpage; loaded via the url
argument or piped in through stdin
. When using stdin
it is required that you use the --fake-url
option in conjunction.
Once Puppeteer has loaded the page all stylesheets with no media
set or with media
set to screen
or all
are loaded again through XHR to avoid browser engine bias when parsing CSS.
The CSS is inlined as per the supplied options and all stylesheets and style elements stripped from the webpage html. You can opt to expose the stripped stylesheets as an array in a script tag through the -e
(--expose-stylesheets
) option.
Install
npm install dr-css-inliner -g
Usage:
dr-css-inliner <url> [options]
Options:
-o, --output [string]
- Write the output to a file. If omitted the output is written tostdout
.-w, --width [value]
- Determines the width of the viewport. Defaults to 1200.-h, --height [value]
- Determines the above-the-fold height. Defaults to the actual document height.-m, --match-media-queries
- Omit media queries that don't match the defined width.-r, --required-selectors [string]
- Force inclusion of required selectors in the form of a comma-separated selector string or an array (as a JSON string) of regexp strings (remember to escape.
,[
and]
etc). Defaults to no required selectors.-s, --strip-resources [string]
- Avoid loading resources (while extracting CSS) matching the string or array (as a JSON string) of strings turned into regexp pattern(s). Used to speed up execution of CSS inlining. Default is no stripping of resources. Warning: when stripping is used caching is not possible.-c, --css-only
- Output the raw required CSS without wrapping it in HTML.-e, --expose-stylesheets [string]
- A variable name (or property on a preexisting variable) to expose an array containing information about the stripped stylesheets in an inline script tag.-t, --insertion-token [string]
- A token (preferably an HTML comment) to control the exact insertion point of the inlined CSS. If omited default insertion is at the first encountered stylesheet.-i, --css-id [string]
- Determines the id attribute of the inline style tag. By default no id is added.-f, --fake-url [url]
- Defines a fake url context. Required when piping in html throughstdin
. Default is null.-dcd, --disk-cache-dir [path]
- Redirect the chroumium cache folder to this path.-u, --user-agent [string]
- Set the user agent.-ihe, --ignore-https-errors
- Ignore HTTPS errors (for example invalid certificates).-b, --browser-timeout [value]
- Set the browser timeout in ms. Defaults to 30000 (30 seconds).-d, --debug
- Prints out an HTML comment in the bottom of the output that exposes some info:time
- The time in ms it took to run the script (not including the puppeteer process itself).loadTime
- The time in ms it took to load the webpage.processingTime
- The time in ms it took to process and return the CSS in the webpage.requests
- An array of urls of all requests made by the webpage. Useful for spotting resources to strip.stripped
- An array of urls of requests aborted by the--strip-resources
option.errors
- An array of errors that ocurred on the page.cssLength
- The length of the inlined CSS in chars.
Examples:
CSS options
Only inline the needed above-the-fold CSS for smaller devices:
dr-css-inliner http://www.mydomain.com/index.html -w 350 -h 480 -m -o index-mobile.html
Inline all needed CSS for the above-the-fold content on all devices (default 1200px and smaller):
dr-css-inliner http://www.mydomain.com/index.html -h 800 -o index-page-top.html
Inline all needed CSS for webpage:
dr-css-inliner http://www.mydomain.com/index.html -o index-full-page.html
Inline all needed CSS for webpage with extra required selectors:
dr-css-inliner http://www.mydomain.com/index.html -r ".foo > .bar, #myId" -o index-full-page.html
Inline all needed CSS for webpage with extra required regexp selector filters:
dr-css-inliner http://www.mydomain.com/index.html -r '["\\.foo > ", "\\.span-\\d+"]' -o index-full-page.html
Output options
The examples listed below use the following index.css
and index.html
samples (unless specified otherwise):
index.css:
.foo {
color: #BADA55;
}
.bar {
color: goldenrod;
}
index.html:
<!doctype html>
<html>
<head>
<title>Foo</title>
<link href="index.css" rel="stylesheet" media="screen" />
<link href="print.css" rel="stylesheet" media="print" />
</head>
<body>
<h1 class="foo">Inlining CSS is in</h1>
</body>
</html>
Doing:
dr-css-inliner index.html
...would get you:
<!doctype html>
<html>
<head>
<title>Foo</title>
<style>
.foo {
color: #BADA55;
}
</style>
</head>
<body>
<h1 class="foo">Inlining CSS is in</h1>
</body>
</html>
Only output CSS
-c, --css-only
dr-css-inliner index.html -c
...would get you:
.foo {
color: #BADA55;
}
Exposing the stripped stylesheets for later consumption
-e, --expose-stylesheets [string]
Single global variable:
dr-css-inliner index.html -e stylesheets
...would get you:
<!doctype html>
<html>
<head>
<title>Foo</title>
<style>
.foo {
color: #BADA55;
}
</style>
<script>
var stylesheets = [{url: "index.css", media: "screen"}, {url: "print.css", media: "print"}];
</script>
</head>
<body>
<h1 class="foo">Inlining CSS is in</h1>
</body>
</html>
Namespaced property:
dr-css-inliner index.html -e myNamespace.stylesheets
provided you had an index.html
like:
<!doctype html>
<html>
<head>
<title>Foo</title>
<script>
var myNamespace = {};
</script>
<link href="index.css" rel="stylesheet" media="screen" />
<link href="print.css" rel="stylesheet" media="print" />
</head>
<body>
<h1 class="foo">Inlining CSS is in</h1>
</body>
</html>
...would get you:
<!doctype html>
<html>
<head>
<title>Foo</title>
<script>
var myNamespace = {};
</script>
<style>
.foo {
color: #BADA55;
}
</style>
<script>
myNamespace.stylesheets = [{url: "index.css", media: "screen"}, {url: "print.css", media: "print"}];
</script>
</head>
<body>
<h1 class="foo">Inlining CSS is in</h1>
</body>
</html>
Controlling where to insert the inlined CSS
-t, --insertion-token [string]
provided you had an index.html
like:
<!doctype html>
<html>
<head>
<title>Foo</title>
<!-- CSS goes here -->
<script>
var myNamespace = {};
</script>
<link href="index.css" rel="stylesheet" media="screen" />
<link href="print.css" rel="stylesheet" media="print" />
</head>
<body>
<h1 class="foo">Inlining CSS is in</h1>
</body>
</html>
dr-css-inliner index.html -t "<!-- CSS goes here -->"
...would get you:
<!doctype html>
<html>
<head>
<title>Foo</title>
<style>
.foo {
color: #BADA55;
}
</style>
<script>
var myNamespace = {};
</script>
</head>
<body>
<h1 class="foo">Inlining CSS is in</h1>
</body>
</html>
Avoid loading unneeded resources
-s, --strip-resources [string]
Doing:
dr-css-inliner index.html -s '["\\.(jpg|gif|png)$","webstat\\.js$"]'
... would avoid loading images and a given web statistic script.
Debug info
-d, --debug
Doing:
dr-css-inliner index.html -d
...would get you:
<!doctype html>
<html>
<head>
<title>Foo</title>
<style>
.foo {
color: #BADA55;
}
</style>
</head>
<body>
<h1 class="foo">Inlining CSS is in</h1>
</body>
</html>
<!--
{"time":300,"loadTime":155,"processingTime":145,"requests":[...],"stripped":[...],"cssLength":5050}
-->
Adding an id to the inlined style tag
-i, --css-id [string]
Doing:
dr-css-inliner index.html -i my-inline-css
...would get you:
<!doctype html>
<html>
<head>
<title>Foo</title>
<style id="my-inline-css">
.foo {
color: #BADA55;
}
</style>
</head>
<body>
<h1 class="foo">Inlining CSS is in</h1>
</body>
</html>
Piping in HTML content through stdin
-f, --fake-url [string]
If you need to parse HTML that is not yet publicly available you can pipe it into dr-css-inliner
. Below is a contrived example (in a real-world example imagine an httpfilter or similar in place of cat
):
cat not-yet-public.html | dr-css-inliner -f http://www.mydomain.com/index.html
All loading of assets will be loaded relative to the fake url - meaning they need to be available already.
Changelog
next
- Ported to Puppeteer
- Fixed first css selector being ignored when the css file starts with a @charset declaration.
bin
config added to package.json.-ihe, --ignore-https-errors
option added.-b, --browser-timeout
option added.-dcd, --disk-cache-dir [path]
option added.-u, --user-agent [string]
option added.
0.6.0
Features:
-o, --output
option added.
Changes:
- Only errors in the inliner will halt execution. Remote script errors are ignored - but logged in
debug.errors
. - the
-x, --allow-cross-domain
option is deprecated and cross domain requests allowed by default. debug.requests
is now populated by default regardless of the--strip-resources
option.
0.5.4
css-inliner
bin removed due to not working on unix.