sw-precache
Advanced tools
Service Worker Precache is a module for generating a service worker that precaches resources. It integrates with your build process. Once configured, it detects all your static resources (HTML, JavaScript, CSS, images, etc.) and generates a hash of each file's contents. Information about each file's URL and versioned hash are stored in the generated service worker file, along with logic to serve those files cache-first, and automatically keep those files up to date when changes are detected in subsequent builds.
Serving your local static resources cache-first means that you can get all the crucial scaffolding for your web app—your App Shell—on the screen without having to wait for any network responses.
The module can be used in JavaScript-based build scripts,
like those written with gulp, and it also provides a
command-line interface. You can use the module
directly, or if you'd prefer, use one of the wrappers
around sw-precache for specific build environments, like
webpack.
It can be used alongside the sw-toolbox
library, which works well when following the App Shell + dynamic content model.
The full documentation is in this README, and the getting started guide provides a quicker jumping off point.
To learn more about the internals of the generated service worker, you can read this deep-dive by Huang Xuan.
Local build integration:
$ npm install --save-dev sw-precache
Global command-line interface:
$ npm install --global sw-precache
Make sure your site is served using HTTPS!
Service worker functionality is only available on pages that are accessed via HTTPS.
(http://localhost will also work, to facilitate testing.) The rationale for this restriction is
outlined in the
"Prefer Secure Origins For Powerful New Features" document.
Incorporate sw-precache into your node-based build script. It should
work well with either gulp or Grunt, or other build scripts that run on
node. In fact, we've provided examples of both in the demo/ directory. Each
build script in demo has a function called writeServiceWorkerFile() that
shows how to use the API. Both scripts generate fully-functional JavaScript code
that takes care of precaching and fetching all the resources your site needs to
function offline. There is also a command-line interface
available, for those using alternate build setups.
Register the service worker JavaScript. The JavaScript that's generated
needs to be registered as the controlling service worker for your pages. This
technically only needs to be done from within a top-level "entry" page for your
site, since the registration includes a scope
which will apply to all pages underneath your top-level page. service-worker-registration.js is a sample
script that illustrates the best practices for registering the generated service
worker and handling the various lifecycle events.
The project's sample gulpfile.js illustrates the full use of sw-precache
in context. (Note that the sample gulpfile.js is the one in the demo folder,
not the one in the root of the project.) You can run the sample by cloning this
repo, using npm install to pull in the
dependencies, changing to the demo/ directory, running `npm bin`/gulp serve-dist, and
then visiting http://localhost:3000.
There's also a sample Gruntfile.js that shows service worker generation in
Grunt. Though, it doesn't run a server on localhost.
Here's a simpler gulp example for a basic use case. It assumes your site's resources are located under
app and that you'd like to cache all your JavaScript, HTML, CSS, and image files.
gulp.task('generate-service-worker', function(callback) {
var swPrecache = require('sw-precache');
var rootDir = 'app';
swPrecache.write(`${rootDir}/service-worker.js`, {
staticFileGlobs: [rootDir + '/**/*.{js,html,css,png,jpg,gif,svg,eot,ttf,woff}'],
stripPrefix: rootDir
}, callback);
});
This task will create app/service-worker.js, which your client pages need to
register before it can take control of your site's
pages. service-worker-registration.js is a ready-to-
use script to handle registration.
Service worker caching should be considered a progressive enhancement. If you follow the model of
conditionally registering a service worker only if it's supported (determined by
if('serviceWorker' in navigator)), you'll get offline support on browsers with service workers and
on browsers that don't support service workers, the offline-specific code will never be called.
There's no overhead/breakage for older browsers if you add sw-precache to your build.
All resources that are precached will be fetched by a service worker running in a separate
thread as soon as the service worker is installed. You should be judicious in what you list in the
dynamicUrlToDependencies and staticFileGlobs options, since listing files that are non-essential
(large images that are not shown on every page, for instance) will result in browsers downloading
more data than is strictly necessary.
Precaching doesn't make sense for all types of resources (see the previous
point). Other caching strategies, like those outlined in the Offline Cookbook, can be used in
conjunction with sw-precache to provide the best experience for your users. If
you do implement additional caching logic, put the code in a separate JavaScript
file and include it using the importScripts() method.
sw-precache uses a cache-first strategy, which results in a copy of
any cached content being returned without consulting the network. A useful
pattern to adopt with this strategy is to display a toast/alert to your users
when there's new content available, and give them an opportunity to reload the
page to pick up that new content (which the service worker will have added to
the cache, and will be available at the next page load). The sample service-worker-registration.js file illustrates
the service worker lifecycle event you can listen for to trigger this message.
For those who would prefer not to use sw-precache as part of a gulp or
Grunt build, there's a command-line interface which supports the
options listed in the API, provided via flags or an
external JavaScript configuration file.
Hypenated flags are converted to camelCase options.
Options starting with --no prefix negate the boolean value. For example, --no-clients-claim sets the value of clientsClaim to false.
Warning: When using sw-precache "by hand", outside of an automated build process, it's your
responsibility to re-run the command each time there's a change to any local resources! If sw-precache
is not run again, the previously cached local resources will be reused indefinitely.
Sensible defaults are assumed for options that are not provided. For example, if you are inside
the top-level directory that contains your site's contents, and you'd like to generate a
service-worker.js file that will automatically precache all of the local files, you can simply run
$ sw-precache
Alternatively, if you'd like to only precache .html files that live within dist/, which is a
subdirectory of the current directory, you could run
$ sw-precache --root=dist --static-file-globs='dist/**/*.html'
Note: Be sure to use quotes around parameter values that have special meanings
to your shell (such as the * characters in the sample command line above,
for example).
Finally, there's support for passing complex configurations using --config <file>.
Any of the options from the file can be overridden via a command-line flag.
We strongly recommend passing it an external JavaScript file defining config via
module.exports.
For example, assume there's a path/to/sw-precache-config.js file that contains:
module.exports = {
staticFileGlobs: [
'app/css/**.css',
'app/**.html',
'app/images/**.*',
'app/js/**.js'
],
stripPrefix: 'app/',
runtimeCaching: [{
urlPattern: /this\\.is\\.a\\.regex/,
handler: 'networkFirst'
}]
};
That file could be passed to the command-line interface, while also setting the
verbose option, via
$ sw-precache --config=path/to/sw-precache-config.js --verbose
This provides the most flexibility, such as providing a regular expression for
the runtimeCaching.urlPattern option.
We also support passing in a JSON file for --config, though this provides
less flexibility:
{
"staticFileGlobs": [
"app/css/**.css",
"app/**.html",
"app/images/**.*",
"app/js/**.js"
],
"stripPrefix": "app/",
"runtimeCaching": [{
"urlPattern": "/express/style/path/(.*)",
"handler": "networkFirst"
}]
}
It's often desireable, even necessary to use precaching and runtime caching together. You may have seen our sw-toolbox tool, which handles runtime caching, and wondered how to use them together. Fortunately, sw-precache handles this for you.
The sw-precache module has the ability to include the sw-toolbox code and configuration alongside its own configuration. Using the runtimeCaching configuration option in sw-precache (see below) is a shortcut that accomplishes what you could do manually by importing sw-toolbox in your service worker and writing your own routing rules.
The sw-precache module exposes two methods: generate and write.
generate takes in options, generates a service worker
from them and passes the result to a callback function, which must
have the following interface:
callback(error, serviceWorkerString)
In the 1.x releases of sw-precache, this was the default and only method
exposed by the module.
Since 2.2.0, generate() also returns a
Promise.
write takes in options, generates a service worker from them,
and writes the service worker to a specified file. This method always
invokes callback(error). If no error was found, the error parameter will
be null
Since 2.2.0, write() also returns a Promise.
Both the generate() and write() methods take the same options.
A string used to distinguish the caches created by different web applications that are served off
of the same origin and path. While serving completely different sites from the same URL is not
likely to be an issue in a production environment, it avoids cache-conflicts when testing various
projects all served off of http://localhost. You may want to set it to, e.g., the name
property from your package.json.
Default: ''
Controls whether or not the generated service worker will call
clients.claim()
inside the activate handler.
Calling clients.claim() allows a newly registered service worker to take
control of a page immediately, instead of having to wait until the next page
navigation.
Default: true
Sets a default filename to return for URL's formatted like directory paths (in
other words, those ending in '/'). sw-precache will take that translation
into account and serve the contents a relative directoryIndex file when
there's no other match for a URL ending in '/'. To turn off this behavior,
set directoryIndex to false or null. To override this behavior for one
or more URLs, use the dynamicUrlToDependencies option to explicitly set up
mappings between a directory URL and a corresponding file.
Default: 'index.html'
It's very important that the requests sw-precache makes to populate your cache
result in the most up-to-date version of a resource at a given URL. Requests
that are fulfilled with out-of-date responses (like those found in your
browser's HTTP cache) can end up being read from the service worker's cache
indefinitely. Jake Archibald's blog post
provides more context about this problem.
In the interest of avoiding that scenario, sw-precache will, by default,
append a cache-busting parameter to the end of each URL it requests when
populating or updating its cache. Developers who are explicitly doing "the right
thing" when it comes to setting HTTP caching headers on their responses might
want to opt out of this cache-busting. For example, if all of your static
resources already include versioning information in their URLs (via a tool like
gulp-rev), and are served with
long-lived HTTP caching headers, then the extra cache-busting URL parameter
is not needed, and can be safely excluded.
dontCacheBustUrlsMatching gives you a way of opting-in to skipping the cache
busting behavior for a subset of your URLs (or all of them, if a catch-all value
like /./ is used).
If set, then the pathname
of each URL that's prefetched will be matched against this value.
If there's a match, then the URL will be prefetched as-is, without an additional
cache-busting URL parameter appended.
Note: Prior to sw-precache v5.0.0, dontCacheBustUrlsMatching matched against
the entire request URL. As of v5.0.0, it only matches against the URL's
pathname.
Default: not set
Maps a dynamic URL string to an array of all the files that URL's contents
depend on. E.g., if the contents of /pages/home are generated server-side via
the templates layout.jade and home.jade, then specify '/pages/home': ['layout.jade', 'home.jade']. The MD5 hash is used to determine whether
/pages/home has changed will depend on the hashes of both layout.jade and
home.jade.
An alternative value for the mapping is supported as well. You can specify
a string or a Buffer instance rather than an array of file names. If you use this option, then the
hash of the string/Buffer will be used to determine whether the URL used as a key has changed.
For example, '/pages/dynamic': dynamicStringValue could be used if the contents of
/pages/dynamic changes whenever the string stored in dynamicStringValue changes.
Default: {}
Determines whether the fetch event handler is included in the generated
service worker code. It is useful to set this to false in development builds,
to ensure that features like live reload still work. Otherwise, the content
would always be served from the service worker cache.
Default: true
sw-precache finds matching cache entries by doing a comparison with the full request URL. It's
common for sites to support URL query parameters that don't affect the site's content and should
be effectively ignored for the purposes of cache matching. One example is the
utm_-prefixed parameters used for tracking
campaign performance. By default, sw-precache will ignore key=value when key matches any of
the regular expressions provided in this option.
To ignore all parameters, use [/./]. To take all parameters into account when matching, use [].
Default: [/^utm_/]
Writes calls to importScripts()
to the resulting service worker to import the specified scripts.
Default: []
Specifies a callback function for logging which resources are being precached and
a precache size. Use function() {} if you'd prefer that nothing is logged.
Within a gulp script, it's recommended that you use gulp-util and pass in gutil.log.
Default: console.log
Sets the maximum allowed size for a file in the precache list.
Default: 2097152 (2 megabytes)
Sets an HTML document to use as a fallback for URLs not found in the sw-precache cache. This
fallback URL needs to be cached via staticFileGlobs or dynamicUrlToDependencies otherwise it
won't work.
// via staticFileGlobs
staticFileGlobs: ['/shell.html']
navigateFallback: '/shell.html'
// via dynamicUrlToDependencies
dynamicUrlToDependencies: {
'/shell': ['/shell.hbs']
},
navigateFallback: '/shell'
This comes in handy when used with a web application that performs client-side URL routing using the History API. It allows any arbitrary URL that the client generates to map to a fallback cached HTML entry. This fallback entry ideally should serve as an "application shell" that is able to load the appropriate resources client-side, based on the request URL.
Note: This is not intended to be used to route failed navigations to a
generic "offline fallback" page. The navigateFallback page is used whether the
browser is online or offline. If you want to implement an "offline fallback",
then using an approach similar to this example
is more appropriate.
Default: ''
Works to limit the effect of navigateFallback, so that the fallback only
applies to requests for URLs with paths that match at least one
RegExp.
This option is useful if you want to fallback to the cached App Shell for certain specific subsections of your site, but not have that behavior apply to all of your site's URLs.
For example, if you would like to have navigateFallback only apply to
navigation requests to URLs whose path begins with /guide/
(e.g. https://example.com/guide/1234), the following configuration could be
used:
navigateFallback: '/shell',
navigateFallbackWhitelist: [/^\/guide\//]
If set to [] (the default), the whitelist will be effectively bypassed, and
navigateFallback will apply to all navigation requests, regardless of URL.
Default: []
Replaces a specified string at the beginning of path URL's at runtime. Use this
option when you are serving static files from a different directory at runtime
than you are at build time. For example, if your local files are under
dist/app/ but your static asset root is at /public/, you'd strip 'dist/app/'
and replace it with '/public/'.
Default: ''
Configures runtime caching for dynamic content. If you use this option, the sw-toolbox
library configured with the caching strategies you specify will automatically be included in
your generated service worker file.
Each Object in the Array needs a urlPattern, which is either a
RegExp
or a string, following the conventions of the sw-toolbox library's
routing configuration. Also required is
a handler, which should be either a string corresponding to one of the
built-in handlers under the toolbox. namespace, or a function corresponding to your custom
request handler.
Optionally, method can be added to specify one of the supported HTTP methods (default: 'get'). There is also
support for options, which corresponds to the same options supported by a
sw-toolbox handler.
For example, the following defines runtime caching behavior for two different URL patterns. It uses a
different handler for each, and specifies a dedicated cache with maximum size for requests
that match /articles/:
runtimeCaching: [{
urlPattern: /^https:\/\/example\.com\/api/,
handler: 'networkFirst'
}, {
urlPattern: /\/articles\//,
handler: 'fastest',
options: {
cache: {
maxEntries: 10,
name: 'articles-cache'
}
}
}]
The sw-precache + sw-toolbox explainer has
more information about how and why you'd use both libraries together.
Default: []
Controls whether or not the generated service worker will call
skipWaiting()
inside the install handler.
By default, when there's an update to a previously installed
service worker, then the new service worker delays activation and stays in a
waiting state until all pages controlled by the old service worker are
unloaded. Calling skipWaiting() allows a newly registered service worker to
bypass the waiting state.
When skipWaiting is true, the new service worker's activate handler will
be called immediately, and any out of date cache entries from the previous
service worker will be deleted. Please keep this in mind if you rely on older
cached resources to be available throughout the page's lifetime, because, for
example, you defer the loading of some resources
until they're needed at runtime.
Default: true
An array of one or more string patterns that will be passed in to
glob.
All files matching these globs will be automatically precached by the generated service worker.
You'll almost always want to specify something for this.
Default: []
Removes a specified string from the beginning of path URL's at runtime. Use this
option when there's a discrepancy between a relative path at build time and
the same path at run time. For example, if all your local files are under
dist/app/ and your web root is also at dist/app/, you'd strip that prefix
from the start of each local file's path in order to get the correct relative
URL.
Default: ''
Maps multiple strings to be stripped and replaced from the beginning of URL paths at runtime.
Use this option when you have multiple discrepancies between relative paths at build time and
the same path at run time.
If stripPrefix and replacePrefix are not equal to '', they are automatically added to this option.
stripPrefixMulti: {
'www-root/public-precached/': 'public/',
'www-root/public/': 'public/'
}
Default: {}
The path to the (lo-dash) template used to
generate service-worker.js. If you need to add additional functionality to the
generated service worker code, it's recommended that you use the
importScripts option to include extra JavaScript rather than
using a different template. But if you do need to change the basic generated
service worker code, please make a copy of the original template,
modify it locally, and use this option to point to your template file.
Default: service-worker.tmpl (in the directory that this module lives in)
Determines whether there's log output for each individual static/dynamic resource that's precached. Even if this is set to false, there will be a final log entry indicating the total size of all precached resources.
Default: false
While it's possible to use the sw-precache module's API directly within any
JavaScript environment, several wrappers have been developed by members of the
community tailored to specific build environments. They include:
There are also several starter kits or scaffolding projects that incorporate
sw-precache into their build process, giving you a full service worker out of
the box. The include:
While there are not always ready-to-use wrappers for specific environments, this list contains some recipes to integrate sw-precache in your workflow:
Thanks to Sindre Sorhus and Addy Osmani for their advice and code reviews. Jake Archibald was kind enough to review the service worker logic.
The team behind sw-toolbox and sw-precache have been busy creating Workbox, which is a collection of libraries and tools that make it easy to build offline web apps. It’s a joining of sw-toolbox and sw-precache with more features and a modern codebase.
For now, it means we’ll continue to support both sw-toolbox and sw-precache with critical bug fixes and releases. However, non-critical bugs are unlikely to be addressed.
We would recommend Workbox for new projects, but there is no immediate need to switch if sw-toolbox / sw-precache meets all your needs in your current project. We will announce a deprecation plan for these modules once Workbox has feature parity with sw-toolbox and sw-precache.
In the meantime, you can get updates by following @workboxjs.
Copyright © 2017 Google, Inc.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
FAQs
Generates a service worker to cache your local App Shell resources.
The npm package sw-precache receives a total of 56,570 weekly downloads. As such, sw-precache popularity was classified as popular.
We found that sw-precache demonstrated a not healthy version release cadence and project activity because the last version was released a year ago. It has 4 open source maintainers collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
RubyGems.org has added a new "maintainer" role that allows for publishing new versions of gems. This new permission type is aimed at improving security for gem owners and the service overall.
Security News
Node.js will be enforcing stricter semver-major PR policies a month before major releases to enhance stability and ensure reliable release candidates.
Security News
Research
Socket's threat research team has detected five malicious npm packages targeting Roblox developers, deploying malware to steal credentials and personal data.