Service Worker Precache

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 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.

Table of Contents

• Install
• Usage
  • Overview
  • Example
  • Considerations
  • Command-line interface
• Runtime Caching
• API
  • Methods
    • generate(options, callback)
    • write(filePath, options, callback)
  • Options Parameter
    • cacheId [String]
    • clientsClaim [Boolean]
    • customOptions [Object]
    • directoryIndex [String]
    • dontCacheBustUrlsMatching [Regex]
    • dynamicUrlToDependencies [Object⟨String,Array⟨String⟩⟩]
    • handleFetch [boolean]
    • ignoreUrlParametersMatching [Array⟨Regex⟩]
    • importScripts [Array⟨String⟩]
    • logger [function]
    • maximumFileSizeToCacheInBytes [Number]
    • navigateFallback [String]
    • navigateFallbackWhitelist [Array⟨RegExp⟩]
    • replacePrefix [String]
    • runtimeCaching [Array⟨Object⟩]
    • skipWaiting [Boolean]
    • staticFileGlobs [Array⟨String⟩]
    • stripPrefix [String]
    • stripPrefixMulti [Object]
    • templateFilePath [String]
    • verbose [boolean]
• Wrappers and Starter Kits
  • CLIs
  • Starter Kits
  • Recipes for writing a custom wrapper
• Acknowledgements
• Future of Service Worker tooling
• License

Install

Local build integration:

$ npm install --save-dev sw-precache

Global command-line interface:

$ npm install --global sw-precache

Usage

Overview

1. 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.

2. 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.

3. 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.

Example

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 path = require('path');
  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.

Considerations

• 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 then 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.

Command-line interface

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.

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"
  }]
}

Runtime Caching

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.

API

Methods

The sw-precache module exposes two methods: generate and write.

generate(options, callback)

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(filePath, options, callback)

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.

Options Parameter

Both the generate() and write() methods take the same options.

cacheId [String]

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: ''

clientsClaim [Boolean]

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

customOptions [Object]

An object that is passed to the template. Using customOptions you can pass arbitrary information to the template. This allows better control over what can be done in a custom template.

Default: null

directoryIndex [String]

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'

dontCacheBustUrlsMatching [Regex]

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: null

dynamicUrlToDependencies [Object&#;String,Array&#;String&#;&#;]

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.

As of v5.1.0, and alternative value for the mapping is supported as well. You can now specify a string rather than an array of file names. If you use this option, then the hash of the string 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: {}

handleFetch [boolean]

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

ignoreUrlParametersMatching [Array&#;Regex&#;]

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_/]

importScripts [Array&#;String&#;]

Writes calls to importScripts() to the resulting service worker to import the specified scripts.

Default: []

logger [function]

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

maximumFileSizeToCacheInBytes [Number]

Sets the maximum allowed size for a file in the precache list.

Default: 2097152 (2 megabytes)

navigateFallback [String]

Sets an HTML document to use as a fallback for URLs not found in the cache. To be effective, this fallback URL should be already cached via staticFileGlobs or dynamicUrlToDependencies.

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: The current implementation searches the request's accept header and triggers the fallback when 'text/html' is found. It does this whether or not the request is a navigation.

Default: ''

navigateFallbackWhitelist [Array&#;RegExp&#;]

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: []

replacePrefix [String]

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: ''

runtimeCaching [Array&#;Object&#;]

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](https://googlechrome.github.io/sw-toolbox/api.html#handlers.

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: []

skipWaiting [Boolean]

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

staticFileGlobs [Array&#;String&#;]

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: []

stripPrefix [String]

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: ''

stripPrefixMulti [Object]

Maps mutliple 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: {}

templateFilePath [String]

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)

verbose [boolean]

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

Wrappers and Starter Kits

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:

• sw-precache-webpack-plugin
• sw-precache-brunch
• grunt-sw-precache
• exhibit-builder-sw-precache

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:

CLIs

• polymer-cli
• create-react-pwa

Starter Kits

• react-redux-universal-hot-example
• Polymer Starter Kit
• Web Starter Kit

Recipes for writing a custom wrapper

While there are not always ready-to-use wrappers for specific environments, this list contains some recipes to integrate sw-precache in your workflow:

• Gradle wrapper for offline JavaDoc
• Brunch starter for Phoenix Framework

Acknowledgements

Thanks to Sindre Sorhus and Addy Osmani for their advice and code reviews. Jake Archibald was kind enough to review the service worker logic.

Future of Service Worker tooling

Both sw-precache and sw-toolbox are actively maintained and we plan to continue supporting them. A large number of production Progressive Web Apps are successfully using them today and we are happy to review issues or PRs related to either project.

In parallel, we are working on the next generation of Service Worker tooling over in sw-helpers. This new work is more modular and will enable a number of libraries with additional capabilities to be built. We will announce further plans around the roadmap for this work in the future.

License

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.