React4XP - API reference
Contents
The react4xp library (Enonic Market) exposes a couple of methods that can be run from XP controllers.
This chapter also covers the SSR engine, compilation, and configuration of those.
Getting started
Install
Instructions on how to install the library locally or insert it into your project are at the library docs at github.
React4xp.render
All-in-one shorthand function for a lot of use cases. Covers both client- and serverside rendering.
Inserts a react component into an (optional) pre-existing HTML string and adds any necessary page contributions to make all work: links to assets, both shared and specific to the entry, and a client-side JS call (render
or hydrate
in the client wrapper) that activates the react component in the browser.
Signature
The signature is analogous to thymeleaf’s familiar render(view, model)
. But there are two extra parameters, and a full XP response object is returned:
{body, pageContributions} = React4xp.render(entry, props, request, options);
Parameters
Name | Type | Description | ||||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
string or component object, mandatory |
|||||||||||||||||||||
|
object, optional |
Data model passed into the react component. JS object must be serializable (no functions can be passed). Corresponds to the model argument in |
||||||||||||||||||||
|
XP request object, optional (but mandatory for react activation) |
Include to detect the rendering mode inside/outside Content Studio: inside Content Studio there should be only a static serverside-rendering, no browser react activation (or client-side rendering), only returning get an HTML visualization with the initial |
||||||||||||||||||||
|
object, optional |
Additional options to control the rendering. All of them are optional within this object:
|
Returns
Returns an XP response object with these main attributes:
Attribute | Type | Description |
---|---|---|
|
string, rendered HTML |
HTML output. The root of this HTML is always a surrounding container HTML that will have a matching-ID target element in it somewhere (an element matching the ID of the clientside call to Inside that matching-ID element, there will be a serverside rendering of the entry (with the initial props from |
|
object |
Regular XP page contributions. Includes everything the browser needs to activate (or client-side render) the react component: script tags with urls to auto-compiled assets for the entry and its dependencies, a client-side react4xp wrapper asset and an activating client-wrapper call. Urls point to react4xp’s own optimized asset services. Also included before this, are any input |
Request and automated behavior
.render
is intended to be convenient to work with and safely wrap around some common corner cases. It automates a little bit of behavior, depending on the request
object argument (which stems from the XP controller):
render
with request
:
React4xp.render(entry, props, request);
React4xp.render(entry, props, request, options); // ...etc etc
If request
is supplied, viewing context is detected from request.mode
: is rendering happening inside or outside of Content Studio?
-
Inside Content Studio,
.render
will always select serverside rendering (no matter whatclientRender
is) and skip JS dependency assets and the clientsiderender
/hydrate
trigger (but still still supply other dependencies, such as CSS). This ensures that a static HTML placeholder rendering is visible inside Content Studio’s browse and edit modes, but keeps react from being activated. This is by design: preventing the possibility that react code might intervene with the UX of Content Studio itself. In order to see working react and interact with your app, preview the page or navigate to a published version: -
Outside Content Studio, the rendering will be activated as a react app (i.e. all pageContributions are rendered). Also,
clientRender
is used, so if this istrue
, serverside rendering is skipped andrender
is called in the client instead ofhydrate
.
Also, when request
is used, .render
will output error messages from SSR in error containers and browser consoles (except in live view mode, where error containers and browser log are more generic: shown without the specific messages).
render
without request
:
React4xp.render(entry, props);
React4xp.render(entry, props, null);
React4xp.render(entry, props, undefined, options); // ...etc etc
Omitting request
from render
has the effect of always rendering as if it’s inside Content Studio (see above). Again, this is by design - trying to make sure that a viewable and safe rendering is always returned, even when viewing context can’t be determined.
However, it also means that clientRender
has no effect (you always get SSR), and there is no call to activate the react app in the browser. Basically, it’s as if JSX is used as a pure, static HTML templating language (same as XP’s Thymeleaf renderer - which shares the same basic signature).
When request
is omitted, .render
will never output error messages from SSR in error containers or browser console.
React4xp object: custom flow
More flexible and controllable than React4xp.render
: create a data-holding react4xp object with the React4xp contructor, manipulate it or extract data from it, combine with other objects, and then later render it to an HTML body string and/or page contributions, separately. This is actually what React4xp.render
does behind the scenes.
Call the two rendering methods from the same react4xp object. Remember, if using clientRender
and/or request
options, they should usually be the same value across the two corresponding calls. A typical (compact) usage example:
exports.get = function(request) => {
// Object constructor:
const myComponent = new React4xp('my-entry');
// ...read myComponent attributes and/or use its setter methods...
// const clientRender = ...true or false...
// Call the rendering methods:
return {
body: myComponent.renderBody({
// clientRender, etc
request
}),
pageContributions: myComponent.renderPageContributions({
// clientRender, etc
request
})
}; // ...etc, etc
}
See the "custom flow syntax" lesson to go more in depth.
Object constructor
const myComponent = new React4xp(entry);
Creates an initial react4xp data object from an entry.
Parameter | Type | Description |
---|---|---|
|
string or component object, mandatory |
Reference to an entry: the react component to be rendered. Direct JsxPath string, or a |
Constructs a react4xp data object, which exposes the attributes and methods below:
Main object attributes
Extract from the object the data that has been generated or set in it.
Name | Type | Description |
---|---|---|
|
string |
Target |
|
string |
jsxPath to the entry. |
|
object |
|
const targetElementId = myComponent.react4xpId;
Setter methods
Use these to set the object’s properties. All of them are optional; if not used, the object will render with empty values or placeholders where needed, along the same logic as for React4xp.render above.
All the setter methods return the data object itself, so that you can use a builder pattern where…
myComponent.firstSetter("a").secondSetter("b").thirdSetter("c");
…is equivalent to:
myComponent.firstSetter("a");
myComponent.secondSetter("b");
myComponent.thirdSetter("c");
The order between the setters doesn’t matter - except for setId
and uniqueId
, which affect each other.
.setProps
myComponent.setProps(props);
Sets props for the entry.
Parameter | Type | Description |
---|---|---|
|
object, mandatory |
|
.setId
myComponent.setId(id);
Sets an ID - directly and literally, so uniqueness is up to you. This ID both identifies this react4xp object (aka. react4xpId
), and crucially, points React to an HTML element (in the body
param, during render
or renderBody
later) which is the target container for rendering the entry into. Phew.
If render or renderBody are called without an ID having been set yet, then a unique random ID will be generated on the fly. This of course implies that there will be no matching-ID element in body . In cases like this (or when there’s no body at all), an empty target element with a matching ID will be generated/inserted, to contain the React rendering. |
If the data object already has an ID, .setId(id)
will overwrite it. If id
is omitted/empty, .setId()
just deletes any previous ID (which has the later effect of giving this a new, unique ID at the time of rendering).
Parameter | Type | Description |
---|---|---|
|
string, optional |
ID of both the target HTML element and the data object itself. |
.uniqueId
myComponent.uniqueId();
Enforces a unique ID, either by itself or after running .setId()
. If the object already has an ID (react4xpId
), a random string will be added to it. If not, the ID will just be the random string.
No parameters.
.setJsxPath
myComponent.setJsxPath(jsxPath);
If you for some reason need to override the JsxPath that was set (or inferred from the component object) in the constructor.
Parameter | Type | Description |
---|---|---|
|
string, mandatory |
New jsxPath to a different entry. |
Rendering methods
These methods perform specific rendering tasks independently, using the data object as a basis, the way it’s set up with the setters and with the entry from the constructor (or the setJsxPath
setter).
Most of these rendering methods will lock down the jsxPath and ID if the react4xp data object, the first time one of them is run. After this, the setters will prevent these from being changed so that another conflicting rendering can’t be performed from the same data object. |
.renderBody
const responseBody = myComponent.renderBody(options);
Similar to React4xp.render above, but renderBody
in itself only renders a static HTML output.
Does not render page contributions. Combine with a corresponding renderPageContributions call from the same data object, or the rendering will not be active in the browser. → See the custom flow syntax examples. |
renderBody
renders based on the state of the data object reached at the time of rendering.
Just like render
does, renderBody
ensures that the output HTML will always contain a matching-ID target element for react-rendering/hydrating the entry into (in the browser). And if serverside rendering is switched on (that is, clientRender
is falsy, or safe context-dependent rendering is enforced by adding request
- see the summary), the target element will contain the static HTML rendering.
Parameters
Parameter | Type | Description | ||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
object, optional |
Options to control the rendering, all of them optional:
Other renderings are not affected, even from the same data object (so you usually want to make sure a different rendering from the same data object uses the same mode). |
Returns
Returns an HTML string ready to return as the body
attribute in an XP response object from the controller.
The root of the returned HTML is always a surrounding container HTML that will have a matching-ID target element in it somewhere (an element matching the data object’s ID (react4xpId
), either from the ID setter methods, or a generated ID if they haven’t been run). This surrounding structure is options.body
, unchanged if that already contained a matching-ID element, or with a new target element generated and inserted at the end if it didn’t have one. If there is no options.body
, the surrounding container is just a generated target element.
Inside that matching-ID element, there will be a serverside rendering of the entry (with the initial props from .setProps
) if options.clientRender
is falsy.
.renderPageContributions
const outputPageContributions = myComponent.renderPageContributions(options);
Similar to React4xp.render above, but only renders the page contributions needed to run and activate the react component in the browser:
-
references to the entry’s own asset,
-
dependency assets,
-
and the react-activating trigger call in the browser (
.render
or.hydrate
, depending on theclientRender
andrequest
options).
Renders based on the state of the data object at the time of rendering.
Does not render any HTML. Run Also, unless you add the |
Parameters:
Parameter | Type | Description | ||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
object, optional |
Options to control the rendering, all of them optional:
|
Returns:
A regular XP page contributions object, ready to be used as the pageContributions
attribute in an XP response object from the controller.
Includes everything the browser needs to activate (or client-side render) the react component: script tags with urls to auto-compiled assets for the entry and its dependencies, a client-side react4xp wrapper asset and an activating trigger call to the client wrapper. Urls point to react4xp’s own optimized asset services. Also included before this, are any input options.pageContributions
.
With a serverside rendering (options.clientRender
is falsy), the client will expect an existing target element with a pre-rendered entry in the response body
, and call hydrate
. If options.clientRender
is truthy, an empty target element is expected in the response body
, and the rendering is left to the client with render
.
Request and automated behavior
The "custom flow" (.renderBody
in tandem with .renderPageContributions
) is intended as a more low-level approach: less hand-holding, more control to the developer for cases where that’s needed.
However, lib-react4xp version 1.6.0 introduced support for a request
option parameter for these methods as well. The main idea is that using request
in both calls will now automate some behavior the same way as calling .render
with request
(see above).
Omitting request
will still work the same way as before, leaving more to developers.
Custom flow with request
const body = myComponent.renderBody({
// clientRender, etc
request
});
const pageContributions = myComponent.renderPageContributions({
// clientRender, etc
request
});
// ...etc, etc
This will act the same way as render
used with a request: viewing context is detected, so inside Content Studio, clientRender
is ignored and you always get SSR, and JS assets and the .hydrate
call is held back so the react component isn’t activated inside Content Studio. And outside Content Studio, you get a fully active render (with optional clientRender
).
As with render
, error message details are held back in live view mode.
Custom flow without request
:
const body = myComponent.renderBody({ /* clientRender, etc */ });
const pageContributions = myComponent.renderPageContributions({ /* clientRender, etc */ });
// ...etc, etc
Contrary to when working with .render
, omitting request
from the custom flow does not enforce a max-safety rendering. Quite the opposite, removing request
will remove all the "safety wheels", so this rendering mode needs a bit of attention to guarantee that everything works everywhere:
-
.renderBody
will takeclientRender
into account in all contexts. What you set it to will take effect.This risks a missing/empty visualization inside Content Studio, since clientRender: true
makes sure no SSR will render a static placeholder. -
And
.renderPageContributions
will render all page contributions in all contexts, including JS dependency assets and thehydrate
/render
browser-side calls.Best case scenario: this might make a clientRender’ed entry visible in Content Studio too. Worst case, it risks intervening with Content Studio’s UX, or even break its functionality, depending on the code used/imported by the entry.
SSR engine
The default running mode of react4xp is serverside rendering (SSR) (although clientRender
can override this).
In a nutshell, source files like JSX are compiled into JS assets that lib-react4xp’s SSR engine runs to render HTML. This output is then delivered to the browser along with dependency code (usually references to necessary assets like CSS, JS etc) - these dependencies are also rendered, as page contributions. Rendering the HTML body and the page contributions happens to two different steps, using either render
(which wraps both steps for convenience) or the "custom flow".
The aim of react4xp is isomorphic rendering: after the react component(s) are serverside rendered, they are activated (hydrated) in the browser, turning them into running, active react apps. It’s the same react code that runs at the server as in the browser: no need to write the same component twice - one for SSR and one for the browser (although occasionally, tweaks are needed to prevent browserspecific code from running on the server).
Nashorn
React4xp can use nashorn as SSR engine. Since most of the world of react SSR is oriented towards node.js, and nashorn does not have full support for the features in node (nor all features expected by a lot of NPM packages that can be imported by react components), react4xp does some polyfilling of the SSR engine at startup.
Renderers
Starting from version 1.5.0, react4xp handles multithreaded rendering. This is done by setting up a number of renderers where each one is ready to answer to rendering requests in parallel, independently.
The number of renderer workers is determined in java but can be overriden.
When a renderer runs into an error during SSR, that renderer is torn down and a new one is initialized (see warmup time below). This happens as far as possible during idle time.
Warmup time
After your react4xp app is (re)started, the first time react4xp is triggered to render something, the engine will initialize. This means the renderers will load the compiled assets necessary for the rendering, into nashorn in memory:
-
nashorn polyfills,
-
react and reactDOM (
externals.*.js
), -
packages from node_modules (
vendors.*.js
), -
dependency assets imported by the react components (aka. chunks),
-
and finally, the entry assets themselves.
This causes some warmup time when starting your app: a noticable delay before the first rendering shows up. This may be just a couple of seconds in total, but it may also take longer. It depends on the size and complexity of the compiled assets involved. This will happen on every restart of the app (and every renderer must be initialized, but they do this in parallel).
But as long as the code runs in nashorn without errors, initialization happens only once (i.e. each asset is loaded once at most, on each renderer). After the warmup, the react apps are ready-to-run from nashorn memory, so repeated renderings after that (even with different props) are fast.
Improving warmup time for development
Since development can involve repeated app/server restarts, here are a couple of ways to improve initial loading time when developing large projects:
-
Set
ssrLazyload = true
(see below). This makes the engine start up and only load the bare minimum of dependency assets instead of preparing all of them at once. Each rendering will also only load/cache the dependencies it needs. The upside of this is that the first rendering (of a component that uses only a subset of the assets) is much faster since there is less to load. The downside is that there will still be assets that haven’t been loaded into the engine yet, which will cause a delay at some other time when they are first rendered. -
Don’t build the react components and dependencies with
BUILD_ENV = development
, but useproduction
(which is the default - see below). Assets built withdevelopment
are much more verbose, and this size difference - although functionally equal - actually makes a difference to nashorn (at the compile-to-bytecode stage).
(Other optimizations and approaches are under consideration, to shorten the warmup time even more).
SSR performance
Apart from when assets are initialized during the engine warmup, each SSR should be fast. However, in cases where you need to improve SSR performance further, it can be done by wrapping the rendering in a cache in the controller.
Be sure to use any value that can change the rendering output - usually from props
(and options
?) or a subset of them - as a key in the cache.
For example:
const cacheLib = require('/lib/cache');
// Set the cache up with a size that's reasonable
// for the most used props combinations
// and the size of the rendered output HTML string:
const cache = cacheLib.newCache({
size: 100,
expire: 3600
});
const makeKey = props => {
// ...return a string that's reliably determined by the relevant values from props
};
exports.get = request => {
const props = {
// ... build props from whatever sources are needed
};
const key = makeKey(props);
// Now render is only called when the key is new.
// If the key is cached before, just returns the output for that key.
return cache.get(
key,
() => React4xp.render(
myEntry,
props,
request,
options
)
);
};
react4xp.properties
The react4xp.properties
file has been removed and it’s settings moved to other files, see the upgrade guide.
react4xp.config.js
Several aspects of react4xp can be configured with a file react4xp.config.js.
If you’re using the starter or your project is based on it, this is already set up and you can skip the following section:
Including react4xp.config.js in your project
Put react4xp.config.js at the root of your XP project. A template can be copied from here, or from node_modules/@enonic/react4xp/examples/react4xp.config.js after installing the react4xp NPM package.
In order to take effect, it must be handled by gradle during the build:
-
Add this to your build.gradle:
apply from: "node_modules/@enonic/react4xp/react4xp.gradle"
This will not only make sure react4xp.config.js is handled, but also give a full react4xp build setup.
-
Of course, this requires that the react4xp NPM package is installed at the time gradle (or the enonic CLI) is run.
If you want to automate that, you can always add this to your gradle.build:
def react4xpGradleFile = new File('node_modules/@enonic/react4xp/react4xp.gradle') if (!react4xpGradleFile.exists()) { def proc = "npm install".execute() proc.in.eachLine { line -> println line } proc.out.close() proc.waitFor() } apply from: 'node_modules/@enonic/react4xp/react4xp.gradle'
-
Or, reverse-engineer the whole thing from react4xp.gradle and make your own gradle adaptation.
entryDirs
entryDirs: ['myComponents', '../otherComponents']
By default, react4xp will look for (ie. sets up webpack to look for) .JSX files to turn into entries below src/main/resources/site/.
Adding comma separated values under entryDirs
adds more folder names/paths (relative to src/main/resources/react4xp/) where JSX files will also become entries.
chunkDirs
By default, react4xp will look for (ie. sets up webpack to look for) resources imported by the entries. These are bundled into separate assets that react4xp automatically loads at both server- and clientside rendering: chunks, by this pattern:
-
If they are react and reactDOM, they are bundled separately into
externals.*.js
(where the * is a content-dependent hash). -
node_modules/@enonic/react-components/ is bundled into
templates.*.js
-
Other packages under node_modules are bundled into
vendors.*.js
-
And everything else that’s not under a
chunkDir
marked here is bundled intoreact4xp.*.js
(Of course, non-JS bundles split out by webpack will have different extensions, such as .css)
The idea is to use chunkDirs
to add a comma-separated list of names/paths of directories (relative to src/main/resources/react4xp/) that will be bundled into chunks of their own. The chunk name will be the name of the last directory in the path:
chunkDirs: ['chunk1', 'bundle2', 'other/stuff']
This example adds these folders as chunkDirs, and anything the entries import from below them is bundled separately into:
-
src/main/resources/react4xp/chunk1/ ➔
chunk1.*.js
-
src/main/resources/react4xp/bundle2/ ➔
bundle2.*.js
-
src/main/resources/react4xp/other/stuff/ ➔
stuff.*.js
webpack.config.react4xp.js
React4xp comes with a minimal set of webpack rules built-in, for compiling react components in JSX files into vanilla JS.
If you need to change/expand this setup, write a custom webpack config file <Project.Dir>/webpack.config.react4xp.js
There can be several reasons to this:
-
Most commonly, the built-in webpack setup is pretty minimal, only adding loaders for compiling react from JSX. It’s likely you will need to add loaders of your own, maybe use additional plugins etc
-
You may want to adjust other aspects of the compilation rules, or even replace the built-in rules entirely
-
The assets that are built during the compilation are the same ones that are run in nashorn and in the browser. It’s possible you may need adjustments here to account for corner cases - but if the problem is missing feature support in nashorn, it’s better to add extra polyfills using nashornPolyfills instead.
Config file shape: syntax variation! Usually, webpack.config.js files tend to have a certain shape, something like:
The extra incoming |
NashornPolyfills (<Project.Dir>/src/main/resources/react4xpNashornPolyfills.es6)
React4xp doesn’t have any ambition to completely polyfill the nashorn engine so that the full feature set of node.js (or modern browsers) is supported.
In the event that your code (or imported packages) rely on functionality that isn’t supported in nashorn, you can add as-vanilla-as-possible JS in <Project.Dir>/src/main/resources/react4xpNashornPolyfills.es6
This will be run as part of the SSR engine initialization, adding functionality before packages or other compiled code is loaded.
For example, until recently Object.assign
wasn’t polyfilled, causing problems for certain packages. That could be added by adding this code chunk to the file referred with nashornPolyfillsSource
.
Command line arguments
Build environment
Use the gradle commandline flag -Pdev
or -Pdevelopment
to enable building in development mode.
enonic project gradle build deploy -Pdev
This switches between react4xp build modes (not to be confused with XP’s run modes).
-
production
: assets are compiled more compact (and faster), with no source maps, and the entire SSR engine is initialized at once. -
development
: assets are compiled for more human-readability, with source maps, making errors easier to track down. IfssrLazyload
hasn’t been set,development
will activate lazy-loading.
Default value is production
.
Verbosity
Use the gradle commandline flag -i or --info to enable a more verbose output when compiling react4xp components, externals and nashornPolyfills.
Application configuration file
$XP_HOME/config/<app.name>.cfg
ssrLazyload
react4xp.ssr.lazyLoad = true
As described above, ssrLazyload
defines the initialization behavior of the renderer workers:
-
If switched on (
true
), only the required assets are loaded into the nashorn engine, speeding up the first load time, but subsequent loading times for other un-initialized assets will be longer. -
If switched off (
false
), the engine will load all assets once and for all, before performing the first rendering. This makes the first warmup time predictably slower (and that may be a substantial difference in a large project), but all subsequent rendering will be spring-loaded and fast.
Default value depends on BUILD_ENV
(or -Pdev
):
-
true
indevelopment
-
false
inproduction
ssrMaxThreads
react4xp.ssr.maxThreads = 5
If this override value is not set, the number of threads (renderers) used for simultaneous SSR requests is determined by java, by…
Runtime.getRuntime().availableProcessors()
…but ssrMaxThreads
sets this number manually instead.
ssrSettings
This value is a comma-separated string of nashorn options:
# Explicit string of settings:
react4xp.ssr.settings = --persistent-code-cache, --class-cache-size=42, --lazy-compilation, --optimistic-types=false
# Default value, turns persistent code cache off:
react4xp.ssr.settings = --optimistic-types=false
react4xp.ssr.settings only effective for Nashorn Engine. |
engineName
Allows to override JavaScript engine name.
# Explicit string of engine name:
react4xp.ssr.engineName = Nashorn
By default, JavaScript engine name is selected in the following order: Graal.js
, Nashorn
, JavaScript
.