[[spring-web-reactive]]
= Web on Reactive Stack
include::attributes.adoc[]
include::page-layout.adoc[]

This part of the documentation covers support for reactive-stack web applications built
on a https://www.reactive-streams.org/[Reactive Streams] API to run on non-blocking
servers, such as Netty, Undertow, and Servlet containers. Individual chapters cover
the <<webflux.adoc#webflux, Spring WebFlux>> framework,
the reactive <<webflux-client, `WebClient`>>, support for <<webflux-test, testing>>,
and <<webflux-reactive-libraries, reactive libraries>>. For Servlet-stack web applications,
see <<web.adoc#spring-web, Web on Servlet Stack>>.

include::web/webflux.adoc[leveloffset=+1]

include::web/webflux-webclient.adoc[leveloffset=+1]


[[webflux-http-interface-client]]
== HTTP Interface Client

The Spring Frameworks lets you define an HTTP service as a Java interface with HTTP
exchange methods. You can then generate a proxy that implements this interface and
performs the exchanges. This helps to simplify HTTP remote access and provides additional
flexibility for to choose an API style such as synchronous or reactive.

See <<integration.adoc#rest-http-interface, REST Endpoints>> for details.


include::web/webflux-websocket.adoc[leveloffset=+1]



[[webflux-test]]
== Testing
[.small]#<<web.adoc#testing, Same in Spring MVC>>#

The `spring-test` module provides mock implementations of `ServerHttpRequest`,
`ServerHttpResponse`, and `ServerWebExchange`.
See <<testing.adoc#mock-objects-web-reactive, Spring Web Reactive>> for a
discussion of mock objects.

<<testing.adoc#webtestclient, `WebTestClient`>> builds on these mock request and
response objects to provide support for testing WebFlux applications without an HTTP
server. You can use the `WebTestClient` for end-to-end integration tests, too.

include::rsocket.adoc[leveloffset=+1]



[[webflux-reactive-libraries]]
== Reactive Libraries

`spring-webflux` depends on `reactor-core` and uses it internally to compose asynchronous
logic and to provide Reactive Streams support. Generally, WebFlux APIs return `Flux` or
`Mono` (since those are used internally) and leniently accept any Reactive Streams
`Publisher` implementation as input. The use of `Flux` versus `Mono` is important, because
it helps to express cardinality -- for example, whether a single or multiple asynchronous
values are expected, and that can be essential for making decisions (for example, when
encoding or decoding HTTP messages).

For annotated controllers, WebFlux transparently adapts to the reactive library chosen by
the application. This is done with the help of the
{api-spring-framework}/core/ReactiveAdapterRegistry.html[`ReactiveAdapterRegistry`], which
provides pluggable support for reactive library and other asynchronous types. The registry
has built-in support for RxJava 3, Kotlin coroutines and SmallRye Mutiny, but you can
register others, too.

For functional APIs (such as <<webflux-fn>>, the `WebClient`, and others), the general rules
for WebFlux APIs apply -- `Flux` and `Mono` as return values and a Reactive Streams
`Publisher` as input. When a `Publisher`, whether custom or from another reactive library,
is provided, it can be treated only as a stream with unknown semantics (0..N). If, however,
the semantics are known, you can wrap it with `Flux` or `Mono.from(Publisher)` instead
of passing the raw `Publisher`.

For example, given a `Publisher` that is not a `Mono`, the Jackson JSON message writer
expects multiple values. If the media type implies an infinite stream (for example,
`application/json+stream`), values are written and flushed individually. Otherwise,
values are buffered into a list and rendered as a JSON array.