8193365: Improve interoperability between HTTP Client's BodyPublisher/BodySubscriber and Flow.Subscriber/Publisher
Reviewed-by: dfuchs
/*
* Copyright (c) 2015, 2017, Oracle and/or its affiliates. All rights reserved.
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
* This code is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License version 2 only, as
* published by the Free Software Foundation. Oracle designates this
* particular file as subject to the "Classpath" exception as provided
* by Oracle in the LICENSE file that accompanied this code.
*
* This code is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
* version 2 for more details (a copy is included in the LICENSE file that
* accompanied this code).
*
* You should have received a copy of the GNU General Public License version
* 2 along with this work; if not, write to the Free Software Foundation,
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
*
* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
* or visit www.oracle.com if you need additional information or have any
* questions.
*/
package jdk.incubator.http;
import java.io.IOException;
import java.io.InputStream;
import java.net.URI;
import jdk.incubator.http.ResponseSubscribers.MultiSubscriberImpl;
import static jdk.incubator.http.internal.common.Utils.unchecked;
import static jdk.incubator.http.internal.common.Utils.charsetFrom;
import java.nio.ByteBuffer;
import java.nio.charset.Charset;
import java.nio.channels.FileChannel;
import java.nio.file.OpenOption;
import java.nio.file.Path;
import java.nio.file.Paths;
import java.nio.file.StandardOpenOption;
import java.security.AccessControlContext;
import java.util.Arrays;
import java.util.List;
import java.util.Objects;
import java.util.Optional;
import java.util.concurrent.CompletableFuture;
import java.util.concurrent.CompletionStage;
import java.util.concurrent.Flow;
import java.util.concurrent.Flow.Subscriber;
import java.util.function.Consumer;
import java.util.function.Function;
import javax.net.ssl.SSLParameters;
/**
* Represents a response to a {@link HttpRequest}.
* {@Incubating}
*
* <p> A {@code HttpResponse} is available when the response status code and
* headers have been received, and typically after the response body has also
* been received. This depends on the response body handler provided when
* sending the request. In all cases, the response body handler is invoked
* before the body is read. This gives applications an opportunity to decide
* how to handle the body.
*
* <p> Methods are provided in this class for accessing the response headers,
* and response body.
*
* <p><b>Response handlers and subscribers</b>
*
* <p> Response bodies are handled at two levels. Application code supplies a
* response handler ({@link BodyHandler}) which may examine the response status
* code and headers, and which then returns a {@link BodySubscriber} to actually
* read (or discard) the body and convert it into some useful Java object type.
* The handler can return one of the pre-defined subscriber types, or a custom
* subscriber, or if the body is to be discarded it can call {@link
* BodySubscriber#discard(Object) discard} and return a subscriber which
* discards the response body. Static implementations of both handlers and
* subscribers are provided in {@linkplain BodyHandler BodyHandler} and
* {@linkplain BodySubscriber BodySubscriber} respectively. In all cases, the
* handler functions provided are convenience implementations which ignore the
* supplied status code and headers and return the relevant pre-defined {@code
* BodySubscriber}.
*
* <p> See {@link BodyHandler} for example usage.
*
* @param <T> the response body type
* @since 9
*/
public abstract class HttpResponse<T> {
/**
* Creates an HttpResponse.
*/
protected HttpResponse() { }
/**
* Returns the status code for this response.
*
* @return the response code
*/
public abstract int statusCode();
/**
* Returns the {@link HttpRequest} corresponding to this response.
*
* <p> This may not be the original request provided by the caller,
* for example, if that request was redirected.
*
* @see #previousResponse()
*
* @return the request
*/
public abstract HttpRequest request();
/**
* Returns an {@code Optional} containing the previous intermediate response
* if one was received. An intermediate response is one that is received
* as a result of redirection or authentication. If no previous response
* was received then an empty {@code Optional} is returned.
*
* @return an Optional containing the HttpResponse, if any.
*/
public abstract Optional<HttpResponse<T>> previousResponse();
/**
* Returns the received response headers.
*
* @return the response headers
*/
public abstract HttpHeaders headers();
/**
* Returns the body. Depending on the type of {@code T}, the returned body
* may represent the body after it was read (such as {@code byte[]}, or
* {@code String}, or {@code Path}) or it may represent an object with
* which the body is read, such as an {@link java.io.InputStream}.
*
* <p> If this {@code HttpResponse} was returned from an invocation of
* {@link #previousResponse()} then this method returns {@code null}
*
* @return the body
*/
public abstract T body();
/**
* Returns the {@link javax.net.ssl.SSLParameters} in effect for this
* response. Returns {@code null} if this is not a HTTPS response.
*
* @return the SSLParameters associated with the response
*/
public abstract SSLParameters sslParameters();
/**
* Returns the {@code URI} that the response was received from. This may be
* different from the request {@code URI} if redirection occurred.
*
* @return the URI of the response
*/
public abstract URI uri();
/**
* Returns the HTTP protocol version that was used for this response.
*
* @return HTTP protocol version
*/
public abstract HttpClient.Version version();
private static String pathForSecurityCheck(Path path) {
return path.toFile().getPath();
}
/** A body handler that is further restricted by a given ACC. */
interface UntrustedBodyHandler<T> extends BodyHandler<T> {
void setAccessControlContext(AccessControlContext acc);
}
/**
* A Path body handler.
*
* Note: Exists mainly too allow setting of the senders ACC post creation of
* the handler.
*/
static class PathBodyHandler implements UntrustedBodyHandler<Path> {
private final Path file;
private final OpenOption[]openOptions;
private volatile AccessControlContext acc;
PathBodyHandler(Path file, OpenOption... openOptions) {
this.file = file;
this.openOptions = openOptions;
}
@Override
public void setAccessControlContext(AccessControlContext acc) {
this.acc = acc;
}
@Override
public BodySubscriber<Path> apply(int statusCode, HttpHeaders headers) {
ResponseSubscribers.PathSubscriber bs = (ResponseSubscribers.PathSubscriber)
BodySubscriber.asFileImpl(file, openOptions);
bs.setAccessControlContext(acc);
return bs;
}
}
// Similar to Path body handler, but for file download. Supports setting ACC.
static class FileDownloadBodyHandler implements UntrustedBodyHandler<Path> {
private final Path directory;
private final OpenOption[]openOptions;
private volatile AccessControlContext acc;
FileDownloadBodyHandler(Path directory, OpenOption... openOptions) {
this.directory = directory;
this.openOptions = openOptions;
}
@Override
public void setAccessControlContext(AccessControlContext acc) {
this.acc = acc;
}
@Override
public BodySubscriber<Path> apply(int statusCode, HttpHeaders headers) {
String dispoHeader = headers.firstValue("Content-Disposition")
.orElseThrow(() -> unchecked(new IOException("No Content-Disposition")));
if (!dispoHeader.startsWith("attachment;")) {
throw unchecked(new IOException("Unknown Content-Disposition type"));
}
int n = dispoHeader.indexOf("filename=");
if (n == -1) {
throw unchecked(new IOException("Bad Content-Disposition type"));
}
int lastsemi = dispoHeader.lastIndexOf(';');
String disposition;
if (lastsemi < n) {
disposition = dispoHeader.substring(n + 9);
} else {
disposition = dispoHeader.substring(n + 9, lastsemi);
}
Path file = Paths.get(directory.toString(), disposition);
ResponseSubscribers.PathSubscriber bs = (ResponseSubscribers.PathSubscriber)
BodySubscriber.asFileImpl(file, openOptions);
bs.setAccessControlContext(acc);
return bs;
}
}
/**
* A handler for response bodies.
* {@Incubating}
*
* <p> This is a function that takes two parameters: the response status code,
* and the response headers, and which returns a {@linkplain BodySubscriber}.
* The function is always called just before the response body is read. Its
* implementation may examine the status code or headers and must decide,
* whether to accept the response body or discard it, and if accepting it,
* exactly how to handle it.
*
* <p> Some pre-defined implementations which do not utilize the status code
* or headers (meaning the body is always accepted) are defined:
* <ul><li>{@link #asByteArray() }</li>
* <li>{@link #asByteArrayConsumer(java.util.function.Consumer)
* asByteArrayConsumer(Consumer)}</li>
* <li>{@link #asString(java.nio.charset.Charset) asString(Charset)}</li>
* <li>{@link #asFile(Path, OpenOption...)
* asFile(Path,OpenOption...)}</li>
* <li>{@link #asFileDownload(java.nio.file.Path,OpenOption...)
* asFileDownload(Path,OpenOption...)}</li>
* <li>{@link #asInputStream() asInputStream()}</li>
* <li>{@link #discard(Object) }</li>
* <li>{@link #buffering(BodyHandler, int)
* buffering(BodyHandler,int)}</li>
* </ul>
*
* <p> These implementations return the equivalent {@link BodySubscriber}.
* Alternatively, the handler can be used to examine the status code
* or headers and return different body subscribers as appropriate.
*
* <p><b>Examples of handler usage</b>
*
* <p> The first example uses one of the predefined handler functions which
* ignores the response headers and status, and always process the response
* body in the same way.
* <pre>
* {@code
* HttpResponse<Path> resp = HttpRequest
* .create(URI.create("http://www.foo.com"))
* .GET()
* .response(BodyHandler.asFile(Paths.get("/tmp/f")));
* }
* </pre>
* Note, that even though these pre-defined handlers ignore the status code
* and headers, this information is still accessible from the
* {@code HttpResponse} when it is returned.
*
* <p> In the second example, the function returns a different subscriber
* depending on the status code.
* <pre>
* {@code
* HttpResponse<Path> resp1 = HttpRequest
* .create(URI.create("http://www.foo.com"))
* .GET()
* .response(
* (status, headers) -> status == 200
* ? BodySubscriber.asFile(Paths.get("/tmp/f"))
* : BodySubscriber.discard(Paths.get("/NULL")));
* }
* </pre>
*
* @param <T> the response body type
*/
@FunctionalInterface
public interface BodyHandler<T> {
/**
* Returns a {@link BodySubscriber BodySubscriber} considering the given
* response status code and headers. This method is always called before
* the body is read and its implementation can decide to keep the body
* and store it somewhere, or else discard it by returning the {@code
* BodySubscriber} returned from {@link BodySubscriber#discard(Object)
* discard}.
*
* @param statusCode the HTTP status code received
* @param responseHeaders the response headers received
* @return a body subscriber
*/
public BodySubscriber<T> apply(int statusCode, HttpHeaders responseHeaders);
/**
* Returns a response body handler that returns a {@link BodySubscriber
* BodySubscriber}{@code <Void>} obtained from {@linkplain
* BodySubscriber#fromSubscriber(Subscriber)}, with the given
* {@code subscriber}.
*
* <p> The response body is not available through this, or the {@code
* HttpResponse} API, but instead all response body is forwarded to the
* given {@code subscriber}, which should make it available, if
* appropriate, through some other mechanism, e.g. an entry in a
* database, etc.
*
* @apiNote This method can be used as an adapter between {@code
* BodySubscriber} and {@code Flow.Subscriber}.
*
* <p> For example:
* <pre> {@code
* TextSubscriber subscriber = new TextSubscriber();
* HttpResponse<Void> response = client.sendAsync(request,
* BodyHandler.fromSubscriber(subscriber)).join();
* System.out.println(response.statusCode());
* }</pre>
*
* @param subscriber the subscriber
* @return a response body handler
*/
public static BodyHandler<Void>
fromSubscriber(Subscriber<? super List<ByteBuffer>> subscriber) {
Objects.requireNonNull(subscriber);
return (status, headers) -> BodySubscriber.fromSubscriber(subscriber,
s -> null);
}
/**
* Returns a response body handler that returns a {@link BodySubscriber
* BodySubscriber}{@code <T>} obtained from {@link
* BodySubscriber#fromSubscriber(Subscriber, Function)}, with the
* given {@code subscriber} and {@code finisher} function.
*
* <p> The given {@code finisher} function is applied after the given
* subscriber's {@code onComplete} has been invoked. The {@code finisher}
* function is invoked with the given subscriber, and returns a value
* that is set as the response's body.
*
* @apiNote This method can be used as an adapter between {@code
* BodySubscriber} and {@code Flow.Subscriber}.
*
* <p> For example:
* <pre> {@code
* TextSubscriber subscriber = ...; // accumulates bytes and transforms them into a String
* HttpResponse<String> response = client.sendAsync(request,
* BodyHandler.fromSubscriber(subscriber, TextSubscriber::getTextResult)).join();
* String text = response.body();
* }</pre>
*
* @param <S> the type of the Subscriber
* @param <T> the type of the response body
* @param subscriber the subscriber
* @param finisher a function to be applied after the subscriber has completed
* @return a response body handler
*/
public static <S extends Subscriber<? super List<ByteBuffer>>,T> BodyHandler<T>
fromSubscriber(S subscriber, Function<S,T> finisher) {
Objects.requireNonNull(subscriber);
Objects.requireNonNull(finisher);
return (status, headers) -> BodySubscriber.fromSubscriber(subscriber,
finisher);
}
/**
* Returns a response body handler which discards the response body and
* uses the given value as a replacement for it.
*
* @param <U> the response body type
* @param value the value of U to return as the body, may be {@code null}
* @return a response body handler
*/
public static <U> BodyHandler<U> discard(U value) {
return (status, headers) -> BodySubscriber.discard(value);
}
/**
* Returns a {@code BodyHandler<String>} that returns a
* {@link BodySubscriber BodySubscriber}{@code <String>} obtained from
* {@link BodySubscriber#asString(Charset) BodySubscriber.asString(Charset)}.
* The body is decoded using the given character set.
*
* @param charset the character set to convert the body with
* @return a response body handler
*/
public static BodyHandler<String> asString(Charset charset) {
Objects.requireNonNull(charset);
return (status, headers) -> BodySubscriber.asString(charset);
}
/**
* Returns a {@code BodyHandler<Path>} that returns a
* {@link BodySubscriber BodySubscriber}{@code <Path>} obtained from
* {@link BodySubscriber#asFile(Path, OpenOption...)
* BodySubscriber.asFile(Path,OpenOption...)}.
*
* <p> When the {@code HttpResponse} object is returned, the body has
* been completely written to the file, and {@link #body()} returns a
* reference to its {@link Path}.
*
* @param file the filename to store the body in
* @param openOptions any options to use when opening/creating the file
* @return a response body handler
* @throws SecurityException If a security manager has been installed
* and it denies {@link SecurityManager#checkWrite(String)
* write access} to the file. The {@link
* SecurityManager#checkDelete(String) checkDelete} method is
* invoked to check delete access if the file is opened with
* the {@code DELETE_ON_CLOSE} option.
*/
public static BodyHandler<Path> asFile(Path file, OpenOption... openOptions) {
Objects.requireNonNull(file);
List<OpenOption> opts = List.of(openOptions);
SecurityManager sm = System.getSecurityManager();
if (sm != null) {
String fn = pathForSecurityCheck(file);
sm.checkWrite(fn);
if (opts.contains(StandardOpenOption.DELETE_ON_CLOSE))
sm.checkDelete(fn);
if (opts.contains(StandardOpenOption.READ))
sm.checkRead(fn);
}
return new PathBodyHandler(file, openOptions);
}
/**
* Returns a {@code BodyHandler<Path>} that returns a
* {@link BodySubscriber BodySubscriber}{@code <Path>} obtained from
* {@link BodySubscriber#asFile(Path) BodySubscriber.asFile(Path)}.
*
* <p> When the {@code HttpResponse} object is returned, the body has
* been completely written to the file, and {@link #body()} returns a
* reference to its {@link Path}.
*
* @param file the file to store the body in
* @return a response body handler
* @throws SecurityException if a security manager has been installed
* and it denies {@link SecurityManager#checkWrite(String)
* write access} to the file
*/
public static BodyHandler<Path> asFile(Path file) {
return BodyHandler.asFile(file, StandardOpenOption.CREATE,
StandardOpenOption.WRITE);
}
/**
* Returns a {@code BodyHandler<Path>} that returns a
* {@link BodySubscriber BodySubscriber}<{@link Path}>
* where the download directory is specified, but the filename is
* obtained from the {@code Content-Disposition} response header. The
* {@code Content-Disposition} header must specify the <i>attachment</i>
* type and must also contain a <i>filename</i> parameter. If the
* filename specifies multiple path components only the final component
* is used as the filename (with the given directory name).
*
* <p> When the {@code HttpResponse} object is returned, the body has
* been completely written to the file and {@link #body()} returns a
* {@code Path} object for the file. The returned {@code Path} is the
* combination of the supplied directory name and the file name supplied
* by the server. If the destination directory does not exist or cannot
* be written to, then the response will fail with an {@link IOException}.
*
* @param directory the directory to store the file in
* @param openOptions open options
* @return a response body handler
* @throws SecurityException If a security manager has been installed
* and it denies {@link SecurityManager#checkWrite(String)
* write access} to the file. The {@link
* SecurityManager#checkDelete(String) checkDelete} method is
* invoked to check delete access if the file is opened with
* the {@code DELETE_ON_CLOSE} option.
*/
//####: check if the dir exists and is writable??
public static BodyHandler<Path> asFileDownload(Path directory,
OpenOption... openOptions) {
Objects.requireNonNull(directory);
List<OpenOption> opts = List.of(openOptions);
SecurityManager sm = System.getSecurityManager();
if (sm != null) {
String fn = pathForSecurityCheck(directory);
sm.checkWrite(fn);
if (opts.contains(StandardOpenOption.DELETE_ON_CLOSE))
sm.checkDelete(fn);
if (opts.contains(StandardOpenOption.READ))
sm.checkRead(fn);
}
return new FileDownloadBodyHandler(directory, openOptions);
}
/**
* Returns a {@code BodyHandler<InputStream>} that returns a
* {@link BodySubscriber BodySubscriber}{@code <InputStream>} obtained
* from {@link BodySubscriber#asInputStream() BodySubscriber.asInputStream}.
*
* <p> When the {@code HttpResponse} object is returned, the response
* headers will have been completely read, but the body may not have
* been fully received yet. The {@link #body()} method returns an
* {@link InputStream} from which the body can be read as it is received.
*
* @apiNote See {@link BodySubscriber#asInputStream()} for more information.
*
* @return a response body handler
*/
public static BodyHandler<InputStream> asInputStream() {
return (status, headers) -> BodySubscriber.asInputStream();
}
/**
* Returns a {@code BodyHandler<Void>} that returns a
* {@link BodySubscriber BodySubscriber}{@code <Void>} obtained from
* {@link BodySubscriber#asByteArrayConsumer(Consumer)
* BodySubscriber.asByteArrayConsumer(Consumer)}.
*
* <p> When the {@code HttpResponse} object is returned, the body has
* been completely written to the consumer.
*
* @param consumer a Consumer to accept the response body
* @return a response body handler
*/
public static BodyHandler<Void> asByteArrayConsumer(Consumer<Optional<byte[]>> consumer) {
Objects.requireNonNull(consumer);
return (status, headers) -> BodySubscriber.asByteArrayConsumer(consumer);
}
/**
* Returns a {@code BodyHandler<byte[]>} that returns a
* {@link BodySubscriber BodySubscriber}<{@code byte[]}> obtained
* from {@link BodySubscriber#asByteArray() BodySubscriber.asByteArray()}.
*
* <p> When the {@code HttpResponse} object is returned, the body has
* been completely written to the byte array.
*
* @return a response body handler
*/
public static BodyHandler<byte[]> asByteArray() {
return (status, headers) -> BodySubscriber.asByteArray();
}
/**
* Returns a {@code BodyHandler<String>} that returns a
* {@link BodySubscriber BodySubscriber}{@code <String>} obtained from
* {@link BodySubscriber#asString(java.nio.charset.Charset)
* BodySubscriber.asString(Charset)}. The body is
* decoded using the character set specified in
* the {@code Content-encoding} response header. If there is no such
* header, or the character set is not supported, then
* {@link java.nio.charset.StandardCharsets#UTF_8 UTF_8} is used.
*
* <p> When the {@code HttpResponse} object is returned, the body has
* been completely written to the string.
*
* @return a response body handler
*/
public static BodyHandler<String> asString() {
return (status, headers) -> BodySubscriber.asString(charsetFrom(headers));
}
/**
* Returns a {@code BodyHandler} which, when invoked, returns a {@linkplain
* BodySubscriber#buffering(BodySubscriber,int) buffering BodySubscriber}
* that buffers data before delivering it to the downstream subscriber.
* These {@code BodySubscriber} instances are created by calling
* {@linkplain BodySubscriber#buffering(BodySubscriber,int)
* BodySubscriber.buffering} with a subscriber obtained from the given
* downstream handler and the {@code bufferSize} parameter.
*
* @param downstreamHandler the downstream handler
* @param bufferSize the buffer size parameter passed to {@linkplain
* BodySubscriber#buffering(BodySubscriber,int) BodySubscriber.buffering}
* @return a body handler
* @throws IllegalArgumentException if {@code bufferSize <= 0}
*/
public static <T> BodyHandler<T> buffering(BodyHandler<T> downstreamHandler,
int bufferSize) {
Objects.requireNonNull(downstreamHandler);
if (bufferSize <= 0)
throw new IllegalArgumentException("must be greater than 0");
return (status, headers) -> BodySubscriber
.buffering(downstreamHandler.apply(status, headers),
bufferSize);
}
}
/**
* A subscriber for response bodies.
* {@Incubating}
*
* <p> The object acts as a {@link Flow.Subscriber}<{@link List}<{@link
* ByteBuffer}>> to the HTTP client implementation, which publishes
* unmodifiable lists of ByteBuffers containing the response body. The Flow
* of data, as well as the order of ByteBuffers in the Flow lists, is a
* strictly ordered representation of the response body. Both the Lists and
* the ByteBuffers, once passed to the subscriber, are no longer used by the
* HTTP client. The subscriber converts the incoming buffers of data to some
* user-defined object type {@code T}.
*
* <p> The {@link #getBody()} method returns a {@link CompletionStage}{@code
* <T>} that provides the response body object. The {@code CompletionStage}
* must be obtainable at any time. When it completes depends on the nature
* of type {@code T}. In many cases, when {@code T} represents the entire
* body after being read then it completes after the body has been read. If
* {@code T} is a streaming type such as {@link java.io.InputStream} then it
* completes before the body has been read, because the calling code uses it
* to consume the data.
*
* @apiNote To ensure that all resources associated with the
* corresponding exchange are properly released, an implementation
* of {@code BodySubscriber} must ensure to {@linkplain
* Flow.Subscription#request request} more data until {@link
* #onComplete() onComplete} or {@link #onError(Throwable) onError}
* are signalled, or {@linkplain Flow.Subscription#request cancel} its
* {@linkplain #onSubscribe(Flow.Subscription) subscription}
* if unable or unwilling to do so.
* Calling {@code cancel} before exhausting the data may cause
* the underlying HTTP connection to be closed and prevent it
* from being reused for subsequent operations.
*
* @param <T> the response body type
*/
public interface BodySubscriber<T>
extends Flow.Subscriber<List<ByteBuffer>> {
/**
* Returns a {@code CompletionStage} which when completed will return
* the response body object.
*
* @return a CompletionStage for the response body
*/
public CompletionStage<T> getBody();
/**
* Returns a body subscriber that forwards all response body to the
* given {@code Flow.Subscriber}. The {@linkplain #getBody()} completion
* stage} of the returned body subscriber completes after one of the
* given subscribers {@code onComplete} or {@code onError} has been
* invoked.
*
* @apiNote This method can be used as an adapter between {@code
* BodySubscriber} and {@code Flow.Subscriber}.
*
* @param <S> the type of the Subscriber
* @param subscriber the subscriber
* @return a body subscriber
*/
public static <S extends Subscriber<? super List<ByteBuffer>>> BodySubscriber<Void>
fromSubscriber(S subscriber) {
return new ResponseSubscribers.SubscriberAdapter<S,Void>(subscriber, s -> null);
}
/**
* Returns a body subscriber that forwards all response body to the
* given {@code Flow.Subscriber}. The {@linkplain #getBody()} completion
* stage} of the returned body subscriber completes after one of the
* given subscribers {@code onComplete} or {@code onError} has been
* invoked.
*
* <p> The given {@code finisher} function is applied after the given
* subscriber's {@code onComplete} has been invoked. The {@code finisher}
* function is invoked with the given subscriber, and returns a value
* that is set as the response's body.
*
* @apiNote This method can be used as an adapter between {@code
* BodySubscriber} and {@code Flow.Subscriber}.
*
* @param <S> the type of the Subscriber
* @param <T> the type of the response body
* @param subscriber the subscriber
* @param finisher a function to be applied after the subscriber has
* completed
* @return a body subscriber
*/
public static <S extends Subscriber<? super List<ByteBuffer>>,T> BodySubscriber<T>
fromSubscriber(S subscriber,
Function<S,T> finisher) {
return new ResponseSubscribers.SubscriberAdapter<S,T>(subscriber, finisher);
}
/**
* Returns a body subscriber which stores the response body as a {@code
* String} converted using the given {@code Charset}.
*
* <p> The {@link HttpResponse} using this subscriber is available after
* the entire response has been read.
*
* @param charset the character set to convert the String with
* @return a body subscriber
*/
public static BodySubscriber<String> asString(Charset charset) {
Objects.requireNonNull(charset);
return new ResponseSubscribers.ByteArraySubscriber<>(
bytes -> new String(bytes, charset)
);
}
/**
* Returns a {@code BodySubscriber} which stores the response body as a
* byte array.
*
* <p> The {@link HttpResponse} using this subscriber is available after
* the entire response has been read.
*
* @return a body subscriber
*/
public static BodySubscriber<byte[]> asByteArray() {
return new ResponseSubscribers.ByteArraySubscriber<>(
Function.identity() // no conversion
);
}
// no security check
private static BodySubscriber<Path> asFileImpl(Path file, OpenOption... openOptions) {
return new ResponseSubscribers.PathSubscriber(file, openOptions);
}
/**
* Returns a {@code BodySubscriber} which stores the response body in a
* file opened with the given options and name. The file will be opened
* with the given options using {@link FileChannel#open(Path,OpenOption...)
* FileChannel.open} just before the body is read. Any exception thrown
* will be returned or thrown from {@link HttpClient#send(HttpRequest,
* BodyHandler) HttpClient::send} or {@link HttpClient#sendAsync(HttpRequest,
* BodyHandler) HttpClient::sendAsync} as appropriate.
*
* <p> The {@link HttpResponse} using this subscriber is available after
* the entire response has been read.
*
* @param file the file to store the body in
* @param openOptions the list of options to open the file with
* @return a body subscriber
* @throws SecurityException If a security manager has been installed
* and it denies {@link SecurityManager#checkWrite(String)
* write access} to the file. The {@link
* SecurityManager#checkDelete(String) checkDelete} method is
* invoked to check delete access if the file is opened with the
* {@code DELETE_ON_CLOSE} option.
*/
public static BodySubscriber<Path> asFile(Path file, OpenOption... openOptions) {
Objects.requireNonNull(file);
List<OpenOption> opts = List.of(openOptions);
SecurityManager sm = System.getSecurityManager();
if (sm != null) {
String fn = pathForSecurityCheck(file);
sm.checkWrite(fn);
if (opts.contains(StandardOpenOption.DELETE_ON_CLOSE))
sm.checkDelete(fn);
if (opts.contains(StandardOpenOption.READ))
sm.checkRead(fn);
}
return asFileImpl(file, openOptions);
}
/**
* Returns a {@code BodySubscriber} which stores the response body in a
* file opened with the given name. Has the same effect as calling
* {@link #asFile(Path, OpenOption...) asFile} with the standard open
* options {@code CREATE} and {@code WRITE}
*
* <p> The {@link HttpResponse} using this subscriber is available after
* the entire response has been read.
*
* @param file the file to store the body in
* @return a body subscriber
* @throws SecurityException if a security manager has been installed
* and it denies {@link SecurityManager#checkWrite(String)
* write access} to the file
*/
public static BodySubscriber<Path> asFile(Path file) {
return asFile(file, StandardOpenOption.CREATE, StandardOpenOption.WRITE);
}
/**
* Returns a {@code BodySubscriber} which provides the incoming body
* data to the provided Consumer of {@code Optional<byte[]>}. Each
* call to {@link Consumer#accept(java.lang.Object) Consumer.accept()}
* will contain a non empty {@code Optional}, except for the final
* invocation after all body data has been read, when the {@code
* Optional} will be empty.
*
* <p> The {@link HttpResponse} using this subscriber is available after
* the entire response has been read.
*
* @param consumer a Consumer of byte arrays
* @return a BodySubscriber
*/
public static BodySubscriber<Void> asByteArrayConsumer(Consumer<Optional<byte[]>> consumer) {
return new ResponseSubscribers.ConsumerSubscriber(consumer);
}
/**
* Returns a {@code BodySubscriber} which streams the response body as
* an {@link InputStream}.
*
* <p> The {@link HttpResponse} using this subscriber is available
* immediately after the response headers have been read, without
* requiring to wait for the entire body to be processed. The response
* body can then be read directly from the {@link InputStream}.
*
* @apiNote To ensure that all resources associated with the
* corresponding exchange are properly released the caller must
* ensure to either read all bytes until EOF is reached, or call
* {@link InputStream#close} if it is unable or unwilling to do so.
* Calling {@code close} before exhausting the stream may cause
* the underlying HTTP connection to be closed and prevent it
* from being reused for subsequent operations.
*
* @return a body subscriber that streams the response body as an
* {@link InputStream}.
*/
public static BodySubscriber<InputStream> asInputStream() {
return new ResponseSubscribers.HttpResponseInputStream();
}
/**
* Returns a response subscriber which discards the response body. The
* supplied value is the value that will be returned from
* {@link HttpResponse#body()}.
*
* @param <U> The type of the response body
* @param value the value to return from HttpResponse.body(), may be {@code null}
* @return a {@code BodySubscriber}
*/
public static <U> BodySubscriber<U> discard(U value) {
return new ResponseSubscribers.NullSubscriber<>(Optional.ofNullable(value));
}
/**
* Returns a {@code BodySubscriber} which buffers data before delivering
* it to the given downstream subscriber. The subscriber guarantees to
* deliver {@code buffersize} bytes of data to each invocation of the
* downstream's {@linkplain #onNext(Object) onNext} method, except for
* the final invocation, just before {@linkplain #onComplete() onComplete}
* is invoked. The final invocation of {@code onNext} may contain fewer
* than {@code buffersize} bytes.
*
* <p> The returned subscriber delegates its {@link #getBody()} method
* to the downstream subscriber.
*
* @param downstream the downstream subscriber
* @param bufferSize the buffer size
* @return a buffering body subscriber
* @throws IllegalArgumentException if {@code bufferSize <= 0}
*/
public static <T> BodySubscriber<T> buffering(BodySubscriber<T> downstream,
int bufferSize) {
if (bufferSize <= 0)
throw new IllegalArgumentException("must be greater than 0");
return new BufferingSubscriber<T>(downstream, bufferSize);
}
}
/**
* A response subscriber for a HTTP/2 multi response.
* {@Incubating}
*
* <p> A multi response comprises a main response, and zero or more additional
* responses. Each additional response is sent by the server in response to
* requests (PUSH_PROMISEs) that the server also generates. Additional responses are
* typically resources that the server expects the client will need which
* are related to the initial request.
* <p>
* Note. Instead of implementing this interface, applications should consider
* first using the mechanism (built on this interface) provided by
* {@link MultiSubscriber#asMap(java.util.function.Function, boolean)
* MultiSubscriber.asMap()} which is a slightly simplified, but also
* general purpose interface.
* <p>
* The server generated requests are also known as <i>push promises</i>.
* The server is permitted to send any number of these requests up to the
* point where the main response is fully received. Therefore, after
* completion of the main response, the final number of additional
* responses is known. Additional responses may be canceled, but given that
* the server does not wait for any acknowledgment before sending the
* response, this must be done quickly to avoid unnecessary data transmission.
*
* <p> {@code MultiSubscriber}s are parameterized with a type {@code U} which
* represents some meaningful aggregate of the responses received. This
* would typically be a collection of response or response body objects.
*
* @param <U> a type representing the aggregated results
* @param <T> a type representing all of the response bodies
*
* @since 9
*/
public interface MultiSubscriber<U,T> {
/**
* Called for the main request from the user. This {@link HttpRequest}
* parameter is the request that was supplied to {@link
* HttpClient#sendAsync(HttpRequest, MultiSubscriber)}. The
* implementation must return an {@link BodyHandler} for the response
* body.
*
* @param request the request
*
* @return an optional body handler
*/
BodyHandler<T> onRequest(HttpRequest request);
/**
* Called for each push promise that is received. The {@link HttpRequest}
* parameter represents the PUSH_PROMISE. The implementation must return
* an {@code Optional} of {@link BodyHandler} for the response body.
* Different handlers (of the same type) can be returned for different
* pushes within the same multi send. If no handler (an empty {@code
* Optional}) is returned, then the push will be canceled. If required,
* the {@code CompletableFuture<Void>} supplied to the {@code
* onFinalPushPromise} parameter of {@link
* #completion(CompletableFuture, CompletableFuture)} can be used to
* determine when the final PUSH_PROMISE is received.
*
* @param pushPromise the push promise
*
* @return an optional body handler
*/
Optional<BodyHandler<T>> onPushPromise(HttpRequest pushPromise);
/**
* Called for each response received. For each request either one of
* onResponse() or onError() is guaranteed to be called, but not both.
*
* <p> Note: The reason for switching to this callback interface rather
* than using CompletableFutures supplied to onRequest() is that there
* is a subtle interaction between those CFs and the CF returned from
* completion() (or when onComplete() was called formerly). The completion()
* CF will not complete until after all of the work done by the onResponse()
* calls is done. Whereas if you just create CF's dependent on a supplied
* CF (to onRequest()) then the implementation has no visibility of the
* dependent CFs and can't guarantee to call onComplete() (or complete
* the completion() CF) after the dependent CFs complete.
*
* @param response the response received
*/
void onResponse(HttpResponse<T> response);
/**
* Called if an error occurs receiving a response. For each request
* either one of onResponse() or onError() is guaranteed to be called,
* but not both.
*
* @param request the main request or subsequent push promise
* @param t the Throwable that caused the error
*/
void onError(HttpRequest request, Throwable t);
/**
* Returns a {@link java.util.concurrent.CompletableFuture}{@code <U>}
* which completes when the aggregate result object itself is available.
* It is expected that the returned {@code CompletableFuture} will depend
* on one of the given {@code CompletableFuture<Void}s which themselves
* complete after all individual responses associated with the multi
* response have completed, or after all push promises have been received.
* This method is called after {@link #onRequest(HttpRequest)} but
* before any other methods.
*
* @implNote Implementations might follow the pattern shown below
* <pre>
* {@code
* CompletableFuture<U> completion(
* CompletableFuture<Void> onComplete,
* CompletableFuture<Void> onFinalPushPromise)
* {
* return onComplete.thenApply((v) -> {
* U u = ... instantiate and populate a U instance
* return u;
* });
* }
* }
* </pre>
*
* @param onComplete a CompletableFuture which completes after all
* responses have been received relating to this multi request.
*
* @param onFinalPushPromise CompletableFuture which completes after all
* push promises have been received.
*
* @return the aggregate CF response object
*/
CompletableFuture<U> completion(CompletableFuture<Void> onComplete,
CompletableFuture<Void> onFinalPushPromise);
/**
* Returns a general purpose handler for multi responses. The aggregated
* result object produced by this handler is a
* {@code Map<HttpRequest,CompletableFuture<HttpResponse<V>>>}. Each
* request (both the original user generated request and each server
* generated push promise) is returned as a key of the map. The value
* corresponding to each key is a
* {@code CompletableFuture<HttpResponse<V>>}.
*
* <p> There are two ways to use these handlers, depending on the value
* of the <i>completion</I> parameter. If completion is true, then the
* aggregated result will be available after all responses have
* themselves completed. If <i>completion</i> is false, then the
* aggregated result will be available immediately after the last push
* promise was received. In the former case, this implies that all the
* CompletableFutures in the map values will have completed. In the
* latter case, they may or may not have completed yet.
*
* <p> The simplest way to use these handlers is to set completion to
* {@code true}, and then all (results) values in the Map will be
* accessible without blocking.
* <p>
* See {@link #asMap(java.util.function.Function, boolean)}
* for a code sample of using this interface.
*
* <p> See {@link #asMap(Function, boolean)} for a code sample of using
* this interface.
*
* @param <V> the body type used for all responses
* @param reqHandler a function invoked for the user's request and each
* push promise
* @param completion {@code true} if the aggregate CompletableFuture
* completes after all responses have been received,
* or {@code false} after all push promises received
*
* @return a MultiSubscriber
*/
public static <V> MultiSubscriber<MultiMapResult<V>,V> asMap(
Function<HttpRequest, Optional<HttpResponse.BodyHandler<V>>> reqHandler,
boolean completion) {
return new MultiSubscriberImpl<V>(reqHandler.andThen(optv -> optv.get()),
reqHandler,
completion);
}
/**
* Returns a general purpose handler for multi responses. This is a
* convenience method which invokes {@link #asMap(Function,boolean)
* asMap(Function, true)} meaning that the aggregate result
* object completes after all responses have been received.
*
* <p><b>Example usage:</b>
* <br>
* <pre>
* {@code
* HttpRequest request = HttpRequest.newBuilder()
* .uri(URI.create("https://www.foo.com/"))
* .GET()
* .build();
*
* HttpClient client = HttpClient.newHttpClient();
*
* Map<HttpRequest,CompletableFuture<HttpResponse<String>>> results = client
* .sendAsync(request, MultiSubscriber.asMap(
* (req) -> Optional.of(HttpResponse.BodyHandler.asString())))
* .join();
* }</pre>
*
* <p> The lambda in this example is the simplest possible implementation,
* where neither the incoming requests are examined, nor the response
* headers, and every push that the server sends is accepted. When the
* join() call returns, all {@code HttpResponse}s and their associated
* body objects are available.
*
* @param <V> the body type used for all responses
* @param reqHandler a function invoked for each push promise and the
* main request
* @return a MultiSubscriber
*/
public static <V> MultiSubscriber<MultiMapResult<V>,V> asMap(
Function<HttpRequest, Optional<HttpResponse.BodyHandler<V>>> reqHandler) {
return asMap(reqHandler, true);
}
}
}