ClientCall.java

/*
 * Copyright 2014 The gRPC Authors
 *
 * 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.
 */

package io.grpc;

import javax.annotation.Nullable;

/**
 * An instance of a call to a remote method. A call will send zero or more
 * request messages to the server and receive zero or more response messages back.
 *
 * <p>Instances are created
 * by a {@link Channel} and used by stubs to invoke their remote behavior.
 *
 * <p>More advanced usages may consume this interface directly as opposed to using a stub. Common
 * reasons for doing so would be the need to interact with flow-control or when acting as a generic
 * proxy for arbitrary operations.
 *
 * <p>{@link #start} must be called prior to calling any other methods, with the exception of
 * {@link #cancel}. Whereas {@link #cancel} must not be followed by any other methods,
 * but can be called more than once, while only the first one has effect.
 *
 * <p>No generic method for determining message receipt or providing acknowledgement is provided.
 * Applications are expected to utilize normal payload messages for such signals, as a response
 * naturally acknowledges its request.
 *
 * <p>Methods are guaranteed to be non-blocking. Not thread-safe except for {@link #request}, which
 * may be called from any thread.
 *
 * <p>There is no interaction between the states on the {@link Listener Listener} and {@link
 * ClientCall}, i.e., if {@link Listener#onClose Listener.onClose()} is called, it has no bearing on
 * the permitted operations on {@code ClientCall} (but it may impact whether they do anything).
 *
 * <p>There is a race between {@link #cancel} and the completion/failure of the RPC in other ways.
 * If {@link #cancel} won the race, {@link Listener#onClose Listener.onClose()} is called with
 * {@link Status#CANCELLED CANCELLED}. Otherwise, {@link Listener#onClose Listener.onClose()} is
 * called with whatever status the RPC was finished. We ensure that at most one is called.
 *
 * <h3>Usage examples</h3>
 * <h4>Simple Unary (1 request, 1 response) RPC</h4>
 * <pre>
 *   call = channel.newCall(unaryMethod, callOptions);
 *   call.start(listener, headers);
 *   call.sendMessage(message);
 *   call.halfClose();
 *   call.request(1);
 *   // wait for listener.onMessage()
 * </pre>
 *
 * <h4>Flow-control in Streaming RPC</h4>
 *
 * <p>The following snippet demonstrates a bi-directional streaming case, where the client sends
 * requests produced by a fictional <code>makeNextRequest()</code> in a flow-control-compliant
 * manner, and notifies gRPC library to receive additional response after one is consumed by
 * a fictional <code>processResponse()</code>.
 *
 * <p><pre>
 *   call = channel.newCall(bidiStreamingMethod, callOptions);
 *   listener = new ClientCall.Listener&lt;FooResponse&gt;() {
 *     &#64;Override
 *     public void onMessage(FooResponse response) {
 *       processResponse(response);
 *       // Notify gRPC to receive one additional response.
 *       call.request(1);
 *     }
 *
 *     &#64;Override
 *     public void onReady() {
 *       while (call.isReady()) {
 *         FooRequest nextRequest = makeNextRequest();
 *         if (nextRequest == null) {  // No more requests to send
 *           call.halfClose();
 *           return;
 *         }
 *         call.sendMessage(nextRequest);
 *       }
 *     }
 *   }
 *   call.start(listener, headers);
 *   // Notify gRPC to receive one response. Without this line, onMessage() would never be called.
 *   call.request(1);
 * </pre>
 *
 * <p>DO NOT MOCK: Use InProcessServerBuilder and make a test server instead.
 *
 * @param <ReqT> type of message sent one or more times to the server.
 * @param <RespT> type of message received one or more times from the server.
 */
public abstract class ClientCall<ReqT, RespT> {
  /**
   * Callbacks for receiving metadata, response messages and completion status from the server.
   *
   * <p>Implementations are free to block for extended periods of time. Implementations are not
   * required to be thread-safe, but they must not be thread-hostile. The caller is free to call
   * an instance from multiple threads, but only one call simultaneously. A single thread may
   * interleave calls to multiple instances, so implementations using ThreadLocals must be careful
   * to avoid leaking inappropriate state (e.g., clearing the ThreadLocal before returning).
   *
   * @param <T> type of message received.
   */
  public abstract static class Listener<T> {

    /**
     * The response headers have been received. Headers always precede messages.
     *
     * <p>Since {@link Metadata} is not thread-safe, the caller must not access (read or write)
     * {@code headers} after this point.
     *
     * @param headers containing metadata sent by the server at the start of the response.
     */
    public void onHeaders(Metadata headers) {}

    /**
     * A response message has been received. May be called zero or more times depending on whether
     * the call response is empty, a single message or a stream of messages.
     *
     * @param message returned by the server
     */
    public void onMessage(T message) {}

    /**
     * The ClientCall has been closed. Any additional calls to the {@code ClientCall} will not be
     * processed by the server. No further receiving will occur and no further notifications will be
     * made.
     *
     * <p>Since {@link Metadata} is not thread-safe, the caller must not access (read or write)
     * {@code trailers} after this point.
     *
     * <p>If {@code status} returns false for {@link Status#isOk()}, then the call failed.
     * An additional block of trailer metadata may be received at the end of the call from the
     * server. An empty {@link Metadata} object is passed if no trailers are received.
     *
     * <p>This method should not throw. If this method throws, there is no way to be notified of the
     * exception. Implementations should therefore be careful of exceptions which can accidentally 
     * leak resources.
     *
     * @param status the result of the remote call.
     * @param trailers metadata provided at call completion.
     */
    public void onClose(Status status, Metadata trailers) {}

    /**
     * This indicates that the ClientCall may now be capable of sending additional messages (via
     * {@link #sendMessage}) without requiring excessive buffering internally. This event is
     * just a suggestion and the application is free to ignore it, however doing so may
     * result in excessive buffering within the ClientCall.
     *
     * <p>Because there is a processing delay to deliver this notification, it is possible for
     * concurrent writes to cause {@code isReady() == false} within this callback. Handle "spurious"
     * notifications by checking {@code isReady()}'s current value instead of assuming it is now
     * {@code true}. If {@code isReady() == false} the normal expectations apply, so there would be
     * <em>another</em> {@code onReady()} callback.
     *
     * <p>If the type of a call is either {@link MethodDescriptor.MethodType#UNARY} or
     * {@link MethodDescriptor.MethodType#SERVER_STREAMING}, this callback may not be fired. Calls
     * that send exactly one message should not await this callback.
     */
    public void onReady() {}
  }

  /**
   * Start a call, using {@code responseListener} for processing response messages.
   *
   * <p>It must be called prior to any other method on this class, except for {@link #cancel} which
   * may be called at any time.
   *
   * <p>Since {@link Metadata} is not thread-safe, the caller must not access (read or write) {@code
   * headers} after this point.
   *
   * @param responseListener receives response messages
   * @param headers which can contain extra call metadata, e.g. authentication credentials.
   * @throws IllegalStateException if a method (including {@code start()}) on this class has been
   *                               called.
   */
  public abstract void start(Listener<RespT> responseListener, Metadata headers);

  /**
   * Requests up to the given number of messages from the call to be delivered to
   * {@link Listener#onMessage(Object)}. No additional messages will be delivered.
   *
   * <p>Message delivery is guaranteed to be sequential in the order received. In addition, the
   * listener methods will not be accessed concurrently. While it is not guaranteed that the same
   * thread will always be used, it is guaranteed that only a single thread will access the listener
   * at a time.
   *
   * <p>If it is desired to bypass inbound flow control, a very large number of messages can be
   * specified (e.g. {@link Integer#MAX_VALUE}).
   *
   * <p>If called multiple times, the number of messages able to delivered will be the sum of the
   * calls.
   *
   * <p>This method is safe to call from multiple threads without external synchronization.
   *
   * @param numMessages the requested number of messages to be delivered to the listener. Must be
   *                    non-negative.
   * @throws IllegalStateException if call is not {@code start()}ed
   * @throws IllegalArgumentException if numMessages is negative
   */
  public abstract void request(int numMessages);

  /**
   * Prevent any further processing for this {@code ClientCall}. No further messages may be sent or
   * will be received. The server is informed of cancellations, but may not stop processing the
   * call. Cancellation is permitted even if previously {@link #halfClose}d. Cancelling an already
   * {@code cancel()}ed {@code ClientCall} has no effect.
   *
   * <p>No other methods on this class can be called after this method has been called.
   *
   * <p>It is recommended that at least one of the arguments to be non-{@code null}, to provide
   * useful debug information. Both argument being null may log warnings and result in suboptimal
   * performance. Also note that the provided information will not be sent to the server.
   *
   * @param message if not {@code null}, will appear as the description of the CANCELLED status
   * @param cause if not {@code null}, will appear as the cause of the CANCELLED status
   */
  public abstract void cancel(@Nullable String message, @Nullable Throwable cause);

  /**
   * Close the call for request message sending. Incoming response messages are unaffected.  This
   * should be called when no more messages will be sent from the client.
   *
   * @throws IllegalStateException if call is already {@code halfClose()}d or {@link #cancel}ed
   */
  public abstract void halfClose();

  /**
   * Send a request message to the server. May be called zero or more times depending on how many
   * messages the server is willing to accept for the operation.
   *
   * @param message message to be sent to the server.
   * @throws IllegalStateException if call is {@link #halfClose}d or explicitly {@link #cancel}ed
   */
  public abstract void sendMessage(ReqT message);

  /**
   * If {@code true}, indicates that the call is capable of sending additional messages
   * without requiring excessive buffering internally. This event is
   * just a suggestion and the application is free to ignore it, however doing so may
   * result in excessive buffering within the call.
   *
   * <p>If {@code false}, {@link Listener#onReady()} will be called after {@code isReady()}
   * transitions to {@code true}.
   *
   * <p>If the type of the call is either {@link MethodDescriptor.MethodType#UNARY} or
   * {@link MethodDescriptor.MethodType#SERVER_STREAMING}, this method may persistently return
   * false. Calls that send exactly one message should not check this method.
   *
   * <p>This abstract class's implementation always returns {@code true}. Implementations generally
   * override the method.
   */
  public boolean isReady() {
    return true;
  }

  /**
   * Enables per-message compression, if an encoding type has been negotiated.  If no message
   * encoding has been negotiated, this is a no-op. By default per-message compression is enabled,
   * but may not have any effect if compression is not enabled on the call.
   */
  @ExperimentalApi("https://github.com/grpc/grpc-java/issues/1703")
  public void setMessageCompression(boolean enabled) {
    // noop
  }

  /**
   * Returns additional properties of the call. May only be called after {@link Listener#onHeaders}
   * or {@link Listener#onClose}. If called prematurely, the implementation may throw {@code
   * IllegalStateException} or return arbitrary {@code Attributes}.
   *
   * @return non-{@code null} attributes
   * @throws IllegalStateException (optional) if called before permitted
   */
  @ExperimentalApi("https://github.com/grpc/grpc-java/issues/2607")
  @Grpc.TransportAttr
  public Attributes getAttributes() {
    return Attributes.EMPTY;
  }
}