View Source GRPC.Server (grpc v0.9.0)
A gRPC server which handles requests by calling user-defined functions.
You should pass a GRPC.Service
in when use the module:
defmodule Greeter.Service do
use GRPC.Service, name: "ping"
rpc :SayHello, Request, Reply
rpc :SayGoodbye, stream(Request), stream(Reply)
end
defmodule Greeter.Server do
use GRPC.Server, service: Greeter.Service
def say_hello(request, _stream) do
Reply.new(message: "Hello")
end
def say_goodbye(request_enum, stream) do
requests = Enum.map request_enum, &(&1)
GRPC.Server.send_reply(stream, reply1)
GRPC.Server.send_reply(stream, reply2)
end
end
Your functions should accept a client request and a GRPC.Server.Stream
.
The request will be a Enumerable.t
(created by Elixir's Stream
) of requests
if it's streaming. If a reply is streaming, you need to call send_reply/2
to send
replies one by one instead of returning reply in the end.
gRPC HTTP/JSON transcoding
Transcoding can be enabled by using the option http_transcode: true
:
defmodule Greeter.Service do
use GRPC.Service, name: "ping"
rpc :SayHello, Request, Reply
rpc :SayGoodbye, stream(Request), stream(Reply)
end
defmodule Greeter.Server do
use GRPC.Server, service: Greeter.Service, http_transcode: true
def say_hello(request, _stream) do
Reply.new(message: "Hello" <> request.name)
end
def say_goodbye(request_enum, stream) do
requests = Enum.map request_enum, &(&1)
GRPC.Server.send_reply(stream, reply1)
GRPC.Server.send_reply(stream, reply2)
end
end
With transcoding enabled gRPC methods can be used over HTTP/1 with JSON i.e
POST localhost/helloworld.Greeter/SayHello`
Content-Type: application/json
{
"message": "gRPC"
}
HTTP/1.1 200 OK
Content-Type: application/json
{
"message": "Hello gRPC"
}
By using option (google.api.http)
annotations in the .proto
file the mapping between
HTTP/JSON to gRPC methods and parameters can be customized:
syntax = "proto3";
import "google/api/annotations.proto";
import "google/protobuf/timestamp.proto";
package helloworld;
service Greeter {
rpc SayHello (HelloRequest) returns (HelloReply) {
option (google.api.http) = {
get: "/v1/greeter/{name}"
};
}
}
message HelloRequest {
string name = 1;
}
message HelloReply {
string message = 1;
}
In addition to the POST localhost/helloworld.Greeter/SayHello
route in the previous examples
this creates an additional route: GET localhost/v1/greeter/:name
GET localhost/v1/greeter/gRPC
Accept: application/json
HTTP/1.1 200 OK
Content-Type: application/json
{
"message": "Hello gRPC"
}
For more comprehensive documentation on annotation usage in .proto
files see
Summary
Functions
Send custom metadata(headers).
Send streaming reply.
Set compressor to compress responses. An accepted compressor will be set if clients use one,
even if set_compressor
is not called. But this can be called to override the chosen.
Set custom metadata(headers).
Set custom trailers, which will be sent in the end.
Types
@type rpc() :: (rpc_req(), GRPC.Server.Stream.t() -> rpc_return())
@type rpc_req() :: struct() | Enumerable.t()
Functions
@spec send_headers(GRPC.Server.Stream.t(), map()) :: GRPC.Server.Stream.t()
Send custom metadata(headers).
You can send headers only once, before that you can set headers using set_headers/2
.
Send streaming reply.
Examples
iex> GRPC.Server.send_reply(stream, reply)
@spec set_compressor(GRPC.Server.Stream.t(), module()) :: GRPC.Server.Stream.t()
Set compressor to compress responses. An accepted compressor will be set if clients use one,
even if set_compressor
is not called. But this can be called to override the chosen.
@spec set_headers(GRPC.Server.Stream.t(), map()) :: GRPC.Server.Stream.t()
Set custom metadata(headers).
You can set headers more than once.
@spec set_trailers(GRPC.Server.Stream.t(), map()) :: GRPC.Server.Stream.t()
Set custom trailers, which will be sent in the end.