/grpc-gateway

gRPC to JSON proxy generator following the gRPC HTTP spec

Primary LanguageGoBSD 3-Clause "New" or "Revised" LicenseBSD-3-Clause

gRPC-Gateway

gRPC to JSON proxy generator following the gRPC HTTP spec

About

The gRPC-Gateway is a plugin of the Google protocol buffers compiler protoc. It reads protobuf service definitions and generates a reverse-proxy server which translates a RESTful HTTP API into gRPC. This server is generated according to the google.api.http annotations in your service definitions.

This helps you provide your APIs in both gRPC and RESTful style at the same time.

Docs

You can read our docs at:

Testimonials

We use the gRPC-Gateway to serve millions of API requests per day, and have been since 2018 and through all of that, we have never had any issues with it.

- William Mill, Ad Hoc

Background

gRPC is great -- it generates API clients and server stubs in many programming languages, it is fast, easy-to-use, bandwidth-efficient and its design is combat-proven by Google. However, you might still want to provide a traditional RESTful JSON API as well. Reasons can range from maintaining backward-compatibility, supporting languages or clients that are not well supported by gRPC, to simply maintaining the aesthetics and tooling involved with a RESTful JSON architecture.

This project aims to provide that HTTP+JSON interface to your gRPC service. A small amount of configuration in your service to attach HTTP semantics is all that's needed to generate a reverse-proxy with this library.

Installation

Compile from source

The following instructions assume you are using Go Modules for dependency management. Use a tool dependency to track the versions of the following executable packages:

// +build tools

package tools

import (
    _ "github.com/monime-lab/grpc-gateway/v2/protoc-gen-grpc-gateway"
    _ "github.com/monime-lab/grpc-gateway/v2/protoc-gen-openapiv2"
    _ "google.golang.org/grpc/cmd/protoc-gen-go-grpc"
    _ "google.golang.org/protobuf/cmd/protoc-gen-go"
)

Run go mod tidy to resolve the versions. Install by running

$ go install \
    github.com/monime-lab/grpc-gateway/v2/protoc-gen-grpc-gateway \
    github.com/monime-lab/grpc-gateway/v2/protoc-gen-openapiv2 \
    google.golang.org/protobuf/cmd/protoc-gen-go \
    google.golang.org/grpc/cmd/protoc-gen-go-grpc

This will place four binaries in your $GOBIN;

  • protoc-gen-grpc-gateway
  • protoc-gen-openapiv2
  • protoc-gen-go
  • protoc-gen-go-grpc

Make sure that your $GOBIN is in your $PATH.

Download the binaries

You may alternatively download the binaries from the GitHub releases page. We generate SLSA3 signatures using the OpenSSF's slsa-framework/slsa-github-generator during the release process. To verify a release binary:

  1. Install the verification tool from slsa-framework/slsa-verifier#installation.
  2. Download the provenance file attestation.intoto.jsonl from the GitHub releases page.
  3. Run the verifier:
slsa-verifier -artifact-path <the-binary> -provenance attestation.intoto.jsonl -source github.com/monime-lab/grpc-gateway -tag <the-tag>

Alternatively, see the section on remotely managed plugin versions below.

Usage

  1. Define your gRPC service using protocol buffers

    your_service.proto:

     syntax = "proto3";
 package your.service.v1;
 option go_package = "github.com/yourorg/yourprotos/gen/go/your/service/v1";

 message StringMessage {
   string value = 1;
 }

 service YourService {
   rpc Echo(StringMessage) returns (StringMessage) {}
 }

  2. Generate gRPC stubs

    This step generates the gRPC stubs that you can use to implement the service and consume from clients:

    Here's an example buf.gen.yaml you can use to generate the stubs with buf:

    version: v1
plugins:
  - name: go
    out: gen/go
    opt:
      - paths=source_relative
  - name: go-grpc
    out: gen/go
    opt:
      - paths=source_relative

    With this file in place, you can generate your files using buf generate.

    For a complete example of using buf generate to generate protobuf stubs, see the boilerplate repo. For more information on generating the stubs with buf, see the official documentation.

    If you are using protoc to generate stubs, here's an example of what a command might look like:

    protoc -I . \
    --go_out ./gen/go/ --go_opt paths=source_relative \
    --go-grpc_out ./gen/go/ --go-grpc_opt paths=source_relative \
    your/service/v1/your_service.proto

  3. Implement your service in gRPC as usual.

  4. Generate reverse-proxy using protoc-gen-grpc-gateway

    At this point, you have 3 options:

    • no further modifications, use the default mapping to HTTP semantics (method, path, etc.)
      • this will work on any .proto file, but will not allow setting HTTP paths, request parameters or similar
    • additional .proto modifications to use a custom mapping
      • relies on parameters in the .proto file to set custom HTTP mappings
    • no .proto modifications, but use an external configuration file
      • relies on an external configuration file to set custom HTTP mappings
      • mostly useful when the source proto file isn't under your control
    1. Using the default mapping

    This requires no additional modification to the .proto file but does require enabling a specific option when executing the plugin. The generate_unbound_methods should be enabled.

    Here's what a buf.gen.yaml file might look like with this option enabled:

    version: v1
plugins:
  - name: go
    out: gen/go
    opt:
      - paths=source_relative
  - name: go-grpc
    out: gen/go
    opt:
      - paths=source_relative
  - name: grpc-gateway
    out: gen/go
    opt:
      - paths=source_relative
      - generate_unbound_methods=true

    With protoc (just the grpc-gateway stubs):

    protoc -I . --grpc-gateway_out ./gen/go \
    --grpc-gateway_opt logtostderr=true \
    --grpc-gateway_opt paths=source_relative \
    --grpc-gateway_opt generate_unbound_methods=true \
    your/service/v1/your_service.proto
    1. With custom annotations

    Add a google.api.http annotation to your .proto file

    your_service.proto:

     syntax = "proto3";
 package your.service.v1;
 option go_package = "github.com/yourorg/yourprotos/gen/go/your/service/v1";
+
+import "google/api/annotations.proto";
+
 message StringMessage {
   string value = 1;
 }

 service YourService {
-  rpc Echo(StringMessage) returns (StringMessage) {}
+  rpc Echo(StringMessage) returns (StringMessage) {
+    option (google.api.http) = {
+      post: "/v1/example/echo"
+      body: "*"
+    };
+  }
 }

    You will need to provide the required third party protobuf files to the protobuf compiler. If you are using buf, this dependency can be added to the deps array in your buf.yaml under the name buf.build/googleapis/googleapis:

    version: v1
name: buf.build/yourorg/myprotos
deps:
  - buf.build/googleapis/googleapis

    Always run buf mod update after adding a dependency to your buf.yaml.

    See a_bit_of_everything.proto for examples of more annotations you can add to customize gateway behavior and generated OpenAPI output.

    Here's what a buf.gen.yaml file might look like:

    version: v1
plugins:
  - name: go
    out: gen/go
    opt:
      - paths=source_relative
  - name: go-grpc
    out: gen/go
    opt:
      - paths=source_relative
  - name: grpc-gateway
    out: gen/go
    opt:
      - paths=source_relative

    If you are using protoc to generate stubs, you need to ensure the required dependencies are available to the compiler at compile time. These can be found by manually cloning and copying the relevant files from the googleapis repository, and providing them to protoc when running. The files you will need are:

    google/api/annotations.proto
google/api/field_behavior.proto
google/api/http.proto
google/api/httpbody.proto

    Here's what a protoc execution might look like:

    protoc -I . --grpc-gateway_out ./gen/go \
    --grpc-gateway_opt logtostderr=true \
    --grpc-gateway_opt paths=source_relative \
    your/service/v1/your_service.proto
    1. External configuration If you do not want to (or cannot) modify the proto file for use with gRPC-Gateway you can alternatively use an external gRPC Service Configuration file. Check our documentation for more information. This is best combined with the standalone=true option to generate a file that can live in its own package, separate from the files generated by the source protobuf file.

    Here's what a buf.gen.yaml file might look like with this option enabled:

    version: v1
plugins:
  - name: go
    out: gen/go
    opt:
      - paths=source_relative
  - name: go-grpc
    out: gen/go
    opt:
      - paths=source_relative
  - name: grpc-gateway
    out: gen/go
    opt:
      - paths=source_relative
      - grpc_api_configuration=path/to/config.yaml
      - standalone=true

    With protoc (just the grpc-gateway stubs):

    protoc -I . --grpc-gateway_out ./gen/go \
    --grpc-gateway_opt logtostderr=true \
    --grpc-gateway_opt paths=source_relative \
    --grpc-gateway_opt grpc_api_configuration=path/to/config.yaml \
    --grpc-gateway_opt standalone=true \
    your/service/v1/your_service.proto

  5. Write an entrypoint for the HTTP reverse-proxy server

    package main

import (
  "context"
  "flag"
  "net/http"

  "github.com/golang/glog"
  "github.com/monime-lab/grpc-gateway/v2/runtime"
  "google.golang.org/grpc"
  "google.golang.org/grpc/credentials/insecure"

  gw "github.com/yourorg/yourrepo/proto/gen/go/your/service/v1/your_service"  // Update
)

var (
  // command-line options:
  // gRPC server endpoint
  grpcServerEndpoint = flag.String("grpc-server-endpoint",  "localhost:9090", "gRPC server endpoint")
)

func run() error {
  ctx := context.Background()
  ctx, cancel := context.WithCancel(ctx)
  defer cancel()

  // Register gRPC server endpoint
  // Note: Make sure the gRPC server is running properly and accessible
  mux := runtime.NewServeMux()
  opts := []grpc.DialOption{grpc.WithTransportCredentials(insecure.NewCredentials())}
  err := gw.RegisterYourServiceHandlerFromEndpoint(ctx, mux,  *grpcServerEndpoint, opts)
  if err != nil {
    return err
  }

  // Start HTTP server (and proxy calls to gRPC server endpoint)
  return http.ListenAndServe(":8081", mux)
}

func main() {
  flag.Parse()
  defer glog.Flush()

  if err := run(); err != nil {
    glog.Fatal(err)
  }
}

  6. (Optional) Generate OpenAPI definitions using protoc-gen-openapiv2

    Here's what a buf.gen.yaml file might look like:

    version: v1
plugins:
  - name: go
    out: gen/go
    opt:
      - paths=source_relative
  - name: go-grpc
    out: gen/go
    opt:
      - paths=source_relative
  - name: grpc-gateway
    out: gen/go
    opt:
      - paths=source_relative
  - name: openapiv2
    out: gen/openapiv2

    To use the custom protobuf annotations supported by protoc-gen-openapiv2, we need another dependency added to our protobuf generation step. If you are using buf, you can add the buf.build/monime-lab/grpc-gateway dependency to your deps array:

    version: v1
name: buf.build/yourorg/myprotos
deps:
  - buf.build/googleapis/googleapis
  - buf.build/monime-lab/grpc-gateway

    With protoc (just the swagger file):

    protoc -I . --openapiv2_out ./gen/openapiv2 \
    --openapiv2_opt logtostderr=true \
    your/service/v1/your_service.proto

    If you are using protoc to generate stubs, you will need to copy the protobuf files from the protoc-gen-openapiv2/options directory of this repository, and providing them to protoc when running.

    Note that this plugin also supports generating OpenAPI definitions for unannotated methods; use the generate_unbound_methods option to enable this.

    It is possible with the HTTP mapping for a gRPC service method to create duplicate mappings with the only difference being constraints on the path parameter.

    /v1/{name=projects/*} and /v1/{name=organizations/*} both become /v1/{name}. When this occurs the plugin will rename the path parameter with a "_1" (or "_2" etc) suffix to differentiate the different operations. So in the above example, the 2nd path would become /v1/{name_1=organizations/*}. This can also cause OpenAPI clients to URL encode the "/" that is part of the path parameter as that is what OpenAPI defines in the specification. To allow gRPC gateway to accept the URL encoded slash and still route the request, use the UnescapingModeAllCharacters or UnescapingModeLegacy (which is the default currently though may change in future versions). See Customizing Your Gateway for more information.

Usage with remote plugins

As an alternative to all of the above, you can use buf with remote plugins to manage plugin versions and generation. An example buf.gen.yaml using remote plugin generation looks like this:

version: v1
plugins:
  - remote: buf.build/library/plugins/go:v1.27.1-1
    out: gen/go
    opt:
      - paths=source_relative
  - remote: buf.build/library/plugins/go-grpc:v1.1.0-2
    out: gen/go
    opt:
      - paths=source_relative
  - remote: buf.build/grpc-ecosystem/plugins/grpc-gateway:v2.6.0-1
    out: gen/go
    opt:
      - paths=source_relative
  - remote: buf.build/grpc-ecosystem/plugins/openapiv2:v2.6.0-1
    out: gen/openapiv2

This requires no local installation of any plugins. Be careful to use the same version of the generator as the runtime library, i.e. if using v2.6.0-1, run

$ go get github.com/monime-lab/grpc-gateway/v2@v2.6.0

To get the same version of the runtime in your go.mod.

Video intro

This GopherCon UK 2019 presentation from our maintainer @JohanBrandhorst provides a good intro to using the gRPC-Gateway. It uses the following boilerplate repo as a base: https://github.com/johanbrandhorst/grpc-gateway-boilerplate.

Parameters and flags

When using buf to generate stubs, flags and parameters are passed through the opt field in your buf.gen.yaml file, for example:

version: v1
plugins:
  - name: grpc-gateway
    out: gen/go
    opt:
      - paths=source_relative
      - grpc_api_configuration=path/to/config.yaml
      - standalone=true

During code generation with protoc, flags to gRPC-Gateway tools must be passed through protoc using one of 2 patterns:

  • as part of the --<tool_suffix>_out protoc parameter: --<tool_suffix>_out=<flags>:<path>
--grpc-gateway_out=logtostderr=true,repeated_path_param_separator=ssv:.
--openapiv2_out=logtostderr=true,repeated_path_param_separator=ssv:.
  • using additional --<tool_suffix>_opt parameters: --<tool_suffix>_opt=<flag>[,<flag>]*
--grpc-gateway_opt logtostderr=true,repeated_path_param_separator=ssv
# or separately
--grpc-gateway_opt logtostderr=true --grpc-gateway_opt repeated_path_param_separator=ssv
--openapiv2_opt logtostderr=true,repeated_path_param_separator=ssv
# or separately
--openapiv2_opt logtostderr=true --openapiv2_opt repeated_path_param_separator=ssv

More examples

More examples are available under the examples directory.

  • proto/examplepb/echo_service.proto, proto/examplepb/a_bit_of_everything.proto, proto/examplepb/unannotated_echo_service.proto: service definition
    • proto/examplepb/echo_service.pb.go, proto/examplepb/a_bit_of_everything.pb.go, proto/examplepb/unannotated_echo_service.pb.go: [generated] stub of the service
    • proto/examplepb/echo_service.pb.gw.go, proto/examplepb/a_bit_of_everything.pb.gw.go, proto/examplepb/uannotated_echo_service.pb.gw.go: [generated] reverse proxy for the service
    • proto/examplepb/unannotated_echo_service.yaml: gRPC API Configuration for unannotated_echo_service.proto
  • server/main.go: service implementation
  • main.go: entrypoint of the generated reverse proxy

To use the same port for custom HTTP handlers (e.g. serving swagger.json), gRPC-Gateway, and a gRPC server, see this example by CoreOS (and its accompanying blog post).

Features

Supported

  • Generating JSON API handlers.
  • Method parameters in the request body.
  • Method parameters in the request path.
  • Method parameters in the query string.
  • Enum fields in the path parameter (including repeated enum fields).
  • Mapping streaming APIs to newline-delimited JSON streams.
  • Mapping HTTP headers with Grpc-Metadata- prefix to gRPC metadata (prefixed with grpcgateway-)
  • Optionally emitting API definitions for OpenAPI (Swagger) v2.
  • Setting gRPC timeouts through inbound HTTP Grpc-Timeout header.
  • Partial support for gRPC API Configuration files as an alternative to annotation.
  • Automatically translating PATCH requests into Field Mask gRPC requests. See the docs for more information.

No plan to support

But patches are welcome.

  • Method parameters in HTTP headers.
  • Handling trailer metadata.
  • Encoding request/response body in XML.
  • True bi-directional streaming.

Mapping gRPC to HTTP

  • How gRPC error codes map to HTTP status codes in the response .
  • HTTP request source IP is added as X-Forwarded-For gRPC request header.
  • HTTP request host is added as X-Forwarded-Host gRPC request header.
  • HTTP Authorization header is added as authorization gRPC request header.
  • Remaining Permanent HTTP header keys (as specified by the IANA here) are prefixed with grpcgateway- and added with their values to gRPC request header.
  • HTTP headers that start with 'Grpc-Metadata-' are mapped to gRPC metadata (prefixed with grpcgateway-).
  • While configurable, the default {un,}marshaling uses protojson.
  • The path template used to map gRPC service methods to HTTP endpoints supports the google.api.http path template syntax. For example, /api/v1/{name=projects/*/topics/*} or /prefix/{path=organizations/**}.

Contribution

See CONTRIBUTING.md.

License

gRPC-Gateway is licensed under the BSD 3-Clause License. See LICENSE.txt for more details.