Google API Compiler

Overview

Google API Compiler (Api Compiler) is an open source tool for processing API specifications. It currently supports OpenAPI specification, Protocol Buffers (proto), and Google API Service Configuration (service config), and can be extended to support other formats.

Google API Compiler parses the input files into object models, processes and validates the models, and generate various outputs, such as:

• Validation warnings and errors.
• Normalized/validated service config.
• API discovery document.
• API reference documentation.
• API client libraries.

Google API Service Configuration

Google API Service Configuration is a specification that defines the surface and behavior of an API service, including interface, types, methods, authentication, discovery, documentation, logging, monitoring and more. It is formally defined by the proto message google.api.Service and works with both REST and RPC APIs. Developers typically create the service config using YAML files, and use the Google API Compiler to generate the proto message.

NOTE: Google API Service Configuration is a rich and mature specification used for Google production services, such as Cloud Logging, Cloud Vision, Cloud Bigtable, IAM, and more.

Used by other tools.

Google API compiler is used by other tools like googleapis/toolkit to read the users API definition and autogenerate client libraries.

Compile Google API Compiler

Clone the Google API Compiler repo

git clone https://github.com/googleapis/api-compiler

Update submodules

git submodule update --recursive --init

Build source code

./gradlew buildGoogleApiConfigGen

For running tests, you need to have protoc in your path. If you don't already have protoc version 3, you can download it from https://github.com/google/protobuf/releases and set a symbolic link to the protoc.

# Example
sudo ln -s  <Path to the downloaded protoc> /usr/local/bin/protoc

Creating service config from proto files

Creating a proto descriptor file

Google API Compiler does not consume the proto files directly. Developers need to use protoc to generate the proto descriptor, then feed it to the Google API Compiler.

# Creates a proto descriptor from proto files using protoc.
protoc <file1.proto> <file2.proto> --include_source_info --descriptor_set_out=out.descriptors

Create service config

# -------------File: myapi.yaml-----------------

# The schema of this file.
type: google.api.Service

# The version of the service config.
config_version: 3

# The service name. It should be the primary DNS name for the service.
name: library-example.googleapis.com

# The official title of this service.
title: Google Example Library API

# The list of API interfaces exposed by the service.
apis:
- name: google.example.library.v1.LibraryService

# Other aspects of the service, such as authentication.
# ...

Executing the Google API Compiler

Once the jar 'gapi-config-gen-with-deps-0.0.0-SNAPSHOT.jar' is built under the build/libs directory, you can execute the jar using the following command:

alias gapi-config-gen='java -jar <path to gapi-config-gen-with-deps-0.0.0-SNAPSHOT.jar>'
DESCRIPTOR_FILE=<PATH TO out.descriptor>
CONFIG_FILE=<path to yaml file>
JSON_FILE_NAME=<json output file name>
BINARY_FILE_NAME=<binary output file name>

gapi-config-gen \
--configs $CONFIG_FILE \
--descriptor $DESCRIPTOR_FILE \
--json_out $JSON_FILE_NAME \
--bin_out $BINARY_FILE_NAME

This command will output the service configuration in different formats:

• Binary file: $BINARY_FILE_NAME
• Json file: $JSON_FILE_NAME

Any of these can be used as input to an Endpoints API server.

Creating service config from an OpenAPI Spec

Validate the OpenAPI Spec and create the service configuration.

alias gapi-config-gen='java -jar <path to gapi-service-config-gen-with-deps-0.0.0-SNAPSHOT.jar>'
OPENAPI_FILE=<OpenAPI Spec filename>
JSON_FILE_NAME=<json output file name>
BINARY_FILE_NAME=<binary output file name>

gapi-config-gen \
--openapi $OPENAPI_FILE \
--json_out $JSON_FILE_NAME \
--bin_out $BINARY_FILE_NAME

This will create the service config:

• Binary file: $BINARY_FILE_NAME
• JSON file: $JSON_FILE_NAME

Any of these can be used as input to an Endpoints API server.