Developer Guide

Developer Guide

Welcome to the opensearch-api-specification developer guide! Glad you want to contribute. Here are the things you need to know while getting started!

Getting Started

Fork the opensearch-api-specification repository to your GitHub account and clone it to your local machine. Whenever you're drafting a change, create a new branch for the change on your fork instead of on the upstream repository.

Specification

The Specification is written in OpenAPI 3, so understanding the OpenAPI 3 specification is a must. If you are new to OpenAPI, you can start by reading the OpenAPI 3 Specification.

File Structure

To make editing the specification easier, we split the OpenAPI spec into multiple files that can be found in the spec directory. The file structure is as follows:

spec
│
├── namespaces
│   ├── _core.yaml
│   ├── cat.yaml
│   ├── cluster.yaml
│   ├── nodes.yaml
│   └── ...
│
├── schemas
│   ├── _common.yaml
│   ├── _common.mapping.yaml
│   ├── _core._common.yaml
│   ├── _core.bulk.yaml
│   ├── cat._common.yaml
│   ├── cat.aliases.yaml
│   └── ...
│
├── _info.yaml
├── _global_parameters.yaml
└── _superseded_operations.yaml

The API Operations are grouped by namespaces in spec/namespaces/ directory. Each file in this directory represents a namespace and holds all paths and operations of the namespace.
The data schemas are grouped by categories in spec/schemas/ directory. Each file in this directory represents a category.

Every .yaml file in the namespaces and schemas folders is a OpenAPI 3 document. This means that you can use any OpenAPI 3 compatible tool to view and edit the files, and IDEs with OpenAPI support will also offer autocomplete and validation in realtime.

Grouping Operations

Each API action is composed of multiple operations. The search action, for example, consists of 4 operations:

GET /_search
POST /_search
GET /{index}/_search
POST /{index}/_search

To group these operations together in the search action, we mark them with the x-operation-group extension with the same value of search. The value of x-operation-group is a string that follows the format [namespace].[action], except for the _core namespace where the namespace is omitted. For example, the search action in the _core namespace will have the x-operation-group value of search, and the create action in the indices namespace will have the x-operation-group value of indices.create.

Note that this extension tells the client generators that these operations serve identical purpose and should be grouped together in the same API method. This extension also tells the generators the namespace and the name of the API method. For example, operations with x-operation-group value of indicies.create will result in client.indices.create() method to be generated, while operation group of search will result in client.search() as it's part of the _core namespace.

For this reason, every operation must be accompanied by the x-operation-group extension, and operations in the same group MUST have identical descriptions, request and response bodies, and query string parameters.

Grouping Schemas

Schemas are grouped by categories to keep their names short, and aid in client generation (where the schemas are translated into data types/classes, and divided into packages/modules). The schema file names can be in one of the following formats:

_common category holds the common schemas that are used across multiple namespaces and features.
_common.<sub_category> category holds the common schemas of a specific sub_category. (e.g. _common.mapping)
<namespace>._common category holds the common schemas of a specific namespace. (e.g. cat._common, _core._common)
<namespace>.<action> category holds the schemas of a specific sub_category of a namespace. (e.g. cat.aliases, _core.search)

Superseded Operations

When an operation is superseded by another operation with identical functionality, that is a rename or a change in the URL, it should be listed in _superseded_operations.yaml file. The merger tool will automatically generate the superseded operation in the OpenAPI spec. The superseded operation will have deprecated: true and x-ignorable: true properties to indicate that it should be ignored by the client generator.

For example, if the _superseded_operations.yaml file contains the following entry:

/_opendistro/_anomaly_detection/{nodeId}/stats/{stat}:
  superseded_by: /_plugins/_anomaly_detection/{nodeId}/stats/{stat}
  operations:
    - GET
    - POST

Then, the merger tool will generate 2 superseded operations:

GET /_opendistro/_anomaly_detection/{nodeId}/stats/{stat}
POST /_opendistro/_anomaly_detection/{nodeId}/stats/{stat}

from their respective superseding operations:

GET /_plugins/_anomaly_detection/{nodeId}/stats/{stat}
POST /_plugins/_anomaly_detection/{nodeId}/stats/{stat}

if and only if the superseding operations exist in the spec. A warning will be printed on the console if they do not.

Note that the path parameter names do not need to match. So, if the actual superseding operations have path of /_plugins/_anomaly_detection/{node_id}/stats/{stat_id}, the merger tool will recognize that it is the same as /_plugins/_anomaly_detection/{nodeId}/stats/{stat} and generate the superseded operations accordingly with the correct path parameter names.

Global Parameters

Certain query parameters are global, and they are accepted by every operation. These parameters are listed in the spec/_global_parameters.yaml. The merger tool will automatically add these parameters to all operations.

OpenAPI Extensions

This repository includes several OpenAPI Specification Extensions to fill in any metadata not natively supported by OpenAPI:

x-operation-group: Used to group operations into API actions.
x-version-added: OpenSearch version when the operation/parameter was added.
x-version-deprecated: OpenSearch version when the operation/parameter was deprecated.
x-version-removed: OpenSearch version when the operation/parameter was removed.
x-deprecation-message: Reason for deprecation and guidance on how to prepare for the next major version.
x-ignorable: Denotes that the operation should be ignored by the client generator. This is used in operation groups where some operations have been replaced by newer ones, but we still keep them in the specs because the server still supports them.
x-global: Denotes that the parameter is a global parameter that is included in every operation. These parameters are listed in the spec/_global_parameters.yaml.
x-default: Contains the default value of a parameter. This is often used to override the default value specified in the schema, or to avoid accidentally changing the default value when updating a shared schema.
x-distributions-included: Contains a list of distributions known to include the API.
x-distributions-excluded: Contains a list of distributions known to exclude the API.

Use opensearch.org for the official distribution in x-distributions-*, amazon-managed for Amazon Managed OpenSearch, and amazon-serverless for Amazon OpenSearch Serverless.

Writing Spec Tests

To assure the correctness of the spec, you must add tests for the spec, when making changes. Check TESTING_GUIDE.md for more information.

Tools

A number of tools have been authored using TypeScript to aid in the development of the specification. These largely center around linting and merging the multi-file spec layout.

Setup

To be able to use or develop the tools, some setup is required:

Install Node.js.
Run npm ci from the repository's root.

Spec Merger

npm run merge -- --help

The merger tool merges the multi-file OpenSearch spec into a single file for programmatic use.

Arguments

--source <path>: The path to the root folder of the multi-file spec, defaults to <repository-root>/spec.
--output <path>: The path to write the final merged spec to, defaults to <repository-root>/build/opensearch-openapi.yaml.
--opensearch-version: An optional target version of OpenSearch, checking values of x-version-added and x-version-removed.

Example

We can take advantage of the default values and simply merge the specification via:

npm run merge

To generate a spec that does not contain any APIs or fields removed in version 2.0 (e.g. document _type fields).

npm run merge -- --opensearch-version=2.0

Spec Linter

npm run lint:spec -- --help

The linter tool validates the OpenSearch multi-file spec, and will print out all the errors and warnings in it.

Arguments

--source <path>: The path to the root folder of the multi-file spec, defaults to <repository-root>/spec.

Example

We can take advantage of the default values and simply lint the specification via:

npm run lint:spec

Dump Cluster Spec

npm run dump-cluster-spec -- --help

The dump-cluster-spec tool connects to an OpenSearch cluster which has the opensearch-api plugin installed and dumps the skeleton OpenAPI specification it provides to a file.

Arguments

--opensearch-url <url>: The URL at which the cluster is accessible, defaults to https://localhost:9200.
--opensearch-insecure: Disable SSL/TLS certificate verification, defaults to performing verification.
--opensearch-username <username>: The username to authenticate with the cluster, defaults to admin, only used when --opensearch-password is set.
--opensearch-password <password>: The password to authenticate with the cluster, also settable via the OPENSEARCH_PASSWORD environment variable.
--output <path>: The path to write the dumped spec to, defaults to <repository-root>/build/opensearch-openapi-CLUSTER.yaml.

Example

You can use this repo's docker image which includes the opensearch-api plugin to spin up a local development cluster with a self-signed certificate (e.g. https://localhost:9200) and security enabled, to then dump the skeleton specification:

OPENSEARCH_PASSWORD='My$3cureP@$$w0rd'

docker build ./coverage --tag opensearch-with-api-plugin
          
docker run \
    --name opensearch \
    --rm -d \
    -p 9200:9200 -p 9600:9600 \
    -e "discovery.type=single-node" \
    -e OPENSEARCH_INITIAL_ADMIN_PASSWORD="$OPENSEARCH_PASSWORD" \
    opensearch-with-api-plugin

OPENSEARCH_PASSWORD="${OPENSEARCH_PASSWORD}" npm run dump-cluster-spec -- --opensearch-insecure

docker stop opensearch

Coverage

npm run coverage:spec -- --help

The coverage tool determines which APIs from the OpenSearch cluster's reference skeleton specification (dumped by the dump-cluster-spec tool) are covered by this specification (as built by the merger tool).

Arguments

--cluster <path>: The path to the cluster's reference skeleton specification, as dumped by dump-cluster-spec, defaults to <repository-root>/build/opensearch-openapi-CLUSTER.yaml.
--specification <path>: The path to the merged specification, as built by merger, defaults to <repository-root>/build/opensearch-openapi.yaml.
--output <path>: The path to write the coverage data to, defaults to <repository-root>/build/coverage.json.

Example

Assuming you've already followed the previous examples to build the merged specification with the merger and dump the cluster's specification with dump-cluster-spec, you can then calculate the API coverage:

npm run coverage:spec

The output file build/coverage.json will now contain data of like below:

{
  "$description": {
    "uncovered": "Endpoints provided by the OpenSearch cluster but DO NOT exist in the specification",
    "covered": "Endpoints both provided by the OpenSearch cluster and exist in the specification",
    "specified_but_not_provided": "Endpoints NOT provided by the OpenSearch cluster but exist in the specification"
  },
  "counts": {
    "uncovered": 552,
    "uncovered_pct": 54.06,
    "covered": 469,
    "covered_pct": 45.94,
    "specified_but_not_provided": 23
  },
  "endpoints": {
    "uncovered": {
      "/_alias": [
        "put"
      ],
      ...
    },
    "covered": {
      "/_mapping": [
        "get"
      ],
      ...
    },
    "specified_but_not_provided": {
      "/_plugins/_knn/{}/stats": [
        "get"
      ],
      ...
    }
  }
}

Testing

Tests

All tools should have tests added in tools/tests, tests are implemented using Jest. They can be run via:

npm run test

Specify the test path to run tests for one of the tools:

npm run jest -- tools/tests/linter/lint.test.ts

The test suite contains unit tests and integration tests. Integration tests, such as these, require a local instance of OpenSearch and are placed into a folder named integ. Unit tests are run in parallel and integration tests are run sequentially using --runInBand. You can run unit tests with npm run test:unit separately from integration tests with npm run test:integ.

Lints

All TypeScript code and YAML files are linted using ESLint, typescript-eslint, and eslint-plugin-yml. Linting can be run via:

npm run lint

If a lint is unavoidable it should only be disabled on a case-by-case basis (e.g. // eslint-disable-next-line @typescript-eslint/dot-notation) and ideally be justified with an accompanying comment or at least in PR review.

ESLint's auto-fixes can be applied by running:

npm run lint--fix

Workflows

Analyze PR Changes

This workflow runs on all pull requests to analyze any potential changes to the specification. It uses the coverage tool and openapi-changes to calculate coverage metrics and provide a report on the changes when comparing with the commit at which the PR was branched off.

Build

This workflow runs on pushes to the main branch and will merge the specification and publish it to GitHub Releases.

Deploy GitHub Pages

This workflow performs a Jekyll build of the main branch to generate the Swagger docs and publish it to GitHub Pages.

Comment on PR

This workflow is triggered by the completion of the workflows such as Analyze PR Changes and downloading a JSON payload artifact which it uses to invoke a template from .github/pr-comment-templates to render a comment which is placed on the original triggering PR.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

DEVELOPER_GUIDE.md

DEVELOPER_GUIDE.md

Developer Guide

Getting Started

Specification

File Structure

Grouping Operations

Grouping Schemas

Superseded Operations

Global Parameters

OpenAPI Extensions

Writing Spec Tests

Tools

Setup

Spec Merger

Arguments

Example

Spec Linter

Arguments

Example

Dump Cluster Spec

Arguments

Example

Coverage

Arguments

Example

Testing

Tests

Lints

Workflows

Analyze PR Changes

Build

Deploy GitHub Pages

Comment on PR

Test Tools (Unit)

Test Tools (Integration)

Validate Spec

Files

DEVELOPER_GUIDE.md

Latest commit

History

DEVELOPER_GUIDE.md

File metadata and controls

Developer Guide

Getting Started

Specification

File Structure

Grouping Operations

Grouping Schemas

Superseded Operations

Global Parameters

OpenAPI Extensions

Writing Spec Tests

Tools

Setup

Arguments

Example

Arguments

Example

Arguments

Example

Arguments

Example

Testing

Tests

Lints

Workflows