graphql-compose-elasticsearch

Warning

This might be an unstable version to experiment fixes and things from graphql-compose-elasticsearch

This module expose Elastic Search REST API via GraphQL.

Elastic Search REST API proxy

Supported all elastic versions that support official elasticsearch-js client. Internally it parses its source code annotations and generates all available methods with params and descriptions to GraphQL Field Config Map. You may put this config map to any GraphQL Schema.

import { GraphQLSchema, GraphQLObjectType } from 'graphql';
import elasticsearch from 'elasticsearch';
import { elasticApiFieldConfig } from 'graphql-compose-elasticsearch';

const schema = new GraphQLSchema({
  query: new GraphQLObjectType({
    name: 'Query',
    fields: {
      elastic50: elasticApiFieldConfig(
        // you may provide existed Elastic Client instance
        new elasticsearch.Client({
          host: 'http://localhost:9200',
          apiVersion: '5.0',
        })
      ),

      // or may provide just config
      elastic24: elasticApiFieldConfig({
        host: 'http://user:pass@localhost:9200',
        apiVersion: '2.4',
      }),

      elastic17: elasticApiFieldConfig({
        host: 'http://user:pass@localhost:9200',
        apiVersion: '1.7',
      }),
    },
  }),
});

Full code example

Live demo of Introspection of Elasticsearch API via Graphiql

ObjectTypeComposer from Elastic mapping

In other side this module is a plugin for graphql-compose, which derives GraphQLType from your elastic mapping generates tons of types, provides all available methods in QueryDSL, Aggregations, Sorting with field autocompletion according to types in your mapping (like Dev Tools Console in Kibana).

Generated ObjectTypeComposer model has several awesome resolvers:

• search - greatly simplified elastic search method. According to GraphQL adaptation and its projection bunch of params setup automatically due your graphql query (eg _source, explain, version, trackScores), other rare fine tuning params moved to opts input field.
• searchConnection - elastic search method that implements Relay Cursor Connection spec for infinite lists. Internally it uses cheap search_after API. One downside, Elastic does not support backward scrolling, so before argument will not work.
• searchPagination - elastic search method that has page and perPage arguments
• findById - get elastic record by id
• updateById - update elastic record by id
• feel free to add your resolver or ask for a new one

import { GraphQLSchema, GraphQLObjectType } from 'graphql';
import elasticsearch from 'elasticsearch';
import { composeWithElastic } from 'graphql-compose-elasticsearch';

const mapping = {
  properties: {
    name: {
      type: 'text',
      fields: {
        keyword: {
          type: 'keyword',
        },
      },
    },
    gender: {
      type: 'keyword',
    },
    skills: {
      type: 'text',
    },
    languages: {
      type: 'keyword',
    },
    location: {
      properties: {
        name: {
          type: 'text',
        },
        point: {
          type: 'geo_point',
        },
      },
    },
    createdAt: {
      type: 'date',
    },
  },
};

const UserTC = composeWithElastic({
  graphqlTypeName: 'UserES',
  elasticIndex: 'user',
  elasticType: 'user',
  elasticMapping: mapping,
  elasticClient: new elasticsearch.Client({
    host: 'http://localhost:9200',
    apiVersion: '5.0',
    log: 'trace',
  }),
  // elastic mapping does not contain information about is fields are arrays or not
  // so provide this information explicitly for obtaining correct types in GraphQL
  pluralFields: ['skills', 'languages'],
});

const Schema = new GraphQLSchema({
  query: new GraphQLObjectType({
    name: 'Query',
    fields: {
      user: UserTC.getResolver('search').getFieldConfig(),
      userPagination: UserTC.getResolver('searchPagination').getFieldConfig(),
      userConnection: UserTC.getResolver('searchConnection').getFieldConfig(),
    },
  }),
});

Full code example

Installation

yarn add graphql graphql-compose elasticsearch graphql-compose-elasticsearch
// or
npm install graphql graphql-compose elasticsearch graphql-compose-elasticsearch --save

Modules graphql, graphql-compose, elasticsearch are in peerDependencies, so should be installed explicitly in your app.

Screenshots

API proxy: Raw search method

API proxy: Getting several raw elastic metric in one request

Mapping: Relay Cursor Connection

Mapping: Generated GraphQL Types and Documentation

FAQ

Creating custom Resolvers

If you need create something special, you may create a custom Resolver. For example, if you need to add a new tag for existing record, do it in the following manner (see full test-case):

ActivitiesEsTC.addResolver({
  name: 'addTag',
  kind: 'mutation',
  type: 'JSON',
  args: {
    id: 'String!',
    tag: 'String!',
  },
  resolve: ({ args }) => {
    return elasticClient.update({
      index: elasticIndex,
      type: elasticType,
      id: args.id,
      body: {
        script: {
          inline: 'ctx._source.tags.add(params.tag)',
          params: { tag: args.tag },
        },
      },
    });
  },
});

License

MIT