elasticsearch-graphql

Elastic Search GraphQL

Schema and query builder for Elastic Search

• Creates a static typed graphql schema from an elastic search mapping
• Transforms your graphql query and creates an elastic search body
• Runs the search on your elastic index
• Returns the results and calls your hits schema

For working example, checkout elasticsearch-graphql-server-example

Compatibility

This package is tested and working on

• graphql version ^0.6.2 (should be okey from version 0.5.x)
• ElasticSearch version ^2.3.1

Usage

var graphql = require('graphql');
var esGraphQL = require('elasticsearch-graphql')
var hitsSchema = require('./schemas/myGraphQLSchema')

var schema = esGraphQL({
  graphql: graphql,
  name: 'ordersSearch',
  mapping: {
    "properties": {
      "id": {
        "type" : "string",
        "index" : "not_analyzed"
      },
      ...
    }
  },
  elastic: {
    host: 'localhost:9200',
    index: 'orders',
    type: 'order',
    query: function(query, context) {
      // Debug or modify the query anyway you want. The context is passed down from graphql
      // Make sure to return the original or modified query

      return query
    }
  },
  hitsSchema: hitsSchema
})

var rootSchema = new graphql.GraphQLSchema({
  query: new graphql.GraphQLObjectType({
    name: 'RootQueryType',
    fields: {
      mySearchData: {
        type: schema.type,
        args: schema.args,
        resolve: schema.resolve
      }
    }
  })
})

Query Builder

It will fetch the current mapping from elasticsearch and create a static typed schema for you. Add the schema to you graphql server and the type helper will lead you. The hits field will resolve to whatever schema you send in. So you can use elasticsearch for searching data and then easily get your real data from anywhere. See full example in in /examples

Example query

{
  ordersSearch(query: "New Order",
    filters: {
      images: {
        createdBy: {
          values: ["Simon Nord", "James Kyburz"],
        }
      }
    }
  ) {
    aggregations {
      status {
        timestamp {
          terms(limit: 5) {
            value,
            count
          }
        }
      }
    }
    hits {
      id
    }
  }
}

TODO

• Support multiple indexes
• Do smarter elasticsearch queries
• Add more options, like query type etc.
• Add tests
• Allow more aggregation types
• Allow more complex filters