Skip to content

danwkennedy/arango-datasouce

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

arango-datasouce

An implementation of Apollo's datasource for ArangoDb

Installation

$ npm install arango-datasource

DataSources

ArangoDataSource

A "general purpose" datasource that's mainly suitable for querying the database using AQL.

Requires passing the target database instance from arango-js

// index.js

const { ArangoDataSource } = require('@danwkennedy/arango-datasource');
const { Database } = require('arango-js');

// initialize the db
const database = new Database('http://my.database.url');

// initialize the server
const server = new ApolloServer({
  typeDefs,
  resolvers,
  cache,
  context,
  dataSources: () => ({
    arangoDataSource: new ArangoDataSource(database)
  })
});

Extend this class to create more targetted DataSources according to your needs:

// UserDataSource.js

const { ArangoDataSource } = require('@danwkennedy/arango-datasource');

module.exports = class UserDataSource extends ArangoDataSource {

  // Pass the user collection to the DataSource
  constructor(db, collection) {
    super(db);
    this.collection = collection;
  }

  // Build the query and call super.query
  async getUsers() {
    const query = aql`
      FOR user in ${this.collection}
      return user
    `;

    return await this.query(query);
  }
}

Basic query caching is available. Cache keys for queries are simply the query object's hash value using object-hash. This type of caching is mainly useful when using a persisted cache across machines (i.e. Redis instead of the default in memory cache) and works best for fetching common data that doesn't change very often.

ArangoDocumentDataSource

Uses the DataLoader class to add batching and caching to fetching Arango Documents by their Id. This is especially useful as a NodeDataSource as ArangoDb's default Id structure prepends the collection name to the _key making it so you don't need to pass the target collection the document datasource.

// index.js

const { ArangoDocumentDataSource } = require('@danwkennedy/arango-datasource');
const { Database } = require('arango-js');

// initialize the db
const database = new Database('http://my.database.url');

// initialize the server
const server = new ApolloServer({
  typeDefs,
  resolvers,
  cache,
  context,
  dataSources: () => ({
    NodeDataSource: new ArangoDocumentDataSource(database)
  })
});

// node/resolver.js

module.exports = {
  Query: {
    node: async (_, { id }, { dataSources }) => {
      return dataSources.NodeDataSource.load(id);
    },
  },
}

Managers

DataSources fetch records from the database. Managers create/update/delete records from the database.

DocumentManager

Manages the lifecylc of documents in a document collection.

// index.js

const { DocumentManager } = require('@danwkennedy/arango-datasource');
const { Database } = require('arango-js');

// initialize the db
const database = new Database('http://my.database.url');
const userCollection = database.collection('users');

// initialize the server
const server = new ApolloServer({
  typeDefs,
  resolvers,
  cache,
  context,
  dataSources: () => ({
    userDocumentManager: new DocumentManager(userCollection)
  })
});

EdgeManager

Manages the lifecycle of edges in a graph.

// index.js

const { EdgeManager } = require('@danwkennedy/arango-datasource');
const { Database } = require('arango-js');

// initialize the db
const database = new Database('http://my.database.url');
const userFavoriteFoodCollection = database.edgeCollection('user_favorite_food');

// initialize the server
const server = new ApolloServer({
  typeDefs,
  resolvers,
  cache,
  context,
  dataSources: () => ({
    userDocumentManager: new EdgeManager(userFavoriteFoodCollection)
  })
});

The main difference between the EdgeManager and the DocumentManager is the EdgeManager requires the _to and _from ids be passed. It also smooths over some difficulties with removing edges where we might not know the edge's id. In that case, we can pass the _from and _to ids and the manager will do the query to find the correct edge.