The Simplest ElasticSearch Node.js Module

Q: There is already an "official" ElasticSearch module, why create a new one...? A: Have you tried using the official client...? Did you enjoy the experience?

We needed an easy way to create, read, update and search our ElasticSearch records from node.js. All the available modules were way too complicated to use for beginners. So we decided to invest the time to create something much simpler!

Creating a record in ElasticSearch from your node.js app using esta is this simple:

var es = require('esta'); // the simplest way to use ElasticSearch in node.js!
es.create({'message':'ElasticSearch is awesome!'}, function(response){
  console.log('record created '+ response.created); // record created true
})

As you are about to discover, there is a much easier way to use ElasticSearch!

##Guide to esta Documentation

Usage:

• Installation
• CONNECT to ElasticSearch Cluster
• CRUD
  • CREATE (Save) a (new) record
  • READ a record
  • UPDATE an (existing) record
  • DELETE a record
• SEARCH for Record(s)
• STATS
• Error handling
• Local/Dev Machine
• (Travis) CI
• (Optional) Use Vagrant to Run ElasticSearch

Philosophy / Background / Detail:

• Only Core Modules
  • Dev Dependencies
• Code Quality
• Contributing
• Module Name
• License

Installation

npm install esta --save

var ES = require('esta');

ES.CONNECT(index, function (response) {
  console.log(response);
  // for more detailed stats see: STATS method below
});

example ES.CONNECT response:

{ status: 200,
  name: 'Ultragirl',
  cluster_name: 'elasticsearch',
  version:
   { number: '1.4.2',
     build_hash: '927caff6f05403e936c20bf4529f144f0c89fd8c',
     build_timestamp: '2014-12-16T14:11:12Z',
     build_snapshot: false,
     lucene_version: '4.10.2' },
  tagline: 'You Know, for Search' }

// define the record you want to store:
var record = {
  date: new Date().toISOString(),
  message: 'Your amazing message goes here'
};
ES.CREATE(record, function(response) {
 // do what ever you like with the response
});

{ _index: 'index',
  _type: 'type',
  _id: '112669114721',
  _version: 1,
  created: true }

• index can be compared to a Database in SQL see: https://www.elasticsearch.org/guide/en/elasticsearch/reference/current/glossary.html#glossary-index
• type is like the table in SQL-world or a collection in other NoSQL systems. see: https://www.elasticsearch.org/guide/en/elasticsearch/reference/current/glossary.html#glossary-type
• id is the unique key for your record. equivalent to the primary-key in a SQL-world

While its optional to set these on your record before asking esta to create it, we highly recommend using index, type and id to organise your records.

// define the record you want to retrieve:
var record = {
  index: 'twitter',
  type: 'tweet',
  id: 1234, // or what ever GUID you want to lookup
};
ES.READ(record, function(response) {
 // do what ever you like with the response
});

{ _index: 'twitter',
  _type: 'tweet',
  _id: '735981868114',
  _version: 1,
  found: true,
  _source: { message: 'My Awesome Message' }
}

When a record does not exist response.found is false. e.g:

{ _index: 'twitter',
  _type: 'tweet',
  _id: '804164689732',
  found: false }

• index we need to know which "database" our record is in
• type "table"
• id the unique key for the record you are looking up.

// define the record you want to store:
var record = {
  index: 'twitter',
  type: 'tweet',
  id: 1234, // or what ever GUID you want
  message: 'Revised message'
};
ES.UPDATE(record, function(response) {
 // do what ever you like with the response
});

{ _index: 'twitter',
  _type: 'tweet',
  _id: '639403095701',
  _version: 2,
  created: false }

• index we need to know which "database" our record is in
• type "table"
• id the unique key for the record you are updating.

Note: UPDATE actually performs an UPSERT UPdate record if already exists or inSERT (create) if its new.

// define the record you want to store:
var record = {
  type: 'tweet',
  index: 'twitter',
  id: 1234, // or what ever GUID you want
  message: 'Revised message'
};
ES.DELETE(record, function(response) {
 // do what ever you like with the response
});

{ found: true,
  _index: 'twitter',
  _type: 'tweet',
  _id: '137167415115',
  _version: 2,
  deleted: true }

• index we need to know which "database" our record is in
• type "table"
• id the unique key for the record you are updating.

Obviously if the record is NOT Found, there is nothing to delete. In that case, the response look like this: (found is false)

{ found: false,
  _index: 'twitter',
  _type: 'tweet',
  _id: '951078315032',
  _version: 1 }

ElasticSearch does not store revisions of your documents by default, we made a "BACKUP" method which stores previous versions of records, when ever they are updated or deleted.

The old versions are stored as different type to avoid polluting the main "table" with copies this type is named: {typename}_bak

For example, if a document has an id abc and its current version is 2 we can find the previous version (v1) of the document by issuing the following read query:

var backup = { index: 'twitter', type: 'tweets_bak', id: 'abc_1'}

// setup query:
var query = {
  index: 'twitter',
  type:  'tweet',
  field: 'text',     // the field we want to search in
  text:  'amazing'   // string we are searching for
};

ES.SEARCH(query, function(response) {
  // console.log(response);
  if(response.hits.total > 0){
    console.log("✓ Search results found: "+ response.hits.total);
  }
});

{ took: 8,
  timed_out: false,
  _shards: { total: 5, successful: 5, failed: 0 },
  hits:
   { total: 924,
     max_score: 0.6355637,
     hits:
      [ [Object],
        [Object],
        etc...
  }
}

• index we need to know which "database" our record is in
• type "table"
• field the field in the record you want to search in.
• text the text you are searching for.

When NO RECORDS are FOUND the response will look this:

{ took: 2,
  timed_out: false,
  _shards: { total: 5, successful: 5, failed: 0 },
  hits: { total: 0, max_score: null, hits: [] } }

We check for if(response.hits.total > 0) { /* use display results */ } else { /* show sad face */} Here's the image we use:

The ES.STATS method exposes the ElasticSearch Instance/Cluster _stats see: https://www.elasticsearch.org/guide/en/elasticsearch/reference/current/indices-stats.html

STATS(function (response) {
  // do something awesome response
});

ElasticSearch returns rich information on cluster health, document count etc. see: #31 for complete STATS output

To help you get started as fast as possible, esta defaults to using your local machine for ElasticSearch.

If you are deploying your App to Heroku there are two ElasticSearch-as-a-Service providers that offer Free entry level service:

• Bonsai: https://addons.heroku.com/bonsai
• SearchBox: https://addons.heroku.com/searchbox

esta supports both of these providers out-of-the-box! as soon as you add the "addon" to your heroku app it "just works!"

Our Travis Build Process includes checks for both Bonsai and SearchBox: See: https://travis-ci.org/dwyl/esta/jobs/58582216#L270

language: node_js
node_js:
  - 6
services:
  - elasticsearch

if you are new to Travis-CI see: https://github.com/docdis/learn-travis

If, like me you prefer not to have Java running on your dev machine (because its chronically insecure) I highly recommend using Vagrant to run a light-weight virtual machine to isolate ElasticSearch and only install Java in the VM.

The other obvious benefit of using Vagrant is that all your fellow developers will have exactly the same (latest) build so there's no risk of version incompatibility. Learn more at: https://github.com/nelsonic/learn-vagrant

I've included a Vagrantfile in this repo which will get you up-and-running with Ubuntu, Node.js & ElasticSearch with a single command: vagrant up

all you need to do is run the following commands in your terminal:

vagrant up
vagrant ssh
sudo service elasticsearch start

If you have any questions, just ask!

We wanted something simpler and thus much easier to extend if you need to! esta is easy to understand. The entire module is 129 lines of clear/clean/commented/DRY code; you can read & understand it all before breakfast! Dive in at /lib. Each method has a corresponding file in /test

We wanted a way of "soft-deleting" records (i.e. avoiding data loss.) If you like the idea of being able to * recover accidentally deleted data*, you will love our DELETE method see: lib/delete.js

There are quite a few modules in the node ecosystem for use with ElasticSearch. However, when I saw how many dependencies the "Official" ElasticSearch Node.js Module https://github.com/elasticsearch/elasticsearch-js had and especially the number of DevDependencies, it made it hard to contribute to the project...

Our aim is to build something that only uses core modules with Stable APIs, so we never have to think about upgrading - it also makes it a lot easier for others to learn how the module works, which invites contribution from the community. Given that ElasticSearch has a REST API we are only using Node's http (core) module. and this is kept DRY (only in one file) see: lib/http_request.js

• Tape for testing: https://github.com/substack/tape
• Istanbul for Code Coverage: https://github.com/nelsonic/learn-istanbul
• Chalk for colors in test output (readability)
• Pre-commit for ensuring all commits pass strict quality checks before being pushed to GitHub. see: https://github.com/nelsonic/learn-pre-commit
• jshint checks code style is consistent: https://github.com/nelsonic/learn-jshint
• CodeClimate for tracking code quality and test coverage: https://github.com/nelsonic/learn-codeclimate

All contributions are welcome. If anything is unclear please create an issue: https://github.com/dwyl/esta/issues

Most of the Node.js developers I've worked with, don't handle errors well. A typical (bad) example:

if(error) {
  console.log(error); // this is worse than useless!
}

So instead of having of having code full of if(err) ... we have deliberately cut out errors from callback functions completely.

Thus, all the methods in this module have the simplified signature:

ES.METHOD(record, function(response){
  // do something with response
});

Instead, we propose using a central error catcher. e.g:

process.on('uncaughtException', function(err) {
  console.log('ERROR: ' + err); // preferably handle errors appropriately
});

or, if you are using Hapi.js we recommend using https://github.com/hapijs/poop

For more on Errors, please read: https://www.joyent.com/developers/node/design/errors

We prefer to have the METHOD names UPPERCASE because it makes them easy to spot and differentiate from your code. If you feel they are a bit "shouty" all methods are available in lowercase too; take your pick! see: https://git.io/pZ6t

Q: Which ElasticSearch Node Module should I use...? A: https://translate.google.com/#auto/en/esta

MIT