Bulk operations about elasticsearch-php

[polyfractal]

There are, please see the documentation here:

http://www.elasticsearch.org/guide/en/elasticsearch/client/php-api/current/_indexing_operations.html#_bulk_indexing

Currently, bulk requests must follow the Elasticsearch Bulk format. In the future, I would like to provide a helper method that builds the bulk format for you (because the syntax is kinda gross)

[insign]

It is terrible :(

[polyfractal]

The bulk syntax was designed for performance, not developer friendliness.

Building the syntax algorithmically is very simple. And honestly, that is all a "helper" method would be doing. Here is some (untested) code to illustrate how simple it is:

//All your docs here
$myDocs = array( ... );  

// This will hold the final bulk
$bulkString = '';

// For brevity, assume docs don't have IDs
$action = array(
  "index" => new \stdClass()
);
$actionString = json_encode($action);

foreach ($myDocs as $doc) {
  $bulkString .= "$actionString\n";
  $bulkString .= json_encode($doc) . "\n";
}

$params['index'] = 'my_index';
$params['type']  = 'my_type';
$params['body']  = $bulkString

$ret = $client->bulk($params);

[insign]

Thank you @polyfractal . Have you idea when It will enter in the main code?

I will really try this. I want a bulk exactly for performace, but I don't think that it need be ugly.

[polyfractal]

Unsure. There are a few bigger issues that I need to deal with first, not quite sure how long they will take.

Could you open a new ticket about the Bulk Helper method so that I don't forget?

[insign]

Sure! After the lunch!

[insign]

A fork (without unit tests) will be acceptable?

[polyfractal]

Do you mean a Pull Request for the bulk helper function?

If yes, it should be implemented in a class outside of the client. The client is supposed to be entirely unopinionated...just a simple translation from the Elasticsearch API to a PHP API. I would recommend adding a new Helper class that lives in /src/Elasticsearch/Helper.php. Then users could do something like:

$helper = new Elasticsearch\Helper();
$helper->bulk()->index($myDoc);

I think there are a few functional requirements to make it useful:

• index()
• update()
• delete()
• create()
• setIndex()
• setType()
• setBulkDocSize()
• setBulkPhysicalSize()
• setBulkTimeout()
• flushBulk()

The bulk helper should also (probably?) return a special bulk object which facilitates working with the results (instead of just an array), so that it could have methods like:

• getSucceeded()
• getFailed()
• getAll()
• retryFailed()

You can see that the helper has quite a bit of functionality to make it really useful. All pretty simple, but a decent amount of code.

Test would be preferred before merging, but we can work on tests after some code has been produced.