A search framework that facilitates reusable code and concepts across search backend libraries.

1.0.0-alpha4 2013-02-15 19:57 UTC


The Search Framework library is a standards compliant PHP project that aims to share concepts, nomenclature, and code across search applications. Its mission is to simplify the process of building best-in-breed search experiences regardless of the source data being collected, search engine being used, or PHP framework the application is built with. The library operates under the assumption that it can't and shouldn't solve all use cases on its own, so it is architected to have pluggable components and extendable events that allow developers write reusable code and optionally contribute the solutions back to the community. It also assumes that PHP is not always the best fit for various operations related to search, so it is designed to integrate with non-PHP tools where appropriate.

What does this library do?

The primary focus of the Search Framework library from a technical standpoint is on the document processing methodology, in other words facilitating data retrieval and preparing it for indexing. The library is architected to support reusable data collections so that the information retrieval and processing code is backend agnostic and able to be shared. It also provides an API to integrate best-in-breed client libraries and a system to achieve parallel indexing. Finally, the library standardizes the search results so that theming systems can handle the responses from search engines in a consistent manner.

Arguably more important than the code, the Search Framework library defines nomenclature and concepts that are intended to be used independent of the underlying technology. The main goals are to help bridge the communication gaps between search projects and facilitate writing interoperable code by adhering to best practices and techniques.

Is the library also a search abstraction layer?

No, absolutely not. Unlike database management systems that are similar enough to abstract 80% of the most common use cases, search engines such as Solr, Elasticsearch, Sphinx, and various proprietary solutions have vastly different capabilities and paradigms making them extremely difficult to abstract. The danger of abstracting complex systems like search engines is the tendency to force all interactions to fit the mold of abstraction layer. At that point the focus is on the tool as opposed to the search related problems the application is trying to solve while at the same time masking the benefits of the search engine. The Search Framework library abstracts only the most basic search operations while allowing the backend clients to do what they do best.

Basic Usage

The code below indexes the "Drupal Planet" RSS feed into Elasticsearch.

use Search\Framework\Indexer;
use Search\Framework\SearchServiceEndpoint;

use Search\Collection\Feed\FeedCollection;     // @see https://github.com/cpliakas/feed-collection
use Search\Engine\Elasticsearch\Elasticsearch; // @see https://github.com/cpliakas/elasticsearch-engine

require 'vendor/autoload.php';

// Instantiate a collection that references the Drupal Planet feed. Collections
// are simply connectors to and models of the source data being indexed.
$drupal_planet = new FeedCollection('feed.drupal');

// Connect to an Elasticsearch server.
$elasticsearch = new Elasticsearch(new SearchEngineEndpoint('local', 'localhost', 'feeds', 9200));

// Instantiate an indexer, attach the collection, and index it.
$indexer = new Indexer($elasticsearch);

How about indexing the data into Solr? Simply swap out the search engine in the code above.

use Search\Engine\Solr\Solr;  // @see https://github.com/cpliakas/solr-search-engine

// Connect a  Solr server and pass it to the indexer.
$solr = new Solr(new SearchEngineEndpoint('local', 'http://localhost', '/solr', 8983));

$indexer = new Indexer($solr);

Optionally use PSR-3 compliant loggers such as Monolog to audit search related events and aid in debugging.

use Monolog\Logger;
use Monolog\Handler\StreamHandler;

$log = new Logger('search');
$log->pushHandler(new StreamHandler('search.log', Logger::INFO));
// Use Logger::DEBUG for development and learning about the application flow.



To install the required libraries, execute the following commands in the directory where this library is extracted.

curl -s https://getcomposer.org/installer | php
php composer.phar install

If curl is not available, replace the first command with the one below:

php -r "eval('?>'.file_get_contents('https://getcomposer.org/installer'));"

Please refer to the Composer tool's installation documentation for more information.