public function setUp()
{
$this->engine = new Engine();
$this->engine->clear();
$this->data = json_decode(file_get_contents(Env::get('TEST_DATASET_PATH')))->data;
foreach ($this->data as $article) {
$this->engine->addDocument(new Document($article->title, $article->content, ''));
}
}
function setup()
{
global $index;
global $engine;
global $store;
$tokenizer = new PorterTokenizer();
$store = new SQLDocumentStore(Env::getPDO(), $tokenizer);
//$store = new MongoDBDocumentStore(ENV::get('MONGO_HOST'), ENV::get('MONGO_PORT'));
$index = new MemcachedDocumentIndex(ENV::get('MEMCACHED_HOST'), ENV::get('MEMCACHED_PORT'));
$ranker = new TFIDFDocumentRanker();
$config = Config::createBuilder()->index($index)->store($store)->tokenizer($tokenizer)->ranker($ranker)->build();
$engine = new Engine($config);
}
public function testFindKeywords()
{
$engine = new Engine(Config::createBuilder()->testConfig()->build());
$dataset = json_decode(file_get_contents(Env::get('TEST_DATASET_PATH')));
foreach ($dataset->data as $article) {
$engine->addDocument(new Document($article->title, $article->content));
}
$results = $engine->findKeywords('In computer engineering, computer architecture is the conceptual design and fundamental operational structure of a computer system.');
$this->assertEquals('computer', $results[0]['keyword']);
$this->assertEquals(count($dataset->data), $engine->size());
$engine->clear('index');
$this->assertNotEquals(0, $engine->size());
$engine->clear('store');
$this->assertEquals(0, $engine->size());
}
function testAddDocument()
{
$this->store->clear();
$tokenizer = new Tokenizer\PorterTokenizer();
$document = new Document('test', 'this is a test document');
$document->id = 1;
$document->tokens = $tokenizer->tokenize($document->content);
$this->store->addDocument($document);
$this->assertEquals(1, $this->store->size());
$d = $this->store->getDocument(1);
$this->assertEquals($document, $d);
if (Env::isSQLite()) {
$this->setExpectedException('Exception', "UNIQUE constraint failed");
} else {
$this->setExpectedException('Exception', "Duplicate entry '1' for key 'id'");
}
$this->store->addDocument($document);
$this->store->clear();
$this->assertEquals(0, $this->store->size());
}
function __construct()
{
$this->index = new Index\MemcachedDocumentIndex(Env::get('MEMCACHED_HOST'), Env::get('MEMCACHED_PORT'));
$this->store = new Store\MongoDBDocumentStore(Env::get('MONGO_HOST'), Env::get('MONGO_PORT'));
}
function init()
{
$this->index = new Index\MongoDBDocumentIndex(Env::get('MONGO_HOST'), Env::get('MONGO_PORT'));
$this->store = new Store\MongoDBDocumentStore(Env::get('MONGO_HOST'), Env::get('MONGO_PORT'));
}
<?php
use Search\Config\Env;
require "setup.php";
// Create the table if it does not exist
try {
$pdo = Env::getPDO();
$query = "CREATE TABLE documents (id int NOT NULL UNIQUE, title VARBINARY (255) NOT NULL, content VARBINARY (2048) NOT NULL);";
$statement = $pdo->prepare($query);
$statement->execute();
} catch (\Exception $e) {
// Do nothing!
}
global $index;
global $engine;
global $store;
setup();
// If index is empty, add test data set.
if ($index->size() === 0) {
$index = $store->buildIndex($index);
}
if ($engine->size() === 0) {
$file = '../../tests/Wikipedia_sample_dataset.json';
$data = json_decode(file_get_contents($file))->data;
foreach ($data as $article) {
$engine->addDocument(new Search\Document($article->title, $article->content, ''));
}
echo json_encode(true);
} else {
echo json_encode(false);
}
