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());
}
/**
* Utils function used to find keywords in a query
* @param $query string to identify keywords in
* @return array of keywords, ordered by the ranker class
*/
public function findKeywords($query)
{
$tokenizer = $this->config->getTokenizer();
$ranker = $this->config->getRanker();
$tokens = $tokenizer->tokenize($query);
$ranker->init($tokens, $this->size());
foreach ($tokens as $token) {
$result = $this->index->search($token);
$ranker->cacheTokenFrequency($token, count($result));
}
$keywords = $ranker->findKeywords($tokens);
// If tokens are stemmed, look up original word
if ($tokenizer instanceof Tokenizer\StemTokenizer) {
$keywords = array_map(function ($token) use($tokenizer) {
$token['keyword'] = $tokenizer->getWord($token['keyword']);
return $token;
}, $keywords);
}
return $keywords;
}
/**
* @expectedException \Exception
* @expectedExceptionMessage Document Ranker not defined
*/
public function testMissingRanker()
{
$config = Config::createBuilder()->index($this->index)->stopWords($this->stopWords)->store($this->store)->tokenizer($this->tokenizer)->build();
}