/**
*
* @param string $queryString
* @return QueryAbstractFactory
*/
public static function factory($queryString)
{
$tokenizer = new GeneralTokenizer();
$tokens = $tokenizer->tokenize($queryString);
if (count($tokens) === 1) {
return new SingleTermQuery($queryString);
} else {
return new MultiTermQuery($queryString);
}
}
/**
*
* @param string $queryString
* @return QueryAbstractFactory
*/
public static function factory($queryString)
{
$tokenizer = new GeneralTokenizer();
$tokens = $tokenizer->tokenize($queryString);
if (in_array($queryString[0], ['"', "'"]) && in_array($queryString[strlen($queryString) - 1], ['"', "'"])) {
return new QuotedQuery($queryString);
} elseif (count($tokens) === 1) {
return new SingleTermQuery($queryString);
} else {
return new MultiTermQuery($queryString);
}
}
/**
* Return an array of tokenized words
* @param string|null $fileId
* @param \TextAnalysis\Tokenizers\TokenizerAbstract
* @return array
*/
public function getWords($fileId = null, $tokenizer = null)
{
if (!$tokenizer) {
$tokenizer = new GeneralTokenizer();
}
$fileIds = [];
if (empty($fileId)) {
$fileIds = $this->getFileIds();
} else {
$fileIds = [$fileId];
}
$words = [];
foreach ($fileIds as $filename) {
$content = file_get_contents($this->getPackage()->getInstallationPath() . $filename);
$words = array_merge($words, $tokenizer->tokenize($content));
unset($content);
}
return $words;
}
public function testLineTokenizer()
{
$tokenizer = new GeneralTokenizer(PHP_EOL);
$this->assertCount(4, $tokenizer->tokenize("This " . PHP_EOL . " has" . PHP_EOL . " some" . PHP_EOL . " words"));
}
