/**
* @param string $text
*
* @return string[]
*/
public function split($text)
{
$this->probabilityCalculator->setAbbreviations($this->getAbbreviations());
$tokens = $this->lexer->run($text);
$probabilities = $this->probabilityCalculator->calculate($tokens);
$sentences = $this->sentenceBuilder->build($probabilities);
return $sentences;
}
public function testAbbreviations()
{
$text = 'Hello Mr. Jones, please turn on the T.V.';
$expected = ['T_CAPITALIZED_WORD<"Hello">', 'T_CAPITALIZED_WORD<"Mr">', 'T_PERIOD<".">', 'T_CAPITALIZED_WORD<"Jones,">', 'T_WORD<"please">', 'T_WORD<"turn">', 'T_WORD<"on">', 'T_WORD<"the">', 'T_CAPITALIZED_WORD<"T">', 'T_PERIOD<".">', 'T_CAPITALIZED_WORD<"V">', 'T_PERIOD<".">', 'T_EOF'];
$lexer = new Lexer();
$tokens = $lexer->run($text);
$actual = $this->getTokensString($tokens);
$this->assertEquals($expected, $actual);
}
/**
* @param string $input
* @param array $expectedResult
* @param array $abbreviations
*/
private function runCalculateTest($input, array $expectedResult, array $abbreviations)
{
$lexer = new Lexer();
$tokens = $lexer->run($input);
$rules = IniConfiguration::loadFile(__DIR__ . '/../../rules/rules.ini')->getRules();
$calc = new ProbabilityCalculator($rules);
$calc->setAbbreviations(new Abbreviations($abbreviations));
$probabilities = $calc->calculate($tokens);
$actual = [];
foreach ($probabilities as $probability) {
$token = $probability->getToken();
if ($token instanceof WordToken || $token instanceof CapitalizedWordToken || $token instanceof WhitespaceToken || $token instanceof EOFToken) {
continue;
}
$actual[] = $token->getName() . ' ' . $probability->getProbability();
}
$this->assertEquals($expectedResult, $actual);
}