/**
* {@inheritDoc}
*/
public function load($path)
{
$data = Yaml::parse($path);
$storage = new Storage();
if (!$data) {
return $storage;
}
foreach ($data as $language => $alphabetInfo) {
$chars = Unicode::ordStr($alphabetInfo['chars']);
$commonCharCodes = array();
if (!empty($alphabetInfo['common_chars'])) {
$commonCharCodes = Unicode::ordStr($alphabetInfo['common_chars']);
}
$multipleCharsCodes = array();
if (!empty($alphabetInfo['multiple_chars'])) {
$multipleChars = explode(' ', $alphabetInfo['multiple_chars']);
foreach ($multipleChars as $multipleChar) {
$multipleCharsCodes[] = Unicode::ordStr($multipleChar);
}
}
$alphabet = new Alphabet($language, $chars, $commonCharCodes, $multipleCharsCodes);
$storage->add($alphabet);
}
return $storage;
}
/**
* Detect languages
*
* @param string $string
*
* @return Languages
*/
public function detect($string)
{
// Remove special chars
$codes = Unicode::ordStr($string);
$languages = new Languages();
foreach ($this->getVisitors() as $visitor) {
$visitor->visit($string, $codes, $languages);
}
return $languages;
}
/**
* Fix section languages by alphabet
*
* @param string $alphabetsFile
* @param string $sectionsFile
*
* @throws \InvalidArgumentException
*/
public static function fixSectionLanguagesByAlphabet($alphabetsFile, $sectionsFile)
{
if (!is_file($alphabetsFile)) {
throw new \InvalidArgumentException(sprintf('File "%s" not found.', $alphabetsFile));
}
if (!is_file($sectionsFile)) {
throw new \InvalidArgumentException(sprintf('File "%s" not found.', $sectionsFile));
}
$sections = Yaml::parse($sectionsFile);
$alphabets = Yaml::parse($alphabetsFile);
if (!$sections || !$alphabets) {
return;
}
foreach ($alphabets as $language => $alphabetInfo) {
$alphabetChars = Unicode::ordStr($alphabetInfo['chars']);
$availableSections = array();
foreach ($alphabetChars as $char) {
foreach ($sections as $section => $sectionInfo) {
list($diapMin, $diapMax) = $sectionInfo['diap'];
if ($char >= $diapMin && $char <= $diapMax) {
$availableSections[] = $section;
}
}
}
if (count($availableSections)) {
foreach ($availableSections as $availableSection) {
if (!is_array($sections[$availableSection]['languages'])) {
$sections[$availableSection]['languages'] = array();
}
if (!in_array($language, $sections[$availableSection]['languages'])) {
$sections[$availableSection]['languages'][] = $language;
}
}
}
}
file_put_contents($sectionsFile, Yaml::dump($sections));
}
