/**
* Парсинг текста
*
* @param string $sText Исходный текст
* @param array $aError Возвращает список возникших ошибок
*
* @return string
*/
public function TextParser($sText, &$aError = null)
{
$sResult = $this->oTextParser->parse($sText, $aError);
return $sResult;
}
/**
* Создание сниппета
*
* @param string $sText
*
* @return string
*/
protected function _makeSnippet($sText)
{
$aError = array();
$sRegexp = $this->_preparePattern();
// * Если задано, то вырезаем все теги
if ($this->bSkipAllTags) {
$sText = strip_tags($sText);
} else {
$sText = $this->oTextParser->parse($sText, $aError);
$sText = str_replace('<br/>', '', $sText);
}
//$sText = str_replace(' ', ' ', $sText);
if (mb_preg_match_all($sRegexp, $sText, $aMatches, PREG_OFFSET_CAPTURE)) {
// * Создаем набор фрагментов текста
$sSnippet = '';
$aFragmentSets = array();
$nFragmentSetsCount = -1;
$nCount = 0;
$aLastSet = array();
$nLastLen = 0;
foreach ($aMatches[0] as $aMatch) {
$sFrTxt = $aMatch[0];
$nFrPos = $aMatch[1];
$nFrLen = mb_strlen($sFrTxt);
// Создаем сеты фрагментов, чтобы близлежащие слова попали в один сет
if ($nFragmentSetsCount == -1 || $nLastLen == 0) {
$aLastSet = array('txt' => $sFrTxt, 'pos' => $nFrPos, 'len' => $nFrLen);
$nLastLen = $nFrPos + $nFrLen;
$aFragmentSets[++$nFragmentSetsCount][] = $aLastSet;
} else {
if ($nFrPos + $nFrLen - $aLastSet['pos'] < $this->nSnippetLength) {
$aFragmentSets[$nFragmentSetsCount][] = array('txt' => $sFrTxt, 'pos' => $nFrPos, 'len' => $nFrLen);
$nLastLen = $nFrPos + $nFrLen - $aLastSet['pos'];
} else {
$aLastSet = array('txt' => $sFrTxt, 'pos' => $nFrPos, 'len' => $nFrLen);
$nLastLen = $nFrPos + $nFrLen;
$aFragmentSets[++$nFragmentSetsCount][] = $aLastSet;
}
}
}
$aFragments = array();
$nPos = 0;
foreach ($aFragmentSets as $aSet) {
$nLen = 0;
foreach ($aSet as $aWord) {
if ($nLen == 0) {
$nLen = $aWord['len'];
$nPos = $aWord['pos'];
} else {
$nLen = $aWord['pos'] + $aWord['len'] - $nPos;
}
}
$aFragments[] = $this->_makeSnippetFragment($sText, $aSet, $nPos, $nLen);
if ($this->nSnippetMaxFragments > 0 && ++$nCount >= $this->nSnippetMaxFragments) {
break;
}
}
foreach ($aFragments as $sFragment) {
$sSnippet .= $this->sSnippetBeforeFragment . $sFragment . $this->sSnippetAfterFragment;
}
} else {
if (mb_strlen($sText) > $this->nSnippetMaxLength) {
$sSnippet = mb_substr($sText, 0, $this->nSnippetMaxLength) . '…';
} else {
$sSnippet = $sText;
}
}
return $sSnippet;
}
