/**
* Splitting string into words.
* Used for indexing, can also be used to find words in query.
*
* @param string String with UTF-8 content to process.
* @return array Array of words in utf-8
* @todo Define visibility
*/
public function split2Words($wordString)
{
// Reset debug string:
$this->debugString = '';
// Then convert the string to lowercase:
if (!$this->lexerConf['casesensitive']) {
$wordString = $this->csObj->conv_case('utf-8', $wordString, 'toLower');
}
// Now, splitting words:
$len = 0;
$start = 0;
$pos = 0;
$words = array();
$this->debugString = '';
while (1) {
list($start, $len) = $this->get_word($wordString, $pos);
if ($len) {
$this->addWords($words, $wordString, $start, $len);
if ($this->debug) {
$this->debugString .= '<span style="color:red">' . htmlspecialchars(substr($wordString, $pos, $start - $pos)) . '</span>' . htmlspecialchars(substr($wordString, $start, $len));
}
$pos = $start + $len;
} else {
break;
}
}
return $words;
}
/**
* Converts a given string to a string that can be used as a URL segment.
* The result is not url-encoded.
*
* @param string $string
* @param string $spaceCharacter
* @return string
*/
public function convertToSafeString($string, $spaceCharacter = '-')
{
$processedTitle = $this->csConvertor->conv_case('utf-8', $string, 'toLower');
$processedTitle = strip_tags($processedTitle);
$processedTitle = preg_replace('/[ \\-+_]+/', $spaceCharacter, $processedTitle);
$processedTitle = $this->csConvertor->specCharsToASCII('utf-8', $processedTitle);
$processedTitle = preg_replace('/[^\\p{L}0-9' . preg_quote($spaceCharacter) . ']/u', '', $processedTitle);
$processedTitle = preg_replace('/' . preg_quote($spaceCharacter) . '{2,}/', $spaceCharacter, $processedTitle);
$processedTitle = trim($processedTitle, $spaceCharacter);
// TODO Post-processing hook here
$processedTitle = strtolower($processedTitle);
return $processedTitle;
}
/**
* Handler for the opening of a tag
*/
public function startHandler($xml_parser, $tag, $attributes)
{
if ((string) $this->xmlCharacterData !== '') {
$this->spellCheckHandler($xml_parser, $this->xmlCharacterData);
$this->xmlCharacterData = '';
}
switch ($tag) {
case 'spellchecker':
break;
case 'br':
case 'BR':
case 'img':
case 'IMG':
case 'hr':
case 'HR':
case 'area':
case 'AREA':
$this->text .= '<' . $this->csConvObj->conv_case($this->parserCharset, $tag, 'toLower') . ' ';
foreach ($attributes as $key => $val) {
$this->text .= $key . '="' . $val . '" ';
}
$this->text .= ' />';
break;
default:
$this->text .= '<' . $this->csConvObj->conv_case($this->parserCharset, $tag, 'toLower') . ' ';
foreach ($attributes as $key => $val) {
$this->text .= $key . '="' . $val . '" ';
}
$this->text .= '>';
}
}
/**
* Check if $value is valid. If it is not valid, needs to add an error
* to result.
*
* @param mixed $value
* @return void
*/
public function isValid($value)
{
if (empty($value) || !is_string($value)) {
return;
}
$allowedOptionsArray = GeneralUtility::trimExplode(',', $this->options['array'], true);
if (!empty($this->options['ignorecase'])) {
$value = $this->charsetConverter->conv_case('utf-8', $value, 'toLower');
foreach ($allowedOptionsArray as &$option) {
$option = $this->charsetConverter->conv_case('utf-8', $option, 'toLower');
}
}
if (!in_array($value, $allowedOptionsArray, !empty($this->options['strict']))) {
$this->addError($this->renderMessage($this->options['errorMessage'][0], $this->options['errorMessage'][1], 'error'), 1442002594);
}
}
/**
* Initialize internal variables, especially selector box values for the search form and search words
*
* @return void
*/
public function initialize()
{
// Indexer configuration from Extension Manager interface:
$this->indexerConfig = unserialize($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf']['indexed_search']);
$this->enableMetaphoneSearch = (bool) $this->indexerConfig['enableMetaphoneSearch'];
$this->storeMetaphoneInfoAsWords = !\TYPO3\CMS\IndexedSearch\Utility\IndexedSearchUtility::isTableUsed('index_words');
$this->timeTracker = GeneralUtility::makeInstance(TimeTracker::class);
// Initialize external document parsers for icon display and other soft operations
if (is_array($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['indexed_search']['external_parsers'])) {
foreach ($GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['indexed_search']['external_parsers'] as $extension => $_objRef) {
$this->external_parsers[$extension] = GeneralUtility::getUserObj($_objRef);
// Init parser and if it returns FALSE, unset its entry again:
if (!$this->external_parsers[$extension]->softInit($extension)) {
unset($this->external_parsers[$extension]);
}
}
}
// Init lexer (used to post-processing of search words)
$lexerObjRef = $GLOBALS['TYPO3_CONF_VARS']['EXTCONF']['indexed_search']['lexer'] ?: \TYPO3\CMS\IndexedSearch\Lexer::class;
$this->lexerObj = GeneralUtility::getUserObj($lexerObjRef);
// If "_sections" is set, this value overrides any existing value.
if ($this->piVars['_sections']) {
$this->piVars['sections'] = $this->piVars['_sections'];
}
// If "_sections" is set, this value overrides any existing value.
if ($this->piVars['_freeIndexUid'] !== '_') {
$this->piVars['freeIndexUid'] = $this->piVars['_freeIndexUid'];
}
// Add previous search words to current
if ($this->piVars['sword_prev_include'] && $this->piVars['sword_prev']) {
$this->piVars['sword'] = trim($this->piVars['sword_prev']) . ' ' . $this->piVars['sword'];
}
$this->piVars['results'] = MathUtility::forceIntegerInRange($this->piVars['results'], 1, 100, $this->defaultResultNumber);
// Make sure that some cropping and markup constants used later are defined
$this->loadSettings();
// Selector-box values defined here:
$this->optValues = array('type' => array('0' => $this->pi_getLL('opt_type_0'), '1' => $this->pi_getLL('opt_type_1'), '2' => $this->pi_getLL('opt_type_2'), '3' => $this->pi_getLL('opt_type_3'), '10' => $this->pi_getLL('opt_type_10'), '20' => $this->pi_getLL('opt_type_20')), 'defOp' => array('0' => $this->pi_getLL('opt_defOp_0'), '1' => $this->pi_getLL('opt_defOp_1')), 'sections' => array('0' => $this->pi_getLL('opt_sections_0'), '-1' => $this->pi_getLL('opt_sections_-1'), '-2' => $this->pi_getLL('opt_sections_-2'), '-3' => $this->pi_getLL('opt_sections_-3')), 'freeIndexUid' => array('-1' => $this->pi_getLL('opt_freeIndexUid_-1'), '-2' => $this->pi_getLL('opt_freeIndexUid_-2'), '0' => $this->pi_getLL('opt_freeIndexUid_0')), 'media' => array('-1' => $this->pi_getLL('opt_media_-1'), '0' => $this->pi_getLL('opt_media_0'), '-2' => $this->pi_getLL('opt_media_-2')), 'order' => array('rank_flag' => $this->pi_getLL('opt_order_rank_flag'), 'rank_freq' => $this->pi_getLL('opt_order_rank_freq'), 'rank_first' => $this->pi_getLL('opt_order_rank_first'), 'rank_count' => $this->pi_getLL('opt_order_rank_count'), 'mtime' => $this->pi_getLL('opt_order_mtime'), 'title' => $this->pi_getLL('opt_order_title'), 'crdate' => $this->pi_getLL('opt_order_crdate')), 'group' => array('sections' => $this->pi_getLL('opt_group_sections'), 'flat' => $this->pi_getLL('opt_group_flat')), 'lang' => array(-1 => $this->pi_getLL('opt_lang_-1'), 0 => $this->pi_getLL('opt_lang_0')), 'desc' => array('0' => $this->pi_getLL('opt_desc_0'), '1' => $this->pi_getLL('opt_desc_1')), 'results' => array('10' => '10', '20' => '20', '50' => '50', '100' => '100'));
// Remove this option if metaphone search is disabled)
if (!$this->enableMetaphoneSearch) {
unset($this->optValues['type']['10']);
}
// Free Index Uid:
if ($this->conf['search.']['defaultFreeIndexUidList']) {
$uidList = GeneralUtility::intExplode(',', $this->conf['search.']['defaultFreeIndexUidList']);
$indexCfgRecords = $this->databaseConnection->exec_SELECTgetRows('uid,title', 'index_config', 'uid IN (' . implode(',', $uidList) . ')' . $this->cObj->enableFields('index_config'), '', '', '', 'uid');
foreach ($uidList as $uidValue) {
if (is_array($indexCfgRecords[$uidValue])) {
$this->optValues['freeIndexUid'][$uidValue] = $indexCfgRecords[$uidValue]['title'];
}
}
}
// Should we use join_pages instead of long lists of uids?
if ($this->conf['search.']['skipExtendToSubpagesChecking']) {
$this->join_pages = 1;
}
// Add media to search in:
if (trim($this->conf['search.']['mediaList']) !== '') {
$mediaList = implode(',', GeneralUtility::trimExplode(',', $this->conf['search.']['mediaList'], true));
}
foreach ($this->external_parsers as $extension => $obj) {
// Skip unwanted extensions
if ($mediaList && !GeneralUtility::inList($mediaList, $extension)) {
continue;
}
if ($name = $obj->searchTypeMediaTitle($extension)) {
$this->optValues['media'][$extension] = $this->pi_getLL('opt_sections_' . $extension, $name);
}
}
// Add operators for various languages
// Converts the operators to lowercase
$this->operator_translate_table[] = array($this->charsetConverter->conv_case('utf-8', $this->pi_getLL('local_operator_AND'), 'toLower'), 'AND');
$this->operator_translate_table[] = array($this->charsetConverter->conv_case('utf-8', $this->pi_getLL('local_operator_OR'), 'toLower'), 'OR');
$this->operator_translate_table[] = array($this->charsetConverter->conv_case('utf-8', $this->pi_getLL('local_operator_NOT'), 'toLower'), 'AND NOT');
// This is the id of the site root. This value may be a commalist of integer (prepared for this)
$this->wholeSiteIdList = (int) $this->frontendController->config['rootLine'][0]['uid'];
// Creating levels for section menu:
// This selects the first and secondary menus for the "sections" selector - so we can search in sections and sub sections.
if ($this->conf['show.']['L1sections']) {
$firstLevelMenu = $this->getMenu($this->wholeSiteIdList);
foreach ($firstLevelMenu as $optionName => $mR) {
if (!$mR['nav_hide']) {
$this->optValues['sections']['rl1_' . $mR['uid']] = trim($this->pi_getLL('opt_RL1') . ' ' . $mR['title']);
if ($this->conf['show.']['L2sections']) {
$secondLevelMenu = $this->getMenu($mR['uid']);
foreach ($secondLevelMenu as $kk2 => $mR2) {
if (!$mR2['nav_hide']) {
$this->optValues['sections']['rl2_' . $mR2['uid']] = trim($this->pi_getLL('opt_RL2') . ' ' . $mR2['title']);
} else {
unset($secondLevelMenu[$kk2]);
}
}
$this->optValues['sections']['rl2_' . implode(',', array_keys($secondLevelMenu))] = $this->pi_getLL('opt_RL2ALL');
}
} else {
unset($firstLevelMenu[$optionName]);
}
}
$this->optValues['sections']['rl1_' . implode(',', array_keys($firstLevelMenu))] = $this->pi_getLL('opt_RL1ALL');
}
// Setting the list of root IDs for the search. Notice, these page IDs MUST have a TypoScript template with root flag on them! Basically this list is used to select on the "rl0" field and page ids are registered as "rl0" only if a TypoScript template record with root flag is there.
// This happens AFTER the use of $this->wholeSiteIdList above because the above will then fetch the menu for the CURRENT site - regardless of this kind of searching here. Thus a general search will lookup in the WHOLE database while a specific section search will take the current sections...
if ($this->conf['search.']['rootPidList']) {
$this->wholeSiteIdList = implode(',', GeneralUtility::intExplode(',', $this->conf['search.']['rootPidList']));
}
// Load the template
$this->templateCode = $this->cObj->fileResource($this->conf['templateFile']);
// Add search languages:
$res = $this->databaseConnection->exec_SELECTquery('*', 'sys_language', '1=1' . $this->cObj->enableFields('sys_language'));
while (false !== ($data = $this->databaseConnection->sql_fetch_assoc($res))) {
$this->optValues['lang'][$data['uid']] = $data['title'];
}
$this->databaseConnection->sql_free_result($res);
// Calling hook for modification of initialized content
if ($hookObj = $this->hookRequest('initialize_postProc')) {
$hookObj->initialize_postProc();
}
// Default values set:
// Setting first values in optValues as default values IF there is not corresponding piVar value set already.
foreach ($this->optValues as $optionName => $optionValue) {
if (!isset($this->piVars[$optionName])) {
reset($optionValue);
$this->piVars[$optionName] = key($optionValue);
}
}
// Blind selectors:
if (is_array($this->conf['blind.'])) {
foreach ($this->conf['blind.'] as $optionName => $optionValue) {
if (is_array($optionValue)) {
foreach ($optionValue as $optionValueSubKey => $optionValueSubValue) {
if (!is_array($optionValueSubValue) && $optionValueSubValue && is_array($this->optValues[substr($optionName, 0, -1)])) {
unset($this->optValues[substr($optionName, 0, -1)][$optionValueSubKey]);
}
}
} elseif ($optionValue) {
// If value is not set, unset the option array
unset($this->optValues[$optionName]);
}
}
}
// This gets the search-words into the $sWArr:
$this->sWArr = $this->getSearchWords($this->piVars['defOp']);
}
/**
* Splits the search word input into an array where each word is represented by an array with key "sword"
* holding the search word and key "oper" holding the SQL operator (eg. AND, OR)
*
* Only words with 2 or more characters are accepted
* Max 200 chars total
* Space is used to split words, "" can be used search for a whole string
* AND, OR and NOT are prefix words, overruling the default operator
* +/|/- equals AND, OR and NOT as operators.
* All search words are converted to lowercase.
*
* $defOp is the default operator. 1=OR, 0=AND
*
* @param bool $defaultOperator If TRUE, the default operator will be OR, not AND
* @return array Search words if any found
*/
protected function getSearchWords($defaultOperator)
{
// Shorten search-word string to max 200 bytes (does NOT take multibyte charsets into account - but never mind,
// shortening the string here is only a run-away feature!)
$searchWords = substr($this->sword, 0, 200);
// Convert to UTF-8 + conv. entities (was also converted during indexing!)
$searchWords = $this->charsetConverter->conv($searchWords, $GLOBALS['TSFE']->metaCharset, 'utf-8');
$searchWords = $this->charsetConverter->entities_to_utf8($searchWords);
$sWordArray = false;
if ($hookObj = $this->hookRequest('getSearchWords')) {
$sWordArray = $hookObj->getSearchWords_splitSWords($searchWords, $defaultOperator);
} else {
// sentence
if ($this->searchData['searchType'] == 20) {
$sWordArray = array(array('sword' => trim($searchWords), 'oper' => 'AND'));
} else {
// case-sensitive. Defines the words, which will be
// operators between words
$operatorTranslateTable = array(array('+', 'AND'), array('|', 'OR'), array('-', 'AND NOT'), array($this->charsetConverter->conv_case('utf-8', LocalizationUtility::translate('localizedOperandAnd', 'IndexedSearch'), 'toLower'), 'AND'), array($this->charsetConverter->conv_case('utf-8', LocalizationUtility::translate('localizedOperandOr', 'IndexedSearch'), 'toLower'), 'OR'), array($this->charsetConverter->conv_case('utf-8', LocalizationUtility::translate('localizedOperandNot', 'IndexedSearch'), 'toLower'), 'AND NOT'));
$swordArray = \TYPO3\CMS\IndexedSearch\Utility\IndexedSearchUtility::getExplodedSearchString($searchWords, $defaultOperator == 1 ? 'OR' : 'AND', $operatorTranslateTable);
if (is_array($swordArray)) {
$sWordArray = $this->procSearchWordsByLexer($swordArray);
}
}
}
return $sWordArray;
}
