/**
* Shorten a string to a given number of characters
*
* The function preserves words, so the result might be a bit shorter or
* longer than the number of characters given. It strips all tags.
*
* @param string $strString The string to shorten
* @param integer $intNumberOfChars The target number of characters
* @param string $strEllipsis An optional ellipsis to append to the shortened string
*
* @return string The shortened string
*/
public static function substr($strString, $intNumberOfChars, $strEllipsis = ' …')
{
$strString = preg_replace('/[\\t\\n\\r]+/', ' ', $strString);
$strString = strip_tags($strString);
if (Utf8::strlen($strString) <= $intNumberOfChars) {
return $strString;
}
$intCharCount = 0;
$arrWords = array();
$arrChunks = preg_split('/\\s+/', $strString);
$blnAddEllipsis = false;
foreach ($arrChunks as $strChunk) {
$intCharCount += Utf8::strlen(static::decodeEntities($strChunk));
if ($intCharCount++ <= $intNumberOfChars) {
$arrWords[] = $strChunk;
continue;
}
// If the first word is longer than $intNumberOfChars already, shorten it
// with Utf8::substr() so the method does not return an empty string.
if (empty($arrWords)) {
$arrWords[] = Utf8::substr($strChunk, 0, $intNumberOfChars);
}
if ($strEllipsis !== false) {
$blnAddEllipsis = true;
}
break;
}
// Deprecated since Contao 4.0, to be removed in Contao 5.0
if ($strEllipsis === true) {
@trigger_error('Passing "true" as third argument to StringUtil::substr() has been deprecated and will no longer work in Contao 5.0. Pass the ellipsis string instead.', E_USER_DEPRECATED);
$strEllipsis = ' …';
}
return implode(' ', $arrWords) . ($blnAddEllipsis ? $strEllipsis : '');
}
/**
* Remove unsafe characters for URL slug
*
* @param string $handle
* @param int $maxlength = Max number of characters of the return value
* @return string $handle
*/
public function slugSafeString($handle, $maxlength = 128)
{
$handle = preg_replace('/[^\\p{L}\\p{Nd}\\-_]+/u', ' ', $handle);
// remove unneeded chars
$handle = preg_replace('/[-\\s]+/', '-', $handle);
// convert spaces to hyphens
return trim(Utf8::substr($handle, 0, $maxlength), '-');
// trim to first $max_length chars
}
/**
* Return the formatted group header as string
*
* @param string $field
* @param mixed $value
* @param integer $mode
*
* @return string
*/
protected function formatCurrentValue($field, $value, $mode)
{
$remoteNew = $value;
// see #3861
if ($GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['inputType'] == 'checkbox' && !$GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['eval']['multiple']) {
$remoteNew = $value != '' ? ucfirst($GLOBALS['TL_LANG']['MSC']['yes']) : ucfirst($GLOBALS['TL_LANG']['MSC']['no']);
} elseif (isset($GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['foreignKey'])) {
$key = explode('.', $GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['foreignKey'], 2);
$objParent = $this->Database->prepare("SELECT " . $key[1] . " AS value FROM " . $key[0] . " WHERE id=?")->limit(1)->execute($value);
if ($objParent->numRows) {
$remoteNew = $objParent->value;
}
} elseif (in_array($mode, array(1, 2))) {
$remoteNew = $value != '' ? ucfirst(Utf8::substr($value, 0, 1)) : '-';
} elseif (in_array($mode, array(3, 4))) {
if (!isset($GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['length'])) {
$GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['length'] = 2;
}
$remoteNew = $value != '' ? ucfirst(Utf8::substr($value, 0, $GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['length'])) : '-';
} elseif (in_array($mode, array(5, 6))) {
$remoteNew = $value != '' ? \Date::parse(\Config::get('dateFormat'), $value) : '-';
} elseif (in_array($mode, array(7, 8))) {
$remoteNew = $value != '' ? date('Y-m', $value) : '-';
$intMonth = $value != '' ? date('m', $value) - 1 : '-';
if (isset($GLOBALS['TL_LANG']['MONTHS'][$intMonth])) {
$remoteNew = $value != '' ? $GLOBALS['TL_LANG']['MONTHS'][$intMonth] . ' ' . date('Y', $value) : '-';
}
} elseif (in_array($mode, array(9, 10))) {
$remoteNew = $value != '' ? date('Y', $value) : '-';
} else {
if ($GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['inputType'] == 'checkbox' && !$GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['eval']['multiple']) {
$remoteNew = $value != '' ? $field : '';
} elseif (is_array($GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['reference'])) {
$remoteNew = $GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['reference'][$value];
} elseif ($GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['eval']['isAssociative'] || array_is_assoc($GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['options'])) {
$remoteNew = $GLOBALS['TL_DCA'][$this->strTable]['fields'][$field]['options'][$value];
} else {
$remoteNew = $value;
}
if (is_array($remoteNew)) {
$remoteNew = $remoteNew[0];
}
if (empty($remoteNew)) {
$remoteNew = '-';
}
}
return $remoteNew;
}
/**
* @covers Patchwork\Utf8::substr
*/
function testSubstr()
{
$b = "deja";
$c = "déjà";
$d = n::normalize($c, n::NFD);
$this->assertTrue($c > $d);
$this->assertSame('국어', u::substr('한국어', 1, 20));
$this->assertSame("de", substr($b, 0, 2));
$this->assertSame("ja", substr($b, -2, 3));
$this->assertSame("ej", substr($b, -3, -1));
$this->assertSame("", substr($b, 1, -3));
$this->assertSame("", substr($c, 5, 0));
// u::substr() returns false here
$this->assertSame("", substr($c, -5, 0));
// u::substr() returns false here
$this->assertSame(false, substr($b, 1, -4));
$this->assertSame("jà", u::substr($c, 2));
$this->assertSame("jà", u::substr($c, -2));
$this->assertSame("dé", u::substr($c, 0, 2));
$this->assertSame("jà", u::substr($c, -2, 3));
$this->assertSame("éj", u::substr($c, -3, -1));
$this->assertSame("", u::substr($c, 1, -3));
$this->assertSame(false, u::substr($c, 5, 0));
// Modelled after grapheme_substr(), not substr() (see above)
$this->assertSame(false, u::substr($c, -5, 0));
// Modelled after grapheme_substr(), not substr() (see above)
$this->assertSame(false, u::substr($c, 1, -4));
$this->assertSame(n::normalize("dé", n::NFD), u::substr($d, 0, 2));
$this->assertSame(n::normalize("jà", n::NFD), u::substr($d, -2, 3));
$this->assertSame(n::normalize("éj", n::NFD), u::substr($d, -3, -1));
$this->assertSame("", u::substr($d, 1, -3));
$this->assertSame(false, u::substr($d, 1, -4));
}
/**
// Gets part of a string.
* @param string $string The input string.
* @param int $start The first position from which the extracted part begins.
* @param int $length The length in character of the extracted part.
* @param string $encoding (optional) The used internally by this function character encoding. If it is omitted, the platform character set will be used by default.
* @return string Returns the part of the string specified by the start and length parameters.
* Note: First character's position is 0. Second character position is 1, and so on.
* This function is aimed at replacing the functions substr() and mb_substr() for human-language strings.
* @link http://php.net/manual/en/function.substr
* @link http://php.net/manual/en/function.mb-substr
*/
function api_substr($string, $start, $length = null, $encoding = null)
{
if (is_null($length)) {
$length = api_strlen($string, $encoding);
}
return Utf8::substr($string, $start, $length);
}
public function test_long_length()
{
$str = 'Iñtërnâtiônàlizætiøn';
$this->assertEquals('Iñtërnâtiônàlizætiøn', u::substr($str, 0, 15536));
}
public function part($start, $length = null)
{
$this->string = u::substr($this->string, $start, $length);
return $this;
}
/**
* Return substring of a string
*
* @param string $str
* @param integer $start
* @param integer $length
*
* @return string
*
* @deprecated Deprecated since Contao 4.0, to be removed in Contao 5.0.
* Use Patchwork\Utf8::substr() instead.
*/
function utf8_substr($str, $start, $length = null)
{
@trigger_error('Using utf8_substr() has been deprecated and will no longer work in Contao 5.0. Use Patchwork\\Utf8::substr() instead.', E_USER_DEPRECATED);
return Utf8::substr($str, $start, $length);
}
/**
* Search the index and return the result object
*
* @param string $strKeywords The keyword string
* @param boolean $blnOrSearch If true, the result can contain any keyword
* @param array $arrPid An optional array of page IDs to limit the result to
* @param integer $intRows An optional maximum number of result rows
* @param integer $intOffset An optional result offset
* @param boolean $blnFuzzy If true, the search will be fuzzy
*
* @return Database\Result The database result object
*
* @throws \Exception If the cleaned keyword string is empty
*/
public static function searchFor($strKeywords, $blnOrSearch = false, $arrPid = array(), $intRows = 0, $intOffset = 0, $blnFuzzy = false)
{
// Clean the keywords
$strKeywords = Utf8::strtolower($strKeywords);
$strKeywords = \StringUtil::decodeEntities($strKeywords);
$strKeywords = preg_replace(array('/\\. /', '/\\.$/', '/: /', '/:$/', '/, /', '/,$/', '/[^\\w\' *+".:,-]/u'), ' ', $strKeywords);
// Check keyword string
if (!strlen($strKeywords)) {
throw new \Exception('Empty keyword string');
}
// Split keywords
$arrChunks = array();
preg_match_all('/"[^"]+"|[\\+\\-]?[^ ]+\\*?/', $strKeywords, $arrChunks);
$arrPhrases = array();
$arrKeywords = array();
$arrWildcards = array();
$arrIncluded = array();
$arrExcluded = array();
foreach ($arrChunks[0] as $strKeyword) {
if (substr($strKeyword, -1) == '*' && strlen($strKeyword) > 1) {
$arrWildcards[] = str_replace('*', '%', $strKeyword);
continue;
}
switch (substr($strKeyword, 0, 1)) {
// Phrases
case '"':
if (($strKeyword = trim(substr($strKeyword, 1, -1))) != false) {
$arrPhrases[] = '[[:<:]]' . str_replace(array(' ', '*'), array('[^[:alnum:]]+', ''), $strKeyword) . '[[:>:]]';
}
break;
// Included keywords
// Included keywords
case '+':
if (($strKeyword = trim(substr($strKeyword, 1))) != false) {
$arrIncluded[] = $strKeyword;
}
break;
// Excluded keywords
// Excluded keywords
case '-':
if (($strKeyword = trim(substr($strKeyword, 1))) != false) {
$arrExcluded[] = $strKeyword;
}
break;
// Wildcards
// Wildcards
case '*':
if (strlen($strKeyword) > 1) {
$arrWildcards[] = str_replace('*', '%', $strKeyword);
}
break;
// Normal keywords
// Normal keywords
default:
$arrKeywords[] = $strKeyword;
break;
}
}
// Fuzzy search
if ($blnFuzzy) {
foreach ($arrKeywords as $strKeyword) {
$arrWildcards[] = '%' . $strKeyword . '%';
}
$arrKeywords = array();
}
// Count keywords
$intPhrases = count($arrPhrases);
$intWildcards = count($arrWildcards);
$intIncluded = count($arrIncluded);
$intExcluded = count($arrExcluded);
$intKeywords = 0;
$arrValues = array();
// Remember found words so we can highlight them later
$strQuery = "SELECT tl_search_index.pid AS sid, GROUP_CONCAT(word) AS matches";
// Get the number of wildcard matches
if (!$blnOrSearch && $intWildcards) {
$strQuery .= ", (SELECT COUNT(*) FROM tl_search_index WHERE (" . implode(' OR ', array_fill(0, $intWildcards, 'word LIKE ?')) . ") AND pid=sid) AS wildcards";
$arrValues = array_merge($arrValues, $arrWildcards);
}
// Count the number of matches
$strQuery .= ", COUNT(*) AS count";
// Get the relevance
$strQuery .= ", SUM(relevance) AS relevance";
// Get meta information from tl_search
$strQuery .= ", tl_search.*";
// see #4506
// Prepare keywords array
$arrAllKeywords = array();
// Get keywords
if (!empty($arrKeywords)) {
$arrAllKeywords[] = implode(' OR ', array_fill(0, count($arrKeywords), 'word=?'));
$arrValues = array_merge($arrValues, $arrKeywords);
$intKeywords += count($arrKeywords);
}
// Get included keywords
if ($intIncluded) {
$arrAllKeywords[] = implode(' OR ', array_fill(0, $intIncluded, 'word=?'));
$arrValues = array_merge($arrValues, $arrIncluded);
$intKeywords += $intIncluded;
}
// Get keywords from phrases
if ($intPhrases) {
foreach ($arrPhrases as $strPhrase) {
$arrWords = explode('[^[:alnum:]]+', Utf8::substr($strPhrase, 7, -7));
$arrAllKeywords[] = implode(' OR ', array_fill(0, count($arrWords), 'word=?'));
$arrValues = array_merge($arrValues, $arrWords);
$intKeywords += count($arrWords);
}
}
// Get wildcards
if ($intWildcards) {
$arrAllKeywords[] = implode(' OR ', array_fill(0, $intWildcards, 'word LIKE ?'));
$arrValues = array_merge($arrValues, $arrWildcards);
}
$strQuery .= " FROM tl_search_index LEFT JOIN tl_search ON(tl_search_index.pid=tl_search.id) WHERE (" . implode(' OR ', $arrAllKeywords) . ")";
// Get phrases
if ($intPhrases) {
$strQuery .= " AND (" . implode($blnOrSearch ? ' OR ' : ' AND ', array_fill(0, $intPhrases, 'tl_search_index.pid IN(SELECT id FROM tl_search WHERE text REGEXP ?)')) . ")";
$arrValues = array_merge($arrValues, $arrPhrases);
}
// Include keywords
if ($intIncluded) {
$strQuery .= " AND tl_search_index.pid IN(SELECT pid FROM tl_search_index WHERE " . implode(' OR ', array_fill(0, $intIncluded, 'word=?')) . ")";
$arrValues = array_merge($arrValues, $arrIncluded);
}
// Exclude keywords
if ($intExcluded) {
$strQuery .= " AND tl_search_index.pid NOT IN(SELECT pid FROM tl_search_index WHERE " . implode(' OR ', array_fill(0, $intExcluded, 'word=?')) . ")";
$arrValues = array_merge($arrValues, $arrExcluded);
}
// Limit results to a particular set of pages
if (!empty($arrPid) && is_array($arrPid)) {
$strQuery .= " AND tl_search_index.pid IN(SELECT id FROM tl_search WHERE pid IN(" . implode(',', array_map('intval', $arrPid)) . "))";
}
$strQuery .= " GROUP BY tl_search_index.pid";
// Make sure to find all words
if (!$blnOrSearch) {
// Number of keywords without wildcards
$strQuery .= " HAVING count >= " . $intKeywords;
// Dynamically add the number of wildcard matches
if ($intWildcards) {
$strQuery .= " + IF(wildcards>" . $intWildcards . ", wildcards, " . $intWildcards . ")";
}
}
// Sort by relevance
$strQuery .= " ORDER BY relevance DESC";
// Return result
$objResultStmt = \Database::getInstance()->prepare($strQuery);
if ($intRows > 0) {
$objResultStmt->limit($intRows, $intOffset);
}
return $objResultStmt->execute($arrValues);
}
