links = $links; } /** * Filter links according to parameters. * * @param string $type Type of filter (eg. tags, permalink, etc.). * @param string $request Filter content. * @param bool $casesensitive Optional: Perform case sensitive filter if true. * @param bool $privateonly Optional: Only returns private links if true. * * @return array filtered link list. */ public function filter($type, $request, $casesensitive = false, $privateonly = false) { switch($type) { case self::$FILTER_HASH: return $this->filterSmallHash($request); break; case self::$FILTER_TEXT: return $this->filterFulltext($request, $privateonly); break; case self::$FILTER_TAG: return $this->filterTags($request, $casesensitive, $privateonly); break; case self::$FILTER_DAY: return $this->filterDay($request); break; default: return $this->noFilter($privateonly); } } /** * Unknown filter, but handle private only. * * @param bool $privateonly returns private link only if true. * * @return array filtered links. */ private function noFilter($privateonly = false) { if (! $privateonly) { krsort($this->links); return $this->links; } $out = array(); foreach ($this->links as $value) { if ($value['private']) { $out[$value['linkdate']] = $value; } } krsort($out); return $out; } /** * Returns the shaare corresponding to a smallHash. * * @param string $smallHash permalink hash. * * @return array $filtered array containing permalink data. */ private function filterSmallHash($smallHash) { $filtered = array(); foreach ($this->links as $l) { if ($smallHash == smallHash($l['linkdate'])) { // Yes, this is ugly and slow $filtered[$l['linkdate']] = $l; return $filtered; } } return $filtered; } /** * Returns the list of links corresponding to a full-text search * * Searches: * - in the URLs, title and description; * - are case-insensitive; * - terms surrounded by quotes " are exact terms search. * - terms starting with a dash - are excluded (except exact terms). * * Example: * print_r($mydb->filterFulltext('hollandais')); * * mb_convert_case($val, MB_CASE_LOWER, 'UTF-8') * - allows to perform searches on Unicode text * - see https://github.com/shaarli/Shaarli/issues/75 for examples * * @param string $searchterms search query. * @param bool $privateonly return only private links if true. * * @return array search results. */ private function filterFulltext($searchterms, $privateonly = false) { $search = mb_convert_case(html_entity_decode($searchterms), MB_CASE_LOWER, 'UTF-8'); $exactRegex = '/"([^"]+)"/'; // Retrieve exact search terms. preg_match_all($exactRegex, $search, $exactSearch); $exactSearch = array_values(array_filter($exactSearch[1])); // Remove exact search terms to get AND terms search. $explodedSearchAnd = explode(' ', trim(preg_replace($exactRegex, '', $search))); $explodedSearchAnd = array_values(array_filter($explodedSearchAnd)); // Filter excluding terms and update andSearch. $excludeSearch = array(); $andSearch = array(); foreach ($explodedSearchAnd as $needle) { if ($needle[0] == '-' && strlen($needle) > 1) { $excludeSearch[] = substr($needle, 1); } else { $andSearch[] = $needle; } } $keys = array('title', 'description', 'url', 'tags'); // Iterate over every stored link. foreach ($this->links as $link) { // ignore non private links when 'privatonly' is on. if (! $link['private'] && $privateonly === true) { continue; } // Iterate over searchable link fields. foreach ($keys as $key) { // Be optimistic $found = true; $haystack = mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8'); // First, we look for exact term search for ($i = 0; $i < count($exactSearch) && $found; $i++) { $found = strpos($haystack, $exactSearch[$i]) !== false; } // Iterate over keywords, if keyword is not found, // no need to check for the others. We want all or nothing. for ($i = 0; $i < count($andSearch) && $found; $i++) { $found = strpos($haystack, $andSearch[$i]) !== false; } // Exclude terms. for ($i = 0; $i < count($excludeSearch) && $found; $i++) { $found = strpos($haystack, $excludeSearch[$i]) === false; } // One of the fields of the link matches, no need to check the other. if ($found) { break; } } if ($found) { $filtered[$link['linkdate']] = $link; } } krsort($filtered); return $filtered; } /** * Returns the list of links associated with a given list of tags * * You can specify one or more tags, separated by space or a comma, e.g. * print_r($mydb->filterTags('linux programming')); * * @param string $tags list of tags separated by commas or blank spaces. * @param bool $casesensitive ignore case if false. * @param bool $privateonly returns private links only. * * @return array filtered links. */ public function filterTags($tags, $casesensitive = false, $privateonly = false) { $searchtags = self::tagsStrToArray($tags, $casesensitive); $filtered = array(); if (empty($searchtags)) { return $filtered; } foreach ($this->links as $link) { // ignore non private links when 'privatonly' is on. if (! $link['private'] && $privateonly === true) { continue; } $linktags = self::tagsStrToArray($link['tags'], $casesensitive); $found = true; for ($i = 0 ; $i < count($searchtags) && $found; $i++) { // Exclusive search, quit if tag found. // Or, tag not found in the link, quit. if (($searchtags[$i][0] == '-' && in_array(substr($searchtags[$i], 1), $linktags)) || ($searchtags[$i][0] != '-') && ! in_array($searchtags[$i], $linktags) ) { $found = false; } } if ($found) { $filtered[$link['linkdate']] = $link; } } krsort($filtered); return $filtered; } /** * Returns the list of articles for a given day, chronologically sorted * * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g. * print_r($mydb->filterDay('20120125')); * * @param string $day day to filter. * * @return array all link matching given day. * * @throws Exception if date format is invalid. */ public function filterDay($day) { if (! checkDateFormat('Ymd', $day)) { throw new Exception('Invalid date format'); } $filtered = array(); foreach ($this->links as $l) { if (startsWith($l['linkdate'], $day)) { $filtered[$l['linkdate']] = $l; } } ksort($filtered); return $filtered; } /** * Convert a list of tags (str) to an array. Also * - handle case sensitivity. * - accepts spaces commas as separator. * * @param string $tags string containing a list of tags. * @param bool $casesensitive will convert everything to lowercase if false. * * @return array filtered tags string. */ public static function tagsStrToArray($tags, $casesensitive) { // We use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek) $tagsOut = $casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER, 'UTF-8'); $tagsOut = str_replace(',', ' ', $tagsOut); return array_filter(explode(' ', trim($tagsOut)), 'strlen'); } }