switch($type) {
case self::$FILTER_HASH:
return $this->filterSmallHash($request);
- break;
+ case self::$FILTER_TAG | self::$FILTER_TEXT:
+ if (!empty($request)) {
+ $filtered = $this->links;
+ if (isset($request[0])) {
+ $filtered = $this->filterTags($request[0], $casesensitive, $privateonly);
+ }
+ if (isset($request[1])) {
+ $lf = new LinkFilter($filtered);
+ $filtered = $lf->filterFulltext($request[1], $privateonly);
+ }
+ return $filtered;
+ }
+ return $this->noFilter($privateonly);
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);
}
*
* Searches:
* - in the URLs, title and description;
- * - are case-insensitive.
+ * - are case-insensitive;
+ * - terms surrounded by quotes " are exact terms search.
+ * - terms starting with a dash - are excluded (except exact terms).
*
* Example:
* print_r($mydb->filterFulltext('hollandais'));
*/
private function filterFulltext($searchterms, $privateonly = false)
{
+ if (empty($searchterms)) {
+ return $this->links;
+ }
+
+ $filtered = array();
$search = mb_convert_case(html_entity_decode($searchterms), MB_CASE_LOWER, 'UTF-8');
- $explodedSearch = explode(' ', trim($search));
- $keys = array('title', 'description', 'url', 'tags');
- $found = true;
- $searchExactPhrase = false;
-
- // Check if we're using double-quotes to search for the exact string
- if ($search[0] == '"' && $search[strlen($search) - 1] == '"') {
- $searchExactPhrase = true;
-
- // Remove the double-quotes as they are not what we search for
- $search = substr($search, 1, -1);
+ $exactRegex = '/"([^"]+)"/';
+ // Retrieve exact search terms.
+ preg_match_all($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) {
continue;
}
- // Iterate over searchable link fields.
+ // Concatenate link fields to search across fields.
+ // Adds a '\' separator for exact search terms.
+ $content = '';
foreach ($keys as $key) {
- // Be optimistic
- $found = true;
-
- // FIXME: Find a better word for where you're searching in
- $haystack = mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8');
-
- // When searching for the phrase, check if it's in the haystack...
- if ( $searchExactPhrase && strpos($haystack, $search) !== false) {
- break;
- }
- else {
- // Iterate over keywords, if keyword is not found,
- // no need to check for the others. We want all or nothing.
- foreach($explodedSearch as $keyword) {
- if(strpos($haystack, $keyword) === false) {
- $found = false;
- break;
- }
- }
- }
-
- // One of the fields of the link matches, no need to check the other.
- if ($found) {
- break;
- }
+ $content .= mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8') . '\\';
+ }
+
+ // Be optimistic
+ $found = true;
+
+ // First, we look for exact term search
+ for ($i = 0; $i < count($exactSearch) && $found; $i++) {
+ $found = strpos($content, $exactSearch[$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++) {
+ $found = strpos($content, $andSearch[$i]) !== false;
+ }
+
+ // Exclude terms.
+ for ($i = 0; $i < count($excludeSearch) && $found; $i++) {
+ $found = strpos($content, $excludeSearch[$i]) === false;
}
-
+
if ($found) {
$filtered[$link['linkdate']] = $link;
}
*/
public function filterTags($tags, $casesensitive = false, $privateonly = false)
{
- $searchtags = $this->tagsStrToArray($tags, $casesensitive);
+ // Implode if array for clean up.
+ $tags = is_array($tags) ? trim(implode(' ', $tags)) : $tags;
+ if (empty($tags)) {
+ return $this->links;
+ }
+
+ $searchtags = self::tagsStrToArray($tags, $casesensitive);
$filtered = array();
+ if (empty($searchtags)) {
+ return $filtered;
+ }
- foreach ($this->links as $l) {
+ foreach ($this->links as $link) {
// ignore non private links when 'privatonly' is on.
- if (! $l['private'] && $privateonly === true) {
+ if (! $link['private'] && $privateonly === true) {
continue;
}
- $linktags = $this->tagsStrToArray($l['tags'], $casesensitive);
+ $linktags = self::tagsStrToArray($link['tags'], $casesensitive);
- if (count(array_intersect($linktags, $searchtags)) == count($searchtags)) {
- $filtered[$l['linkdate']] = $l;
+ $found = true;
+ for ($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);
* Convert a list of tags (str) to an array. Also
* - handle case sensitivity.
* - accepts spaces commas as separator.
- * - remove private tags for loggedout users.
*
* @param string $tags string containing a list of tags.
* @param bool $casesensitive will convert everything to lowercase if false.
*
* @return array filtered tags string.
*/
- public function tagsStrToArray($tags, $casesensitive)
+ public static function tagsStrToArray($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(',', ' ', $tagsOut);
- return explode(' ', trim($tagsOut));
+ return array_filter(explode(' ', trim($tagsOut)), 'strlen');
}
}
projects / github / shaarli / Shaarli.git / blobdiff