public static $FILTER_DAY = 'FILTER_DAY';
/**
- * @var array all available links.
+ * @var string Allowed characters for hashtags (regex syntax).
+ */
+ public static $HASHTAG_CHARS = '\p{Pc}\p{N}\p{L}\p{Mn}';
+
+ /**
+ * @var LinkDB all available links.
*/
private $links;
/**
- * @param array $links initialization.
+ * @param LinkDB $links initialization.
*/
public function __construct($links)
{
* Filter links according to parameters.
*
* @param string $type Type of filter (eg. tags, permalink, etc.).
- * @param string $request Filter content.
+ * @param mixed $request Filter content.
* @param bool $casesensitive Optional: Perform case sensitive filter if true.
- * @param bool $privateonly Optional: Only returns private links if true.
+ * @param string $visibility Optional: return only all/private/public links
*
* @return array filtered link list.
*/
- public function filter($type, $request, $casesensitive = false, $privateonly = false)
+ public function filter($type, $request, $casesensitive = false, $visibility = 'all')
{
+ if (! in_array($visibility, ['all', 'public', 'private'])) {
+ $visibility = 'all';
+ }
+
switch($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, $visibility);
+ }
+ if (isset($request[1])) {
+ $lf = new LinkFilter($filtered);
+ $filtered = $lf->filterFulltext($request[1], $visibility);
+ }
+ return $filtered;
+ }
+ return $this->noFilter($visibility);
case self::$FILTER_TEXT:
- return $this->filterFulltext($request, $privateonly);
- break;
+ return $this->filterFulltext($request, $visibility);
case self::$FILTER_TAG:
- return $this->filterTags($request, $casesensitive, $privateonly);
- break;
+ return $this->filterTags($request, $casesensitive, $visibility);
case self::$FILTER_DAY:
return $this->filterDay($request);
- break;
default:
- return $this->noFilter($privateonly);
+ return $this->noFilter($visibility);
}
}
/**
* Unknown filter, but handle private only.
*
- * @param bool $privateonly returns private link only if true.
+ * @param string $visibility Optional: return only all/private/public links
*
* @return array filtered links.
*/
- private function noFilter($privateonly = false)
+ private function noFilter($visibility = 'all')
{
- if (! $privateonly) {
- krsort($this->links);
+ if ($visibility === 'all') {
return $this->links;
}
$out = array();
- foreach ($this->links as $value) {
- if ($value['private']) {
- $out[$value['linkdate']] = $value;
+ foreach ($this->links as $key => $value) {
+ if ($value['private'] && $visibility === 'private') {
+ $out[$key] = $value;
+ } else if (! $value['private'] && $visibility === 'public') {
+ $out[$key] = $value;
}
}
- krsort($out);
return $out;
}
* @param string $smallHash permalink hash.
*
* @return array $filtered array containing permalink data.
+ *
+ * @throws LinkNotFoundException if the smallhash doesn't match any link.
*/
private function filterSmallHash($smallHash)
{
$filtered = array();
- foreach ($this->links as $l) {
- if ($smallHash == smallHash($l['linkdate'])) {
+ foreach ($this->links as $key => $l) {
+ if ($smallHash == $l['shorturl']) {
// Yes, this is ugly and slow
- $filtered[$l['linkdate']] = $l;
+ $filtered[$key] = $l;
return $filtered;
}
}
+
+ if (empty($filtered)) {
+ throw new LinkNotFoundException();
+ }
+
return $filtered;
}
*
* Searches:
* - in the URLs, title and description;
- * - are case-insensitive.
+ * - are case-insensitive;
+ * - terms surrounded by quotes " are exact terms search.
+ * - terms starting with a dash - are excluded (except exact terms).
*
* Example:
* print_r($mydb->filterFulltext('hollandais'));
* - see https://github.com/shaarli/Shaarli/issues/75 for examples
*
* @param string $searchterms search query.
- * @param bool $privateonly return only private links if true.
+ * @param string $visibility Optional: return only all/private/public links.
*
* @return array search results.
*/
- private function filterFulltext($searchterms, $privateonly = false)
+ private function filterFulltext($searchterms, $visibility = 'all')
{
+ if (empty($searchterms)) {
+ return $this->noFilter($visibility);
+ }
+
+ $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 $link) {
+ foreach ($this->links as $id => $link) {
// ignore non private links when 'privatonly' is on.
- if (! $link['private'] && $privateonly === true) {
- continue;
+ if ($visibility !== 'all') {
+ if (! $link['private'] && $visibility === 'private') {
+ continue;
+ } else if ($link['private'] && $visibility === 'public') {
+ 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;
+ $filtered[$id] = $link;
}
}
- krsort($filtered);
return $filtered;
}
*
* @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.
+ * @param string $visibility Optional: return only all/private/public links.
*
* @return array filtered links.
*/
- public function filterTags($tags, $casesensitive = false, $privateonly = false)
+ public function filterTags($tags, $casesensitive = false, $visibility = 'all')
{
- $searchtags = $this->tagsStrToArray($tags, $casesensitive);
+ // Implode if array for clean up.
+ $tags = is_array($tags) ? trim(implode(' ', $tags)) : $tags;
+ if ($tags === false) {
+ return $this->filterUntagged($visibility);
+ }
+ if (empty($tags)) {
+ return $this->noFilter($visibility);
+ }
+
+ $searchtags = self::tagsStrToArray($tags, $casesensitive);
$filtered = array();
+ if (empty($searchtags)) {
+ return $filtered;
+ }
- foreach ($this->links as $l) {
+ foreach ($this->links as $key => $link) {
// ignore non private links when 'privatonly' is on.
- if (! $l['private'] && $privateonly === true) {
- continue;
+ if ($visibility !== 'all') {
+ if (! $link['private'] && $visibility === 'private') {
+ continue;
+ } else if ($link['private'] && $visibility === 'public') {
+ 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] == '-'
+ && $this->searchTagAndHashTag(substr($searchtags[$i], 1), $linktags, $link['description']))
+ || ($searchtags[$i][0] != '-')
+ && ! $this->searchTagAndHashTag($searchtags[$i], $linktags, $link['description'])
+ ) {
+ $found = false;
+ }
}
- $linktags = $this->tagsStrToArray($l['tags'], $casesensitive);
+ if ($found) {
+ $filtered[$key] = $link;
+ }
+ }
+ return $filtered;
+ }
+
+ /**
+ * Return only links without any tag.
+ *
+ * @param string $visibility return only all/private/public links.
+ *
+ * @return array filtered links.
+ */
+ public function filterUntagged($visibility)
+ {
+ $filtered = [];
+ foreach ($this->links as $key => $link) {
+ if ($visibility !== 'all') {
+ if (! $link['private'] && $visibility === 'private') {
+ continue;
+ } else if ($link['private'] && $visibility === 'public') {
+ continue;
+ }
+ }
- if (count(array_intersect($linktags, $searchtags)) == count($searchtags)) {
- $filtered[$l['linkdate']] = $l;
+ if (empty(trim($link['tags']))) {
+ $filtered[$key] = $link;
}
}
- krsort($filtered);
+
return $filtered;
}
}
$filtered = array();
- foreach ($this->links as $l) {
- if (startsWith($l['linkdate'], $day)) {
- $filtered[$l['linkdate']] = $l;
+ foreach ($this->links as $key => $l) {
+ if ($l['created']->format('Ymd') == $day) {
+ $filtered[$key] = $l;
}
}
- ksort($filtered);
- return $filtered;
+
+ // sort by date ASC
+ return array_reverse($filtered, true);
+ }
+
+ /**
+ * Check if a tag is found in the taglist, or as an hashtag in the link description.
+ *
+ * @param string $tag Tag to search.
+ * @param array $taglist List of tags for the current link.
+ * @param string $description Link description.
+ *
+ * @return bool True if found, false otherwise.
+ */
+ protected function searchTagAndHashTag($tag, $taglist, $description)
+ {
+ if (in_array($tag, $taglist)) {
+ return true;
+ }
+
+ if (preg_match('/(^| )#'. $tag .'([^'. self::$HASHTAG_CHARS .']|$)/mui', $description) > 0) {
+ return true;
+ }
+
+ return false;
}
/**
* 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 preg_split('/\s+/', $tagsOut, -1, PREG_SPLIT_NO_EMPTY);
}
}
+
+class LinkNotFoundException extends Exception
+{
+ protected $message = 'The link you are trying to reach does not exist or has been deleted.';
+}