X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=application%2FLinkFilter.php;h=d4fe28df6d4968af4ac673110022ab35d2be9488;hb=5ebc1d504bc8a8f29f49a8a4fc1c421f78677b2a;hp=e2ef94ea78ff5476cca3d058ff72538060358539;hpb=bedd176a5406003631da42366736fd5ebae29135;p=github%2Fshaarli%2FShaarli.git diff --git a/application/LinkFilter.php b/application/LinkFilter.php index e2ef94ea..d4fe28df 100644 --- a/application/LinkFilter.php +++ b/application/LinkFilter.php @@ -27,6 +27,11 @@ class LinkFilter */ public static $FILTER_DAY = 'FILTER_DAY'; + /** + * @var string Allowed characters for hashtags (regex syntax). + */ + public static $HASHTAG_CHARS = '\p{Pc}\p{N}\p{L}\p{Mn}'; + /** * @var array all available links. */ @@ -44,7 +49,7 @@ class LinkFilter * 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 $privateonly Optional: Only returns private links if true. * @@ -55,16 +60,25 @@ class LinkFilter switch($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); } @@ -101,6 +115,8 @@ class LinkFilter * @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) { @@ -112,6 +128,11 @@ class LinkFilter return $filtered; } } + + if (empty($filtered)) { + throw new LinkNotFoundException(); + } + return $filtered; } @@ -138,6 +159,11 @@ class LinkFilter */ private function filterFulltext($searchterms, $privateonly = false) { + if (empty($searchterms)) { + return $this->links; + } + + $filtered = array(); $search = mb_convert_case(html_entity_decode($searchterms), MB_CASE_LOWER, 'UTF-8'); $exactRegex = '/"([^"]+)"/'; // Retrieve exact search terms. @@ -169,35 +195,32 @@ class LinkFilter 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; - - $haystack = mb_convert_case($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; - } + $content .= mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8') . '\\'; + } - // Iterate over keywords, if keyword is not found, - // no need to check for the others. We want all or nothing. - for ($i = 0; $i < count($andSearch) && $found; $i++) { - $found = strpos($haystack, $andSearch[$i]) !== false; - } + // Be optimistic + $found = true; - // Exclude terms. - for ($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 ($found) { - break; - } + // First, we look for exact term search + for ($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']] = $link; } @@ -221,6 +244,12 @@ class LinkFilter */ public function filterTags($tags, $casesensitive = false, $privateonly = false) { + // 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)) { @@ -239,8 +268,10 @@ class LinkFilter 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], $linktags) + if (($searchtags[$i][0] == '-' + && $this->searchTagAndHashTag(substr($searchtags[$i], 1), $linktags, $link['description'])) + || ($searchtags[$i][0] != '-') + && ! $this->searchTagAndHashTag($searchtags[$i], $linktags, $link['description']) ) { $found = false; } @@ -282,6 +313,28 @@ class LinkFilter return $filtered; } + /** + * 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. @@ -298,6 +351,11 @@ class LinkFilter $tagsOut = $casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER, 'UTF-8'); $tagsOut = str_replace(',', ' ', $tagsOut); - return array_filter(explode(' ', trim($tagsOut)), 'strlen'); + return array_values(array_filter(explode(' ', trim($tagsOut)), 'strlen')); } } + +class LinkNotFoundException extends Exception +{ + protected $message = 'The link you are trying to reach does not exist or has been deleted.'; +}