X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=application%2FLinkFilter.php;h=8f147974e9d5cf232f3cf3f509b3401fcf5e6b84;hb=1004742f09b55ff781c13745781b9a7e90986faa;hp=0e887d3805a2fa7ef2024f5da2d6b155dc4a870e;hpb=3e395a6bc63cba16b0772a382f6cbac0ce4ab906;p=github%2Fshaarli%2FShaarli.git

diff --git a/application/LinkFilter.php b/application/LinkFilter.php
index 0e887d38..8f147974 100644
--- a/application/LinkFilter.php
+++ b/application/LinkFilter.php
@@ -52,35 +52,47 @@ class LinkFilter
      * @param mixed  $request       Filter content.
      * @param bool   $casesensitive Optional: Perform case sensitive filter if true.
      * @param string $visibility    Optional: return only all/private/public links
+     * @param string $untaggedonly  Optional: return only untagged links. Applies only if $type includes FILTER_TAG
      *
      * @return array filtered link list.
      */
-    public function filter($type, $request, $casesensitive = false, $visibility = 'all')
+    public function filter($type, $request, $casesensitive = false, $visibility = 'all', $untaggedonly = false)
     {
         if (! in_array($visibility, ['all', 'public', 'private'])) {
             $visibility = 'all';
         }
 
-        switch($type) {
+        switch ($type) {
             case self::$FILTER_HASH:
                 return $this->filterSmallHash($request);
-            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], $visibility);
+            case self::$FILTER_TAG | self::$FILTER_TEXT: // == "vuotext"
+                $noRequest = empty($request) || (empty($request[0]) && empty($request[1]));
+                if ($noRequest) {
+                    if ($untaggedonly) {
+                        return $this->filterUntagged($visibility);
                     }
-                    return $filtered;
+                    return $this->noFilter($visibility);
                 }
-                return $this->noFilter($visibility);
+                if ($untaggedonly) {
+                    $filtered = $this->filterUntagged($visibility);
+                } else {
+                    $filtered = $this->links;
+                }
+                if (!empty($request[0])) {
+                    $filtered = (new LinkFilter($filtered))->filterTags($request[0], $casesensitive, $visibility);
+                }
+                if (!empty($request[1])) {
+                    $filtered = (new LinkFilter($filtered))->filterFulltext($request[1], $visibility);
+                }
+                return $filtered;
             case self::$FILTER_TEXT:
                 return $this->filterFulltext($request, $visibility);
             case self::$FILTER_TAG:
-                return $this->filterTags($request, $casesensitive, $visibility);
+                if ($untaggedonly) {
+                    return $this->filterUntagged($visibility);
+                } else {
+                    return $this->filterTags($request, $casesensitive, $visibility);
+                }
             case self::$FILTER_DAY:
                 return $this->filterDay($request);
             default:
@@ -105,7 +117,7 @@ class LinkFilter
         foreach ($this->links as $key => $value) {
             if ($value['private'] && $visibility === 'private') {
                 $out[$key] = $value;
-            } else if (! $value['private'] && $visibility === 'public') {
+            } elseif (! $value['private'] && $visibility === 'public') {
                 $out[$key] = $value;
             }
         }
@@ -193,12 +205,11 @@ class LinkFilter
 
         // Iterate over every stored link.
         foreach ($this->links as $id => $link) {
-
             // ignore non private links when 'privatonly' is on.
             if ($visibility !== 'all') {
                 if (! $link['private'] && $visibility === 'private') {
                     continue;
-                } else if ($link['private'] && $visibility === 'public') {
+                } elseif ($link['private'] && $visibility === 'public') {
                     continue;
                 }
             }
@@ -237,6 +248,51 @@ class LinkFilter
         return $filtered;
     }
 
+    /**
+     * generate a regex fragment out of a tag
+     * @param string $tag to to generate regexs from. may start with '-' to negate, contain '*' as wildcard
+     * @return string generated regex fragment
+     */
+    private static function tag2regex($tag)
+    {
+        $len = strlen($tag);
+        if (!$len || $tag === "-" || $tag === "*") {
+            // nothing to search, return empty regex
+            return '';
+        }
+        if ($tag[0] === "-") {
+            // query is negated
+            $i = 1; // use offset to start after '-' character
+            $regex = '(?!'; // create negative lookahead
+        } else {
+            $i = 0; // start at first character
+            $regex = '(?='; // use positive lookahead
+        }
+        $regex .= '.*(?:^| )'; // before tag may only be a space or the beginning
+        // iterate over string, separating it into placeholder and content
+        for (; $i < $len; $i++) {
+            if ($tag[$i] === '*') {
+                // placeholder found
+                $regex .= '[^ ]*?';
+            } else {
+                // regular characters
+                $offset = strpos($tag, '*', $i);
+                if ($offset === false) {
+                    // no placeholder found, set offset to end of string
+                    $offset = $len;
+                }
+                // subtract one, as we want to get before the placeholder or end of string
+                $offset -= 1;
+                // we got a tag name that we want to search for. escape any regex characters to prevent conflicts.
+                $regex .= preg_quote(substr($tag, $i, $offset - $i + 1), '/');
+                // move $i on
+                $i = $offset;
+            }
+        }
+        $regex .= '(?:$| ))'; // after the tag may only be a space or the end
+        return $regex;
+    }
+
     /**
      * Returns the list of links associated with a given list of tags
      *
@@ -251,49 +307,60 @@ class LinkFilter
      */
     public function filterTags($tags, $casesensitive = false, $visibility = 'all')
     {
-        // Implode if array for clean up.
-        $tags = is_array($tags) ? trim(implode(' ', $tags)) : $tags;
-        if ($tags === false) {
-            return $this->filterUntagged($visibility);
+        // get single tags (we may get passed an array, even though the docs say different)
+        $inputTags = $tags;
+        if (!is_array($tags)) {
+            // we got an input string, split tags
+            $inputTags = preg_split('/(?:\s+)|,/', $inputTags, -1, PREG_SPLIT_NO_EMPTY);
         }
-        if (empty($tags)) {
+
+        if (!count($inputTags)) {
+            // no input tags
             return $this->noFilter($visibility);
         }
 
-        $searchtags = self::tagsStrToArray($tags, $casesensitive);
-        $filtered = array();
-        if (empty($searchtags)) {
-            return $filtered;
+        // build regex from all tags
+        $re = '/^' . implode(array_map("self::tag2regex", $inputTags)) . '.*$/';
+        if (!$casesensitive) {
+            // make regex case insensitive
+            $re .= 'i';
         }
 
+        // create resulting array
+        $filtered = array();
+
+        // iterate over each link
         foreach ($this->links as $key => $link) {
-            // ignore non private links when 'privatonly' is on.
+            // check level of visibility
+            // ignore non private links when 'privateonly' is on.
             if ($visibility !== 'all') {
                 if (! $link['private'] && $visibility === 'private') {
                     continue;
-                } else if ($link['private'] && $visibility === 'public') {
+                } elseif ($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;
+            $search = $link['tags']; // build search string, start with tags of current link
+            if (strlen(trim($link['description'])) && strpos($link['description'], '#') !== false) {
+                // description given and at least one possible tag found
+                $descTags = array();
+                // find all tags in the form of #tag in the description
+                preg_match_all(
+                    '/(?<![' . self::$HASHTAG_CHARS . '])#([' . self::$HASHTAG_CHARS . ']+?)\b/sm',
+                    $link['description'],
+                    $descTags
+                );
+                if (count($descTags[1])) {
+                    // there were some tags in the description, add them to the search string
+                    $search .= ' ' . implode(' ', $descTags[1]);
                 }
+            };
+            // match regular expression with search string
+            if (!preg_match($re, $search)) {
+                // this entry does _not_ match our regex
+                continue;
             }
-
-            if ($found) {
-                $filtered[$key] = $link;
-            }
+            $filtered[$key] = $link;
         }
         return $filtered;
     }
@@ -312,7 +379,7 @@ class LinkFilter
             if ($visibility !== 'all') {
                 if (! $link['private'] && $visibility === 'private') {
                     continue;
-                } else if ($link['private'] && $visibility === 'public') {
+                } elseif ($link['private'] && $visibility === 'public') {
                     continue;
                 }
             }
@@ -354,28 +421,6 @@ class LinkFilter
         return array_reverse($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.
@@ -398,5 +443,11 @@ class LinkFilter
 
 class LinkNotFoundException extends Exception
 {
-    protected $message = 'The link you are trying to reach does not exist or has been deleted.';
+    /**
+     * LinkNotFoundException constructor.
+     */
+    public function __construct()
+    {
+        $this->message =  t('The link you are trying to reach does not exist or has been deleted.');
+    }
 }