X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=application%2FLinkFilter.php;h=91c79905280851473878059a7e6d8e57dd9b3265;hb=f24896b237e40718fb6eaa2869592eb0855a47fd;hp=ceb47d16c21133e670c20066d126aa23019ed816;hpb=1e7331126d81a5759ab91c221f7e0f164aeebfb5;p=github%2Fshaarli%2FShaarli.git

diff --git a/application/LinkFilter.php b/application/LinkFilter.php
index ceb47d16..91c79905 100644
--- a/application/LinkFilter.php
+++ b/application/LinkFilter.php
@@ -1,5 +1,7 @@
 <?php
 
+use Shaarli\Bookmark\LinkDB;
+
 /**
  * Class LinkFilter.
  *
@@ -28,12 +30,17 @@ class LinkFilter
     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)
     {
@@ -44,54 +51,79 @@ 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.
+     * @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, $privateonly = false)
+    public function filter($type, $request, $casesensitive = false, $visibility = 'all', $untaggedonly = false)
     {
-        switch($type) {
+        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: // == "vuotext"
+                $noRequest = empty($request) || (empty($request[0]) && empty($request[1]));
+                if ($noRequest) {
+                    if ($untaggedonly) {
+                        return $this->filterUntagged($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, $privateonly);
-                break;
+                return $this->filterFulltext($request, $visibility);
             case self::$FILTER_TAG:
-                return $this->filterTags($request, $casesensitive, $privateonly);
-                break;
+                if ($untaggedonly) {
+                    return $this->filterUntagged($visibility);
+                } else {
+                    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;
+            } elseif (! $value['private'] && $visibility === 'public') {
+                $out[$key] = $value;
             }
         }
 
-        krsort($out);
         return $out;
     }
 
@@ -101,17 +133,24 @@ 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)
     {
         $filtered = array();
-        foreach ($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;
     }
 
@@ -120,7 +159,9 @@ class LinkFilter
      *
      * 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'));
@@ -130,71 +171,130 @@ class LinkFilter
      *  - 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[] = $needle;
+            }
         }
-        // Iterate over every stored link.
-        foreach ($this->links as $link) {
 
+        $keys = array('title', 'description', 'url', 'tags');
+
+        // Iterate over every stored link.
+        foreach ($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'] && $visibility === 'private') {
+                    continue;
+                } elseif ($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;
     }
 
+    /**
+     * 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
      *
@@ -203,42 +303,94 @@ class LinkFilter
      *
      * @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')
     {
-        $searchtags = self::tagsStrToArray($tags, $casesensitive);
-        $filtered = array();
-        if (empty($searchtags)) {
-            return $filtered;
+        // 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);
         }
 
-        foreach ($this->links as $link) {
-            // ignore non private links when 'privatonly' is on.
-            if (! $link['private'] && $privateonly === true) {
+        if (!count($inputTags)) {
+            // no input tags
+            return $this->noFilter($visibility);
+        }
+
+        // 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 => $link) {
+            // check level of visibility
+            // ignore non private links when 'privateonly' is on.
+            if ($visibility !== 'all') {
+                if (! $link['private'] && $visibility === 'private') {
+                    continue;
+                } elseif ($link['private'] && $visibility === 'public') {
+                    continue;
+                }
+            }
+            $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;
             }
+            $filtered[$key] = $link;
+        }
+        return $filtered;
+    }
 
-            $linktags = self::tagsStrToArray($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;
+    /**
+     * 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'] && $visibility === 'private') {
+                    continue;
+                } elseif ($link['private'] && $visibility === 'public') {
+                    continue;
                 }
             }
 
-            if ($found) {
-                $filtered[$link['linkdate']] = $link;
+            if (empty(trim($link['tags']))) {
+                $filtered[$key] = $link;
             }
         }
-        krsort($filtered);
+
         return $filtered;
     }
 
@@ -261,13 +413,14 @@ class LinkFilter
         }
 
         $filtered = array();
-        foreach ($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);
     }
 
     /**
@@ -279,13 +432,24 @@ class LinkFilter
      * @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');
+        return preg_split('/\s+/', $tagsOut, -1, PREG_SPLIT_NO_EMPTY);
+    }
+}
+
+class LinkNotFoundException extends Exception
+{
+    /**
+     * LinkNotFoundException constructor.
+     */
+    public function __construct()
+    {
+        $this->message =  t('The link you are trying to reach does not exist or has been deleted.');
     }
 }