Merge pull request #947 from thewilli/wildcardsearch

wildcard tag search support

2 files changed, 167 insertions, 47 deletions

diff --git a/application/LinkFilter.php b/application/LinkFilter.php
index 95519528..99ecd1e2 100644
--- a/application/LinkFilter.php
+++ b/application/LinkFilter.php
@@ -250,6 +250,51 @@ class LinkFilter
     }
 
     /**
+     * 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 $regex;
+    }
+
+    /**
      * Returns the list of links associated with a given list of tags
      *
      * You can specify one or more tags, separated by space or a comma, e.g.
@@ -263,20 +308,32 @@ class LinkFilter
      */
     public function filterTags($tags, $casesensitive = false, $visibility = 'all')
     {
-        // Implode if array for clean up.
-        $tags = is_array($tags) ? trim(implode(' ', $tags)) : $tags;
-        if (empty($tags)) {
+        // 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+)|,/', $inputTags, -1, PREG_SPLIT_NO_EMPTY);
+        }
+
+        if(!count($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;
@@ -284,25 +341,27 @@ class LinkFilter
                     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;
     }
@@ -364,28 +423,6 @@ class LinkFilter
     }
 
     /**
-     * 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.
diff --git a/tests/api/controllers/GetLinksTest.php b/tests/api/controllers/GetLinksTest.php
index 4cb70224..d22ed3bf 100644
--- a/tests/api/controllers/GetLinksTest.php
+++ b/tests/api/controllers/GetLinksTest.php
@@ -367,6 +367,89 @@ class GetLinksTest extends \PHPUnit_Framework_TestCase
         $this->assertEquals(1, count($data));
         $this->assertEquals(41, $data[0]['id']);
         $this->assertEquals(self::NB_FIELDS_LINK, count($data[0]));
+
+        // wildcard: placeholder at the start
+        $env = Environment::mock([
+            'REQUEST_METHOD' => 'GET',
+            'QUERY_STRING' => 'searchtags=*Tuff',
+        ]);
+        $request = Request::createFromEnvironment($env);
+        $response = $this->controller->getLinks($request, new Response());
+        $this->assertEquals(200, $response->getStatusCode());
+        $data = json_decode((string) $response->getBody(), true);
+        $this->assertEquals(2, count($data));
+        $this->assertEquals(41, $data[0]['id']);
+
+        // wildcard: placeholder at the end
+        $env = Environment::mock([
+            'REQUEST_METHOD' => 'GET',
+            'QUERY_STRING' => 'searchtags=c*',
+        ]);
+        $request = Request::createFromEnvironment($env);
+        $response = $this->controller->getLinks($request, new Response());
+        $this->assertEquals(200, $response->getStatusCode());
+        $data = json_decode((string) $response->getBody(), true);
+        $this->assertEquals(4, count($data));
+        $this->assertEquals(6, $data[0]['id']);
+
+        // wildcard: placeholder at the middle
+        $env = Environment::mock([
+            'REQUEST_METHOD' => 'GET',
+            'QUERY_STRING' => 'searchtags=w*b',
+        ]);
+        $request = Request::createFromEnvironment($env);
+        $response = $this->controller->getLinks($request, new Response());
+        $this->assertEquals(200, $response->getStatusCode());
+        $data = json_decode((string) $response->getBody(), true);
+        $this->assertEquals(4, count($data));
+        $this->assertEquals(6, $data[0]['id']);
+
+        // wildcard: match all
+        $env = Environment::mock([
+            'REQUEST_METHOD' => 'GET',
+            'QUERY_STRING' => 'searchtags=*',
+        ]);
+        $request = Request::createFromEnvironment($env);
+        $response = $this->controller->getLinks($request, new Response());
+        $this->assertEquals(200, $response->getStatusCode());
+        $data = json_decode((string) $response->getBody(), true);
+        $this->assertEquals(9, count($data));
+        $this->assertEquals(41, $data[0]['id']);
+
+        // wildcard: optional ('*' does not need to expand)
+        $env = Environment::mock([
+            'REQUEST_METHOD' => 'GET',
+            'QUERY_STRING' => 'searchtags=*stuff*',
+        ]);
+        $request = Request::createFromEnvironment($env);
+        $response = $this->controller->getLinks($request, new Response());
+        $this->assertEquals(200, $response->getStatusCode());
+        $data = json_decode((string) $response->getBody(), true);
+        $this->assertEquals(2, count($data));
+        $this->assertEquals(41, $data[0]['id']);
+
+        // wildcard: exclusions
+        $env = Environment::mock([
+            'REQUEST_METHOD' => 'GET',
+            'QUERY_STRING' => 'searchtags=*a*+-*e*',
+        ]);
+        $request = Request::createFromEnvironment($env);
+        $response = $this->controller->getLinks($request, new Response());
+        $this->assertEquals(200, $response->getStatusCode());
+        $data = json_decode((string) $response->getBody(), true);
+        $this->assertEquals(1, count($data));
+        $this->assertEquals(41, $data[0]['id']); // finds '#hashtag' in descr.
+
+        // wildcard: exclude all
+        $env = Environment::mock([
+            'REQUEST_METHOD' => 'GET',
+            'QUERY_STRING' => 'searchtags=-*',
+        ]);
+        $request = Request::createFromEnvironment($env);
+        $response = $this->controller->getLinks($request, new Response());
+        $this->assertEquals(200, $response->getStatusCode());
+        $data = json_decode((string) $response->getBody(), true);
+        $this->assertEquals(0, count($data));
     }
 
     /**