Hashtag system

  * Hashtag are auto-linked with a filter search
  * Supports unicode
  * Compatible with markdown (excluded in code blocks)

4 files changed, 108 insertions, 57 deletions

diff --git a/application/LinkDB.php b/application/LinkDB.php
index b1072e07..929a6b0f 100644
--- a/application/LinkDB.php
+++ b/application/LinkDB.php
@@ -409,7 +409,7 @@ You use the community supported version of the original Shaarli project, by Seba
         $searchterm = !empty($filterRequest['searchterm']) ? escape($filterRequest['searchterm']) : '';
 
         // Search tags + fullsearch.
-        if (empty($type) && ! empty($searchtags) && ! empty($searchterm)) {
+        if (! empty($searchtags) && ! empty($searchterm)) {
             $type = LinkFilter::$FILTER_TAG | LinkFilter::$FILTER_TEXT;
             $request = array($searchtags, $searchterm);
         }
diff --git a/application/LinkFilter.php b/application/LinkFilter.php
index e693b284..d4fe28df 100644
--- a/application/LinkFilter.php
+++ b/application/LinkFilter.php
@@ -28,6 +28,11 @@ 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.
      */
     private $links;
@@ -263,8 +268,10 @@ class LinkFilter
             for ($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;
                 }
@@ -307,6 +314,28 @@ 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', $description) > 0) {
+            return true;
+        }
+
+        return false;
+    }
+
+    /**
      * Convert a list of tags (str) to an array. Also
      * - handle case sensitivity.
      * - accepts spaces commas as separator.
diff --git a/application/LinkUtils.php b/application/LinkUtils.php
index da04ca97..eeca631f 100644
--- a/application/LinkUtils.php
+++ b/application/LinkUtils.php
@@ -91,5 +91,80 @@ function count_private($links)
     foreach ($links as $link) {
         $cpt = $link['private'] == true ? $cpt + 1 : $cpt;
     }
+
     return $cpt;
 }
+
+/**
+ * In a string, converts URLs to clickable links.
+ *
+ * @param string $text       input string.
+ * @param string $redirector if a redirector is set, use it to gerenate links.
+ *
+ * @return string returns $text with all links converted to HTML links.
+ *
+ * @see Function inspired from http://www.php.net/manual/en/function.preg-replace.php#85722
+ */
+function text2clickable($text, $redirector = '')
+{
+    $regex = '!(((?:https?|ftp|file)://|apt:|magnet:)\S+[[:alnum:]]/?)!si';
+
+    if (empty($redirector)) {
+        return preg_replace($regex, '<a href="$1">$1</a>', $text);
+    }
+    // Redirector is set, urlencode the final URL.
+    return preg_replace_callback(
+        $regex,
+        function ($matches) use ($redirector) {
+            return '<a href="' . $redirector . urlencode($matches[1]) .'">'. $matches[1] .'</a>';
+        },
+        $text
+    );
+}
+
+/**
+ * Auto-link hashtags.
+ *
+ * @param string $description Given description.
+ * @param string $indexUrl    Root URL.
+ *
+ * @return string Description with auto-linked hashtags.
+ */
+function hashtag_autolink($description, $indexUrl = '')
+{
+    /*
+     * To support unicode: http://stackoverflow.com/a/35498078/1484919
+     * \p{Pc} - to match underscore
+     * \p{N} - numeric character in any script
+     * \p{L} - letter from any language
+     * \p{Mn} - any non marking space (accents, umlauts, etc)
+     */
+    $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}]+)/mui';
+    $replacement = '$1<a href="'. $indexUrl .'?addtag=$2" title="Hashtag $2">#$2</a>';
+    return preg_replace($regex, $replacement, $description);
+}
+
+/**
+ * This function inserts &nbsp; where relevant so that multiple spaces are properly displayed in HTML
+ * even in the absence of <pre>  (This is used in description to keep text formatting).
+ *
+ * @param string $text input text.
+ *
+ * @return string formatted text.
+ */
+function space2nbsp($text)
+{
+    return preg_replace('/(^| ) /m', '$1&nbsp;', $text);
+}
+
+/**
+ * Format Shaarli's description
+ *
+ * @param string $description shaare's description.
+ * @param string $redirector  if a redirector is set, use it to gerenate links.
+ *
+ * @return string formatted description.
+ */
+function format_description($description, $redirector = '', $indexUrl = '') {
+    return nl2br(space2nbsp(hashtag_autolink(text2clickable($description, $redirector), $indexUrl)));
+}
diff --git a/application/Utils.php b/application/Utils.php
index da521cce..7d7eaffd 100644
--- a/application/Utils.php
+++ b/application/Utils.php
@@ -198,59 +198,6 @@ function is_session_id_valid($sessionId)
 }
 
 /**
- * In a string, converts URLs to clickable links.
- *
- * @param string $text       input string.
- * @param string $redirector if a redirector is set, use it to gerenate links.
- *
- * @return string returns $text with all links converted to HTML links.
- *
- * @see Function inspired from http://www.php.net/manual/en/function.preg-replace.php#85722
- */
-function text2clickable($text, $redirector)
-{
-    $regex = '!(((?:https?|ftp|file)://|apt:|magnet:)\S+[[:alnum:]]/?)!si';
-
-    if (empty($redirector)) {
-        return preg_replace($regex, '<a href="$1">$1</a>', $text);
-    }
-    // Redirector is set, urlencode the final URL.
-    return preg_replace_callback(
-        $regex,
-        function ($matches) use ($redirector) {
-            return '<a href="' . $redirector . urlencode($matches[1]) .'">'. $matches[1] .'</a>';
-        },
-        $text
-    );
-}
-
-/**
- * This function inserts &nbsp; where relevant so that multiple spaces are properly displayed in HTML
- * even in the absence of <pre>  (This is used in description to keep text formatting).
- *
- * @param string $text input text.
- *
- * @return string formatted text.
- */
-function space2nbsp($text)
-{
-    return preg_replace('/(^| ) /m', '$1&nbsp;', $text);
-}
-
-/**
- * Format Shaarli's description
- * TODO: Move me to ApplicationUtils when it's ready.
- *
- * @param string $description shaare's description.
- * @param string $redirector  if a redirector is set, use it to gerenate links.
- *
- * @return string formatted description.
- */
-function format_description($description, $redirector = false) {
-    return nl2br(space2nbsp(text2clickable($description, $redirector)));
-}
-
-/**
  * Sniff browser language to set the locale automatically.
  * Note that is may not work on your server if the corresponding locale is not installed.
  *
@@ -273,4 +220,4 @@ function autoLocale($headerLocale)
         }
     }
     setlocale(LC_ALL, $attempts);
-}
\ No newline at end of file
+}