tag-related fixes

1. move tag sanitization to feedparser common item class
2. enforce length limit on tags when parsing
3. support multiple tags passed via one dc:subject and other such elements, parse them as a comma-separated list
4. sort resulting tag list to prevent different order between feed updates
5. remove some duplicate code related to tag validation
6. allow + symbol in tags

3 files changed, 39 insertions, 8 deletions

diff --git a/classes/feeditem/atom.php b/classes/feeditem/atom.php
index a962b59f2..a03080981 100755
--- a/classes/feeditem/atom.php
+++ b/classes/feeditem/atom.php
@@ -103,20 +103,20 @@ class FeedItem_Atom extends FeedItem_Common {
 
 	function get_categories() {
 		$categories = $this->elem->getElementsByTagName("category");
-		$cats = array();
+		$cats = [];
 
 		foreach ($categories as $cat) {
 			if ($cat->hasAttribute("term"))
-				array_push($cats, trim($cat->getAttribute("term")));
+				array_push($cats, $cat->getAttribute("term"));
 		}
 
 		$categories = $this->xpath->query("dc:subject", $this->elem);
 
 		foreach ($categories as $cat) {
-			array_push($cats, clean(trim($cat->nodeValue)));
+			array_push($cats, $cat->nodeValue);
 		}
 
-		return $cats;
+		return $this->normalize_categories($cats);
 	}
 
 	function get_enclosures() {
diff --git a/classes/feeditem/common.php b/classes/feeditem/common.php
index 3193ed273..f208f4a48 100755
--- a/classes/feeditem/common.php
+++ b/classes/feeditem/common.php
@@ -162,4 +162,35 @@ abstract class FeedItem_Common extends FeedItem {
 		}
 	}
 
+	static function normalize_categories($cats) {
+
+		$tmp = [];
+
+		foreach ($cats as $rawcat) {
+			$tmp = array_merge($tmp, explode(",", $rawcat));
+		}
+
+		$tmp = array_map(function($srccat) {
+			$cat = clean(trim(mb_strtolower($srccat)));
+
+			// we don't support numeric tags
+			if (is_numeric($cat))
+				$cat = 't:' . $cat;
+
+			$cat = preg_replace('/[,\'\"]/', "", $cat);
+
+			if (DB_TYPE == "mysql") {
+				$cat = preg_replace('/[\x{10000}-\x{10FFFF}]/u', "\xEF\xBF\xBD", $cat);
+			}
+
+			if (mb_strlen($cat) > 250)
+				$cat = mb_substr($cat, 0, 250);
+
+			return $cat;
+		}, $tmp);
+
+		asort($tmp);
+
+		return array_unique($tmp);
+	}
 }
diff --git a/classes/feeditem/rss.php b/classes/feeditem/rss.php
index 916c73ec4..1f7953c51 100755
--- a/classes/feeditem/rss.php
+++ b/classes/feeditem/rss.php
@@ -97,19 +97,19 @@ class FeedItem_RSS extends FeedItem_Common {
 
 	function get_categories() {
 		$categories = $this->elem->getElementsByTagName("category");
-		$cats = array();
+		$cats = [];
 
 		foreach ($categories as $cat) {
-			array_push($cats, trim($cat->nodeValue));
+			array_push($cats, $cat->nodeValue);
 		}
 
 		$categories = $this->xpath->query("dc:subject", $this->elem);
 
 		foreach ($categories as $cat) {
-			array_push($cats, clean(trim($cat->nodeValue)));
+			array_push($cats, $cat->nodeValue);
 		}
 
-		return $cats;
+		return $this->normalize_categories($cats);
 	}
 
 	function get_enclosures() {