parser: properly support tag subtrees instead of text content for article content

3 files changed, 19 insertions, 6 deletions

diff --git a/classes/feeditem/atom.php b/classes/feeditem/atom.php
index dfac7149f..e132789ba 100644
--- a/classes/feeditem/atom.php
+++ b/classes/feeditem/atom.php
@@ -75,7 +75,7 @@ class FeedItem_Atom extends FeedItem_Common {
 				}
 			}
 
-			return $content->nodeValue;
+			return $this->subtree_or_text($content);
 		}
 	}
 
@@ -95,7 +95,7 @@ class FeedItem_Atom extends FeedItem_Common {
 				}
 			}
 
-			return $content->nodeValue;
+			return $this->subtree_or_text($content);
 		}
 
 	}
diff --git a/classes/feeditem/common.php b/classes/feeditem/common.php
index 80bebf8fb..070692d7f 100644
--- a/classes/feeditem/common.php
+++ b/classes/feeditem/common.php
@@ -70,6 +70,17 @@ abstract class FeedItem_Common extends FeedItem {
 		}
 	}
 
+	function count_children($node) {
+		return $node->getElementsByTagName("*")->length;
+	}
+
+	function subtree_or_text($node) {
+		if ($this->count_children($node) == 0) {
+			return $node->nodeValue;
+		} else {
+			return $node->c14n();
+		}
+	}
 
 }
 ?>
diff --git a/classes/feeditem/rss.php b/classes/feeditem/rss.php
index 27a364b81..080e4083a 100644
--- a/classes/feeditem/rss.php
+++ b/classes/feeditem/rss.php
@@ -71,17 +71,19 @@ class FeedItem_RSS extends FeedItem_Common {
 		$contentB = $this->elem->getElementsByTagName("description")->item(0);
 
 		if ($contentA && !$contentB) {
-			return $contentA->nodeValue;
+			return $this->subtree_or_text($contentA);
 		}
 
 
 		if ($contentB && !$contentA) {
-			return $contentB->nodeValue;
+			return $this->subtree_or_text($contentB);
 		}
 
 		if ($contentA && $contentB) {
-			return mb_strlen($contentA->nodeValue) > mb_strlen($contentB->nodeValue) ?
-				$contentA->nodeValue : $contentB->nodeValue;
+			$resultA = $this->subtree_or_text($contentA);
+			$resultB = $this->subtree_or_text($contentB);
+
+			return mb_strlen($resultA) > mb_strlen($resultB) ? $resultA : $resultB;
 		}
 	}