From e1b9ea114e17aa08805eed8f013bb74b337936a2 Mon Sep 17 00:00:00 2001
From: Pedro Amorim <pedro.boucanovaamorim@3ds.com>
Date: Mon, 9 Apr 2018 10:49:30 +0200
Subject: Fix notice non-object on $parentOfTopCandidate for tumblr.com

PHP notice on DOMElement $parentOfTopCandidate.
  Trying to get property of non-object in serc/Readability.php line 1000
  Trying to get property of non-object in serc/Readability.php line 1009

Reproduced with this url:
  https://clipartx.tumblr.com/post/172752750628/orange-swirl-burnt-orange-orange

Config:
  $config = new Configuration;
  $config->setWordThreshold(5)
    ->setSummonCthulhu(true)
    ->setFixRelativeURLs(true)
    ->setOriginalURL($url);
---
 src/Readability.php | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

(limited to 'src')

diff --git a/src/Readability.php b/src/Readability.php
index 93fc810..2623bb2 100644
--- a/src/Readability.php
+++ b/src/Readability.php
@@ -997,7 +997,7 @@ class Readability
             $MINIMUM_TOPCANDIDATES = 3;
             if (count($alternativeCandidateAncestors) >= $MINIMUM_TOPCANDIDATES) {
                 $parentOfTopCandidate = $topCandidate->parentNode;
-                while ($parentOfTopCandidate->nodeName !== 'body') {
+                while ($parentOfTopCandidate->nodeName !== 'body' && $parentOfTopCandidate->nodeType === XML_ELEMENT_NODE) {
                     $listsContainingThisAncestor = 0;
                     for ($ancestorIndex = 0; $ancestorIndex < count($alternativeCandidateAncestors) && $listsContainingThisAncestor < $MINIMUM_TOPCANDIDATES; $ancestorIndex++) {
                         $listsContainingThisAncestor += (int)in_array($parentOfTopCandidate, $alternativeCandidateAncestors[$ancestorIndex]);
@@ -1047,7 +1047,10 @@ class Readability
             // If the top candidate is the only child, use parent instead. This will help sibling
             // joining logic when adjacent content is actually located in parent's sibling node.
             $parentOfTopCandidate = $topCandidate->parentNode;
-            while ($parentOfTopCandidate->nodeName !== 'body' && count($parentOfTopCandidate->getChildren(true)) === 1) {
+            while ($parentOfTopCandidate->nodeName !== 'body'
+                   && $parentOfTopCandidate->nodeType === XML_ELEMENT_NODE
+                   && count($parentOfTopCandidate->getChildren(true)) === 1
+            ) {
                 $topCandidate = $parentOfTopCandidate;
                 $parentOfTopCandidate = $topCandidate->parentNode;
             }
-- 
cgit v1.2.3