From d6556fbc040b23e553ad0af5976d069491c881e8 Mon Sep 17 00:00:00 2001
From: Andres Rey <andreskrey@gmail.com>
Date: Fri, 21 Oct 2016 23:36:25 +0100
Subject: Added function to search for topCandidate

---
 src/HTMLParser.php  | 49 +++++++++++++++++++++++++++++++++++++++++++++++++
 src/Readability.php |  2 +-
 2 files changed, 50 insertions(+), 1 deletion(-)

(limited to 'src')

diff --git a/src/HTMLParser.php b/src/HTMLParser.php
index 2f54114..4e74b05 100644
--- a/src/HTMLParser.php
+++ b/src/HTMLParser.php
@@ -339,5 +339,54 @@ class HTMLParser
 
         $topCandidate = isset($topCandidates[0]) ? $topCandidates[0] : null;
         $neededToCreateTopCandidate = false;
+
+        /*
+         * If we still have no top candidate, just use the body as a last resort.
+         * We also have to copy the body node so it is something we can modify.
+         */
+
+        if ($topCandidate === null || $topCandidate->tagNameEqualsTo('body')) {
+            //TODO
+        } elseif ($topCandidate) {
+            /*
+             * Because of our bonus system, parents of candidates might have scores
+             * themselves. They get half of the node. There won't be nodes with higher
+             * scores than our topCandidate, but if we see the score going *up* in the first
+             * few steps up the tree, that's a decent sign that there might be more content
+             * lurking in other places that we want to unify in. The sibling stuff
+             * below does some of that - but only if we've looked high enough up the DOM
+             * tree.
+             */
+
+            $parentOfTopCandidate = $topCandidate->getParent();
+            $lastScore = $topCandidate->getContentScore();
+
+            // The scores shouldn't get too low.
+            $scoreThreshold = $lastScore / 3;
+
+            while ($parentOfTopCandidate) {
+                /** @var Readability $parentOfTopCandidate */
+                $parentScore = $parentOfTopCandidate->getContentScore();
+                if ($parentScore < $scoreThreshold) {
+                    break;
+                }
+
+                if ($parentScore > $lastScore) {
+                    // Alright! We found a better parent to use.
+                    $topCandidate = $parentOfTopCandidate;
+                    break;
+                }
+                $lastScore = $parentOfTopCandidate->getContentScore();
+                $parentOfTopCandidate = $parentOfTopCandidate->getParent();
+            }
+        }
+
+        /*
+         * Now that we have the top candidate, look through its siblings for content
+         * that might also be related. Things like preambles, content split by ads
+         * that we removed, etc.
+         */
+
+
     }
 }
diff --git a/src/Readability.php b/src/Readability.php
index a15b664..8a40a7c 100644
--- a/src/Readability.php
+++ b/src/Readability.php
@@ -244,6 +244,6 @@ class Readability extends Element implements ReadabilityInterface
      */
     public function getTextContent()
     {
-        return $this->node->getChildrenAsString();
+        return $this->getChildrenAsString();
     }
 }
-- 
cgit v1.2.3