Added horrible way to restore score. Should be gone in next versions (or whe I learn to code properly)

2 files changed, 32 insertions, 0 deletions

diff --git a/src/HTMLParser.php b/src/HTMLParser.php
index bd7774b..49c220d 100644
--- a/src/HTMLParser.php
+++ b/src/HTMLParser.php
@@ -406,6 +406,18 @@ class HTMLParser
         }
 
         /*
+         * TODO This is an horrible hack because I don't know how to properly pass by reference.
+         * When candidates are added to the $candidates array, they lose the reference to the original object
+         * and on each loop, the object inside $candidates doesn't get updated. This function restores the score
+         * by getting it of the data-readability tag. This should be fixed using proper references and good coding
+         * practices (which I lack)
+         */
+
+        foreach ($candidates as $candidate) {
+            $candidate->reloadScore();
+        }
+
+        /*
          * After we've calculated scores, loop through all of the possible
          * candidate nodes we found and find the one with the highest score.
          */
diff --git a/src/Readability.php b/src/Readability.php
index 6bc9603..a6390c9 100644
--- a/src/Readability.php
+++ b/src/Readability.php
@@ -433,4 +433,24 @@ class Readability extends Element implements ReadabilityInterface
     {
         return $this->initialized;
     }
+
+    /**
+     * Reloads the score stores in the data-readability tag.
+     *
+     * @return int|bool
+     */
+    public function reloadScore()
+    {
+        if (method_exists($this->node, 'getAttribute')) {
+            if ($this->node->hasAttribute('data-readability')) {
+                $this->initialized = true;
+                $score = $this->node->getAttribute('data-readability');
+                $this->setContentScore($score);
+
+                return $score;
+            }
+        }
+
+        return false;
+    }
 }