From 871e40ac8aef298449578b16aafe65259dedcbd8 Mon Sep 17 00:00:00 2001
From: Eric Heikes <eheikes@gmail.com>
Date: Wed, 13 Sep 2017 23:16:31 -0500
Subject: [PATCH] Clean up the HTML before searching for the <title>.

Strips <script> and <style> tags from the page.
Fixes sites like Kotaku.
---
 toreadapi.php | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/toreadapi.php b/toreadapi.php
index 2063869..d8f2f5e 100644
--- a/toreadapi.php
+++ b/toreadapi.php
@@ -257,7 +257,9 @@ function postEntry() {
         }
 
         // Save the <title>.
-        if (preg_match("#<title[^>]*>(.*)</title>#iU", $result, $matches)
+        $cleanedHtml = preg_replace('#<script.*</script>#isU', '', $result);
+        $cleanedHtml = preg_replace('#<style.*</style>#isU', '', $cleanedHtml);
+        if (preg_match("#<title[^>]*>(.*)</title>#iU", $cleanedHtml, $matches)
         and $matches[1] != "")
         {
           $title = mb_substr($matches[1], 0, $maxTitleLength, 'UTF-8');