From: Andrew Dolgov <noreply@fakecake.org>
Date: Tue, 7 Jul 2015 07:15:08 +0000 (+0300)
Subject: af_readability: add a workaround for meta charset html pages
X-Git-Tag: 16.3~244
X-Git-Url: https://git.wh0rd.org/?a=commitdiff_plain;h=b7d1306b197bc7ae60df706f81d1f5665ee04bed;p=tt-rss.git

af_readability: add a workaround for meta charset html pages
---

diff --git a/plugins/af_readability/init.php b/plugins/af_readability/init.php
index 6cdd28fa..15b88d32 100644
--- a/plugins/af_readability/init.php
+++ b/plugins/af_readability/init.php
@@ -101,6 +101,19 @@ class Af_Readability extends Plugin {
 		$tmp = fetch_file_contents($article["link"]);
 
 		if ($tmp) {
+			$tmpdoc = new DOMDocument("1.0", "UTF-8");
+			$tmpdoc->loadHTML($tmp);
+
+			if ($tmpdoc->encoding != 'UTF-8') {
+				$tmpxpath = new DOMXPath($tmpdoc);
+
+				foreach ($tmpxpath->query("//meta") as $elem) {
+					$elem->parentNode->removeChild($elem);
+				}
+
+				$tmp = $tmpdoc->saveHTML();
+			}
+
 			$r = new Readability($tmp, $article["link"]);
 
 			if ($r->init()) {