From 941c6c50ade6edf0c142c156ba8d43e33ea3879b Mon Sep 17 00:00:00 2001
From: Neil Minton <neil@archive.org>
Date: Thu, 6 Aug 2020 19:23:50 -0400
Subject: [PATCH] Strip quotes from URL value.

---
 .../main/java/org/archive/modules/extractor/ExtractorHTML.java | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/modules/src/main/java/org/archive/modules/extractor/ExtractorHTML.java b/modules/src/main/java/org/archive/modules/extractor/ExtractorHTML.java
index ebcaf47a..b26eed3f 100644
--- a/modules/src/main/java/org/archive/modules/extractor/ExtractorHTML.java
+++ b/modules/src/main/java/org/archive/modules/extractor/ExtractorHTML.java
@@ -981,7 +981,8 @@ public class ExtractorHTML extends ContentExtractor implements InitializingBean
         } else if ("refresh".equalsIgnoreCase(httpEquiv) && content != null) {
             int urlIndex = content.indexOf("=") + 1;
             if(urlIndex>0) {
-                String refreshUri = content.substring(urlIndex);
+                // strip any quotes ("') characters from the URL value.
+                String refreshUri = TextUtils.replaceAll("[\"']", content.substring(urlIndex), "");
                 try {
                     int max = getExtractorParameters().getMaxOutlinks();
                     addRelativeToBase(curi, max, refreshUri,