From 0354be75d8ee28c6003d01c22e6e56862f07a998 Mon Sep 17 00:00:00 2001
From: Kirk Mayo <kirk@stripetheweb.com>
Date: Thu, 15 Dec 2016 10:34:59 +1300
Subject: [PATCH 1/2] Added options to bypass cache and follow redirected links

---
 README.md                      | 21 +++++++++++++++++++++
 code/tasks/CurlLinkChecker.php | 20 ++++++++++++++------
 2 files changed, 35 insertions(+), 6 deletions(-)

diff --git a/README.md b/README.md
index 9bae495..bee33bf 100644
--- a/README.md
+++ b/README.md
@@ -80,3 +80,24 @@ file in mysite/_config
         - 401
         - 403
         - 501
+
+## Follow 301 redirects ##
+
+You may want to follow a redirected URL a example of this would be redirecting from http to https
+can give you a false poitive as the http code of 301 will be returned which will be classed
+as a working link.
+
+To allow redirects to be followed setup the following config in your config.yml
+
+    # Follow 301 redirects
+    CurlLinkChecker:
+      FollowLocation: 1
+
+## Bypass cache ##
+
+By default the task will attempt to cache any results the cache can be bypassed with the
+following config in config.yml.
+
+    # Bypass SS_Cache
+    CurlLinkChecker:
+      BypassCache: 1
diff --git a/code/tasks/CurlLinkChecker.php b/code/tasks/CurlLinkChecker.php
index a84439d..487285b 100644
--- a/code/tasks/CurlLinkChecker.php
+++ b/code/tasks/CurlLinkChecker.php
@@ -28,22 +28,30 @@ class CurlLinkChecker implements LinkChecker {
 		// Skip non-external links
 		if(!preg_match('/^https?[^:]*:\/\//', $href)) return null;
 
-		// Check if we have a cached result
-		$cacheKey = md5($href);
-		$result = $this->getCache()->load($cacheKey);
-		if($result !== false) return $result;
+		if (!Config::inst()->get('CurlLinkChecker', 'BypassCache')) {
+			// Check if we have a cached result
+			$cacheKey = md5($href);
+			$result = $this->getCache()->load($cacheKey);
+			if($result !== false) return $result;
+		}
 
 		// No cached result so just request
 		$handle = curl_init($href);
 		curl_setopt($handle, CURLOPT_RETURNTRANSFER, TRUE);
+		// do we want to follow any redirect locations eg http to https
+		if (Config::inst()->get('CurlLinkChecker', 'FollowLocation')) {
+			curl_setopt($handle, CURLOPT_FOLLOWLOCATION, TRUE);
+		}
 		curl_setopt($handle, CURLOPT_CONNECTTIMEOUT, 5);
 		curl_setopt($handle, CURLOPT_TIMEOUT, 10);
 		curl_exec($handle);
 		$httpCode = curl_getinfo($handle, CURLINFO_HTTP_CODE);
 		curl_close($handle);
 
-		// Cache result
-		$this->getCache()->save($httpCode, $cacheKey);
+		if (!Config::inst()->get('CurlLinkChecker', 'BypassCache')) {
+			// Cache result
+			$this->getCache()->save($httpCode, $cacheKey);
+		}
 		return $httpCode;
 	}
 }

From 9e9946ef647db603c8be0f5bd69e37c35ac1680a Mon Sep 17 00:00:00 2001
From: Kirk Mayo <kirk.mayo@tec.govt.nz>
Date: Sat, 3 Feb 2018 13:55:22 +1300
Subject: [PATCH 2/2] Added config annotations to CurlLinkChecker

---
 code/tasks/CurlLinkChecker.php | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/code/tasks/CurlLinkChecker.php b/code/tasks/CurlLinkChecker.php
index 487285b..ac05542 100644
--- a/code/tasks/CurlLinkChecker.php
+++ b/code/tasks/CurlLinkChecker.php
@@ -5,6 +5,24 @@
  */
 class CurlLinkChecker implements LinkChecker {
 
+	/**
+	 * If we want to follow redirects a 301 http code for example
+	 * Set via YAML file
+	 *
+	 * @config
+	 * @var boolean
+	 */
+	private static $FollowLocation = false;
+
+	/**
+	 * If we want to bypass the cache
+	 * Set via YAML file
+	 *
+	 * @config
+	 * @var boolean
+	 */
+	private static $BypassCache = false;
+
 	/**
 	 * Return cache
 	 *