NEW: Adding link tracking css class and using built in reporting

README.md

@@ -38,12 +38,6 @@ Add the following code to the mysite config to run the job every 24 hours (86400
 `Config::inst()->update('CheckExternalLinks', 'QueuedJob', 86400);`
 
 
-## Disable the Broken external link menu
-
-To disable the *Broken Ext. Links* menu add the following code to mysite/_config.php
-
-`CMSMenu::remove_menu_item('BrokenExternalLinksAdmin');`
-
 ## TODO ##
 
 Fix setting the class attribute of broken links to ss-broken

code/jobs/CheckExternalLinksJob.php

@@ -9,7 +9,7 @@ class CheckExternalLinksJob extends AbstractQueuedJob {
 	public static $regenerate_time = 43200;
 
 	public function __construct() {
-		$this->pagesToProcess = DB::query('SELECT "ID" FROM "SiteTree_Live" WHERE "ShowInSearch"=1')->column();
+		$this->pagesToProcess = SiteTree::get();
 		$this->currentStep = 0;
 		$this->totalSteps = count($this->pagesToProcess);
 	}
@@ -49,7 +49,7 @@ class CheckExternalLinksJob extends AbstractQueuedJob {
 		$restart = $this->currentStep == 0;
 
 		if ($restart) {
-			$this->pagesToProcess = DB::query('SELECT "ID" FROM SiteTree_Live WHERE ShowInSearch=1')->column();
+			$this->pagesToProcess = SiteTree::get();
 		}
 	}
 
@@ -63,6 +63,10 @@ class CheckExternalLinksJob extends AbstractQueuedJob {
 	public function process() {
 		$task = new CheckExternalLinks();
 		$task->run();
+		$data = $this->getJobData();
+		$completedPages = $task->getCompletedPages();
+		$totalPages = $task->getTotalPages();
+		$this->addMessage("$completedPages/$totalPages pages completed");
 		$this->completeJob();
 	}
 

code/model/BrokenExternalLinks.php

@@ -1,44 +0,0 @@
-<?php
-
-class BrokenExternalLinks extends DataObject {
-
-	private static $db = array(
-		'Link' => 'Varchar(2083)', // 2083 is the maximum length of a URL in Internet Explorer.
-		'HTTPCode' =>'Int'
-	);
-
-	private static $has_one = array(
-		'Page' => 'Page'
-	);
-
-	public static $summary_fields = array(
-		'Page.Title' => 'Page',
-		'HTTPCode' => 'HTTP Code',
-		'Created' => 'Created'
-	);
-
-	public static $searchable_fields = array(
-		'HTTPCode' => array('title' => 'HTTP Code')
-	);
-
-	function canEdit($member = false) {
-		return false;
-	}
-
-}
-
-class BrokenExternalLinksAdmin extends ModelAdmin {
-
-	public static $url_segment = 'broken-external-links-admin';
-
-	public static $managed_models = array(
-		'BrokenExternalLinks'
-	);
-
-	public static $menu_title = 'Broken Ext. links';
-
-	public function init() {
-		parent::init();
-	}
-
-}

code/tasks/CheckExternalLinks.php

@@ -7,26 +7,46 @@ class CheckExternalLinks extends BuildTask {
 
 	protected $enabled = true;
 
+	private $completedPages;
+	private $totalPages;
+
+	public function getCompletedPages() {
+		return $this->completedPages;
+	}
+
+	public function getTotalPages() {
+		return $this->totalPages;
+	}
+
 	function run($request) {
-		// clear broken external link table
+		$pages = Versioned::get_by_stage('SiteTree', 'Live');
-		$table = 'BrokenExternalLinks';
-		if(method_exists(DB::getConn(), 'clearTable')) DB::getConn()->clearTable($table);
-		else DB::query("TRUNCATE \"$table\"");
-		$pages = SiteTree::get();
 		foreach ($pages as $page) {
+			++$this->totalPages;
+
 			$htmlValue = Injector::inst()->create('HTMLValue', $page->Content);
+			if (!$htmlValue->isValid()) {
+				continue;
+			}
 
 			// Populate link tracking for internal links & links to asset files.
 			if($links = $htmlValue->getElementsByTagName('a')) foreach($links as $link) {
+				$class = $link->getAttribute('class');
+				$pos = stripos($class, 'ss-broken');
+				if ($pos !== false && $page->HasBrokenLink == 1) continue;
+
 				$href = Director::makeRelative($link->getAttribute('href'));
 				if ($href == 'admin/') continue;
 
-				// ignore SiteTree and assets links as they will be caught by SiteTreeLinkTracking
+				// ignore SiteTree, anchor and assets links as they will be caught
-				if(preg_match('/\[sitetree_link,id=([0-9]+)\]/i', $href, $matches)) {
+				// by SiteTreeLinkTracking
+				if(preg_match('/\[(file_link|sitetree_link),id=([0-9]+)\]/i', $href, $matches)) {
+					continue;
+				} else if (isset($href[0]) && $href[0] == '#') {
 					continue;
 				} else if(substr($href, 0, strlen(ASSETS_DIR) + 1) == ASSETS_DIR.'/') {
 					continue;
 				}
+
 				if($href && function_exists('curl_init')) {
 					$handle = curl_init($href);
 					curl_setopt($handle, CURLOPT_RETURNTRANSFER, TRUE);
@@ -36,31 +56,30 @@ class CheckExternalLinks extends BuildTask {
 					if (($httpCode < 200 || $httpCode > 302)
 						|| ($href == '' || $href[0] == '/'))
 					{
-						$brokenLink = new BrokenExternalLinks();			
+						// set the broken link class
-						$brokenLink->PageID = $page->ID;
+						$class = ($class && stripos($class, 'ss-broken')) ?
-						$brokenLink->Link = $href;
+							$class . ' ss-broken' : 'ss-broken';
-						$brokenLink->HTTPCode = $httpCode;
+						$link->setAttribute('class', ($class ? $class : 'ss-broken'));
-						$brokenLink->write();
+						$htmlValue->__call('saveHTML', array());
 
-						// TODO set the broken link class
+						$page->Content = $htmlValue->getContent();
-						/*
+						$page->write();
-						$class = $link->getAttribute('class');
-						$class = ($class) ? $class . 'ss-broken' : 'ss-broken';
-						$link->setAttribute('class', ($class ? "$class ss-broken" : 'ss-broken'));
-						*/
 
-						// use raw sql query to set broken link as calling the dataobject write
+						if (!$page->HasBrokenLink) {
-						// method will reset the links if no broken internal links are found
+							// bypass the ORM as syncLinkTracking does not allow you
-						$query = "UPDATE \"SiteTree\" SET \"HasBrokenLink\" = 1 ";	
+							// to update HasBrokenLink to true
-						$query .= "WHERE \"ID\" = " . (int)$page->ID;	
+							$query = "UPDATE \"SiteTree_Live\" SET \"HasBrokenLink\" = 1 ";
-						$result = DB::query($query);
+							$query .= "WHERE \"ID\" = " . (int)$page->ID;   
-						if (!$result) {
+							$result = DB::query($query);
-							// error updating hasBrokenLink
+							if (!$result) {
+								$this->debugMessage('Error updating HasBrokenLink');
+							}
 						}
 
 					}
 				}
 			}
+			++$this->completedPages;
 		}
 
 		// run this again if queued jobs exists and is a valid int