NEW: Start of external broken links module

README.md

@@ -0,0 +1,34 @@
+# External links
+
+## Introduction
+
+The external links module is a task and ModelAdmin to track and to report on broken external links.
+
+## Maintainer Contact
+
+	* Kirk Mayo kirk (at) silverstripe (dot) com
+
+## Requirements
+
+	* SilverStripe 3.0 +
+
+## Features
+
+* Add external links to broken links reports
+* Add a model admin for external broken links
+* Add a task to track external broken links
+
+## Installation
+
+ 1. Download the module form GitHub (Composer support to be added)
+ 2. Extract the file (if you are on windows try 7-zip for extracting tar.gz files
+ 3. Make sure the folder after being extracted is named 'externallinks'
+ 4. Place this directory in your sites root directory. This is the one with framework and cms in it.
+ 5. Run in your browser - `/dev/build` to rebuild the database.
+ 6. You should see a new menu called *Broken Ext. Links*
+
+## 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');`

_config.php

@@ -0,0 +1 @@
+<?php

code/BrokenExternalLinks.php

@@ -0,0 +1,44 @@
+<?php
+
+class BrokenExternalLinks extends DataObject {
+
+	private static $db = array(
+		'Link' => 'Varchar',
+		'HTTPCode' =>'Int'
+	);
+
+	private static $has_one = array(
+		'Page' => 'Page'
+	);
+
+	public static $summary_fields = array(
+		'Page.Title' => 'Page',
+		'HTTPCode' => 'HTTP Code',
+		'Created' => 'Created'
+	);
+
+	public static $searchable_fields = array(
+		'HTTPCode'
+	);
+
+	function canEdit($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();
+	}
+
+}

tasks/CheckExternalLinks.php

@@ -0,0 +1,66 @@
+<?php
+
+class CheckExternalLinks extends BuildTask {
+	protected $title = 'Checking broken External links in the SiteTree';
+
+	protected $description = 'A task that records external broken links in the SiteTree';
+
+	protected $enabled = true;
+
+	function run($request) {
+		// clear broken external link table
+		$table = 'BrokenExternalLinks';
+		if(method_exists(DB::getConn(), 'clearTable')) DB::getConn()->clearTable($table);
+		else DB::query("TRUNCATE \"$table\"");
+		$pages = SiteTree::get();
+		foreach ($pages as $page) {
+			$htmlValue = Injector::inst()->create('HTMLValue', $page->Content);
+
+			// Populate link tracking for internal links & links to asset files.
+			if($links = $htmlValue->getElementsByTagName('a')) foreach($links as $link) {
+				$href = Director::makeRelative($link->getAttribute('href'));
+				if ($href == 'admin/') continue;
+
+				// ignore SiteTree and assets links as they will be caught by SiteTreeLinkTracking
+				if(preg_match('/\[sitetree_link,id=([0-9]+)\]/i', $href, $matches)) {
+					return;
+				} else if(substr($href, 0, strlen(ASSETS_DIR) + 1) == ASSETS_DIR.'/') {
+					return;
+				}
+				if($href && function_exists('curl_init')) {
+					$handle = curl_init($href);
+					curl_setopt($handle, CURLOPT_RETURNTRANSFER, TRUE);
+					$response = curl_exec($handle);
+					$httpCode = curl_getinfo($handle, CURLINFO_HTTP_CODE);  
+					curl_close($handle);
+					if (($httpCode < 200 || $httpCode > 302)
+						|| ($href == '' || $href[0] == '/'))
+					{
+						$brokenLink = new BrokenExternalLinks();			
+						$brokenLink->PageID = $page->ID;
+						$brokenLink->Link = $href;
+						$brokenLink->HTTPCode = $httpCode;
+						$brokenLink->write();
+
+						// set the broken link class
+						/*
+						$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
+						// method will reset the links if no broken internal links are found
+						$query = 'UPDATE SiteTree SET HasBrokenLink = 1 ';	
+						$query .= 'WHERE ID = ' . (int)$page->ID;	
+						$result = DB::query($query);
+						if (!$result) {
+							// error updating hasBrokenLink
+						}
+
+					}
+				}
+			}
+		}
+	}
+}