silverstripe / silverstripe-externallinks

code/model/BrokenExternalPageTrackStatus.php

Doc Comments +3 added lines

@@ -120,6 +120,9 @@
         return $status;
     }
 
+    /**
+     * @param string $message
+     */
     public function updateJobInfo($message)
     {
         $this->JobInfo = $message;

Indentation +109 added lines, -109 removed lines

@@ -10,130 +10,130 @@
  */
 class BrokenExternalPageTrackStatus extends DataObject
 {
-    private static $db = array(
-        'Status' => 'Enum("Completed, Running", "Running")',
-        'JobInfo' => 'Varchar(255)'
-    );
+	private static $db = array(
+		'Status' => 'Enum("Completed, Running", "Running")',
+		'JobInfo' => 'Varchar(255)'
+	);
 
-    private static $has_many = array(
-        'TrackedPages' => 'BrokenExternalPageTrack',
-        'BrokenLinks' => 'BrokenExternalLink'
-    );
+	private static $has_many = array(
+		'TrackedPages' => 'BrokenExternalPageTrack',
+		'BrokenLinks' => 'BrokenExternalLink'
+	);
 
-    /**
-     * Get the latest track status
-     *
-     * @return self
-     */
-    public static function get_latest()
-    {
-        return self::get()
-            ->sort('ID', 'DESC')
-            ->first();
-    }
+	/**
+	 * Get the latest track status
+	 *
+	 * @return self
+	 */
+	public static function get_latest()
+	{
+		return self::get()
+			->sort('ID', 'DESC')
+			->first();
+	}
 
-    /**
-     * Gets the list of Pages yet to be checked
-     *
-     * @return DataList
-     */
-    public function getIncompletePageList()
-    {
-        $pageIDs = $this
-            ->getIncompleteTracks()
-            ->column('PageID');
-        if ($pageIDs) {
-            return Versioned::get_by_stage('SiteTree', 'Stage')
-            ->byIDs($pageIDs);
-        }
-    }
+	/**
+	 * Gets the list of Pages yet to be checked
+	 *
+	 * @return DataList
+	 */
+	public function getIncompletePageList()
+	{
+		$pageIDs = $this
+			->getIncompleteTracks()
+			->column('PageID');
+		if ($pageIDs) {
+			return Versioned::get_by_stage('SiteTree', 'Stage')
+			->byIDs($pageIDs);
+		}
+	}
 
-    /**
-     * Get the list of incomplete BrokenExternalPageTrack
-     *
-     * @return DataList
-     */
-    public function getIncompleteTracks()
-    {
-        return $this
-            ->TrackedPages()
-            ->filter('Processed', 0);
-    }
+	/**
+	 * Get the list of incomplete BrokenExternalPageTrack
+	 *
+	 * @return DataList
+	 */
+	public function getIncompleteTracks()
+	{
+		return $this
+			->TrackedPages()
+			->filter('Processed', 0);
+	}
 
-    /**
-     * Get total pages count
-     */
-    public function getTotalPages()
-    {
-        return $this->TrackedPages()->count();
-    }
+	/**
+	 * Get total pages count
+	 */
+	public function getTotalPages()
+	{
+		return $this->TrackedPages()->count();
+	}
 
-    /**
-     * Get completed pages count
-     */
-    public function getCompletedPages()
-    {
-        return $this
-            ->TrackedPages()
-            ->filter('Processed', 1)
-            ->count();
-    }
+	/**
+	 * Get completed pages count
+	 */
+	public function getCompletedPages()
+	{
+		return $this
+			->TrackedPages()
+			->filter('Processed', 1)
+			->count();
+	}
 
-    /**
-     * Returns the latest run, or otherwise creates a new one
-     *
-     * @return self
-     */
-    public static function get_or_create()
-    {
-        // Check the current status
-        $status = self::get_latest();
-        if ($status && $status->Status == 'Running') {
-            $status->updateStatus();
-            return $status;
-        }
+	/**
+	 * Returns the latest run, or otherwise creates a new one
+	 *
+	 * @return self
+	 */
+	public static function get_or_create()
+	{
+		// Check the current status
+		$status = self::get_latest();
+		if ($status && $status->Status == 'Running') {
+			$status->updateStatus();
+			return $status;
+		}
 
-        return self::create_status();
-    }
+		return self::create_status();
+	}
 
-    /*
+	/*
      * Create and prepare a new status
      *
      * @return self
      */
-    public static function create_status()
-    {
-        // If the script is to be started create a new status
-        $status = self::create();
-        $status->updateJobInfo('Creating new tracking object');
+	public static function create_status()
+	{
+		// If the script is to be started create a new status
+		$status = self::create();
+		$status->updateJobInfo('Creating new tracking object');
 
-        // Setup all pages to test
-        $pageIDs = Versioned::get_by_stage('SiteTree', 'Stage')
-            ->column('ID');
-        foreach ($pageIDs as $pageID) {
-            $trackPage = BrokenExternalPageTrack::create();
-            $trackPage->PageID = $pageID;
-            $trackPage->StatusID = $status->ID;
-            $trackPage->write();
-        }
+		// Setup all pages to test
+		$pageIDs = Versioned::get_by_stage('SiteTree', 'Stage')
+			->column('ID');
+		foreach ($pageIDs as $pageID) {
+			$trackPage = BrokenExternalPageTrack::create();
+			$trackPage->PageID = $pageID;
+			$trackPage->StatusID = $status->ID;
+			$trackPage->write();
+		}
 
-        return $status;
-    }
+		return $status;
+	}
 
-    public function updateJobInfo($message)
-    {
-        $this->JobInfo = $message;
-        $this->write();
-    }
+	public function updateJobInfo($message)
+	{
+		$this->JobInfo = $message;
+		$this->write();
+	}
 
-    /**
-     * Self check status
-     */
-    public function updateStatus()
-    {
-        if ($this->CompletedPages == $this->TotalPages) {
-            $this->Status = 'Completed';
-            $this->updateJobInfo('Setting to completed');
-        }
-    }
+	/**
+	 * Self check status
+	 */
+	public function updateStatus()
+	{
+		if ($this->CompletedPages == $this->TotalPages) {
+			$this->Status = 'Completed';
+			$this->updateJobInfo('Setting to completed');
+		}
+	}
 }

code/controllers/CMSExternalLinks.php

Indentation +44 added lines, -44 removed lines

@@ -2,59 +2,59 @@
 
 class CMSExternalLinks_Controller extends Controller
 {
-    private static $allowed_actions = array('getJobStatus', 'start');
+	private static $allowed_actions = array('getJobStatus', 'start');
 
-    /*
+	/*
      * Respond to Ajax requests for info on a running job
      *
      * @return string JSON string detailing status of the job
      */
-    public function getJobStatus()
-    {
-        // Set headers
-        HTTP::set_cache_age(0);
-        HTTP::add_cache_headers($this->response);
-        $this->response
-            ->addHeader('Content-Type', 'application/json')
-            ->addHeader('Content-Encoding', 'UTF-8')
-            ->addHeader('X-Content-Type-Options', 'nosniff');
+	public function getJobStatus()
+	{
+		// Set headers
+		HTTP::set_cache_age(0);
+		HTTP::add_cache_headers($this->response);
+		$this->response
+			->addHeader('Content-Type', 'application/json')
+			->addHeader('Content-Encoding', 'UTF-8')
+			->addHeader('X-Content-Type-Options', 'nosniff');
 
-        // Format status
-        $track = BrokenExternalPageTrackStatus::get_latest();
-        if ($track) {
-            return json_encode(array(
-            'TrackID' => $track->ID,
-            'Status' => $track->Status,
-            'Completed' => $track->getCompletedPages(),
-            'Total' => $track->getTotalPages()
-        ));
-        }
-    }
+		// Format status
+		$track = BrokenExternalPageTrackStatus::get_latest();
+		if ($track) {
+			return json_encode(array(
+			'TrackID' => $track->ID,
+			'Status' => $track->Status,
+			'Completed' => $track->getCompletedPages(),
+			'Total' => $track->getTotalPages()
+		));
+		}
+	}
 
 
-    /*
+	/*
      * Starts a broken external link check
      */
-    public function start()
-    {
-        // return if the a job is already running
-        $status = BrokenExternalPageTrackStatus::get_latest();
-        if ($status && $status->Status == 'Running') {
-            return;
-        }
+	public function start()
+	{
+		// return if the a job is already running
+		$status = BrokenExternalPageTrackStatus::get_latest();
+		if ($status && $status->Status == 'Running') {
+			return;
+		}
 
-        // Create a new job
-        if (class_exists('QueuedJobService')) {
-            // Force the creation of a new run
-            BrokenExternalPageTrackStatus::create_status();
-            $checkLinks = new CheckExternalLinksJob();
-            singleton('QueuedJobService')->queueJob($checkLinks);
-        } else {
-            //TODO this hangs as it waits for the connection to be released
-            // should return back and continue processing
-            // http://us3.php.net/manual/en/features.connection-handling.php
-            $task = CheckExternalLinksTask::create();
-            $task->runLinksCheck();
-        }
-    }
+		// Create a new job
+		if (class_exists('QueuedJobService')) {
+			// Force the creation of a new run
+			BrokenExternalPageTrackStatus::create_status();
+			$checkLinks = new CheckExternalLinksJob();
+			singleton('QueuedJobService')->queueJob($checkLinks);
+		} else {
+			//TODO this hangs as it waits for the connection to be released
+			// should return back and continue processing
+			// http://us3.php.net/manual/en/features.connection-handling.php
+			$task = CheckExternalLinksTask::create();
+			$task->runLinksCheck();
+		}
+	}
 }

code/jobs/CheckExternalLinksJob.php

Indentation +24 added lines, -24 removed lines

@@ -1,7 +1,7 @@ 
 <?php
 
 if (!class_exists('AbstractQueuedJob')) {
-    return;
+	return;
 }
 
 /**
@@ -10,30 +10,30 @@ 
  */
 class CheckExternalLinksJob extends AbstractQueuedJob implements QueuedJob
 {
-    public function getTitle()
-    {
-        return _t('CheckExternalLiksJob.TITLE', 'Checking for external broken links');
-    }
+	public function getTitle()
+	{
+		return _t('CheckExternalLiksJob.TITLE', 'Checking for external broken links');
+	}
 
-    public function getJobType()
-    {
-        return QueuedJob::QUEUED;
-    }
+	public function getJobType()
+	{
+		return QueuedJob::QUEUED;
+	}
 
-    public function getSignature()
-    {
-        return md5(get_class($this));
-    }
+	public function getSignature()
+	{
+		return md5(get_class($this));
+	}
 
-    /**
-     * Check an individual page
-     */
-    public function process()
-    {
-        $task = CheckExternalLinksTask::create();
-        $track = $task->runLinksCheck(1);
-        $this->currentStep = $track->CompletedPages;
-        $this->totalSteps = $track->TotalPages;
-        $this->isComplete = $track->Status === 'Completed';
-    }
+	/**
+	 * Check an individual page
+	 */
+	public function process()
+	{
+		$task = CheckExternalLinksTask::create();
+		$track = $task->runLinksCheck(1);
+		$this->currentStep = $track->CompletedPages;
+		$this->totalSteps = $track->TotalPages;
+		$this->isComplete = $track->Status === 'Completed';
+	}
 }

code/model/BrokenExternalLink.php

Indentation +55 added lines, -55 removed lines

@@ -8,66 +8,66 @@
  */
 class BrokenExternalLink extends DataObject
 {
-    private static $db = array(
-        'Link' => 'Varchar(2083)', // 2083 is the maximum length of a URL in Internet Explorer.
-        'HTTPCode' =>'Int'
-    );
+	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(
-        'Track' => 'BrokenExternalPageTrack',
-        'Status' => 'BrokenExternalPageTrackStatus'
-    );
+	private static $has_one = array(
+		'Track' => 'BrokenExternalPageTrack',
+		'Status' => 'BrokenExternalPageTrackStatus'
+	);
 
-    private static $summary_fields = array(
-        'Created' => 'Checked',
-        'Link' => 'External Link',
-        'HTTPCodeDescription' => 'HTTP Error Code',
-        'Page.Title' => 'Page link is on'
-    );
+	private static $summary_fields = array(
+		'Created' => 'Checked',
+		'Link' => 'External Link',
+		'HTTPCodeDescription' => 'HTTP Error Code',
+		'Page.Title' => 'Page link is on'
+	);
 
-    private static $searchable_fields = array(
-        'HTTPCode' => array('title' => 'HTTP Code')
-    );
+	private static $searchable_fields = array(
+		'HTTPCode' => array('title' => 'HTTP Code')
+	);
 
-    /**
-     * @return SiteTree
-     */
-    public function Page()
-    {
-        return $this->Track()->Page();
-    }
+	/**
+	 * @return SiteTree
+	 */
+	public function Page()
+	{
+		return $this->Track()->Page();
+	}
 
-    public function canEdit($member = false)
-    {
-        return false;
-    }
+	public function canEdit($member = false)
+	{
+		return false;
+	}
 
-    public function canView($member = false)
-    {
-        $member = $member ? $member : Member::currentUser();
-        $codes = array('content-authors', 'administrators');
-        return Permission::checkMember($member, $codes);
-    }
+	public function canView($member = false)
+	{
+		$member = $member ? $member : Member::currentUser();
+		$codes = array('content-authors', 'administrators');
+		return Permission::checkMember($member, $codes);
+	}
 
-    /**
-     * Retrieve a human readable description of a response code
-     *
-     * @return string
-     */
-    public function getHTTPCodeDescription()
-    {
-        $code = $this->HTTPCode;
-        if (empty($code)) {
-            // Assume that $code = 0 means there was no response
-            $description = _t('BrokenExternalLink.NOTAVAILABLE', 'Server Not Available');
-        } elseif (
-            ($descriptions = Config::inst()->get('SS_HTTPResponse', 'status_codes'))
-            && isset($descriptions[$code])
-        ) {
-            $description = $descriptions[$code];
-        } else {
-            $description = _t('BrokenExternalLink.UNKNOWNRESPONSE', 'Unknown Response Code');
-        }
-        return sprintf("%d (%s)", $code, $description);
-    }
+	/**
+	 * Retrieve a human readable description of a response code
+	 *
+	 * @return string
+	 */
+	public function getHTTPCodeDescription()
+	{
+		$code = $this->HTTPCode;
+		if (empty($code)) {
+			// Assume that $code = 0 means there was no response
+			$description = _t('BrokenExternalLink.NOTAVAILABLE', 'Server Not Available');
+		} elseif (
+			($descriptions = Config::inst()->get('SS_HTTPResponse', 'status_codes'))
+			&& isset($descriptions[$code])
+		) {
+			$description = $descriptions[$code];
+		} else {
+			$description = _t('BrokenExternalLink.UNKNOWNRESPONSE', 'Unknown Response Code');
+		}
+		return sprintf("%d (%s)", $code, $description);
+	}
 }

code/model/BrokenExternalPageTrack.php

Indentation +18 added lines, -18 removed lines

@@ -5,25 +5,25 @@
  */
 class BrokenExternalPageTrack extends DataObject
 {
-    private static $db = array(
-        'Processed' => 'Boolean'
-    );
+	private static $db = array(
+		'Processed' => 'Boolean'
+	);
 
-    private static $has_one = array(
-        'Page' => 'SiteTree',
-        'Status' => 'BrokenExternalPageTrackStatus'
-    );
+	private static $has_one = array(
+		'Page' => 'SiteTree',
+		'Status' => 'BrokenExternalPageTrackStatus'
+	);
 
-    private static $has_many = array(
-        'BrokenLinks' => 'BrokenExternalLink'
-    );
+	private static $has_many = array(
+		'BrokenLinks' => 'BrokenExternalLink'
+	);
 
-    /**
-     * @return SiteTree
-     */
-    public function Page()
-    {
-        return Versioned::get_by_stage('SiteTree', 'Stage')
-            ->byID($this->PageID);
-    }
+	/**
+	 * @return SiteTree
+	 */
+	public function Page()
+	{
+		return Versioned::get_by_stage('SiteTree', 'Stage')
+			->byID($this->PageID);
+	}
 }

code/reports/BrokenExternalLinksReport.php

Indentation +72 added lines, -72 removed lines

@@ -8,83 +8,83 @@
 
 class BrokenExternalLinksReport extends SS_Report
 {
-    /**
-     * Returns the report title
-     *
-     * @return string
-     */
-    public function title()
-    {
-        return _t('ExternalBrokenLinksReport.EXTERNALBROKENLINKS', "External broken links report");
-    }
+	/**
+	 * Returns the report title
+	 *
+	 * @return string
+	 */
+	public function title()
+	{
+		return _t('ExternalBrokenLinksReport.EXTERNALBROKENLINKS', "External broken links report");
+	}
 
-    public function columns()
-    {
-        return array(
-            "Created" => "Checked",
-            'Link' => array(
-                'title' => 'External Link',
-                'formatting' => function ($value, $item) {
-                    return sprintf(
-                        '<a target="_blank" href="%s">%s</a>',
-                        Convert::raw2att($item->Link),
-                        Convert::raw2xml($item->Link)
-                    );
-                }
-            ),
-            'HTTPCodeDescription' => 'HTTP Error Code',
-            "Title" => array(
-                "title" => 'Page link is on',
-                'formatting' => function ($value, $item) {
-                    $page = $item->Page();
-                    return sprintf(
-                        '<a href="%s">%s</a>',
-                        Convert::raw2att($page->CMSEditLink()),
-                        Convert::raw2xml($page->Title)
-                    );
-                }
-            )
-        );
-    }
+	public function columns()
+	{
+		return array(
+			"Created" => "Checked",
+			'Link' => array(
+				'title' => 'External Link',
+				'formatting' => function ($value, $item) {
+					return sprintf(
+						'<a target="_blank" href="%s">%s</a>',
+						Convert::raw2att($item->Link),
+						Convert::raw2xml($item->Link)
+					);
+				}
+			),
+			'HTTPCodeDescription' => 'HTTP Error Code',
+			"Title" => array(
+				"title" => 'Page link is on',
+				'formatting' => function ($value, $item) {
+					$page = $item->Page();
+					return sprintf(
+						'<a href="%s">%s</a>',
+						Convert::raw2att($page->CMSEditLink()),
+						Convert::raw2xml($page->Title)
+					);
+				}
+			)
+		);
+	}
 
-    /**
-     * Alias of columns(), to support the export to csv action
-     * in {@link GridFieldExportButton} generateExportFileData method.
-     * @return array
-     */
-    public function getColumns()
-    {
-        return $this->columns();
-    }
+	/**
+	 * Alias of columns(), to support the export to csv action
+	 * in {@link GridFieldExportButton} generateExportFileData method.
+	 * @return array
+	 */
+	public function getColumns()
+	{
+		return $this->columns();
+	}
 
-    public function sourceRecords()
-    {
-        $track = BrokenExternalPageTrackStatus::get_latest();
-        if ($track) {
-            return $track->BrokenLinks();
-        }
-        return new ArrayList();
-    }
+	public function sourceRecords()
+	{
+		$track = BrokenExternalPageTrackStatus::get_latest();
+		if ($track) {
+			return $track->BrokenLinks();
+		}
+		return new ArrayList();
+	}
 
-    public function getCMSFields()
-    {
-        Requirements::javascript('externallinks/javascript/BrokenExternalLinksReport.js');
-        $fields = parent::getCMSFields();
+	public function getCMSFields()
+	{
+		Requirements::javascript('externallinks/javascript/BrokenExternalLinksReport.js');
+		$fields = parent::getCMSFields();
 
-        $reportResultSpan = '</ br></ br><h3 id="ReportHolder"></h3>';
-        $reportResult = new LiteralField('ResultTitle', $reportResultSpan);
-        $fields->push($reportResult);
+		$reportResultSpan = '</ br></ br><h3 id="ReportHolder"></h3>';
+		$reportResult = new LiteralField('ResultTitle', $reportResultSpan);
+		$fields->push($reportResult);
 
-        $button = '<button id="externalLinksReport" type="button">%s</button>';
-        $runReportButton = new LiteralField(
-            'runReport',
-            sprintf(
-                $button,
-                _t('ExternalBrokenLinksReport.RUNREPORT', 'Create new report')
-            )
-        );
-        $fields->push($runReportButton);
+		$button = '<button id="externalLinksReport" type="button">%s</button>';
+		$runReportButton = new LiteralField(
+			'runReport',
+			sprintf(
+				$button,
+				_t('ExternalBrokenLinksReport.RUNREPORT', 'Create new report')
+			)
+		);
+		$fields->push($runReportButton);
 
-        return $fields;
-    }
+		return $fields;
+	}
 }

Spacing +2 added lines, -2 removed lines

@@ -24,7 +24,7 @@ 
             "Created" => "Checked",
             'Link' => array(
                 'title' => 'External Link',
-                'formatting' => function ($value, $item) {
+                'formatting' => function($value, $item) {
                     return sprintf(
                         '<a target="_blank" href="%s">%s</a>',
                         Convert::raw2att($item->Link),
@@ -35,7 +35,7 @@ 
             'HTTPCodeDescription' => 'HTTP Error Code',
             "Title" => array(
                 "title" => 'Page link is on',
-                'formatting' => function ($value, $item) {
+                'formatting' => function($value, $item) {
                     $page = $item->Page();
                     return sprintf(
                         '<a href="%s">%s</a>',

code/tasks/CheckExternalLinksTask.php

Indentation +207 added lines, -207 removed lines

@@ -2,211 +2,211 @@
 
 class CheckExternalLinksTask extends BuildTask
 {
-    private static $dependencies = array(
-        'LinkChecker' => '%$LinkChecker'
-    );
-
-    /**
-     * @var bool
-     */
-    protected $silent = false;
-
-    /**
-     * @var LinkChecker
-     */
-    protected $linkChecker;
-
-    protected $title = 'Checking broken External links in the SiteTree';
-
-    protected $description = 'A task that records external broken links in the SiteTree';
-
-    protected $enabled = true;
-
-    /**
-     * Log a message
-     *
-     * @param string $message
-     */
-    protected function log($message)
-    {
-        if (!$this->silent) {
-            Debug::message($message);
-        }
-    }
-
-    public function run($request)
-    {
-        $this->runLinksCheck();
-    }
-    /**
-     * Turn on or off message output
-     *
-     * @param bool $silent
-     */
-    public function setSilent($silent)
-    {
-        $this->silent = $silent;
-    }
-
-    /**
-     * @param LinkChecker $linkChecker
-     */
-    public function setLinkChecker(LinkChecker $linkChecker)
-    {
-        $this->linkChecker = $linkChecker;
-    }
-
-    /**
-     * @return LinkChecker
-     */
-    public function getLinkChecker()
-    {
-        return $this->linkChecker;
-    }
-
-    /**
-     * Check the status of a single link on a page
-     *
-     * @param BrokenExternalPageTrack $pageTrack
-     * @param DOMNode $link
-     */
-    protected function checkPageLink(BrokenExternalPageTrack $pageTrack, DOMNode $link)
-    {
-        $class = $link->getAttribute('class');
-        $href = $link->getAttribute('href');
-        $markedBroken = preg_match('/\b(ss-broken)\b/', $class);
-
-        // Check link
-        $httpCode = $this->linkChecker->checkLink($href);
-        if ($httpCode === null) {
-            return;
-        } // Null link means uncheckable, such as an internal link
-
-        // If this code is broken then mark as such
-        if ($foundBroken = $this->isCodeBroken($httpCode)) {
-            // Create broken record
-            $brokenLink = new BrokenExternalLink();
-            $brokenLink->Link = $href;
-            $brokenLink->HTTPCode = $httpCode;
-            $brokenLink->TrackID = $pageTrack->ID;
-            $brokenLink->StatusID = $pageTrack->StatusID; // Slight denormalisation here for performance reasons
-            $brokenLink->write();
-        }
-
-        // Check if we need to update CSS class, otherwise return
-        if ($markedBroken == $foundBroken) {
-            return;
-        }
-        if ($foundBroken) {
-            $class .= ' ss-broken';
-        } else {
-            $class = preg_replace('/\s*\b(ss-broken)\b\s*/', ' ', $class);
-        }
-        $link->setAttribute('class', trim($class));
-    }
-
-    /**
-     * Determine if the given HTTP code is "broken"
-     *
-     * @param int $httpCode
-     * @return bool True if this is a broken code
-     */
-    protected function isCodeBroken($httpCode)
-    {
-        // Null represents no request attempted
-        if ($httpCode === null) {
-            return false;
-        }
-
-        // do we have any whitelisted codes
-        $ignoreCodes = Config::inst()->get('CheckExternalLinks', 'IgnoreCodes');
-        if (is_array($ignoreCodes) && in_array($httpCode, $ignoreCodes)) {
-            return false;
-        }
-
-        // Check if code is outside valid range
-        return $httpCode < 200 || $httpCode > 302;
-    }
-
-    /**
-     * Runs the links checker and returns the track used
-     *
-     * @param int $limit Limit to number of pages to run, or null to run all
-     * @return BrokenExternalPageTrackStatus
-     */
-    public function runLinksCheck($limit = null)
-    {
-        // Check the current status
-        $status = BrokenExternalPageTrackStatus::get_or_create();
-
-        // Calculate pages to run
-        $pageTracks = $status->getIncompleteTracks();
-        if ($limit) {
-            $pageTracks = $pageTracks->limit($limit);
-        }
-
-        // Check each page
-        foreach ($pageTracks as $pageTrack) {
-            // Flag as complete
-            $pageTrack->Processed = 1;
-            $pageTrack->write();
-
-            // Check value of html area
-            $page = $pageTrack->Page();
-            $this->log("Checking {$page->Title}");
-            $htmlValue = Injector::inst()->create('HTMLValue', $page->Content);
-            if (!$htmlValue->isValid()) {
-                continue;
-            }
-
-            // Check each link
-            $links = $htmlValue->getElementsByTagName('a');
-            foreach ($links as $link) {
-                $this->checkPageLink($pageTrack, $link);
-            }
-
-            // Update content of page based on link fixes / breakages
-            $htmlValue->saveHTML();
-            $page->Content = $htmlValue->getContent();
-            $page->write();
-
-            // Once all links have been created for this page update HasBrokenLinks
-            $count = $pageTrack->BrokenLinks()->count();
-            $this->log("Found {$count} broken links");
-            if ($count) {
-                // Bypass the ORM as syncLinkTracking does not allow you to update HasBrokenLink to true
-                DB::query(sprintf(
-                    'UPDATE "SiteTree" SET "HasBrokenLink" = 1 WHERE "ID" = \'%d\'',
-                    intval($pageTrack->ID)
-                ));
-            }
-        }
-
-        $status->updateJobInfo('Updating completed pages');
-        $status->updateStatus();
-        return $status;
-    }
-
-    private function updateCompletedPages($trackID = 0)
-    {
-        $noPages = BrokenExternalPageTrack::get()
-            ->filter(array(
-                'TrackID' => $trackID,
-                'Processed' => 1
-            ))
-            ->count();
-        $track = BrokenExternalPageTrackStatus::get_latest();
-        $track->CompletedPages = $noPages;
-        $track->write();
-        return $noPages;
-    }
-
-    private function updateJobInfo($message)
-    {
-        $track = BrokenExternalPageTrackStatus::get_latest();
-        if ($track) {
-            $track->JobInfo = $message;
-            $track->write();
-        }
-    }
+	private static $dependencies = array(
+		'LinkChecker' => '%$LinkChecker'
+	);
+
+	/**
+	 * @var bool
+	 */
+	protected $silent = false;
+
+	/**
+	 * @var LinkChecker
+	 */
+	protected $linkChecker;
+
+	protected $title = 'Checking broken External links in the SiteTree';
+
+	protected $description = 'A task that records external broken links in the SiteTree';
+
+	protected $enabled = true;
+
+	/**
+	 * Log a message
+	 *
+	 * @param string $message
+	 */
+	protected function log($message)
+	{
+		if (!$this->silent) {
+			Debug::message($message);
+		}
+	}
+
+	public function run($request)
+	{
+		$this->runLinksCheck();
+	}
+	/**
+	 * Turn on or off message output
+	 *
+	 * @param bool $silent
+	 */
+	public function setSilent($silent)
+	{
+		$this->silent = $silent;
+	}
+
+	/**
+	 * @param LinkChecker $linkChecker
+	 */
+	public function setLinkChecker(LinkChecker $linkChecker)
+	{
+		$this->linkChecker = $linkChecker;
+	}
+
+	/**
+	 * @return LinkChecker
+	 */
+	public function getLinkChecker()
+	{
+		return $this->linkChecker;
+	}
+
+	/**
+	 * Check the status of a single link on a page
+	 *
+	 * @param BrokenExternalPageTrack $pageTrack
+	 * @param DOMNode $link
+	 */
+	protected function checkPageLink(BrokenExternalPageTrack $pageTrack, DOMNode $link)
+	{
+		$class = $link->getAttribute('class');
+		$href = $link->getAttribute('href');
+		$markedBroken = preg_match('/\b(ss-broken)\b/', $class);
+
+		// Check link
+		$httpCode = $this->linkChecker->checkLink($href);
+		if ($httpCode === null) {
+			return;
+		} // Null link means uncheckable, such as an internal link
+
+		// If this code is broken then mark as such
+		if ($foundBroken = $this->isCodeBroken($httpCode)) {
+			// Create broken record
+			$brokenLink = new BrokenExternalLink();
+			$brokenLink->Link = $href;
+			$brokenLink->HTTPCode = $httpCode;
+			$brokenLink->TrackID = $pageTrack->ID;
+			$brokenLink->StatusID = $pageTrack->StatusID; // Slight denormalisation here for performance reasons
+			$brokenLink->write();
+		}
+
+		// Check if we need to update CSS class, otherwise return
+		if ($markedBroken == $foundBroken) {
+			return;
+		}
+		if ($foundBroken) {
+			$class .= ' ss-broken';
+		} else {
+			$class = preg_replace('/\s*\b(ss-broken)\b\s*/', ' ', $class);
+		}
+		$link->setAttribute('class', trim($class));
+	}
+
+	/**
+	 * Determine if the given HTTP code is "broken"
+	 *
+	 * @param int $httpCode
+	 * @return bool True if this is a broken code
+	 */
+	protected function isCodeBroken($httpCode)
+	{
+		// Null represents no request attempted
+		if ($httpCode === null) {
+			return false;
+		}
+
+		// do we have any whitelisted codes
+		$ignoreCodes = Config::inst()->get('CheckExternalLinks', 'IgnoreCodes');
+		if (is_array($ignoreCodes) && in_array($httpCode, $ignoreCodes)) {
+			return false;
+		}
+
+		// Check if code is outside valid range
+		return $httpCode < 200 || $httpCode > 302;
+	}
+
+	/**
+	 * Runs the links checker and returns the track used
+	 *
+	 * @param int $limit Limit to number of pages to run, or null to run all
+	 * @return BrokenExternalPageTrackStatus
+	 */
+	public function runLinksCheck($limit = null)
+	{
+		// Check the current status
+		$status = BrokenExternalPageTrackStatus::get_or_create();
+
+		// Calculate pages to run
+		$pageTracks = $status->getIncompleteTracks();
+		if ($limit) {
+			$pageTracks = $pageTracks->limit($limit);
+		}
+
+		// Check each page
+		foreach ($pageTracks as $pageTrack) {
+			// Flag as complete
+			$pageTrack->Processed = 1;
+			$pageTrack->write();
+
+			// Check value of html area
+			$page = $pageTrack->Page();
+			$this->log("Checking {$page->Title}");
+			$htmlValue = Injector::inst()->create('HTMLValue', $page->Content);
+			if (!$htmlValue->isValid()) {
+				continue;
+			}
+
+			// Check each link
+			$links = $htmlValue->getElementsByTagName('a');
+			foreach ($links as $link) {
+				$this->checkPageLink($pageTrack, $link);
+			}
+
+			// Update content of page based on link fixes / breakages
+			$htmlValue->saveHTML();
+			$page->Content = $htmlValue->getContent();
+			$page->write();
+
+			// Once all links have been created for this page update HasBrokenLinks
+			$count = $pageTrack->BrokenLinks()->count();
+			$this->log("Found {$count} broken links");
+			if ($count) {
+				// Bypass the ORM as syncLinkTracking does not allow you to update HasBrokenLink to true
+				DB::query(sprintf(
+					'UPDATE "SiteTree" SET "HasBrokenLink" = 1 WHERE "ID" = \'%d\'',
+					intval($pageTrack->ID)
+				));
+			}
+		}
+
+		$status->updateJobInfo('Updating completed pages');
+		$status->updateStatus();
+		return $status;
+	}
+
+	private function updateCompletedPages($trackID = 0)
+	{
+		$noPages = BrokenExternalPageTrack::get()
+			->filter(array(
+				'TrackID' => $trackID,
+				'Processed' => 1
+			))
+			->count();
+		$track = BrokenExternalPageTrackStatus::get_latest();
+		$track->CompletedPages = $noPages;
+		$track->write();
+		return $noPages;
+	}
+
+	private function updateJobInfo($message)
+	{
+		$track = BrokenExternalPageTrackStatus::get_latest();
+		if ($track) {
+			$track->JobInfo = $message;
+			$track->write();
+		}
+	}
 }

code/tasks/CurlLinkChecker.php

Indentation +43 added lines, -43 removed lines

@@ -5,51 +5,51 @@
  */
 class CurlLinkChecker implements LinkChecker
 {
-    /**
-     * Return cache
-     *
-     * @return Zend_Cache_Frontend
-     */
-    protected function getCache()
-    {
-        return SS_Cache::factory(
-            __CLASS__,
-            'Output',
-            array('automatic_serialization' => true)
-        );
-    }
+	/**
+	 * Return cache
+	 *
+	 * @return Zend_Cache_Frontend
+	 */
+	protected function getCache()
+	{
+		return SS_Cache::factory(
+			__CLASS__,
+			'Output',
+			array('automatic_serialization' => true)
+		);
+	}
 
-    /**
-     * Determine the http status code for a given link
-     *
-     * @param string $href URL to check
-     * @return int HTTP status code, or null if not checkable (not a link)
-     */
-    public function checkLink($href)
-    {
-        // Skip non-external links
-        if (!preg_match('/^https?[^:]*:\/\//', $href)) {
-            return null;
-        }
+	/**
+	 * Determine the http status code for a given link
+	 *
+	 * @param string $href URL to check
+	 * @return int HTTP status code, or null if not checkable (not a link)
+	 */
+	public function checkLink($href)
+	{
+		// 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;
-        }
+		// 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($handle, CURLOPT_RETURNTRANSFER, true);
-        curl_setopt($handle, CURLOPT_CONNECTTIMEOUT, 5);
-        curl_setopt($handle, CURLOPT_TIMEOUT, 10);
-        curl_exec($handle);
-        $httpCode = curl_getinfo($handle, CURLINFO_HTTP_CODE);
-        curl_close($handle);
+		// No cached result so just request
+		$handle = curl_init($href);
+		curl_setopt($handle, CURLOPT_RETURNTRANSFER, true);
+		curl_setopt($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);
-        return $httpCode;
-    }
+		// Cache result
+		$this->getCache()->save($httpCode, $cacheKey);
+		return $httpCode;
+	}
 }

code/tasks/LinkChecker.php

Indentation +7 added lines, -7 removed lines

@@ -5,11 +5,11 @@
  */
 interface LinkChecker
 {
-    /**
-     * Determine the http status code for a given link
-     *
-     * @param string $href URL to check
-     * @return int HTTP status code, or null if not checkable (not a link)
-     */
-    public function checkLink($href);
+	/**
+	 * Determine the http status code for a given link
+	 *
+	 * @param string $href URL to check
+	 * @return int HTTP status code, or null if not checkable (not a link)
+	 */
+	public function checkLink($href);
 }