silverstripe / silverstripe-externallinks

src/Model/BrokenExternalPageTrack.php

Indentation +19 added lines, -19 removed lines

@@ -13,27 +13,27 @@
  */
 class BrokenExternalPageTrack extends DataObject
 {
-    private static $table_name = 'BrokenExternalPageTrack';
+	private static $table_name = 'BrokenExternalPageTrack';
 
-    private static $db = array(
-        'Processed' => 'Boolean'
-    );
+	private static $db = array(
+		'Processed' => 'Boolean'
+	);
 
-    private static $has_one = array(
-        'Page' => SiteTree::class,
-        'Status' => BrokenExternalPageTrackStatus::class
-    );
+	private static $has_one = array(
+		'Page' => SiteTree::class,
+		'Status' => BrokenExternalPageTrackStatus::class
+	);
 
-    private static $has_many = array(
-        'BrokenLinks' => BrokenExternalLink::class
-    );
+	private static $has_many = array(
+		'BrokenLinks' => BrokenExternalLink::class
+	);
 
-    /**
-     * @return SiteTree
-     */
-    public function Page()
-    {
-        return Versioned::get_by_stage(SiteTree::class, 'Stage')
-            ->byID($this->PageID);
-    }
+	/**
+	 * @return SiteTree
+	 */
+	public function Page()
+	{
+		return Versioned::get_by_stage(SiteTree::class, 'Stage')
+			->byID($this->PageID);
+	}
 }

src/Tasks/LinkChecker.php

Indentation +7 added lines, -7 removed lines

@@ -8,11 +8,11 @@
 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);
 }

src/Jobs/CheckExternalLinksJob.php

Indentation +27 added lines, -27 removed lines

@@ -7,7 +7,7 @@ 
 use SilverStripe\ExternalLinks\Tasks\CheckExternalLinksTask;
 
 if (!class_exists(AbstractQueuedJob::class)) {
-    return;
+	return;
 }
 
 /**
@@ -17,30 +17,30 @@ 
 class CheckExternalLinksJob extends AbstractQueuedJob implements QueuedJob
 {
 
-    public function getTitle()
-    {
-        return _t(__CLASS__ . '.TITLE', 'Checking for external broken links');
-    }
-
-    public function getJobType()
-    {
-        return QueuedJob::QUEUED;
-    }
-
-    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 = $track->Status === 'Completed';
-    }
+	public function getTitle()
+	{
+		return _t(__CLASS__ . '.TITLE', 'Checking for external broken links');
+	}
+
+	public function getJobType()
+	{
+		return QueuedJob::QUEUED;
+	}
+
+	public function getSignature()
+	{
+		return md5(get_class($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';
+	}
 }

tests/ExternalLinksTest.php

Indentation +80 added lines, -80 removed lines

@@ -16,95 +16,95 @@
 class ExternalLinksTest extends SapphireTest
 {
 
-    protected static $fixture_file = 'ExternalLinksTest.yml';
+	protected static $fixture_file = 'ExternalLinksTest.yml';
 
-    protected static $extra_dataobjects = array(
-        ExternalLinksTestPage::class
-    );
+	protected static $extra_dataobjects = array(
+		ExternalLinksTestPage::class
+	);
 
-    protected function setUp()
-    {
-        parent::setUp();
+	protected function setUp()
+	{
+		parent::setUp();
 
-        // Stub link checker
-        $checker = new PretendLinkChecker;
-        Injector::inst()->registerService($checker, LinkChecker::class);
-    }
+		// Stub link checker
+		$checker = new PretendLinkChecker;
+		Injector::inst()->registerService($checker, LinkChecker::class);
+	}
 
-    public function testLinks()
-    {
-        // Run link checker
-        $task = CheckExternalLinksTask::create();
-        $task->setSilent(true); // Be quiet during the test!
-        $task->runLinksCheck();
+	public function testLinks()
+	{
+		// Run link checker
+		$task = CheckExternalLinksTask::create();
+		$task->setSilent(true); // Be quiet during the test!
+		$task->runLinksCheck();
 
-        // Get all links checked
-        $status = BrokenExternalPageTrackStatus::get_latest();
-        $this->assertEquals('Completed', $status->Status);
-        $this->assertEquals(5, $status->TotalPages);
-        $this->assertEquals(5, $status->CompletedPages);
+		// Get all links checked
+		$status = BrokenExternalPageTrackStatus::get_latest();
+		$this->assertEquals('Completed', $status->Status);
+		$this->assertEquals(5, $status->TotalPages);
+		$this->assertEquals(5, $status->CompletedPages);
 
-        // Check all pages have had the correct HTML adjusted
-        for ($i = 1; $i <= 5; $i++) {
-            $page = $this->objFromFixture(ExternalLinksTestPage::class, 'page'.$i);
-            $this->assertNotEmpty($page->Content);
-            $this->assertEquals(
-                $page->ExpectedContent,
-                $page->Content,
-                "Assert that the content of page{$i} has been updated"
-            );
-        }
+		// Check all pages have had the correct HTML adjusted
+		for ($i = 1; $i <= 5; $i++) {
+			$page = $this->objFromFixture(ExternalLinksTestPage::class, 'page'.$i);
+			$this->assertNotEmpty($page->Content);
+			$this->assertEquals(
+				$page->ExpectedContent,
+				$page->Content,
+				"Assert that the content of page{$i} has been updated"
+			);
+		}
 
-        // Check that the correct report of broken links is generated
-        $links = $status
-            ->BrokenLinks()
-            ->sort('Link');
+		// Check that the correct report of broken links is generated
+		$links = $status
+			->BrokenLinks()
+			->sort('Link');
 
-        $this->assertEquals(4, $links->count());
-        $this->assertEquals(
-            array(
-                'http://www.broken.com',
-                'http://www.broken.com/url/thing',
-                'http://www.broken.com/url/thing',
-                'http://www.nodomain.com'
-            ),
-            array_values($links->map('ID', 'Link')->toArray())
-        );
+		$this->assertEquals(4, $links->count());
+		$this->assertEquals(
+			array(
+				'http://www.broken.com',
+				'http://www.broken.com/url/thing',
+				'http://www.broken.com/url/thing',
+				'http://www.nodomain.com'
+			),
+			array_values($links->map('ID', 'Link')->toArray())
+		);
 
-        // Check response codes are correct
-        $expected = array(
-            'http://www.broken.com' => 403,
-            'http://www.broken.com/url/thing' => 404,
-            'http://www.nodomain.com' => 0
-        );
-        $actual = $links->map('Link', 'HTTPCode')->toArray();
-        $this->assertEquals($expected, $actual);
+		// Check response codes are correct
+		$expected = array(
+			'http://www.broken.com' => 403,
+			'http://www.broken.com/url/thing' => 404,
+			'http://www.nodomain.com' => 0
+		);
+		$actual = $links->map('Link', 'HTTPCode')->toArray();
+		$this->assertEquals($expected, $actual);
 
-        // Check response descriptions are correct
-        i18n::set_locale('en_NZ');
-        $expected = array(
-            'http://www.broken.com' => '403 (Forbidden)',
-            'http://www.broken.com/url/thing' => '404 (Not Found)',
-            'http://www.nodomain.com' => '0 (Server Not Available)'
-        );
-        $actual = $links->map('Link', 'HTTPCodeDescription')->toArray();
-        $this->assertEquals($expected, $actual);
-    }
+		// Check response descriptions are correct
+		i18n::set_locale('en_NZ');
+		$expected = array(
+			'http://www.broken.com' => '403 (Forbidden)',
+			'http://www.broken.com/url/thing' => '404 (Not Found)',
+			'http://www.nodomain.com' => '0 (Server Not Available)'
+		);
+		$actual = $links->map('Link', 'HTTPCodeDescription')->toArray();
+		$this->assertEquals($expected, $actual);
+	}
 
-    /**
-     * Test that broken links appears in the reports list
-     */
-    public function testReportExists()
-    {
-        $reports = Report::get_reports();
-        $reportNames = array();
-        foreach ($reports as $report) {
-            $reportNames[] = get_class($report);
-        }
-        $this->assertContains(
-            BrokenExternalLinksReport::class,
-            $reportNames,
-            'BrokenExternalLinksReport is in reports list'
-        );
-    }
+	/**
+	 * Test that broken links appears in the reports list
+	 */
+	public function testReportExists()
+	{
+		$reports = Report::get_reports();
+		$reportNames = array();
+		foreach ($reports as $report) {
+			$reportNames[] = get_class($report);
+		}
+		$this->assertContains(
+			BrokenExternalLinksReport::class,
+			$reportNames,
+			'BrokenExternalLinksReport is in reports list'
+		);
+	}
 }

src/Model/BrokenExternalPageTrackStatus.php

Indentation +132 added lines, -132 removed lines

@@ -18,136 +18,136 @@
  */
 class BrokenExternalPageTrackStatus extends DataObject
 {
-    private static $table_name = 'BrokenExternalPageTrackStatus';
-
-    private static $db = array(
-        'Status' => 'Enum("Completed, Running", "Running")',
-        'JobInfo' => 'Varchar(255)'
-    );
-
-    private static $has_many = array(
-        'TrackedPages' => BrokenExternalPageTrack::class,
-        'BrokenLinks' => BrokenExternalLink::class
-    );
-
-    /**
-     * Get the latest track status
-     *
-     * @return BrokenExternalPageTrackStatus
-     */
-    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::class, 'Stage')
-            ->byIDs($pageIDs);
-        }
-    }
-
-    /**
-     * Get the list of incomplete BrokenExternalPageTrack
-     *
-     * @return DataList
-     */
-    public function getIncompleteTracks()
-    {
-        return $this
-            ->TrackedPages()
-            ->filter('Processed', 0);
-    }
-
-    /**
-     * Get total pages count
-     *
-     * @return int
-     */
-    public function getTotalPages()
-    {
-        return $this->TrackedPages()->count();
-    }
-
-    /**
-     * Get completed pages count
-     *
-     * @return int
-     */
-    public function getCompletedPages()
-    {
-        return $this
-            ->TrackedPages()
-            ->filter('Processed', 1)
-            ->count();
-    }
-
-    /**
-     * Returns the latest run, or otherwise creates a new one
-     *
-     * @return BrokenExternalPageTrackStatus
-     */
-    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();
-    }
-
-    /**
-     * Create and prepare a new status
-     *
-     * @return BrokenExternalPageTrackStatus
-     */
-    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
-        $pageIDs = Versioned::get_by_stage(SiteTree::class, 'Stage')
-            ->column('ID');
-        foreach ($pageIDs as $pageID) {
-            $trackPage = BrokenExternalPageTrack::create();
-            $trackPage->PageID = $pageID;
-            $trackPage->StatusID = $status->ID;
-            $trackPage->write();
-        }
-
-        return $status;
-    }
-
-    public function updateJobInfo($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');
-        }
-    }
+	private static $table_name = 'BrokenExternalPageTrackStatus';
+
+	private static $db = array(
+		'Status' => 'Enum("Completed, Running", "Running")',
+		'JobInfo' => 'Varchar(255)'
+	);
+
+	private static $has_many = array(
+		'TrackedPages' => BrokenExternalPageTrack::class,
+		'BrokenLinks' => BrokenExternalLink::class
+	);
+
+	/**
+	 * Get the latest track status
+	 *
+	 * @return BrokenExternalPageTrackStatus
+	 */
+	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::class, 'Stage')
+			->byIDs($pageIDs);
+		}
+	}
+
+	/**
+	 * Get the list of incomplete BrokenExternalPageTrack
+	 *
+	 * @return DataList
+	 */
+	public function getIncompleteTracks()
+	{
+		return $this
+			->TrackedPages()
+			->filter('Processed', 0);
+	}
+
+	/**
+	 * Get total pages count
+	 *
+	 * @return int
+	 */
+	public function getTotalPages()
+	{
+		return $this->TrackedPages()->count();
+	}
+
+	/**
+	 * Get completed pages count
+	 *
+	 * @return int
+	 */
+	public function getCompletedPages()
+	{
+		return $this
+			->TrackedPages()
+			->filter('Processed', 1)
+			->count();
+	}
+
+	/**
+	 * Returns the latest run, or otherwise creates a new one
+	 *
+	 * @return BrokenExternalPageTrackStatus
+	 */
+	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();
+	}
+
+	/**
+	 * Create and prepare a new status
+	 *
+	 * @return BrokenExternalPageTrackStatus
+	 */
+	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
+		$pageIDs = Versioned::get_by_stage(SiteTree::class, 'Stage')
+			->column('ID');
+		foreach ($pageIDs as $pageID) {
+			$trackPage = BrokenExternalPageTrack::create();
+			$trackPage->PageID = $pageID;
+			$trackPage->StatusID = $status->ID;
+			$trackPage->write();
+		}
+
+		return $status;
+	}
+
+	public function updateJobInfo($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');
+		}
+	}
 }

tests/Stubs/PretendLinkChecker.php

Indentation +20 added lines, -20 removed lines

@@ -6,24 +6,24 @@
 
 class PretendLinkChecker implements LinkChecker, TestOnly
 {
-    public function checkLink($href)
-    {
-        switch ($href) {
-            case 'http://www.working.com':
-                return 200;
-            case 'http://www.broken.com':
-                return 403;
-            case 'http://www.nodomain.com':
-                return 0;
-            case '/internal/link':
-            case '[sitetree_link,id=9999]':
-            case 'home':
-            case 'broken-internal':
-            case '[sitetree_link,id=1]':
-                return null;
-            case 'http://www.broken.com/url/thing':
-            default:
-                return 404;
-        }
-    }
+	public function checkLink($href)
+	{
+		switch ($href) {
+			case 'http://www.working.com':
+				return 200;
+			case 'http://www.broken.com':
+				return 403;
+			case 'http://www.nodomain.com':
+				return 0;
+			case '/internal/link':
+			case '[sitetree_link,id=9999]':
+			case 'home':
+			case 'broken-internal':
+			case '[sitetree_link,id=1]':
+				return null;
+			case 'http://www.broken.com/url/thing':
+			default:
+				return 404;
+		}
+	}
 }

tests/Stubs/ExternalLinksTestPage.php

Indentation +4 added lines, -4 removed lines

@@ -7,9 +7,9 @@
 
 class ExternalLinksTestPage extends Page implements TestOnly
 {
-    private static $table_name = 'ExternalLinksTestPage';
+	private static $table_name = 'ExternalLinksTestPage';
 
-    private static $db = array(
-        'ExpectedContent' => 'HTMLText'
-    );
+	private static $db = array(
+		'ExpectedContent' => 'HTMLText'
+	);
 }

src/Tasks/CheckExternalLinksTask.php

Indentation +220 added lines, -220 removed lines

@@ -17,224 +17,224 @@
 
 class CheckExternalLinksTask extends BuildTask
 {
-    private static $dependencies = [
-        'LinkChecker' => '%$' . LinkChecker::class
-    ];
-
-    private static $segment = 'CheckExternalLinksTask';
-
-    /**
-     * Define a list of HTTP response codes that should not be treated as "broken", where they usually
-     * might be.
-     *
-     * @config
-     * @var array
-     */
-    private static $ignore_codes = [];
-
-    /**
-     * @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 = $this->config()->get('ignore_codes');
-        if (is_array($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) {
-                $siteTreeTable = DataObject::getSchema()->tableName(SiteTree::class);
-                // Bypass the ORM as syncLinkTracking does not allow you to update HasBrokenLink to true
-                DB::query(sprintf(
-                    'UPDATE "%s" SET "HasBrokenLink" = 1 WHERE "ID" = \'%d\'',
-                    $siteTreeTable,
-                    intval($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 $dependencies = [
+		'LinkChecker' => '%$' . LinkChecker::class
+	];
+
+	private static $segment = 'CheckExternalLinksTask';
+
+	/**
+	 * Define a list of HTTP response codes that should not be treated as "broken", where they usually
+	 * might be.
+	 *
+	 * @config
+	 * @var array
+	 */
+	private static $ignore_codes = [];
+
+	/**
+	 * @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 = $this->config()->get('ignore_codes');
+		if (is_array($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) {
+				$siteTreeTable = DataObject::getSchema()->tableName(SiteTree::class);
+				// Bypass the ORM as syncLinkTracking does not allow you to update HasBrokenLink to true
+				DB::query(sprintf(
+					'UPDATE "%s" SET "HasBrokenLink" = 1 WHERE "ID" = \'%d\'',
+					$siteTreeTable,
+					intval($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();
+		}
+	}
 }

src/Model/BrokenExternalLink.php

Indentation +54 added lines, -54 removed lines

@@ -17,68 +17,68 @@
  */
 class BrokenExternalLink extends DataObject
 {
-    private static $table_name = 'BrokenExternalLink';
+	private static $table_name = 'BrokenExternalLink';
 
-    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::class,
-        'Status' => BrokenExternalPageTrackStatus::class
-    );
+	private static $has_one = array(
+		'Track' => BrokenExternalPageTrack::class,
+		'Status' => BrokenExternalPageTrackStatus::class
+	);
 
-    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 : Security::getCurrentUser();
-        $codes = array('content-authors', 'administrators');
-        return Permission::checkMember($member, $codes);
-    }
+	public function canView($member = false)
+	{
+		$member = $member ? $member : Security::getCurrentUser();
+		$codes = array('content-authors', 'administrators');
+		return Permission::checkMember($member, $codes);
+	}
 
-    /**
-     * Retrieve a human readable description of a response code
-     *
-     * @return string
-     */
-    public function getHTTPCodeDescription()
-    {
-        $code = $this->HTTPCode;
+	/**
+	 * Retrieve a human readable description of a response code
+	 *
+	 * @return string
+	 */
+	public function getHTTPCodeDescription()
+	{
+		$code = $this->HTTPCode;
 
-        try {
-            $response = HTTPResponse::create('', $code);
-            // Assume that $code = 0 means there was no response
-            $description = $code ?
-                $response->getStatusDescription() :
-                _t(__CLASS__ . '.NOTAVAILABLE', 'Server Not Available');
-        } catch (InvalidArgumentException $e) {
-            $description = _t(__CLASS__ . '.UNKNOWNRESPONSE', 'Unknown Response Code');
-        }
+		try {
+			$response = HTTPResponse::create('', $code);
+			// Assume that $code = 0 means there was no response
+			$description = $code ?
+				$response->getStatusDescription() :
+				_t(__CLASS__ . '.NOTAVAILABLE', 'Server Not Available');
+		} catch (InvalidArgumentException $e) {
+			$description = _t(__CLASS__ . '.UNKNOWNRESPONSE', 'Unknown Response Code');
+		}
 
-        return sprintf("%d (%s)", $code, $description);
-    }
+		return sprintf("%d (%s)", $code, $description);
+	}
 }