silverstripe / silverstripe-externallinks

src/Tasks/CheckExternalLinksTask.php

Indentation +225 added lines, -225 removed lines

@@ -18,229 +18,229 @@
 
 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();
-            try {
-                $page->write();
-            } catch (ValidationException $ex) {
-                $this->log("Exception caught for {$page->Title}, skipping. Message: " . $ex->getMessage());
-                continue;
-            }
-
-            // 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();
+			try {
+				$page->write();
+			} catch (ValidationException $ex) {
+				$this->log("Exception caught for {$page->Title}, skipping. Message: " . $ex->getMessage());
+				continue;
+			}
+
+			// 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();
+		}
+	}
 }

Spacing +2 added lines, -2 removed lines

@@ -19,7 +19,7 @@ 
 class CheckExternalLinksTask extends BuildTask
 {
     private static $dependencies = [
-        'LinkChecker' => '%$' . LinkChecker::class
+        'LinkChecker' => '%$'.LinkChecker::class
     ];
 
     private static $segment = 'CheckExternalLinksTask';
@@ -198,7 +198,7 @@ 
             try {
                 $page->write();
             } catch (ValidationException $ex) {
-                $this->log("Exception caught for {$page->Title}, skipping. Message: " . $ex->getMessage());
+                $this->log("Exception caught for {$page->Title}, skipping. Message: ".$ex->getMessage());
                 continue;
             }
 

src/Model/BrokenExternalPageTrackStatus.php

Indentation +149 added lines, -149 removed lines

@@ -17,153 +17,153 @@
  */
 class BrokenExternalPageTrackStatus extends DataObject implements i18nEntityProvider
 {
-    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();
-    }
-
-    /**
-     * Returns the list of provided translations for this object
-     *
-     * @return array
-     */
-    public function provideI18nEntities()
-    {
-        return [
-            __CLASS__ . '.SINGULARNAME' => 'Broken External Page Track Status',
-            __CLASS__ . '.PLURALNAME' => 'Broken External Page Track Statuses',
-            __CLASS__ . '.PLURALS' => [
-              'one' => 'A Broken External Page Track Status',
-              'other' => '{count} Broken External Page Track Statuses',
-            ],
-        ];
-    }
-
-    /**
-     * 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();
+	}
+
+	/**
+	 * Returns the list of provided translations for this object
+	 *
+	 * @return array
+	 */
+	public function provideI18nEntities()
+	{
+		return [
+			__CLASS__ . '.SINGULARNAME' => 'Broken External Page Track Status',
+			__CLASS__ . '.PLURALNAME' => 'Broken External Page Track Statuses',
+			__CLASS__ . '.PLURALS' => [
+			  'one' => 'A Broken External Page Track Status',
+			  'other' => '{count} Broken External Page Track Statuses',
+			],
+		];
+	}
+
+	/**
+	 * 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');
+		}
+	}
 }

Spacing +3 added lines, -3 removed lines

@@ -49,9 +49,9 @@
     public function provideI18nEntities()
     {
         return [
-            __CLASS__ . '.SINGULARNAME' => 'Broken External Page Track Status',
-            __CLASS__ . '.PLURALNAME' => 'Broken External Page Track Statuses',
-            __CLASS__ . '.PLURALS' => [
+            __CLASS__.'.SINGULARNAME' => 'Broken External Page Track Status',
+            __CLASS__.'.PLURALNAME' => 'Broken External Page Track Statuses',
+            __CLASS__.'.PLURALS' => [
               'one' => 'A Broken External Page Track Status',
               'other' => '{count} Broken External Page Track Statuses',
             ],

tests/Model/BrokenExternalLinkTest.php

Indentation +24 added lines, -24 removed lines

@@ -7,29 +7,29 @@
 
 class BrokenExternalLinkTest extends SapphireTest
 {
-    /**
-     * @param int $httpCode
-     * @param string $expected
-     * @dataProvider httpCodeProvider
-     */
-    public function testGetHTTPCodeDescription($httpCode, $expected)
-    {
-        $link = new BrokenExternalLink();
-        $link->HTTPCode = $httpCode;
-        $this->assertSame($expected, $link->getHTTPCodeDescription());
-    }
+	/**
+	 * @param int $httpCode
+	 * @param string $expected
+	 * @dataProvider httpCodeProvider
+	 */
+	public function testGetHTTPCodeDescription($httpCode, $expected)
+	{
+		$link = new BrokenExternalLink();
+		$link->HTTPCode = $httpCode;
+		$this->assertSame($expected, $link->getHTTPCodeDescription());
+	}
 
-    /**
-     * @return array[]
-     */
-    public function httpCodeProvider()
-    {
-        return [
-            [200, '200 (OK)'],
-            [302, '302 (Found)'],
-            [404, '404 (Not Found)'],
-            [500, '500 (Internal Server Error)'],
-            [789, '789 (Unknown Response Code)'],
-        ];
-    }
+	/**
+	 * @return array[]
+	 */
+	public function httpCodeProvider()
+	{
+		return [
+			[200, '200 (OK)'],
+			[302, '302 (Found)'],
+			[404, '404 (Not Found)'],
+			[500, '500 (Internal Server Error)'],
+			[789, '789 (Unknown Response Code)'],
+		];
+	}
 }