feat(webcal): only update modified and deleted events from webcal calendars

Signed-off-by: Anna Larch <anna@nextcloud.com>

3 files changed, 323 insertions, 172 deletions

diff --git a/apps/dav/lib/CalDAV/CalDavBackend.php b/apps/dav/lib/CalDAV/CalDavBackend.php
index 8fff6902f9a..0d12b8efead 100644
--- a/apps/dav/lib/CalDAV/CalDavBackend.php
+++ b/apps/dav/lib/CalDAV/CalDavBackend.php
@@ -946,6 +946,43 @@ class CalDavBackend extends AbstractBackend implements SyncSupport, Subscription
 	}
 
 	/**
+	 * Returns all calendar objects with limited metadata for a calendar
+	 *
+	 * Every item contains an array with the following keys:
+	 *   * id - the table row id
+	 *   * etag - An arbitrary string
+	 *   * uri - a unique key which will be used to construct the uri. This can
+	 *     be any arbitrary string.
+	 *   * calendardata - The iCalendar-compatible calendar data
+	 *
+	 * @param mixed $calendarId
+	 * @param int $calendarType
+	 * @return array
+	 */
+	public function getLimitedCalendarObjects(int $calendarId, int $calendarType = self::CALENDAR_TYPE_CALENDAR):array {
+		$query = $this->db->getQueryBuilder();
+		$query->select(['id','uid', 'etag', 'uri', 'calendardata'])
+			->from('calendarobjects')
+			->where($query->expr()->eq('calendarid', $query->createNamedParameter($calendarId)))
+			->andWhere($query->expr()->eq('calendartype', $query->createNamedParameter($calendarType)))
+			->andWhere($query->expr()->isNull('deleted_at'));
+		$stmt = $query->executeQuery();
+
+		$result = [];
+		while (($row = $stmt->fetch()) !== false) {
+			$result[$row['uid']] = [
+				'id' => $row['id'],
+				'etag' => $row['etag'],
+				'uri' => $row['uri'],
+				'calendardata' => $row['calendardata'],
+			];
+		}
+		$stmt->closeCursor();
+
+		return $result;
+	}
+
+	/**
 	 * Delete all of an user's shares
 	 *
 	 * @param string $principaluri
@@ -3265,6 +3302,45 @@ class CalDavBackend extends AbstractBackend implements SyncSupport, Subscription
 	}
 
 	/**
+	 * @param int $subscriptionId
+	 * @param array<int> $calendarObjectIds
+	 * @param array<string> $calendarObjectUris
+	 */
+	public function purgeCachedEventsForSubscription(int $subscriptionId, array $calendarObjectIds, array $calendarObjectUris): void {
+		if(empty($calendarObjectUris)) {
+			return;
+		}
+
+		$this->atomic(function () use ($subscriptionId, $calendarObjectIds, $calendarObjectUris) {
+			foreach (array_chunk($calendarObjectIds, 1000) as $chunk) {
+				$query = $this->db->getQueryBuilder();
+				$query->delete($this->dbObjectPropertiesTable)
+					->where($query->expr()->eq('calendarid', $query->createNamedParameter($subscriptionId)))
+					->andWhere($query->expr()->eq('calendartype', $query->createNamedParameter(self::CALENDAR_TYPE_SUBSCRIPTION)))
+					->andWhere($query->expr()->in('id', $query->createNamedParameter($chunk, IQueryBuilder::PARAM_INT_ARRAY), IQueryBuilder::PARAM_INT_ARRAY))
+					->executeStatement();
+
+				$query = $this->db->getQueryBuilder();
+				$query->delete('calendarobjects')
+					->where($query->expr()->eq('calendarid', $query->createNamedParameter($subscriptionId)))
+					->andWhere($query->expr()->eq('calendartype', $query->createNamedParameter(self::CALENDAR_TYPE_SUBSCRIPTION)))
+					->andWhere($query->expr()->in('id', $query->createNamedParameter($chunk, IQueryBuilder::PARAM_INT_ARRAY), IQueryBuilder::PARAM_INT_ARRAY))
+					->executeStatement();
+			}
+
+			foreach (array_chunk($calendarObjectUris, 1000) as $chunk) {
+				$query = $this->db->getQueryBuilder();
+				$query->delete('calendarchanges')
+					->where($query->expr()->eq('calendarid', $query->createNamedParameter($subscriptionId)))
+					->andWhere($query->expr()->eq('calendartype', $query->createNamedParameter(self::CALENDAR_TYPE_SUBSCRIPTION)))
+					->andWhere($query->expr()->in('uri', $query->createNamedParameter($chunk, IQueryBuilder::PARAM_STR_ARRAY), IQueryBuilder::PARAM_STR_ARRAY))
+					->executeStatement();
+			}
+			$this->addChanges($subscriptionId, $calendarObjectUris, 3, self::CALENDAR_TYPE_SUBSCRIPTION);
+		}, $this->db);
+	}
+
+	/**
 	 * Move a calendar from one user to another
 	 *
 	 * @param string $uriName
diff --git a/apps/dav/lib/CalDAV/WebcalCaching/Connection.php b/apps/dav/lib/CalDAV/WebcalCaching/Connection.php
new file mode 100644
index 00000000000..9d626f2db6d
--- /dev/null
+++ b/apps/dav/lib/CalDAV/WebcalCaching/Connection.php
@@ -0,0 +1,170 @@
+<?php
+
+declare(strict_types=1);
+
+/**
+ * SPDX-FileCopyrightText: 2024 Nextcloud GmbH and Nextcloud contributors
+ * SPDX-License-Identifier: AGPL-3.0-or-later
+ */
+namespace OCA\DAV\CalDAV\WebcalCaching;
+
+use Exception;
+use GuzzleHttp\HandlerStack;
+use GuzzleHttp\Middleware;
+use OCP\Http\Client\IClientService;
+use OCP\Http\Client\LocalServerException;
+use OCP\IAppConfig;
+use Psr\Http\Message\RequestInterface;
+use Psr\Http\Message\ResponseInterface;
+use Psr\Log\LoggerInterface;
+use Sabre\DAV\Xml\Property\Href;
+use Sabre\VObject\Reader;
+
+class Connection {
+	public function __construct(private IClientService $clientService,
+		private IAppConfig $config,
+		private LoggerInterface $logger) {
+	}
+
+	/**
+	 * gets webcal feed from remote server
+	 */
+	public function queryWebcalFeed(array $subscription, array &$mutations): ?string {
+		$client = $this->clientService->newClient();
+
+		$didBreak301Chain = false;
+		$latestLocation = null;
+
+		$handlerStack = HandlerStack::create();
+		$handlerStack->push(Middleware::mapRequest(function (RequestInterface $request) {
+			return $request
+				->withHeader('Accept', 'text/calendar, application/calendar+json, application/calendar+xml')
+				->withHeader('User-Agent', 'Nextcloud Webcal Service');
+		}));
+		$handlerStack->push(Middleware::mapResponse(function (ResponseInterface $response) use (&$didBreak301Chain, &$latestLocation) {
+			if (!$didBreak301Chain) {
+				if ($response->getStatusCode() !== 301) {
+					$didBreak301Chain = true;
+				} else {
+					$latestLocation = $response->getHeader('Location');
+				}
+			}
+			return $response;
+		}));
+
+		$allowLocalAccess = $this->config->getValueString('dav', 'webcalAllowLocalAccess', 'no');
+		$subscriptionId = $subscription['id'];
+		$url = $this->cleanURL($subscription['source']);
+		if ($url === null) {
+			return null;
+		}
+
+		try {
+			$params = [
+				'allow_redirects' => [
+					'redirects' => 10
+				],
+				'handler' => $handlerStack,
+				'nextcloud' => [
+					'allow_local_address' => $allowLocalAccess === 'yes',
+				]
+			];
+
+			$user = parse_url($subscription['source'], PHP_URL_USER);
+			$pass = parse_url($subscription['source'], PHP_URL_PASS);
+			if ($user !== null && $pass !== null) {
+				$params['auth'] = [$user, $pass];
+			}
+
+			$response = $client->get($url, $params);
+			$body = $response->getBody();
+
+			if ($latestLocation !== null) {
+				$mutations['{http://calendarserver.org/ns/}source'] = new Href($latestLocation);
+			}
+
+			$contentType = $response->getHeader('Content-Type');
+			$contentType = explode(';', $contentType, 2)[0];
+			switch ($contentType) {
+				case 'application/calendar+json':
+					try {
+						$jCalendar = Reader::readJson($body, Reader::OPTION_FORGIVING);
+					} catch (Exception $ex) {
+						// In case of a parsing error return null
+						$this->logger->warning("Subscription $subscriptionId could not be parsed", ['exception' => $ex]);
+						return null;
+					}
+					return $jCalendar->serialize();
+
+				case 'application/calendar+xml':
+					try {
+						$xCalendar = Reader::readXML($body);
+					} catch (Exception $ex) {
+						// In case of a parsing error return null
+						$this->logger->warning("Subscription $subscriptionId could not be parsed", ['exception' => $ex]);
+						return null;
+					}
+					return $xCalendar->serialize();
+
+				case 'text/calendar':
+				default:
+					try {
+						$vCalendar = Reader::read($body);
+					} catch (Exception $ex) {
+						// In case of a parsing error return null
+						$this->logger->warning("Subscription $subscriptionId could not be parsed", ['exception' => $ex]);
+						return null;
+					}
+					return $vCalendar->serialize();
+			}
+		} catch (LocalServerException $ex) {
+			$this->logger->warning("Subscription $subscriptionId was not refreshed because it violates local access rules", [
+				'exception' => $ex,
+			]);
+
+			return null;
+		} catch (Exception $ex) {
+			$this->logger->warning("Subscription $subscriptionId could not be refreshed due to a network error", [
+				'exception' => $ex,
+			]);
+
+			return null;
+		}
+	}
+
+	/**
+	 * This method will strip authentication information and replace the
+	 * 'webcal' or 'webcals' protocol scheme
+	 *
+	 * @param string $url
+	 * @return string|null
+	 */
+	private function cleanURL(string $url): ?string {
+		$parsed = parse_url($url);
+		if ($parsed === false) {
+			return null;
+		}
+
+		if (isset($parsed['scheme']) && $parsed['scheme'] === 'http') {
+			$scheme = 'http';
+		} else {
+			$scheme = 'https';
+		}
+
+		$host = $parsed['host'] ?? '';
+		$port = isset($parsed['port']) ? ':' . $parsed['port'] : '';
+		$path = $parsed['path'] ?? '';
+		$query = isset($parsed['query']) ? '?' . $parsed['query'] : '';
+		$fragment = isset($parsed['fragment']) ? '#' . $parsed['fragment'] : '';
+
+		$cleanURL = "$scheme://$host$port$path$query$fragment";
+		// parse_url is giving some weird results if no url and no :// is given,
+		// so let's test the url again
+		$parsedClean = parse_url($cleanURL);
+		if ($parsedClean === false || !isset($parsedClean['host'])) {
+			return null;
+		}
+
+		return $cleanURL;
+	}
+}
diff --git a/apps/dav/lib/CalDAV/WebcalCaching/RefreshWebcalService.php b/apps/dav/lib/CalDAV/WebcalCaching/RefreshWebcalService.php
index 209365ea5e4..102571c8ef9 100644
--- a/apps/dav/lib/CalDAV/WebcalCaching/RefreshWebcalService.php
+++ b/apps/dav/lib/CalDAV/WebcalCaching/RefreshWebcalService.php
@@ -8,19 +8,11 @@ declare(strict_types=1);
  */
 namespace OCA\DAV\CalDAV\WebcalCaching;
 
-use Exception;
-use GuzzleHttp\HandlerStack;
-use GuzzleHttp\Middleware;
 use OCA\DAV\CalDAV\CalDavBackend;
-use OCP\Http\Client\IClientService;
-use OCP\Http\Client\LocalServerException;
-use OCP\IConfig;
-use Psr\Http\Message\RequestInterface;
-use Psr\Http\Message\ResponseInterface;
+use OCP\AppFramework\Utility\ITimeFactory;
 use Psr\Log\LoggerInterface;
 use Sabre\DAV\Exception\BadRequest;
 use Sabre\DAV\PropPatch;
-use Sabre\DAV\Xml\Property\Href;
 use Sabre\VObject\Component;
 use Sabre\VObject\DateTimeParser;
 use Sabre\VObject\InvalidDataException;
@@ -33,25 +25,15 @@ use function count;
 
 class RefreshWebcalService {
 
-	private CalDavBackend $calDavBackend;
-
-	private IClientService $clientService;
-
-	private IConfig $config;
-
-	/** @var LoggerInterface */
-	private LoggerInterface $logger;
-
 	public const REFRESH_RATE = '{http://apple.com/ns/ical/}refreshrate';
 	public const STRIP_ALARMS = '{http://calendarserver.org/ns/}subscribed-strip-alarms';
 	public const STRIP_ATTACHMENTS = '{http://calendarserver.org/ns/}subscribed-strip-attachments';
 	public const STRIP_TODOS = '{http://calendarserver.org/ns/}subscribed-strip-todos';
 
-	public function __construct(CalDavBackend $calDavBackend, IClientService $clientService, IConfig $config, LoggerInterface $logger) {
-		$this->calDavBackend = $calDavBackend;
-		$this->clientService = $clientService;
-		$this->config = $config;
-		$this->logger = $logger;
+	public function __construct(private CalDavBackend $calDavBackend,
+		private LoggerInterface $logger,
+		private Connection $connection,
+		private ITimeFactory $time) {
 	}
 
 	public function refreshSubscription(string $principalUri, string $uri) {
@@ -61,11 +43,25 @@ class RefreshWebcalService {
 			return;
 		}
 
-		$webcalData = $this->queryWebcalFeed($subscription, $mutations);
+		// Check the refresh rate if there is any
+		if(!empty($subscription['{http://apple.com/ns/ical/}refreshrate'])) {
+			// add the refresh interval to the lastmodified timestamp
+			$refreshInterval = new \DateInterval($subscription['{http://apple.com/ns/ical/}refreshrate']);
+			$updateTime = $this->time->getDateTime();
+			$updateTime->setTimestamp($subscription['lastmodified'])->add($refreshInterval);
+			if($updateTime->getTimestamp() > $this->time->getTime()) {
+				return;
+			}
+		}
+
+
+		$webcalData = $this->connection->queryWebcalFeed($subscription, $mutations);
 		if (!$webcalData) {
 			return;
 		}
 
+		$localData = $this->calDavBackend->getLimitedCalendarObjects((int) $subscription['id'], CalDavBackend::CALENDAR_TYPE_SUBSCRIPTION);
+
 		$stripTodos = ($subscription[self::STRIP_TODOS] ?? 1) === 1;
 		$stripAlarms = ($subscription[self::STRIP_ALARMS] ?? 1) === 1;
 		$stripAttachments = ($subscription[self::STRIP_ATTACHMENTS] ?? 1) === 1;
@@ -73,14 +69,10 @@ class RefreshWebcalService {
 		try {
 			$splitter = new ICalendar($webcalData, Reader::OPTION_FORGIVING);
 
-			// we wait with deleting all outdated events till we parsed the new ones
-			// in case the new calendar is broken and `new ICalendar` throws a ParseException
-			// the user will still see the old data
-			$this->calDavBackend->purgeAllCachedEventsForSubscription($subscription['id']);
-
 			while ($vObject = $splitter->getNext()) {
 				/** @var Component $vObject */
 				$compName = null;
+				$uid = null;
 
 				foreach ($vObject->getComponents() as $component) {
 					if ($component->name === 'VTIMEZONE') {
@@ -95,21 +87,62 @@ class RefreshWebcalService {
 					if ($stripAttachments) {
 						unset($component->{'ATTACH'});
 					}
+
+					$uid = $component->{ 'UID' }->getValue();
 				}
 
 				if ($stripTodos && $compName === 'VTODO') {
 					continue;
 				}
 
-				$objectUri = $this->getRandomCalendarObjectUri();
-				$calendarData = $vObject->serialize();
+				if (!isset($uid)) {
+					continue;
+				}
+
+				$denormalized = $this->calDavBackend->getDenormalizedData($vObject->serialize());
+				// Find all identical sets and remove them from the update
+				if (isset($localData[$uid]) && $denormalized['etag'] === $localData[$uid]['etag']) {
+					unset($localData[$uid]);
+					continue;
+				}
+
+				$vObjectCopy = clone $vObject;
+				$identical = isset($localData[$uid]) && $this->compareWithoutDtstamp($vObjectCopy, $localData[$uid]);
+				if ($identical) {
+					unset($localData[$uid]);
+					continue;
+				}
+
+				// Find all modified sets and update them
+				if (isset($localData[$uid]) && $denormalized['etag'] !== $localData[$uid]['etag']) {
+					$this->calDavBackend->updateCalendarObject($subscription['id'], $localData[$uid]['uri'], $vObject->serialize(), CalDavBackend::CALENDAR_TYPE_SUBSCRIPTION);
+					unset($localData[$uid]);
+					continue;
+				}
+
+				// Only entirely new events get created here
 				try {
-					$this->calDavBackend->createCalendarObject($subscription['id'], $objectUri, $calendarData, CalDavBackend::CALENDAR_TYPE_SUBSCRIPTION);
+					$objectUri = $this->getRandomCalendarObjectUri();
+					$this->calDavBackend->createCalendarObject($subscription['id'], $objectUri, $vObject->serialize(), CalDavBackend::CALENDAR_TYPE_SUBSCRIPTION);
 				} catch (NoInstancesException | BadRequest $ex) {
 					$this->logger->error('Unable to create calendar object from subscription {subscriptionId}', ['exception' => $ex, 'subscriptionId' => $subscription['id'], 'source' => $subscription['source']]);
 				}
 			}
 
+			$ids = array_map(static function ($dataSet): int {
+				return (int) $dataSet['id'];
+			}, $localData);
+			$uris = array_map(static function ($dataSet): string {
+				return $dataSet['uri'];
+			}, $localData);
+
+			if(!empty($ids) && !empty($uris)) {
+				// Clean up on aisle 5
+				// The only events left over in the $localData array should be those that don't exist upstream
+				// All deleted VObjects from upstream are removed
+				$this->calDavBackend->purgeCachedEventsForSubscription($subscription['id'], $ids, $uris);
+			}
+
 			$newRefreshRate = $this->checkWebcalDataForRefreshRate($subscription, $webcalData);
 			if ($newRefreshRate) {
 				$mutations[self::REFRESH_RATE] = $newRefreshRate;
@@ -139,111 +172,6 @@ class RefreshWebcalService {
 		return $subscriptions[0];
 	}
 
-	/**
-	 * gets webcal feed from remote server
-	 */
-	private function queryWebcalFeed(array $subscription, array &$mutations): ?string {
-		$client = $this->clientService->newClient();
-
-		$didBreak301Chain = false;
-		$latestLocation = null;
-
-		$handlerStack = HandlerStack::create();
-		$handlerStack->push(Middleware::mapRequest(function (RequestInterface $request) {
-			return $request
-				->withHeader('Accept', 'text/calendar, application/calendar+json, application/calendar+xml')
-				->withHeader('User-Agent', 'Nextcloud Webcal Service');
-		}));
-		$handlerStack->push(Middleware::mapResponse(function (ResponseInterface $response) use (&$didBreak301Chain, &$latestLocation) {
-			if (!$didBreak301Chain) {
-				if ($response->getStatusCode() !== 301) {
-					$didBreak301Chain = true;
-				} else {
-					$latestLocation = $response->getHeader('Location');
-				}
-			}
-			return $response;
-		}));
-
-		$allowLocalAccess = $this->config->getAppValue('dav', 'webcalAllowLocalAccess', 'no');
-		$subscriptionId = $subscription['id'];
-		$url = $this->cleanURL($subscription['source']);
-		if ($url === null) {
-			return null;
-		}
-
-		try {
-			$params = [
-				'allow_redirects' => [
-					'redirects' => 10
-				],
-				'handler' => $handlerStack,
-				'nextcloud' => [
-					'allow_local_address' => $allowLocalAccess === 'yes',
-				]
-			];
-
-			$user = parse_url($subscription['source'], PHP_URL_USER);
-			$pass = parse_url($subscription['source'], PHP_URL_PASS);
-			if ($user !== null && $pass !== null) {
-				$params['auth'] = [$user, $pass];
-			}
-
-			$response = $client->get($url, $params);
-			$body = $response->getBody();
-
-			if ($latestLocation) {
-				$mutations['{http://calendarserver.org/ns/}source'] = new Href($latestLocation);
-			}
-
-			$contentType = $response->getHeader('Content-Type');
-			$contentType = explode(';', $contentType, 2)[0];
-			switch ($contentType) {
-				case 'application/calendar+json':
-					try {
-						$jCalendar = Reader::readJson($body, Reader::OPTION_FORGIVING);
-					} catch (Exception $ex) {
-						// In case of a parsing error return null
-						$this->logger->warning("Subscription $subscriptionId could not be parsed", ['exception' => $ex]);
-						return null;
-					}
-					return $jCalendar->serialize();
-
-				case 'application/calendar+xml':
-					try {
-						$xCalendar = Reader::readXML($body);
-					} catch (Exception $ex) {
-						// In case of a parsing error return null
-						$this->logger->warning("Subscription $subscriptionId could not be parsed", ['exception' => $ex]);
-						return null;
-					}
-					return $xCalendar->serialize();
-
-				case 'text/calendar':
-				default:
-					try {
-						$vCalendar = Reader::read($body);
-					} catch (Exception $ex) {
-						// In case of a parsing error return null
-						$this->logger->warning("Subscription $subscriptionId could not be parsed", ['exception' => $ex]);
-						return null;
-					}
-					return $vCalendar->serialize();
-			}
-		} catch (LocalServerException $ex) {
-			$this->logger->warning("Subscription $subscriptionId was not refreshed because it violates local access rules", [
-				'exception' => $ex,
-			]);
-
-			return null;
-		} catch (Exception $ex) {
-			$this->logger->warning("Subscription $subscriptionId could not be refreshed due to a network error", [
-				'exception' => $ex,
-			]);
-
-			return null;
-		}
-	}
 
 	/**
 	 * check if:
@@ -304,47 +232,24 @@ class RefreshWebcalService {
 	}
 
 	/**
-	 * This method will strip authentication information and replace the
-	 * 'webcal' or 'webcals' protocol scheme
+	 * Returns a random uri for a calendar-object
 	 *
-	 * @param string $url
-	 * @return string|null
+	 * @return string
 	 */
-	private function cleanURL(string $url): ?string {
-		$parsed = parse_url($url);
-		if ($parsed === false) {
-			return null;
-		}
+	public function getRandomCalendarObjectUri():string {
+		return UUIDUtil::getUUID() . '.ics';
+	}
 
-		if (isset($parsed['scheme']) && $parsed['scheme'] === 'http') {
-			$scheme = 'http';
-		} else {
-			$scheme = 'https';
+	private function compareWithoutDtstamp(Component $vObject, array $calendarObject): bool {
+		foreach ($vObject->getComponents() as $component) {
+			unset($component->{'DTSTAMP'});
 		}
 
-		$host = $parsed['host'] ?? '';
-		$port = isset($parsed['port']) ? ':' . $parsed['port'] : '';
-		$path = $parsed['path'] ?? '';
-		$query = isset($parsed['query']) ? '?' . $parsed['query'] : '';
-		$fragment = isset($parsed['fragment']) ? '#' . $parsed['fragment'] : '';
-
-		$cleanURL = "$scheme://$host$port$path$query$fragment";
-		// parse_url is giving some weird results if no url and no :// is given,
-		// so let's test the url again
-		$parsedClean = parse_url($cleanURL);
-		if ($parsedClean === false || !isset($parsedClean['host'])) {
-			return null;
+		$localVobject = Reader::read($calendarObject['calendardata']);
+		foreach ($localVobject->getComponents() as $component) {
+			unset($component->{'DTSTAMP'});
 		}
 
-		return $cleanURL;
-	}
-
-	/**
-	 * Returns a random uri for a calendar-object
-	 *
-	 * @return string
-	 */
-	public function getRandomCalendarObjectUri():string {
-		return UUIDUtil::getUUID() . '.ics';
+		return strcasecmp($localVobject->serialize(), $vObject->serialize()) === 0;
 	}
 }