@@ -43,149 +43,149 @@ |
||
| 43 | 43 | use Psr\Log\LoggerInterface; |
| 44 | 44 | |
| 45 | 45 | class LinkReferenceProvider implements IReferenceProvider { |
| 46 | - public const MAX_PREVIEW_SIZE = 1024 * 1024; |
|
| 47 | - |
|
| 48 | - public const ALLOWED_CONTENT_TYPES = [ |
|
| 49 | - 'image/png', |
|
| 50 | - 'image/jpg', |
|
| 51 | - 'image/jpeg', |
|
| 52 | - 'image/gif', |
|
| 53 | - 'image/svg+xml', |
|
| 54 | - 'image/webp' |
|
| 55 | - ]; |
|
| 56 | - |
|
| 57 | - private IClientService $clientService; |
|
| 58 | - private LoggerInterface $logger; |
|
| 59 | - private SystemConfig $systemConfig; |
|
| 60 | - private IAppDataFactory $appDataFactory; |
|
| 61 | - private IURLGenerator $urlGenerator; |
|
| 62 | - private Limiter $limiter; |
|
| 63 | - private IUserSession $userSession; |
|
| 64 | - private IRequest $request; |
|
| 65 | - |
|
| 66 | - public function __construct(IClientService $clientService, LoggerInterface $logger, SystemConfig $systemConfig, IAppDataFactory $appDataFactory, IURLGenerator $urlGenerator, Limiter $limiter, IUserSession $userSession, IRequest $request) { |
|
| 67 | - $this->clientService = $clientService; |
|
| 68 | - $this->logger = $logger; |
|
| 69 | - $this->systemConfig = $systemConfig; |
|
| 70 | - $this->appDataFactory = $appDataFactory; |
|
| 71 | - $this->urlGenerator = $urlGenerator; |
|
| 72 | - $this->limiter = $limiter; |
|
| 73 | - $this->userSession = $userSession; |
|
| 74 | - $this->request = $request; |
|
| 75 | - } |
|
| 76 | - |
|
| 77 | - public function matchReference(string $referenceText): bool { |
|
| 78 | - if ($this->systemConfig->getValue('reference_opengraph', true) !== true) { |
|
| 79 | - return false; |
|
| 80 | - } |
|
| 81 | - |
|
| 82 | - return (bool)preg_match(IURLGenerator::URL_REGEX, $referenceText); |
|
| 83 | - } |
|
| 84 | - |
|
| 85 | - public function resolveReference(string $referenceText): ?IReference { |
|
| 86 | - if ($this->matchReference($referenceText)) { |
|
| 87 | - $reference = new Reference($referenceText); |
|
| 88 | - $this->fetchReference($reference); |
|
| 89 | - return $reference; |
|
| 90 | - } |
|
| 91 | - |
|
| 92 | - return null; |
|
| 93 | - } |
|
| 94 | - |
|
| 95 | - private function fetchReference(Reference $reference): void { |
|
| 96 | - try { |
|
| 97 | - $user = $this->userSession->getUser(); |
|
| 98 | - if ($user) { |
|
| 99 | - $this->limiter->registerUserRequest('opengraph', 10, 120, $user); |
|
| 100 | - } else { |
|
| 101 | - $this->limiter->registerAnonRequest('opengraph', 10, 120, $this->request->getRemoteAddress()); |
|
| 102 | - } |
|
| 103 | - } catch (RateLimitExceededException $e) { |
|
| 104 | - return; |
|
| 105 | - } |
|
| 106 | - |
|
| 107 | - $client = $this->clientService->newClient(); |
|
| 108 | - try { |
|
| 109 | - $headResponse = $client->head($reference->getId(), [ 'timeout' => 10 ]); |
|
| 110 | - } catch (\Exception $e) { |
|
| 111 | - $this->logger->debug('Failed to perform HEAD request to get target metadata', ['exception' => $e]); |
|
| 112 | - return; |
|
| 113 | - } |
|
| 114 | - $linkContentLength = $headResponse->getHeader('Content-Length'); |
|
| 115 | - if (is_numeric($linkContentLength) && (int) $linkContentLength > 5 * 1024 * 1024) { |
|
| 116 | - $this->logger->debug('Skip resolving links pointing to content length > 5 MB'); |
|
| 117 | - return; |
|
| 118 | - } |
|
| 119 | - $linkContentType = $headResponse->getHeader('Content-Type'); |
|
| 120 | - $expectedContentType = 'text/html'; |
|
| 121 | - $suffixedExpectedContentType = $expectedContentType . ';'; |
|
| 122 | - $startsWithSuffixed = substr($linkContentType, 0, strlen($suffixedExpectedContentType)) === $suffixedExpectedContentType; |
|
| 123 | - // check the header begins with the expected content type |
|
| 124 | - if ($linkContentType !== $expectedContentType && !$startsWithSuffixed) { |
|
| 125 | - $this->logger->debug('Skip resolving links pointing to content type that is not "text/html"'); |
|
| 126 | - return; |
|
| 127 | - } |
|
| 128 | - try { |
|
| 129 | - $response = $client->get($reference->getId(), [ 'timeout' => 10 ]); |
|
| 130 | - } catch (\Exception $e) { |
|
| 131 | - $this->logger->debug('Failed to fetch link for obtaining open graph data', ['exception' => $e]); |
|
| 132 | - return; |
|
| 133 | - } |
|
| 134 | - |
|
| 135 | - $responseBody = (string)$response->getBody(); |
|
| 136 | - |
|
| 137 | - // OpenGraph handling |
|
| 138 | - $consumer = new Consumer(); |
|
| 139 | - $consumer->useFallbackMode = true; |
|
| 140 | - $object = $consumer->loadHtml($responseBody); |
|
| 141 | - |
|
| 142 | - $reference->setUrl($reference->getId()); |
|
| 143 | - |
|
| 144 | - if ($object->title) { |
|
| 145 | - $reference->setTitle($object->title); |
|
| 146 | - } |
|
| 147 | - |
|
| 148 | - if ($object->description) { |
|
| 149 | - $reference->setDescription($object->description); |
|
| 150 | - } |
|
| 151 | - |
|
| 152 | - if ($object->images) { |
|
| 153 | - try { |
|
| 154 | - $host = parse_url($object->images[0]->url, PHP_URL_HOST); |
|
| 155 | - if ($host === false || $host === null) { |
|
| 156 | - $this->logger->warning('Could not detect host of open graph image URI for ' . $reference->getId()); |
|
| 157 | - } else { |
|
| 158 | - $appData = $this->appDataFactory->get('core'); |
|
| 159 | - try { |
|
| 160 | - $folder = $appData->getFolder('opengraph'); |
|
| 161 | - } catch (NotFoundException $e) { |
|
| 162 | - $folder = $appData->newFolder('opengraph'); |
|
| 163 | - } |
|
| 164 | - $response = $client->get($object->images[0]->url, ['timeout' => 10]); |
|
| 165 | - $contentType = $response->getHeader('Content-Type'); |
|
| 166 | - $contentLength = $response->getHeader('Content-Length'); |
|
| 167 | - |
|
| 168 | - if (in_array($contentType, self::ALLOWED_CONTENT_TYPES, true) && $contentLength < self::MAX_PREVIEW_SIZE) { |
|
| 169 | - $stream = Utils::streamFor($response->getBody()); |
|
| 170 | - $bodyStream = new LimitStream($stream, self::MAX_PREVIEW_SIZE, 0); |
|
| 171 | - $reference->setImageContentType($contentType); |
|
| 172 | - $folder->newFile(md5($reference->getId()), $bodyStream->getContents()); |
|
| 173 | - $reference->setImageUrl($this->urlGenerator->linkToRouteAbsolute('core.Reference.preview', ['referenceId' => md5($reference->getId())])); |
|
| 174 | - } |
|
| 175 | - } |
|
| 176 | - } catch (GuzzleException $e) { |
|
| 177 | - $this->logger->info('Failed to fetch and store the open graph image for ' . $reference->getId(), ['exception' => $e]); |
|
| 178 | - } catch (\Throwable $e) { |
|
| 179 | - $this->logger->error('Failed to fetch and store the open graph image for ' . $reference->getId(), ['exception' => $e]); |
|
| 180 | - } |
|
| 181 | - } |
|
| 182 | - } |
|
| 183 | - |
|
| 184 | - public function getCachePrefix(string $referenceId): string { |
|
| 185 | - return $referenceId; |
|
| 186 | - } |
|
| 187 | - |
|
| 188 | - public function getCacheKey(string $referenceId): ?string { |
|
| 189 | - return null; |
|
| 190 | - } |
|
| 46 | + public const MAX_PREVIEW_SIZE = 1024 * 1024; |
|
| 47 | + |
|
| 48 | + public const ALLOWED_CONTENT_TYPES = [ |
|
| 49 | + 'image/png', |
|
| 50 | + 'image/jpg', |
|
| 51 | + 'image/jpeg', |
|
| 52 | + 'image/gif', |
|
| 53 | + 'image/svg+xml', |
|
| 54 | + 'image/webp' |
|
| 55 | + ]; |
|
| 56 | + |
|
| 57 | + private IClientService $clientService; |
|
| 58 | + private LoggerInterface $logger; |
|
| 59 | + private SystemConfig $systemConfig; |
|
| 60 | + private IAppDataFactory $appDataFactory; |
|
| 61 | + private IURLGenerator $urlGenerator; |
|
| 62 | + private Limiter $limiter; |
|
| 63 | + private IUserSession $userSession; |
|
| 64 | + private IRequest $request; |
|
| 65 | + |
|
| 66 | + public function __construct(IClientService $clientService, LoggerInterface $logger, SystemConfig $systemConfig, IAppDataFactory $appDataFactory, IURLGenerator $urlGenerator, Limiter $limiter, IUserSession $userSession, IRequest $request) { |
|
| 67 | + $this->clientService = $clientService; |
|
| 68 | + $this->logger = $logger; |
|
| 69 | + $this->systemConfig = $systemConfig; |
|
| 70 | + $this->appDataFactory = $appDataFactory; |
|
| 71 | + $this->urlGenerator = $urlGenerator; |
|
| 72 | + $this->limiter = $limiter; |
|
| 73 | + $this->userSession = $userSession; |
|
| 74 | + $this->request = $request; |
|
| 75 | + } |
|
| 76 | + |
|
| 77 | + public function matchReference(string $referenceText): bool { |
|
| 78 | + if ($this->systemConfig->getValue('reference_opengraph', true) !== true) { |
|
| 79 | + return false; |
|
| 80 | + } |
|
| 81 | + |
|
| 82 | + return (bool)preg_match(IURLGenerator::URL_REGEX, $referenceText); |
|
| 83 | + } |
|
| 84 | + |
|
| 85 | + public function resolveReference(string $referenceText): ?IReference { |
|
| 86 | + if ($this->matchReference($referenceText)) { |
|
| 87 | + $reference = new Reference($referenceText); |
|
| 88 | + $this->fetchReference($reference); |
|
| 89 | + return $reference; |
|
| 90 | + } |
|
| 91 | + |
|
| 92 | + return null; |
|
| 93 | + } |
|
| 94 | + |
|
| 95 | + private function fetchReference(Reference $reference): void { |
|
| 96 | + try { |
|
| 97 | + $user = $this->userSession->getUser(); |
|
| 98 | + if ($user) { |
|
| 99 | + $this->limiter->registerUserRequest('opengraph', 10, 120, $user); |
|
| 100 | + } else { |
|
| 101 | + $this->limiter->registerAnonRequest('opengraph', 10, 120, $this->request->getRemoteAddress()); |
|
| 102 | + } |
|
| 103 | + } catch (RateLimitExceededException $e) { |
|
| 104 | + return; |
|
| 105 | + } |
|
| 106 | + |
|
| 107 | + $client = $this->clientService->newClient(); |
|
| 108 | + try { |
|
| 109 | + $headResponse = $client->head($reference->getId(), [ 'timeout' => 10 ]); |
|
| 110 | + } catch (\Exception $e) { |
|
| 111 | + $this->logger->debug('Failed to perform HEAD request to get target metadata', ['exception' => $e]); |
|
| 112 | + return; |
|
| 113 | + } |
|
| 114 | + $linkContentLength = $headResponse->getHeader('Content-Length'); |
|
| 115 | + if (is_numeric($linkContentLength) && (int) $linkContentLength > 5 * 1024 * 1024) { |
|
| 116 | + $this->logger->debug('Skip resolving links pointing to content length > 5 MB'); |
|
| 117 | + return; |
|
| 118 | + } |
|
| 119 | + $linkContentType = $headResponse->getHeader('Content-Type'); |
|
| 120 | + $expectedContentType = 'text/html'; |
|
| 121 | + $suffixedExpectedContentType = $expectedContentType . ';'; |
|
| 122 | + $startsWithSuffixed = substr($linkContentType, 0, strlen($suffixedExpectedContentType)) === $suffixedExpectedContentType; |
|
| 123 | + // check the header begins with the expected content type |
|
| 124 | + if ($linkContentType !== $expectedContentType && !$startsWithSuffixed) { |
|
| 125 | + $this->logger->debug('Skip resolving links pointing to content type that is not "text/html"'); |
|
| 126 | + return; |
|
| 127 | + } |
|
| 128 | + try { |
|
| 129 | + $response = $client->get($reference->getId(), [ 'timeout' => 10 ]); |
|
| 130 | + } catch (\Exception $e) { |
|
| 131 | + $this->logger->debug('Failed to fetch link for obtaining open graph data', ['exception' => $e]); |
|
| 132 | + return; |
|
| 133 | + } |
|
| 134 | + |
|
| 135 | + $responseBody = (string)$response->getBody(); |
|
| 136 | + |
|
| 137 | + // OpenGraph handling |
|
| 138 | + $consumer = new Consumer(); |
|
| 139 | + $consumer->useFallbackMode = true; |
|
| 140 | + $object = $consumer->loadHtml($responseBody); |
|
| 141 | + |
|
| 142 | + $reference->setUrl($reference->getId()); |
|
| 143 | + |
|
| 144 | + if ($object->title) { |
|
| 145 | + $reference->setTitle($object->title); |
|
| 146 | + } |
|
| 147 | + |
|
| 148 | + if ($object->description) { |
|
| 149 | + $reference->setDescription($object->description); |
|
| 150 | + } |
|
| 151 | + |
|
| 152 | + if ($object->images) { |
|
| 153 | + try { |
|
| 154 | + $host = parse_url($object->images[0]->url, PHP_URL_HOST); |
|
| 155 | + if ($host === false || $host === null) { |
|
| 156 | + $this->logger->warning('Could not detect host of open graph image URI for ' . $reference->getId()); |
|
| 157 | + } else { |
|
| 158 | + $appData = $this->appDataFactory->get('core'); |
|
| 159 | + try { |
|
| 160 | + $folder = $appData->getFolder('opengraph'); |
|
| 161 | + } catch (NotFoundException $e) { |
|
| 162 | + $folder = $appData->newFolder('opengraph'); |
|
| 163 | + } |
|
| 164 | + $response = $client->get($object->images[0]->url, ['timeout' => 10]); |
|
| 165 | + $contentType = $response->getHeader('Content-Type'); |
|
| 166 | + $contentLength = $response->getHeader('Content-Length'); |
|
| 167 | + |
|
| 168 | + if (in_array($contentType, self::ALLOWED_CONTENT_TYPES, true) && $contentLength < self::MAX_PREVIEW_SIZE) { |
|
| 169 | + $stream = Utils::streamFor($response->getBody()); |
|
| 170 | + $bodyStream = new LimitStream($stream, self::MAX_PREVIEW_SIZE, 0); |
|
| 171 | + $reference->setImageContentType($contentType); |
|
| 172 | + $folder->newFile(md5($reference->getId()), $bodyStream->getContents()); |
|
| 173 | + $reference->setImageUrl($this->urlGenerator->linkToRouteAbsolute('core.Reference.preview', ['referenceId' => md5($reference->getId())])); |
|
| 174 | + } |
|
| 175 | + } |
|
| 176 | + } catch (GuzzleException $e) { |
|
| 177 | + $this->logger->info('Failed to fetch and store the open graph image for ' . $reference->getId(), ['exception' => $e]); |
|
| 178 | + } catch (\Throwable $e) { |
|
| 179 | + $this->logger->error('Failed to fetch and store the open graph image for ' . $reference->getId(), ['exception' => $e]); |
|
| 180 | + } |
|
| 181 | + } |
|
| 182 | + } |
|
| 183 | + |
|
| 184 | + public function getCachePrefix(string $referenceId): string { |
|
| 185 | + return $referenceId; |
|
| 186 | + } |
|
| 187 | + |
|
| 188 | + public function getCacheKey(string $referenceId): ?string { |
|
| 189 | + return null; |
|
| 190 | + } |
|
| 191 | 191 | } |
@@ -79,7 +79,7 @@ discard block |
||
| 79 | 79 | return false; |
| 80 | 80 | } |
| 81 | 81 | |
| 82 | - return (bool)preg_match(IURLGenerator::URL_REGEX, $referenceText); |
|
| 82 | + return (bool) preg_match(IURLGenerator::URL_REGEX, $referenceText); |
|
| 83 | 83 | } |
| 84 | 84 | |
| 85 | 85 | public function resolveReference(string $referenceText): ?IReference { |
@@ -106,7 +106,7 @@ discard block |
||
| 106 | 106 | |
| 107 | 107 | $client = $this->clientService->newClient(); |
| 108 | 108 | try { |
| 109 | - $headResponse = $client->head($reference->getId(), [ 'timeout' => 10 ]); |
|
| 109 | + $headResponse = $client->head($reference->getId(), ['timeout' => 10]); |
|
| 110 | 110 | } catch (\Exception $e) { |
| 111 | 111 | $this->logger->debug('Failed to perform HEAD request to get target metadata', ['exception' => $e]); |
| 112 | 112 | return; |
@@ -118,7 +118,7 @@ discard block |
||
| 118 | 118 | } |
| 119 | 119 | $linkContentType = $headResponse->getHeader('Content-Type'); |
| 120 | 120 | $expectedContentType = 'text/html'; |
| 121 | - $suffixedExpectedContentType = $expectedContentType . ';'; |
|
| 121 | + $suffixedExpectedContentType = $expectedContentType.';'; |
|
| 122 | 122 | $startsWithSuffixed = substr($linkContentType, 0, strlen($suffixedExpectedContentType)) === $suffixedExpectedContentType; |
| 123 | 123 | // check the header begins with the expected content type |
| 124 | 124 | if ($linkContentType !== $expectedContentType && !$startsWithSuffixed) { |
@@ -126,13 +126,13 @@ discard block |
||
| 126 | 126 | return; |
| 127 | 127 | } |
| 128 | 128 | try { |
| 129 | - $response = $client->get($reference->getId(), [ 'timeout' => 10 ]); |
|
| 129 | + $response = $client->get($reference->getId(), ['timeout' => 10]); |
|
| 130 | 130 | } catch (\Exception $e) { |
| 131 | 131 | $this->logger->debug('Failed to fetch link for obtaining open graph data', ['exception' => $e]); |
| 132 | 132 | return; |
| 133 | 133 | } |
| 134 | 134 | |
| 135 | - $responseBody = (string)$response->getBody(); |
|
| 135 | + $responseBody = (string) $response->getBody(); |
|
| 136 | 136 | |
| 137 | 137 | // OpenGraph handling |
| 138 | 138 | $consumer = new Consumer(); |
@@ -153,7 +153,7 @@ discard block |
||
| 153 | 153 | try { |
| 154 | 154 | $host = parse_url($object->images[0]->url, PHP_URL_HOST); |
| 155 | 155 | if ($host === false || $host === null) { |
| 156 | - $this->logger->warning('Could not detect host of open graph image URI for ' . $reference->getId()); |
|
| 156 | + $this->logger->warning('Could not detect host of open graph image URI for '.$reference->getId()); |
|
| 157 | 157 | } else { |
| 158 | 158 | $appData = $this->appDataFactory->get('core'); |
| 159 | 159 | try { |
@@ -174,9 +174,9 @@ discard block |
||
| 174 | 174 | } |
| 175 | 175 | } |
| 176 | 176 | } catch (GuzzleException $e) { |
| 177 | - $this->logger->info('Failed to fetch and store the open graph image for ' . $reference->getId(), ['exception' => $e]); |
|
| 177 | + $this->logger->info('Failed to fetch and store the open graph image for '.$reference->getId(), ['exception' => $e]); |
|
| 178 | 178 | } catch (\Throwable $e) { |
| 179 | - $this->logger->error('Failed to fetch and store the open graph image for ' . $reference->getId(), ['exception' => $e]); |
|
| 179 | + $this->logger->error('Failed to fetch and store the open graph image for '.$reference->getId(), ['exception' => $e]); |
|
| 180 | 180 | } |
| 181 | 181 | } |
| 182 | 182 | } |