+ // is it a timestamp?
+ if (false !== filter_var($date, FILTER_VALIDATE_INT)) {
+ $date = '@' . $date;
+ }
+
+ try {
+ // is it already a DateTime?
+ // (it's inside the try/catch in case of fail to be parse time string)
+ if (!$date instanceof \DateTime) {
+ $date = new \DateTime($date);
+ }
+
+ $entry->setPublishedAt($date);
+ } catch (\Exception $e) {
+ $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $entry->getUrl(), 'date' => $value]);
+ }
+ }
+
+ /**
+ * Helper to extract and save host from entry url.
+ *
+ * @param Entry $entry
+ */
+ public function setEntryDomainName(Entry $entry)
+ {
+ $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
+ if (false !== $domainName) {
+ $entry->setDomainName($domainName);
+ }
+ }
+
+ /**
+ * Helper to set a default title using:
+ * - url basename, if applicable
+ * - hostname.
+ *
+ * @param Entry $entry
+ */
+ public function setDefaultEntryTitle(Entry $entry)
+ {
+ $url = parse_url($entry->getUrl());
+ $path = pathinfo($url['path'], PATHINFO_BASENAME);
+
+ if (empty($path)) {
+ $path = $url['host'];
+ }
+
+ $entry->setTitle($path);
+ }
+
+ /**
+ * Try to sanitize the title of the fetched content from wrong character encodings and invalid UTF-8 character.
+ *
+ * @param $title
+ * @param $contentType
+ *
+ * @return string
+ */
+ private function sanitizeContentTitle($title, $contentType)
+ {
+ if ('application/pdf' === $contentType) {
+ $title = $this->convertPdfEncodingToUTF8($title);
+ }
+
+ return $this->sanitizeUTF8Text($title);
+ }
+
+ /**
+ * If the title from the fetched content comes from a PDF, then its very possible that the character encoding is not
+ * UTF-8. This methods tries to identify the character encoding and translate the title to UTF-8.
+ *
+ * @param $title
+ *
+ * @return string (maybe contains invalid UTF-8 character)
+ */
+ private function convertPdfEncodingToUTF8($title)
+ {
+ // first try UTF-8 because its easier to detect its present/absence
+ foreach (['UTF-8', 'UTF-16BE', 'WINDOWS-1252'] as $encoding) {
+ if (mb_check_encoding($title, $encoding)) {
+ return mb_convert_encoding($title, 'UTF-8', $encoding);
+ }
+ }
+
+ return $title;
+ }
+
+ /**
+ * Remove invalid UTF-8 characters from the given string.
+ *
+ * @param string $rawText
+ *
+ * @return string
+ */
+ private function sanitizeUTF8Text($rawText)
+ {
+ if (mb_check_encoding($rawText, 'UTF-8')) {
+ return $rawText;
+ }
+
+ return iconv('UTF-8', 'UTF-8//IGNORE', $rawText);
+ }
+
+ /**
+ * Stock entry with fetched or imported content.
+ * Will fall back to OpenGraph data if available.
+ *
+ * @param Entry $entry Entry to stock
+ * @param array $content Array with at least title, url & html
+ */
+ private function stockEntry(Entry $entry, array $content)
+ {
+ $this->updateOriginUrl($entry, $content['url']);
+
+ $this->setEntryDomainName($entry);
+
+ if (!empty($content['title'])) {
+ $entry->setTitle($content['title']);
+ } elseif (!empty($content['open_graph']['og_title'])) {
+ $entry->setTitle($content['open_graph']['og_title']);