protected $mimeGuesser;
protected $fetchingErrorMessage;
protected $eventDispatcher;
+ protected $storeArticleHeaders;
- public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage)
+ public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage, $storeArticleHeaders = false)
{
$this->graby = $graby;
$this->tagger = $tagger;
$this->logger = $logger;
$this->mimeGuesser = new MimeTypeExtensionGuesser();
$this->fetchingErrorMessage = $fetchingErrorMessage;
+ $this->storeArticleHeaders = $storeArticleHeaders;
}
/**
$this->stockEntry($entry, $content);
}
+ /**
+ * Use a Symfony validator to ensure the language is well formatted.
+ *
+ * @param Entry $entry
+ * @param string $value Language to validate and save
+ */
+ public function updateLanguage(Entry $entry, $value)
+ {
+ // some lang are defined as fr-FR, es-ES.
+ // replacing - by _ might increase language support
+ $value = str_replace('-', '_', $value);
+
+ $errors = $this->validator->validate(
+ $value,
+ (new LocaleConstraint())
+ );
+
+ if (0 === \count($errors)) {
+ $entry->setLanguage($value);
+
+ return;
+ }
+
+ $this->logger->warning('Language validation failed. ' . (string) $errors);
+ }
+
+ /**
+ * Use a Symfony validator to ensure the preview picture is a real url.
+ *
+ * @param Entry $entry
+ * @param string $value URL to validate and save
+ */
+ public function updatePreviewPicture(Entry $entry, $value)
+ {
+ $errors = $this->validator->validate(
+ $value,
+ (new UrlConstraint())
+ );
+
+ if (0 === \count($errors)) {
+ $entry->setPreviewPicture($value);
+
+ return;
+ }
+
+ $this->logger->warning('PreviewPicture validation failed. ' . (string) $errors);
+ }
+
+ /**
+ * Update date.
+ *
+ * @param Entry $entry
+ * @param string $value Date to validate and save
+ */
+ public function updatePublishedAt(Entry $entry, $value)
+ {
+ $date = $value;
+
+ // is it a timestamp?
+ if (false !== filter_var($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);
+ }
+
/**
* Stock entry with fetched or imported content.
* Will fall back to OpenGraph data if available.
*/
private function stockEntry(Entry $entry, array $content)
{
- $title = $content['title'];
- if (!$title && !empty($content['open_graph']['og_title'])) {
- $title = $content['open_graph']['og_title'];
+ $entry->setUrl($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']);
}
$html = $content['html'];
}
}
- $entry->setUrl($content['url']);
- $entry->setTitle($title);
$entry->setContent($html);
- $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
-
- if (!empty($content['date'])) {
- $date = $content['date'];
-
- // is it a timestamp?
- if (filter_var($date, FILTER_VALIDATE_INT) !== false) {
- $date = '@' . $content['date'];
- }
+ $entry->setReadingTime(Utils::getReadingTime($html));
- try {
- $entry->setPublishedAt(new \DateTime($date));
- } catch (\Exception $e) {
- $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $content['url'], 'date' => $content['date']]);
- }
+ if (!empty($content['status'])) {
+ $entry->setHttpStatus($content['status']);
}
- if (!empty($content['authors']) && is_array($content['authors'])) {
+ if (!empty($content['authors']) && \is_array($content['authors'])) {
$entry->setPublishedBy($content['authors']);
}
- if (!empty($content['all_headers'])) {
+ if (!empty($content['all_headers']) && $this->storeArticleHeaders) {
$entry->setHeaders($content['all_headers']);
}
- $this->validateAndSetLanguage(
- $entry,
- isset($content['language']) ? $content['language'] : null
- );
+ if (!empty($content['date'])) {
+ $this->updatePublishedAt($entry, $content['date']);
+ }
- $this->validateAndSetPreviewPicture(
- $entry,
- isset($content['open_graph']['og_image']) ? $content['open_graph']['og_image'] : null
- );
+ if (!empty($content['language'])) {
+ $this->updateLanguage($entry, $content['language']);
+ }
- // if content is an image, define it as a preview too
- if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
- $this->validateAndSetPreviewPicture(
- $entry,
- $content['url']
- );
+ if (!empty($content['open_graph']['og_image'])) {
+ $this->updatePreviewPicture($entry, $content['open_graph']['og_image']);
}
- $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : '');
- $entry->setReadingTime(Utils::getReadingTime($html));
+ // if content is an image, define it as a preview too
+ if (!empty($content['content_type']) && \in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
+ $this->updatePreviewPicture($entry, $content['url']);
+ }
- $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
- if (false !== $domainName) {
- $entry->setDomainName($domainName);
+ if (!empty($content['content_type'])) {
+ $entry->setMimetype($content['content_type']);
}
try {
{
return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
}
-
- /**
- * Use a Symfony validator to ensure the language is well formatted.
- *
- * @param Entry $entry
- * @param string $value Language to validate
- */
- private function validateAndSetLanguage($entry, $value)
- {
- // some lang are defined as fr-FR, es-ES.
- // replacing - by _ might increase language support
- $value = str_replace('-', '_', $value);
-
- $errors = $this->validator->validate(
- $value,
- (new LocaleConstraint())
- );
-
- if (0 === count($errors)) {
- $entry->setLanguage($value);
-
- return;
- }
-
- $this->logger->warning('Language validation failed. ' . (string) $errors);
- }
-
- /**
- * Use a Symfony validator to ensure the preview picture is a real url.
- *
- * @param Entry $entry
- * @param string $value URL to validate
- */
- private function validateAndSetPreviewPicture($entry, $value)
- {
- $errors = $this->validator->validate(
- $value,
- (new UrlConstraint())
- );
-
- if (0 === count($errors)) {
- $entry->setPreviewPicture($value);
-
- return;
- }
-
- $this->logger->warning('PreviewPicture validation failed. ' . (string) $errors);
- }
}