use Graby\Graby;
use Psr\Log\LoggerInterface;
+use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
+use Symfony\Component\Validator\Constraints\Locale as LocaleConstraint;
+use Symfony\Component\Validator\Constraints\Url as UrlConstraint;
+use Symfony\Component\Validator\Validator\ValidatorInterface;
use Wallabag\CoreBundle\Entity\Entry;
use Wallabag\CoreBundle\Tools\Utils;
-use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
/**
* This kind of proxy class take care of getting the content from an url
{
protected $graby;
protected $tagger;
+ protected $validator;
protected $logger;
protected $mimeGuesser;
protected $fetchingErrorMessage;
protected $eventDispatcher;
- public function __construct(Graby $graby, RuleBasedTagger $tagger, LoggerInterface $logger, $fetchingErrorMessage)
+ public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage)
{
$this->graby = $graby;
$this->tagger = $tagger;
+ $this->validator = $validator;
$this->logger = $logger;
$this->mimeGuesser = new MimeTypeExtensionGuesser();
$this->fetchingErrorMessage = $fetchingErrorMessage;
}
/**
- * Fetch content using graby and hydrate given entry with results information.
- * In case we couldn't find content, we'll try to use Open Graph data.
- *
- * We can also force the content, in case of an import from the v1 for example, so the function won't
- * fetch the content from the website but rather use information given with the $content parameter.
+ * Update entry using either fetched or provided content.
*
- * @param Entry $entry Entry to update
- * @param string $url Url to grab content for
- * @param array $content An array with AT LEAST keys title, html, url, language & content_type to skip the fetchContent from the url
- *
- * @return Entry
+ * @param Entry $entry Entry to update
+ * @param string $url Url of the content
+ * @param array $content Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url
+ * @param bool $disableContentUpdate Whether to skip trying to fetch content using Graby
*/
- public function updateEntry(Entry $entry, $url, array $content = [])
+ public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false)
{
- // do we have to fetch the content or the provided one is ok?
- if (empty($content) || false === $this->validateContent($content)) {
+ if (!empty($content['html'])) {
+ $content['html'] = $this->graby->cleanupHtml($content['html'], $url);
+ }
+
+ if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) {
$fetchedContent = $this->graby->fetchContent($url);
// when content is imported, we have information in $content
}
}
+ // be sure to keep the url in case of error
+ // so we'll be able to refetch it in the future
+ $content['url'] = !empty($content['url']) ? $content['url'] : $url;
+
+ $this->stockEntry($entry, $content);
+ }
+
+ /**
+ * 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)
+ {
$title = $content['title'];
- if (!$title && isset($content['open_graph']['og_title'])) {
+ if (!$title && !empty($content['open_graph']['og_title'])) {
$title = $content['open_graph']['og_title'];
}
if (false === $html) {
$html = $this->fetchingErrorMessage;
- if (isset($content['open_graph']['og_description'])) {
+ if (!empty($content['open_graph']['og_description'])) {
$html .= '<p><i>But we found a short description: </i></p>';
$html .= $content['open_graph']['og_description'];
}
}
- $entry->setUrl($content['url'] ?: $url);
+ $entry->setUrl($content['url']);
$entry->setTitle($title);
$entry->setContent($html);
$entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
- if (isset($content['date']) && null !== $content['date'] && '' !== $content['date']) {
- $entry->setPublishedAt(new \DateTime($content['date']));
+ if (!empty($content['date'])) {
+ $date = $content['date'];
+
+ // is it a timestamp?
+ if (filter_var($date, FILTER_VALIDATE_INT) !== false) {
+ $date = '@' . $content['date'];
+ }
+
+ 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['authors'])) {
+ if (!empty($content['authors']) && is_array($content['authors'])) {
$entry->setPublishedBy($content['authors']);
}
$entry->setHeaders($content['all_headers']);
}
- $entry->setLanguage(isset($content['language']) ? $content['language'] : '');
+ $this->validateAndSetLanguage(
+ $entry,
+ isset($content['language']) ? $content['language'] : null
+ );
+
+ $this->validateAndSetPreviewPicture(
+ $entry,
+ isset($content['open_graph']['og_image']) ? $content['open_graph']['og_image'] : null
+ );
+
+ // if content is an image, define it as a preview too
+ if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
+ $this->validateAndSetPreviewPicture(
+ $entry,
+ $content['url']
+ );
+ }
+
$entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : '');
$entry->setReadingTime(Utils::getReadingTime($html));
$entry->setDomainName($domainName);
}
- if (isset($content['open_graph']['og_image']) && $content['open_graph']['og_image']) {
- $entry->setPreviewPicture($content['open_graph']['og_image']);
- }
-
- // if content is an image define as a preview too
- if (isset($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
- $entry->setPreviewPicture($content['url']);
- }
-
try {
$this->tagger->tag($entry);
} catch (\Exception $e) {
$this->logger->error('Error while trying to automatically tag an entry.', [
- 'entry_url' => $url,
+ 'entry_url' => $content['url'],
'error_msg' => $e->getMessage(),
]);
}
-
- return $entry;
}
/**
- * Validate that the given content as enough value to be used
- * instead of fetch the content from the url.
+ * Validate that the given content has at least a title, an html and a url.
*
* @param array $content
*
*/
private function validateContent(array $content)
{
- return isset($content['title']) && isset($content['html']) && isset($content['url']) && isset($content['language']) && isset($content['content_type']);
+ 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);
}
}
projects / github / wallabag / wallabag.git / blobdiff