* </body>
* @endverbatim
*/
- public static function getSiteinfo(string $url, bool $do_oembed = true, int $count = 1)
+ public static function getSiteinfo(string $url, bool $do_oembed = true, int $count = 1): array
{
if (empty($url)) {
return [
];
if ($count > 10) {
- Logger::notice('Endless loop detected', ['url' => $url]);
+ Logger::warning('Endless loop detected', ['url' => $url]);
return $siteinfo;
}
// Expected form: Content-Type: text/html; charset=ISO-8859-4
if (preg_match('/charset=([a-z0-9-_.\/]+)/i', $curlResult->getContentType(), $matches)) {
$charset = trim(trim(trim(array_pop($matches)), ';,'));
+ } else {
+ // Then in body that gets precedence
+ // Expected forms:
+ // - <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
+ // - <meta charset="utf-8">
+ // - <meta charset=utf-8>
+ // - <meta charSet="utf-8">
+ // We escape <style> and <script> tags since they can contain irrelevant charset information
+ // (see https://github.com/friendica/friendica/issues/9251#issuecomment-698636806)
+ Strings::performWithEscapedBlocks($body, '#<(?:style|script).*?</(?:style|script)>#ism', function ($body) use (&$charset) {
+ if (preg_match('/charset=["\']?([a-z0-9-_.\/]+)/i', $body, $matches)) {
+ $charset = trim(trim(trim(array_pop($matches)), ';,'));
+ }
+ });
}
- // Then in body that gets precedence
- // Expected forms:
- // - <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
- // - <meta charset="utf-8">
- // - <meta charset=utf-8>
- // - <meta charSet="utf-8">
- // We escape <style> and <script> tags since they can contain irrelevant charset information
- // (see https://github.com/friendica/friendica/issues/9251#issuecomment-698636806)
- Strings::performWithEscapedBlocks($body, '#<(?:style|script).*?</(?:style|script)>#ism', function ($body) use (&$charset) {
- if (preg_match('/charset=["\']?([a-z0-9-_.\/]+)/i', $body, $matches)) {
- $charset = trim(trim(trim(array_pop($matches)), ';,'));
- }
- });
-
$siteinfo['charset'] = $charset;
if ($charset && strtoupper($charset) != 'UTF-8') {
* Convert tags from CSV to an array
*
* @param string $string Tags
+ *
* @return array with formatted Hashtags
*/
public static function convertTagsToArray(string $string): array
*
* @param string $tag The pure tag name
* @param int $k Counter for internal use
+ *
* @return void
*/
private static function arrAddHashes(string &$tag, int $k)
*
* @param array $siteinfo
* @param array $jsonld
+ *
* @return array siteinfo
*/
private static function parseParts(array $siteinfo, array $jsonld): array
*
* @param array $siteinfo
* @param array $jsonld
+ *
* @return array siteinfo
*/
private static function parseJsonLd(array $siteinfo, array $jsonld): array
*
* @param array $siteinfo
* @param array $jsonld
+ *
* @return array siteinfo
*/
private static function parseJsonLdAuthor(array $siteinfo, array $jsonld): array
*
* @param array $siteinfo
* @param array $jsonld
+ *
* @return array siteinfo
*/
private static function parseJsonLdArticle(array $siteinfo, array $jsonld): array
*
* @param array $siteinfo
* @param array $jsonld
+ *
* @return array siteinfo
*/
private static function parseJsonLdWebPage(array $siteinfo, array $jsonld): array
}
$content = JsonLD::fetchElement($jsonld, 'description');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['text'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'image');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['image'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'thumbnailUrl');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['image'] = trim($content);
}
*
* @param array $siteinfo
* @param array $jsonld
+ *
* @return array siteinfo
*/
private static function parseJsonLdWebSite(array $siteinfo, array $jsonld): array
$jsonldinfo = [];
$content = JsonLD::fetchElement($jsonld, 'name');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['publisher_name'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'description');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['publisher_description'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'url');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['publisher_url'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'thumbnailUrl');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['image'] = trim($content);
}
*
* @param array $siteinfo
* @param array $jsonld
+ *
* @return array siteinfo
*/
private static function parseJsonLdWebOrganization(array $siteinfo, array $jsonld): array
$jsonldinfo = [];
$content = JsonLD::fetchElement($jsonld, 'name');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['publisher_name'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'description');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['publisher_description'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'url');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['publisher_url'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'logo', 'url', '@type', 'ImageObject');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['publisher_img'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'brand', 'name', '@type', 'Organization');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['publisher_name'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'brand', 'url', '@type', 'Organization');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['publisher_url'] = trim($content);
}
*
* @param array $siteinfo
* @param array $jsonld
+ *
* @return array siteinfo
*/
private static function parseJsonLdWebPerson(array $siteinfo, array $jsonld): array
$jsonldinfo = [];
$content = JsonLD::fetchElement($jsonld, 'name');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['author_name'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'description');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['author_description'] = trim($content);
}
}
$content = JsonLD::fetchElement($jsonld, 'url');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$jsonldinfo['author_url'] = trim($content);
}
*
* @param array $siteinfo
* @param array $jsonld
+ *
* @return array siteinfo
*/
private static function parseJsonLdMediaObject(array $siteinfo, array $jsonld, string $name): array
$media = [];
$content = JsonLD::fetchElement($jsonld, 'caption');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$media['caption'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'url');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$media['url'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'mainEntityOfPage');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$media['main'] = Strings::compareLink($content, $siteinfo['url']);
}
$content = JsonLD::fetchElement($jsonld, 'description');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$media['description'] = trim($content);
}
}
$content = JsonLD::fetchElement($jsonld, 'contentUrl');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$media['content'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'embedUrl');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$media['embed'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'height');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$media['height'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'width');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$media['width'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'image');
- if (!empty($content)) {
+ if (!empty($content) && is_string($content)) {
$media['image'] = trim($content);
}