use DOMXPath;
use Friendica\Content\OEmbed;
use Friendica\Core\Addon;
+use Friendica\Core\Logger;
use Friendica\Database\DBA;
use Friendica\Object\Image;
-
-require_once 'include/dba.php';
+use Friendica\Util\Strings;
/**
* @brief Class with methods for extracting certain content from an url
}
$parsed_url = DBA::selectFirst('parsed_url', ['content'],
- ['url' => normalise_link($url), 'guessing' => !$no_guessing, 'oembed' => $do_oembed]
+ ['url' => Strings::normaliseLink($url), 'guessing' => !$no_guessing, 'oembed' => $do_oembed]
);
if (!empty($parsed_url['content'])) {
$data = unserialize($parsed_url['content']);
DBA::insert(
'parsed_url',
[
- 'url' => normalise_link($url), 'guessing' => !$no_guessing,
+ 'url' => Strings::normaliseLink($url), 'guessing' => !$no_guessing,
'oembed' => $do_oembed, 'content' => serialize($data),
'created' => DateTimeFormat::utcNow()
],
*/
public static function getSiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1)
{
- $a = get_app();
+ $a = \get_app();
$siteinfo = [];
}
if ($count > 10) {
- logger('Endless loop detected for ' . $url, LOGGER_DEBUG);
+ Logger::log('Endless loop detected for ' . $url, Logger::DEBUG);
return $siteinfo;
}
$siteinfo['url'] = $url;
$siteinfo['type'] = 'link';
- $data = Network::curl($url);
- if (!$data['success']) {
+ $curlResult = Network::curl($url);
+ if (!$curlResult->isSuccess()) {
return $siteinfo;
}
// If the file is too large then exit
- if ($data['info']['download_content_length'] > 1000000) {
+ if (defaults($curlResult->getInfo(), 'download_content_length', 0) > 1000000) {
return $siteinfo;
}
// If it isn't a HTML file then exit
- if (($data['info']['content_type'] != '') && !strstr(strtolower($data['info']['content_type']), 'html')) {
+ if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
return $siteinfo;
}
- $header = $data['header'];
- $body = $data['body'];
+ $header = $curlResult->getHeader();
+ $body = $curlResult->getBody();
if ($do_oembed) {
$oembed_data = OEmbed::fetchURL($url);
$siteinfo['type'] = $oembed_data->type;
}
- if (($oembed_data->type == 'link') && ($siteinfo['type'] != 'photo')) {
+ // See https://github.com/friendica/friendica/pull/5763#discussion_r217913178
+ if ($siteinfo['type'] != 'photo') {
if (isset($oembed_data->title)) {
$siteinfo['title'] = trim($oembed_data->title);
}
$charset = trim(trim(trim(array_pop($matches)), ';,'));
}
- if ($charset == '') {
- $charset = 'utf-8';
- }
+ if ($charset && strtoupper($charset) != 'UTF-8') {
+ // See https://github.com/friendica/friendica/issues/5470#issuecomment-418351211
+ $charset = str_ireplace('latin-1', 'latin1', $charset);
- if (($charset != '') && (strtoupper($charset) != 'UTF-8')) {
- logger('detected charset ' . $charset, LOGGER_DEBUG);
+ Logger::log('detected charset ' . $charset, Logger::DEBUG);
$body = iconv($charset, 'UTF-8//TRANSLIT', $body);
}
$siteinfo['type'] = 'link';
}
- if ((@$siteinfo['image'] == '') && !$no_guessing) {
+ if (empty($siteinfo['image']) && !$no_guessing) {
$list = $xpath->query('//img[@src]');
foreach ($list as $node) {
$img_tag = [];
}
}
- logger('Siteinfo for ' . $url . ' ' . print_r($siteinfo, true), LOGGER_DEBUG);
+ Logger::log('Siteinfo for ' . $url . ' ' . print_r($siteinfo, true), Logger::DEBUG);
Addon::callHooks('getsiteinfo', $siteinfo);
$complete = $schemearr["scheme"]."://".$schemearr["host"];
- if (@$schemearr["port"] != "") {
+ if (!empty($schemearr["port"])) {
$complete .= ":".$schemearr["port"];
}
- if (strpos($urlarr["path"], "/") !== 0) {
- $complete .= "/";
- }
+ if (!empty($urlarr["path"])) {
+ if (strpos($urlarr["path"], "/") !== 0) {
+ $complete .= "/";
+ }
- $complete .= $urlarr["path"];
+ $complete .= $urlarr["path"];
+ }
- if (@$urlarr["query"] != "") {
+ if (!empty($urlarr["query"])) {
$complete .= "?".$urlarr["query"];
}
- if (@$urlarr["fragment"] != "") {
+ if (!empty($urlarr["fragment"])) {
$complete .= "#".$urlarr["fragment"];
}