Friendica/friendica-github
1
0
Fork 0
mirror of https://github.com/friendica/friendica synced 2024-12-23 08:40:16 +00:00
Code Issues Projects Releases Packages Wiki Activity

Check media links when fetching page data

Browse source
This commit is contained in:
Michael 2021-03-21 14:24:42 +00:00
parent fd1ae3f1fb
commit 2a68ad9b26
1 changed files with 112 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

129
src/Util/ParseUrl.php
View file

@ -478,20 +478,14 @@ class ParseUrl
$siteinfo['type'] = 'link';
}
if (!empty($siteinfo['image']) && empty($siteinfo['images'])) {
$src = self::completeUrl($siteinfo['image'], $url);
if (!empty($siteinfo['image'])) {
$siteinfo['images'] = $siteinfo['images'] ?? [];
array_unshift($siteinfo['images'], ['url' => $siteinfo['image']]);
unset($siteinfo['image']);
$photodata = Images::getInfoFromURLCached($src);
if (($photodata) && ($photodata[0] > 10) && ($photodata[1] > 10)) {
$siteinfo['images'][] = ['src' => $src,
'width' => $photodata[0],
'height' => $photodata[1]];
}
}
$siteinfo = self::checkMedia($url, $siteinfo);
if (!empty($siteinfo['text']) && mb_strlen($siteinfo['text']) > self::MAX_DESC_COUNT) {
$siteinfo['text'] = mb_substr($siteinfo['text'], 0, self::MAX_DESC_COUNT) . '…';
$pos = mb_strrpos($siteinfo['text'], '.');
@ -507,6 +501,98 @@ class ParseUrl
return $siteinfo;
}
/**
* Check the attached media elements.
* Fix existing data and add missing data.
*
* @param string $page_url
* @param array $siteinfo
* @return void
*/
private static function checkMedia(string $page_url, array $siteinfo)
{
if (!empty($siteinfo['images'])) {
array_walk($siteinfo['images'], function (&$image) use ($page_url) {
// According to the specifications someone could place a picture url into the content field as well.
// But this doesn't seem to happen in the wild, so we don't cover it here.
$image['url'] = self::completeUrl($image['url'], $page_url);
$photodata = Images::getInfoFromURLCached($image['url']);
if (!empty($photodata) && ($photodata[0] > 50) && ($photodata[1] > 50)) {
$image['src'] = $image['url'];
$image['width'] = $photodata[0];
$image['height'] = $photodata[1];
$image['contenttype'] = $photodata['mime'];
unset($image['url']);
ksort($image);
} else {
$image = [];
}
});
$siteinfo['images'] = array_values(array_filter($siteinfo['images']));
}
foreach (['audio', 'video'] as $element) {
if (!empty($siteinfo[$element])) {
array_walk($siteinfo[$element], function (&$media) use ($page_url, &$siteinfo) {
$url = '';
$embed = '';
$content = '';
$contenttype = '';
foreach (['embed', 'content', 'url'] as $field) {
if (!empty($media[$field])) {
$media[$field] = self::completeUrl($media[$field], $page_url);
$type = self::getContentType($media[$field]);
if ($type[0] == 'text') {
if ($field == 'embed') {
$embed = $media[$field];
} else {
$url = $media[$field];
}
} elseif (!empty($type[0])) {
$content = $media[$field];
$contenttype = implode('/', $type);
}
}
unset($media[$field]);
}
foreach (['image', 'preview'] as $field) {
if (!empty($media[$field])) {
$media[$field] = self::completeUrl($media[$field], $page_url);
}
}
if (!empty($url)) {
$media['url'] = $url;
}
if (!empty($embed)) {
$media['embed'] = $embed;
if (!empty($media['main'])) {
$siteinfo['embed'] = $embed;
}
}
if (!empty($content)) {
$media['src'] = $content;
}
if (!empty($contenttype)) {
$media['contenttype'] = $contenttype;
}
if (empty($url) && empty($content) && empty($embed)) {
$media = [];
}
ksort($media);
});
$siteinfo[$element] = array_values(array_filter($siteinfo[$element]));
}
if (empty($siteinfo[$element])) {
unset($siteinfo[$element]);
}
}
return $siteinfo;
}
/**
* Convert tags from CSV to an array
*
@ -1058,7 +1144,12 @@ class ParseUrl
$content = JsonLD::fetchElement($jsonld, 'url');
if (!empty($content)) {
$media['src'] = trim($content);
$media['url'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'mainEntityOfPage');
if (!empty($content)) {
$media['main'] = Strings::compareLink($content, $siteinfo['url']);
}
$content = JsonLD::fetchElement($jsonld, 'description');
@ -1091,16 +1182,20 @@ class ParseUrl
$media['width'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'thumbnailUrl');
if (!empty($content)) {
$media['preview'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'image');
if (!empty($content)) {
$media['image'] = trim($content);
}
$content = JsonLD::fetchElement($jsonld, 'thumbnailUrl');
if (!empty($content) && (($media['image'] ?? '') != trim($content))) {
if (!empty($media['image'])) {
$media['preview'] = trim($content);
} else {
$media['image'] = trim($content);
}
}
Logger::info('Fetched Media information', ['url' => $siteinfo['url'], 'fetched' => $media]);
$siteinfo[$name][] = $media;
return $siteinfo;