[Embed] Retrieve remote thumbnails and other improvements

This commit is contained in:
Diogo Peralta Cordeiro 2021-08-13 19:36:34 +01:00 committed by Hugo Sales
parent 7a0a6f1f22
commit fb28a3656a
Signed by: someonewithpc
GPG Key ID: 7D0C7EAFC9D835A0
2 changed files with 163 additions and 258 deletions

View File

@ -52,7 +52,6 @@ use App\Util\Common;
use App\Util\Exception\DuplicateFoundException; use App\Util\Exception\DuplicateFoundException;
use App\Util\Exception\NotFoundException; use App\Util\Exception\NotFoundException;
use App\Util\Exception\ServerException; use App\Util\Exception\ServerException;
use App\Util\Exception\TemporaryFileException;
use App\Util\Formatting; use App\Util\Formatting;
use App\Util\TemporaryFile; use App\Util\TemporaryFile;
use Embed\Embed as LibEmbed; use Embed\Embed as LibEmbed;
@ -92,6 +91,7 @@ class Embed extends Plugin
* *
* @return bool * @return bool
* *
*
*/ */
public function onAddRoute(RouteLoader $m): bool public function onAddRoute(RouteLoader $m): bool
{ {
@ -103,17 +103,16 @@ class Embed extends Plugin
/** /**
* Insert oembed and opengraph tags in all HTML head elements * Insert oembed and opengraph tags in all HTML head elements
*/ */
public function onShowHeadElements(Request $request, array &$result) public function onShowHeadElements(Request $request, array &$result): bool
{ {
$matches = []; $matches = [];
preg_match(',/?([^/]+)/?(.*),', $request->getPathInfo(), $matches); preg_match(',/?([^/]+)/?(.*),', $request->getPathInfo(), $matches);
switch ($matches[1]) { $url = match ($matches[1]) {
case 'attachment': 'attachment' => "{$matches[1]}/{$matches[2]}",
$url = "{$matches[1]}/{$matches[2]}"; default => null,
break; };
}
if (isset($url)) { if (is_null($url)) {
foreach (['xml', 'json'] as $format) { foreach (['xml', 'json'] as $format) {
$result[] = [ $result[] = [
'link' => [ 'link' => [
@ -127,6 +126,36 @@ class Embed extends Plugin
return Event::next; return Event::next;
} }
/**
* Show this attachment enhanced with the corresponding Embed data, if available
*
* @param array $vars
* @param array $res
*
* @return bool
*/
public function onViewRemoteUrl(array $vars, array &$res): bool
{
$remote_url = $vars['remote_url'];
try {
$embed = Cache::get('attachment-embed-' . $remote_url->getId(),
fn () => DB::findOneBy('attachment_embed', ['remoteurl_id' => $remote_url->getId()]));
} catch (DuplicateFoundException $e) {
Log::warning($e);
return Event::next;
} catch (NotFoundException) {
Log::debug("Embed doesn\\'t have a representation for the remote_url id={$remote_url->getId()}. Must have been stored before the plugin was enabled.");
return Event::next;
}
$attributes = $embed->getImageHTMLAttributes(['class' => 'u-photo embed']);
$res[] = Formatting::twigRenderFile('embed/embedView.html.twig',
['embed' => $embed, 'attributes' => $attributes, 'remote_url' => $remote_url]);
return Event::stop;
}
/** /**
* Save embedding information for an Attachment, if applicable. * Save embedding information for an Attachment, if applicable.
* *
@ -134,15 +163,14 @@ class Embed extends Plugin
* @param Note $note * @param Note $note
* *
* @throws DuplicateFoundException * @throws DuplicateFoundException
* @throws ServerException
* @throws TemporaryFileException
* *
* @return bool * @return bool
*/ */
public function onNewRemoteURLFromNote(RemoteURL $remote_url, Note $note): bool public function onNewRemoteURLFromNote(RemoteURL $remote_url, Note $note): bool
{ {
// Only handle text mime // Only handle text mime
if ($remote_url->getMimetypeMajor() !== 'text') { $mimetype = $remote_url->getMimetype();
if (!(Formatting::startsWith($mimetype, 'text/html') || Formatting::startsWith($mimetype, 'application/xhtml+xml'))) {
return Event::next; return Event::next;
} }
@ -152,21 +180,41 @@ class Embed extends Plugin
return Event::next; return Event::next;
} }
$mimetype = $remote_url->getMimetype(); // If an attachment already exist, do not create an Embed for it. Some other plugin must have done things
$remote_url_to_attachment = DB::find('remoteurl_to_attachment', ['remoteurl_id' => $remote_url->getId()]);
if (Formatting::startsWith($mimetype, 'text/html') || Formatting::startsWith($mimetype, 'application/xhtml+xml')) { if (!is_null($remote_url_to_attachment)) {
$attachment_id = $remote_url_to_attachment->getAttachmentId();
try { try {
$embed_data = $this->getEmbed($remote_url->getRemoteUrl()); $attachment = DB::findOneBy('attachment', ['id' => $attachment_id]);
$attachment->livesIncrementAndGet();
return Event::next;
} catch (DuplicateFoundException | NotFoundException $e) {
Log::error($e);
}
}
// Create an Embed representation for this URL
$embed_data = $this->getEmbedLibMetadata($remote_url->getRemoteUrl());
$embed_data['remoteurl_id'] = $remote_url->getId(); $embed_data['remoteurl_id'] = $remote_url->getId();
// Create attachment $img_data = $this->downloadThumbnail($embed_data['thumbnail_url']);
switch ($img_data) {
case null: // URL isn't usable
$embed_data['thumbnail_url'] = null;
// no break
case false: // Thumbnail isn't acceptable
DB::persist($attachment = Attachment::create(['mimetype' => $remote_url->getMimetype()]));
Event::handle('AttachmentStoreNew', [&$attachment]);
break;
default: // String is valid image data
$temp_file = new TemporaryFile();
$temp_file->write($img_data);
$attachment = GSFile::sanitizeAndStoreFileAsAttachment($temp_file);
$embed_data['attachment_id'] = $attachment->getId();
}
$embed_data['attachment_id'] = $attachment->getId(); $embed_data['attachment_id'] = $attachment->getId();
DB::persist(Entity\AttachmentEmbed::create($embed_data)); DB::persist(Entity\AttachmentEmbed::create($embed_data));
DB::flush(); DB::flush();
} catch (Exception $e) { return Event::stop;
Log::warning($e);
}
}
return Event::next;
} }
/** /**
@ -181,11 +229,8 @@ class Embed extends Plugin
* *
* @return array * @return array
*/ */
public function getEmbed(string $url): array private function getEmbedLibMetadata(string $url): array
{ {
Log::info('Checking for remote URL metadata for ' . $url);
try {
Log::info("Trying to find Embed data for {$url} with 'oscarotero/Embed'"); Log::info("Trying to find Embed data for {$url} with 'oscarotero/Embed'");
$embed = new LibEmbed(); $embed = new LibEmbed();
$info = $embed->get($url); $info = $embed->get($url);
@ -204,12 +249,8 @@ class Embed extends Plugin
// Check thumbnail URL validity // Check thumbnail URL validity
$metadata['thumbnail_url'] = $thumbnail_url; $metadata['thumbnail_url'] = $thumbnail_url;
} catch (Exception $e) {
Log::info("Failed to find Embed data for {$url} with 'oscarotero/Embed', got exception: " . $e->getMessage());
}
$metadata = self::normalize($metadata); return self::normalizeEmbedLibMetadata($metadata);
return $metadata;
} }
/** /**
@ -219,7 +260,7 @@ class Embed extends Plugin
* *
* @return array * @return array
*/ */
public static function normalize(array $metadata): array private static function normalizeEmbedLibMetadata(array $metadata): array
{ {
if (isset($metadata['thumbnail_url'])) { if (isset($metadata['thumbnail_url'])) {
// sometimes sites serve the path, not the full URL, for images // sometimes sites serve the path, not the full URL, for images
@ -241,81 +282,13 @@ class Embed extends Plugin
} }
/** /**
* Show this attachment enhanced with the corresponding Embed data, if available * @param string $url
* *
* @param array $vars * @return bool true if allowed by the lists, false otherwise
* @param array $res
*
* @return bool
*/ */
public function onViewAttachmentText(array $vars, array &$res): bool private function allowedRemoteUrl(string $url): bool
{
$attachment = $vars['attachment'];
try {
$embed = Cache::get('attachment-embed-' . $attachment->getId(),
fn () => DB::findOneBy('attachment_embed', ['attachment_id' => $attachment->getId()]));
} catch (DuplicateFoundException $e) {
Log::warning($e);
return Event::next;
} catch (NotFoundException) {
return Event::next;
}
if (is_null($embed) && empty($embed->getAuthorName()) && empty($embed->getProviderName())) {
Log::debug('Embed doesn\'t have a representation for the attachment #' . $attachment->getId());
return Event::next;
}
$attributes = $embed->getImageHTMLAttributes(['class' => 'u-photo embed']);
$res[] = Formatting::twigRenderString(<<<END
<article class="h-entry embed">
<header>
{% if attributes != false %}
<img class="u-photo embed" width="{{attributes['width']}}" height="{{attributes['height']}}" src="{{attributes['src']}}" />
{% endif %}
<h5 class="p-name embed">
<a class="u-url" href="{{attachment.getRemoteUrl()}}">{{embed.getTitle() | escape}}</a>
</h5>
<div class="p-author embed">
{% if embed.getAuthorName() is not null %}
<div class="fn vcard author">
{% if embed.getAuthorUrl() is null %}
<p>{{embed.getAuthorName()}}</p>
{% else %}
<a href="{{embed.getAuthorUrl()}}" class="url">{{embed.getAuthorName()}}</a>
{% endif %}
</div>
{% endif %}
{% if embed.getProviderName() is not null %}
<div class="fn vcard">
{% if embed.getProviderUrl() is null %}
<p>{{embed.getProviderName()}}</p>
{% else %}
<a href="{{embed.getProviderUrl()}}" class="url">{{embed.getProviderName()}}</a>
{% endif %}
</div>
{% endif %}
</div>
</header>
<div class="p-summary embed">
{{ embed.getHtml() | escape }}
</div>
</article>
END, ['embed' => $embed, 'attributes' => $attributes, 'attachment' => $attachment]);
return Event::stop;
}
/**
* @throws ServerException if check is made but fails
*
* @return bool false on no check made, provider name on success
* @return false|string on no check made, provider name on success
*
*
*/
protected function checkWhitelist(string $url): string | bool
{ {
return true;
if ($this->check_whitelist ?? false) { if ($this->check_whitelist ?? false) {
return false; // indicates "no check made" return false; // indicates "no check made"
} }
@ -327,156 +300,53 @@ END, ['embed' => $embed, 'attributes' => $attributes, 'attachment' => $attachmen
} }
} }
throw new ServerException(_m('Domain not in remote thumbnail source whitelist: {host}', ['host' => $host])); return false;
} }
/** /**
* Check the file size of a remote file using a HEAD request and checking * Private helper that:
* the content-length variable returned. This isn't 100% foolproof but is * - checks if given URL is valid and is in fact an image (basic test), returns null if not;
* reliable enough for our purposes. * - checks if respects file quota and whitelist/blacklist, returns false if not;
* - downloads the thumbnail, returns a string if successful.
* *
* @param string $url * @param string $remote_url URL to the remote thumbnail
* @param null|array $headers - if we already made a request
* *
* @return null|int the file size if it succeeds, false otherwise. * @return null|bool|string
*/ */
private function getRemoteFileSize(string $url, ?array $headers = null): ?int private function downloadThumbnail(string $remote_url): bool|string|null
{ {
try { // Is this a valid URL?
if ($headers === null) { if (!Common::isValidHttpUrl($remote_url)) {
if (!Common::isValidHttpUrl($url)) { Log::debug("Invalid URL ({$remote_url}) in Embed->downloadThumbnail.");
Log::error('Invalid URL in Embed::getRemoteFileSize()'); return null;
}
// Is this URL trusted?
if (!$this->allowedRemoteUrl($remote_url)) {
Log::info("Blocked URL ({$remote_url}) in Embed->downloadThumbnail.");
return false; return false;
} }
$head = HTTPClient::head($url);
// Validate if the URL really does point to a remote image
$head = HTTPClient::head($remote_url);
$headers = $head->getHeaders(); $headers = $head->getHeaders();
$headers = array_change_key_case($headers, CASE_LOWER); $headers = array_change_key_case($headers, CASE_LOWER);
} if (empty($headers['content-type']) || GSFile::mimetypeMajor($headers['content-type'][0]) !== 'image') {
return $headers['content-length'][0] ?? false; Log::debug("URL ({$remote_url}) doesn't point to an image (content-type: " . (!empty($headers['content-type'][0]) ? $headers['content-type'][0] : 'not available') . ') in Embed->downloadThumbnail.');
} catch (Exception $e) { return null;
Loog::error($e);
return false;
}
} }
/** // Does it respect the file quota?
* A private helper function that uses a HEAD request to check the mimetype $file_size = $headers['content-length'][0];
* of a remote URL to see if it's an image.
*
* @param mixed $url
* @param null|mixed $headers
*
* @return bool true if the remote URL is an image, or false otherwise.
*/
private function isRemoteImage(string $url, ?array $headers = null): bool
{
try {
if ($headers === null) {
if (!Common::isValidHttpUrl($url)) {
Log::error('Invalid URL in Embed::getRemoteFileSize()');
return false;
}
$head = HTTPClient::head($url);
$headers = $head->getHeaders();
$headers = array_change_key_case($headers, CASE_LOWER);
}
return !empty($headers['content-type']) && GSFile::mimetypeMajor($headers['content-type'][0]) === 'image';
} catch (Exception $e) {
Log::error($e);
return false;
}
}
/**
* Validate that $imgData is a valid image, place it in its folder and resize
*
* @param $imgData - The image data to validate
* @param null|array $headers - The headers possible previous request to $url
*/
protected function validateAndWriteImage($imgData, ?array $headers = null): array
{
$file = new TemporaryFile();
$file->write($imgData);
Event::handle('HashFile', [$file->getRealPath(), &$hash]);
$filepath = Common::config('storage', 'dir') . "embed/{$hash}" . Common::config('thumbnail', 'extension');
$width = Common::config('plugin_embed', 'width');
$height = Common::config('plugin_embed', 'height');
$smart_crop = Common::config('plugin_embed', 'smart_crop');
Event::handle('ResizeImagePath', [$file->getRealPath(), $filepath, &$width, &$height, $smart_crop, &$mimetype]);
unset($file);
if (!is_null($headers) && array_key_exists('content-disposition', $headers) && preg_match('/^.+; filename="(.+?)"$/', $headers['content-disposition'][0], $matches) === 1) {
$original_name = $matches[1];
}
return [$filepath, $width, $height, $original_name ?? null, $mimetype];
}
/**
* Fetch, Validate and Write a remote image from url to temporary file
*
* @param Attachment $attachment
* @param string $media_url URL for the actual media representation
*
* @throws Exception
*
* @return array|bool
*/
protected function fetchValidateWriteRemoteImage(Attachment $attachment, string $media_url): array | bool
{
if ($attachment->hasFilename() && file_exists($attachment->getPath())) {
throw new AlreadyFulfilledException(_m('A thumbnail seems to already exist for remote file with id=={id}', ['id' => $attachment->getId()]));
}
if (Formatting::startsWith($media_url, 'file://')) {
$filename = Formatting::removePrefix($media_url, 'file://');
$info = getimagesize($filename);
$filename = basename($filename);
$width = $info[0];
$height = $info[1];
} else {
$this->checkWhitelist($media_url);
$head = HTTPClient::head($media_url);
$headers = $head->getHeaders();
$headers = array_change_key_case($headers, CASE_LOWER);
try {
$is_image = $this->isRemoteImage($media_url, $headers);
if ($is_image == true) {
$file_size = $this->getRemoteFileSize($media_url, $headers);
$max_size = Common::config('attachments', 'file_quota'); $max_size = Common::config('attachments', 'file_quota');
if (($file_size != false) && ($file_size > $max_size)) { if ($file_size > $max_size) {
throw new \Exception("Wanted to store remote thumbnail of size {$file_size} but the upload limit is {$max_size} so we aborted."); Log::debug("Went to download remote thumbnail of size {$file_size} but the upload limit is {$max_size} so we aborted in Embed->downloadThumbnail.");
}
} else {
return false; return false;
} }
} catch (Exception $err) {
Log::debug('Could not determine size of remote image, aborted local storage.');
throw $err;
}
// First we download the file to memory and test whether it's actually an image file // Download and return the file
Log::debug('Downloading remote thumbnail for file id==' . $attachment->getId() . " with thumbnail URL: {$media_url}"); Log::debug("Downloading remote thumbnail from URL: {$remote_url} in Embed->downloadThumbnail.");
return HTTPClient::get($remote_url)->getContent();
try {
$imgData = HTTPClient::get($media_url)->getContent();
if (isset($imgData)) {
[$filepath, $width, $height, $original_name, $mimetype] = $this->validateAndWriteImage($imgData, $headers);
} else {
throw new UnsupportedMediaException(_m('HTTPClient returned an empty result'));
}
} catch (UnsupportedMediaException $e) {
// Couldn't find anything that looks like an image, nothing to do
Log::debug($e);
return false;
}
}
return [$filepath, $width, $height, $original_name, $mimetype];
} }
/** /**
@ -485,6 +355,8 @@ END, ['embed' => $embed, 'attributes' => $attributes, 'attachment' => $attachmen
* *
* @param &$versions array inherited from parent * @param &$versions array inherited from parent
* *
* @throws ServerException
*
* @return bool true hook value * @return bool true hook value
*/ */
public function onPluginVersion(array &$versions): bool public function onPluginVersion(array &$versions): bool

View File

@ -0,0 +1,33 @@
<article class="h-entry embed">
<header>
{% if attributes != false %}
<img class="u-photo embed" width="{{attributes['width']}}" height="{{attributes['height']}}" src="{{attributes['src']}}" />
{% endif %}
<h5 class="p-name embed">
<a class="u-url" href="{{attachment.getRemoteUrl()}}">{{embed.getTitle() | escape}}</a>
</h5>
<div class="p-author embed">
{% if embed.getAuthorName() is not null %}
<div class="fn vcard author">
{% if embed.getAuthorUrl() is null %}
<p>{{embed.getAuthorName()}}</p>
{% else %}
<a href="{{embed.getAuthorUrl()}}" class="url">{{embed.getAuthorName()}}</a>
{% endif %}
</div>
{% endif %}
{% if embed.getProviderName() is not null %}
<div class="fn vcard">
{% if embed.getProviderUrl() is null %}
<p>{{embed.getProviderName()}}</p>
{% else %}
<a href="{{embed.getProviderUrl()}}" class="url">{{embed.getProviderName()}}</a>
{% endif %}
</div>
{% endif %}
</div>
</header>
<div class="p-summary embed">
{{ embed.getHtml() | escape }}
</div>
</article>