GNUsocial/gnu-social
4
0
Fork 1
Code Issues Releases Wiki Activity

[ATTACHMENTS][EVENT] Add onHashFile event, which can be used to deduplicate files

Browse Source 
Currently, we simply hash the contents of the file with sha256, but in the future we can use something smarter,
which could find visual feature similarity between images
This commit is contained in:
Hugo Sales
2021-04-27 20:53:59 +00:00
parent 60a9085e56
commit 34059a8d3d
2 changed files with 20 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
src/Core/GSFile.php
View File

@@ -43,9 +43,9 @@ class GSFile
bool $is_local = true,
int $actor_id = null): Attachment
{
Event::handle('HashFile', [$file->getPathname(), &$hash]);
// The following properly gets the mimetype with `file` or other
// available methods, so should be safe
$hash = hash_file(Attachment::FILEHASH_ALGO, $sfile->getPathname());
$mimetype = $sfile->getMimeType();
Event::handle('AttachmentValidation', [&$sfile, &$mimetype]);
$attachment = Attachment::create([
@@ -70,16 +70,17 @@ class GSFile
public static function validateAndStoreURL(string $url): Attachment
{
if (Common::isValidHttpUrl($url)) {
HTTPClient::head($url);
$head = HTTPClient::head($url);
$headers = $head->getHeaders();
$headers = array_change_key_case($headers, CASE_LOWER);
$attachment = Attachment::create([
'remote_url' => $match[0],
'remote_url_hash' => hash('sha256', $match[0]),
'mimetype' => $headers['content-type'],
'remote_url' => $url,
'remote_url_hash' => hash(Attachment::URLHASH_ALGO, $url),
'mimetype' => $headers['content-type'][0],
'is_local' => false,
]);
DB::persist($attachment);
Event::handle('AttachmentStoreNew', [&$at]);
Event::handle('AttachmentStoreNew', [&$attachment]);
return $attachment;
} else {
throw new \InvalidArgumentException();