Files
dredgegram/app/Services/MediaCacher.php
T
dredgy fadbc8cd5d
linter / quality (push) Canceled after 0s
tests / ci (8.3) (push) Canceled after 0s
tests / ci (8.4) (push) Canceled after 0s
tests / ci (8.5) (push) Canceled after 0s
Initial Commit
2026-08-29 14:39:18 +10:00

125 lines
4.2 KiB
PHP

<?php
namespace App\Services;
use Illuminate\Support\Facades\File;
use Illuminate\Support\Facades\Http;
use Illuminate\Support\Facades\Log;
use Illuminate\Support\Facades\Storage;
use Throwable;
/**
* Shared logic for downloading a model's Instagram-hosted media into local
* storage and rewriting its url columns to point at the cached copy.
* Instagram's CDN urls are signed and expire, so caching is a one-way trip:
* once a field holds a local url it's left alone on every later call, which
* makes cache() safe to call repeatedly. It also repairs any url that was
* cached as an absolute address before switching to relative ones, without
* re-downloading — the file is already on disk.
*/
abstract class MediaCacher
{
private const INSTAGRAM_HOST_SUFFIXES = ['fbcdn.net', 'cdninstagram.com'];
/**
* True if the url still points at Instagram's own CDN (i.e. hasn't been
* cached locally yet). Public so callers writing scraped data back to
* an existing row can check it before deciding whether to overwrite a
* field a cache pass already replaced.
*/
public function isInstagramUrl(?string $url): bool
{
if (! $url) {
return false;
}
$host = parse_url($url, PHP_URL_HOST);
if (! $host) {
return false;
}
foreach (self::INSTAGRAM_HOST_SUFFIXES as $suffix) {
if (str_ends_with($host, $suffix)) {
return true;
}
}
return false;
}
/**
* Figures out what, if anything, a url field needs to become: null if
* it's already a relative path we generated (nothing to do), a
* rewritten relative path if it's an absolute url pointing at our own
* /storage/ (cached before the relative-path fix, no download needed),
* a freshly downloaded local path if it's still on Instagram's CDN, or
* null for anything else we don't recognize.
*/
protected function resolveUrl(?string $url, string $pathWithoutExtension): ?string
{
if (! $url || str_starts_with($url, '/storage/')) {
return null;
}
if (($pos = strpos($url, '/storage/')) !== false) {
return substr($url, $pos);
}
if ($this->isInstagramUrl($url)) {
return $this->download($url, $pathWithoutExtension);
}
return null;
}
private function download(string $url, string $pathWithoutExtension): ?string
{
// Stream the response straight to disk via Guzzle's `sink` option
// instead of buffering the whole body in memory — a several-MB
// video/reel can easily exceed PHP's default memory_limit if
// fetched the normal way (Http::get()->body() reads it all into a
// string first). The extension isn't known until we see the
// response, so this writes to a temp file and renames it after.
$tempPath = $pathWithoutExtension.'.downloading-'.uniqid();
$absoluteTempPath = Storage::disk('public')->path($tempPath);
File::ensureDirectoryExists(dirname($absoluteTempPath));
try {
$response = Http::withHeaders(['User-Agent' => 'Mozilla/5.0'])
->timeout(60)
->withOptions(['sink' => $absoluteTempPath])
->get($url);
} catch (Throwable $e) {
Log::warning(static::class.": failed to fetch {$url}: {$e->getMessage()}");
@unlink($absoluteTempPath);
return null;
}
if (! $response->successful()) {
Log::warning(static::class.": got {$response->status()} fetching {$url}");
@unlink($absoluteTempPath);
return null;
}
$path = $pathWithoutExtension.'.'.$this->guessExtension($url, $response->header('Content-Type'));
Storage::disk('public')->move($tempPath, $path);
return '/storage/'.$path;
}
private function guessExtension(string $url, ?string $contentType): string
{
$fromUrl = pathinfo(parse_url($url, PHP_URL_PATH) ?? '', PATHINFO_EXTENSION);
if ($fromUrl && strlen($fromUrl) <= 4) {
return strtolower($fromUrl);
}
return str_contains((string) $contentType, 'video') ? 'mp4' : 'jpg';
}
}