URLs not excluded from static cache in a multi-site environment

[aerni]

Bug description

I've noticed that the static:warm command isn't excluding URLs of sites that use a domain that's different from APP_URL.

It's important to note that this issue doesn't actually affect the statically cached pages. It only affects the URLs that are considered for warming by the static:warm command.

Example

Consider the following setup:

config/statamic/sites.php

'sites' => [

    'gounity' => [
        'name' => 'goUnity',
        'locale' => 'de_DE',
        'url' => env('APP_URL_GOUNITY'),
    ],

    'gomagazin' => [
        'name' => 'goMagazin',
        'locale' => 'de_DE',
        'url' => env('APP_URL_GOMAGAZIN'),
    ],

],

config/statamic/static_caching.php

'exclude' => [
    'urls' => [
        '/story/*',
    ],
],

.env

APP_URL_GOUNITY=https://go-unity.test
APP_URL_GOMAGAZIN=https://gomagazin.test
APP_URL="${APP_URL_GOUNITY}"

When calling the static:warm command, it will correctly exclude the /story/* URLs in the default gounity site. However, it will still warm all the URLs in the gomagazin site.

What's happening

The URLs that should be excluded from the static:warm command are rejected here:

cms/src/Console/Commands/StaticWarm.php

Lines 170 to 172 in 024aad4

	->reject(function ($uri) use ($cacher) {
	return $cacher->isExcluded($uri);
	})

The issue is that the baseUrl of the DefaultUrlExcluder class will resolve to the URL of the current request as per the getBaseUrl() method in the AbstractCacher class.

cms/src/StaticCaching/Cachers/AbstractCacher.php

Lines 49 to 62 in 024aad4

	public function getBaseUrl()
	{
	// Check 'base_url' for backward compatibility.
	if (! $baseUrl = $this->config('base_url')) {
	// This could potentially just be Site::current()->absoluteUrl() but at the
	// moment that method gets the URL based on the request. For now, we will
	// manually get it from the config, as to not break any existing sites.
	$baseUrl = Str::startsWith($url = Site::current()->url(), '/')
	? Str::removeRight(config('app.url'), '/').$url
	: $url;
	}
	return rtrim($baseUrl, '/');
	}

The $baseUrl will always be Site::current(), which is resolved from request()->getUri(). When running the static:warm command, request()->getUri() will resolve to the APP_URL and not the URL of the site that the URL that is about to be cached belongs to.

cms/src/Sites/Sites.php

Lines 72 to 77 in 024aad4

	private function findByCurrentUrl()
	{
	return $this->findByUrl(
	$this->currentUrlCallback ? call_user_func($this->currentUrlCallback) : request()->getUri()
	);
	}

And this leads to the isExcluded() method returning false on any site that doesn't use the APP_URL as base.

cms/src/StaticCaching/DefaultUrlExcluder.php

Lines 28 to 47 in 024aad4

	public function isExcluded(string $url): bool
	{
	// Query strings should be ignored.
	$url = explode('?', $url)[0];
	$url = Str::removeLeft($url, $this->baseUrl);
	foreach ($this->exclusions as $excluded) {
	if (Str::endsWith($excluded, '*')) {
	$prefix = Str::removeRight($excluded, '*');
	if (Str::startsWith($url, $prefix) && ! (Str::endsWith($prefix, '/') && $url === $prefix)) {
	return true;
	}
	} elseif (Str::removeRight($url, '/') === Str::removeRight($excluded, '/')) {
	return true;
	}
	}
	return false;

Workaround

You can work around this issue by explicitly excluding absolute URLs.

'exclude' => [
    'urls' => [
        '/story/*',
        env('APP_URL_GOMAGAZIN') . '/story/*',
    ],
],

How to reproduce

1. Create a multi-site with different domains
2. Exclude a URL from the static cache
3. Run the static:warm command
4. Notice the URL only being excluded in the default site

Logs

No response

Environment

Environment
Application Name: goUnity
Laravel Version: 10.48.10
PHP Version: 8.3.7
Composer Version: 2.7.6
Environment: local
Debug Mode: ENABLED
URL: go-unity.test
Maintenance Mode: OFF

Cache
Config: NOT CACHED
Events: NOT CACHED
Routes: NOT CACHED
Views: CACHED

Drivers
Broadcasting: log
Cache: statamic
Database: mysql
Logs: stack / single, flare
Mail: smtp
Queue: redis
Session: file

Locales
Installed
LaravelLang\Actions\Plugin: 1.8.2
LaravelLang\Attributes\Plugin: 2.10.5
LaravelLang\HttpStatuses\Plugin: 3.8.3
LaravelLang\Lang\Plugin: 13.12.0
Locales Version: 1.9.0
Protected
Publisher Version: 15.0.2

Statamic
Addons: 11
Antlers: runtime
Sites: 2 (goUnity, goMagazin)
Stache Watcher: Enabled
Static Caching: full
Version: 4.58.2 PRO

Statamic Addons
aerni/advanced-seo: 2.4.7
aerni/livewire-forms: 8.1.0
aerni/social-links: 3.2.0
jacksleight/statamic-bard-texstyle: 3.2.2
jonassiewertsen/statamic-jobs: 1.5.0
jonassiewertsen/statamic-livewire: 2.12.0
marcorieser/tailwind-merge-statamic: 1.1.0
rias/statamic-redirect: 3.7.1
spatie/statamic-responsive-images: 4.1.1
stillat/antlers-components: 2.2.0
stillat/relationships: 2.2.0

Installation

Fresh statamic/statamic site via CLI

Additional details

This issue sounds very similar to #7456.

[jasonvarga]

Thanks for the detailed info. 👍

[aerni]

No worries. I just edited the issue, as it wasn't quite correct. It should be good now.

I also found this related issue #4732 and this PR #5065

[aerni]

I already worked out a fix. Will PR soon.