[Feature] Domain custom resolve (#22)

* initial commit for support custom domain resolving * Fix styling * wip * Use correct config * Test data * wip * Make curlopt_resolve option array * wip * wip * wip * Disable cache temporarily * Fix styling * wip * Fix styling * revert changes * Fix styling * Check if ip is not null * test * wip * Remove file extension validating as it is not working as expected * Remove test code * Add tests * Fix styling --------- Co-authored-by: markvaneijk <[email protected]> Co-authored-by: Bas van Dinther <[email protected]> Co-authored-by: Baspa <[email protected]>
vormkracht10 · May 29, 2023 · 45d24a1 · 45d24a1
1 parent 453625d
commit 45d24a1
Show file tree

Hide file tree

Showing 5 changed files with 76 additions and 2 deletions.
diff --git a/composer.json b/composer.json
@@ -20,6 +20,7 @@
         "guzzlehttp/guzzle": "^7.5",
         "illuminate/contracts": "^9.0|^10.0",
         "j0k3r/php-readability": "^2.0",
+        "jeremykendall/php-domain-parser": "^6.3",
         "spatie/laravel-package-tools": "^1.13",
         "symfony/dom-crawler": "^6.2",
         "symfony/finder": "^6.2",

diff --git a/config/seo.php b/config/seo.php
@@ -78,6 +78,20 @@
         'vapor-ui/*',
     ],
 
+    /*
+    |--------------------------------------------------------------------------
+    | Domains (DNS resolving)
+    |--------------------------------------------------------------------------
+    |
+    | Here you can add domains and a corresponding IP address
+    | Can be used for example to bypass certain DNS layers like the Cloudflare proxy,
+    | or resolve a domain to localhost.
+    |
+    */
+    'resolve' => [
+        'example.com' => '127.0.0.1',
+    ],
+
     /*
     |--------------------------------------------------------------------------
     | Database

diff --git a/src/Checks/Content/BrokenLinkCheck.php b/src/Checks/Content/BrokenLinkCheck.php
@@ -49,10 +49,9 @@ public function validateContent(Crawler $crawler): bool
         $content = collect($content)->filter(fn ($value) => $value !== null)
             ->map(fn ($link) => addBaseIfRelativeUrl($link, $this->url))
             ->filter(function ($link) {
-                // Filter out all links that are mailto, tel or have a file extension
+                // Filter out all links that are mailto or tel
                 if (preg_match('/^mailto:/msi', $link) ||
                     preg_match('/^tel:/msi', $link) ||
-                    preg_match('/\.[a-z]{2,4}$/msi', $link) ||
                     filter_var($link, FILTER_VALIDATE_URL) === false
                 ) {
                     return false;

diff --git a/src/helpers.php b/src/helpers.php
@@ -30,6 +30,7 @@ function getRemoteStatus(string $url): int
             $options = [
                 CURLOPT_RETURNTRANSFER => true,
                 CURLOPT_HEADER => true,
+                CURLOPT_NOBODY => true,
                 CURLOPT_TIMEOUT => 30,
                 CURLOPT_FOLLOWLOCATION => true,
             ];
@@ -44,6 +45,18 @@ function getRemoteStatus(string $url): int
                 $options[CURLOPT_SSL_VERIFYSTATUS] = false;
             }
 
+            $domain = parse_url($url, PHP_URL_HOST);
+
+            if (in_array($domain, array_keys(config('seo.resolve')))) {
+                $port = str_contains($url, 'https://') ? 443 : 80;
+
+                $ipAddress = config('seo.resolve')[$domain];
+
+                if (! empty($ipAddress)) {
+                    $options[CURLOPT_RESOLVE] = ["{$domain}:{$port}:{$ipAddress}"];
+                }
+            }
+
             curl_setopt_array($handle, $options);
             curl_exec($handle);
 
@@ -70,6 +83,7 @@ function getRemoteFileSize(string $url): int
             $handle = curl_init($url);
 
             if (! $handle) {
+
                 return 0;
             }
 
@@ -90,6 +104,18 @@ function getRemoteFileSize(string $url): int
                 $options[CURLOPT_SSL_VERIFYSTATUS] = false;
             }
 
+            $domain = parse_url($url, PHP_URL_HOST);
+
+            if (in_array($domain, array_keys(config('seo.resolve')))) {
+                $port = str_contains($url, 'https://') ? 443 : 80;
+
+                $ipAddress = config('seo.resolve')[$domain];
+
+                if (! empty($ipAddress)) {
+                    $options[CURLOPT_RESOLVE] = ["{$domain}:{$port}:{$ipAddress}"];
+                }
+            }
+
             curl_setopt_array($handle, $options);
 
             $data = curl_exec($handle);

diff --git a/tests/Checks/Content/BrokenLinkCheckTest.php b/tests/Checks/Content/BrokenLinkCheckTest.php
@@ -55,3 +55,37 @@
 
     $this->assertFalse($check->check(Http::get('vormkracht10.nl'), $crawler));
 });
+
+it('can bypass DNS layers using DNS resolving', function () {
+    $check = new BrokenLinkCheck();
+    $crawler = new Crawler();
+
+    Http::fake([
+        'vormkracht10.nl' => Http::response('<html><head></head><body><a href="https://vormkracht10.nl">Vormkracht10</a></body></html>', 200),
+    ]);
+
+    $crawler->addHtmlContent(Http::get('vormkracht10.nl')->body());
+
+    config(['seo.resolve' => [
+        'vormkracht10.nl' => '240.0.0.0',
+    ]]);
+
+    $this->assertFalse($check->check(Http::get('vormkracht10.nl'), $crawler));
+});
+
+it('cannot bypass DNS layers using a fake IP when DNS resolving', function () {
+    $check = new BrokenLinkCheck();
+    $crawler = new Crawler();
+
+    config(['seo.resolve' => [
+        'vormkracht10.nl' => '8.8.8.8',
+    ]]);
+
+    Http::fake([
+        'vormkracht10.nl' => Http::response('<html><head></head><body><a href="https://vormkracht10.nl">Vormkracht10</a></body></html>', 200),
+    ]);
+
+    $crawler->addHtmlContent(Http::get('vormkracht10.nl')->body());
+
+    $this->assertTrue($check->check(Http::get('vormkracht10.nl'), $crawler));
+});