From cb3622128fccec6e7fef9a688b232f10da510c07 Mon Sep 17 00:00:00 2001
From: Bas van Dinther <bas@vormkracht10.nl>
Date: Fri, 13 Jan 2023 13:18:43 +0100
Subject: [PATCH] Check if a relative url is used when performing broken link
 checks (#17)

* Check if relative url and try to build an absolute url

* Fix styling

* Add tests

* Fix styling

Co-authored-by: Baspa <Baspa@users.noreply.github.com>
---
 src/Checks/Content/BrokenImageCheck.php      |  1 +
 src/Checks/Content/BrokenLinkCheck.php       | 27 +++++++++++---------
 src/Seo.php                                  |  4 +++
 src/Traits/PerformCheck.php                  |  4 +++
 src/helpers.php                              | 19 ++++++++++++++
 tests/Checks/Content/BrokenLinkCheckTest.php | 13 ++++++++++
 6 files changed, 56 insertions(+), 12 deletions(-)

diff --git a/src/Checks/Content/BrokenImageCheck.php b/src/Checks/Content/BrokenImageCheck.php
index a5dae2c4..852cb2ed 100644
--- a/src/Checks/Content/BrokenImageCheck.php
+++ b/src/Checks/Content/BrokenImageCheck.php
@@ -47,6 +47,7 @@ public function validateContent(Crawler $crawler): bool
         }
 
         $content = collect($content)->filter(fn ($value) => $value !== null)
+            ->map(fn ($link) => addBaseIfRelativeUrl($link, $this->url))
             ->filter(fn ($link) => isBrokenLink($link))->toArray();
 
         $this->actualValue = $content;
diff --git a/src/Checks/Content/BrokenLinkCheck.php b/src/Checks/Content/BrokenLinkCheck.php
index 76772b99..ebcfe8d3 100644
--- a/src/Checks/Content/BrokenLinkCheck.php
+++ b/src/Checks/Content/BrokenLinkCheck.php
@@ -46,18 +46,21 @@ public function validateContent(Crawler $crawler): bool
             return true;
         }
 
-        $content = collect($content)->filter(fn ($value) => $value !== null)->filter(function ($link) {
-            // Filter out all links that are mailto, tel or have a file extension
-            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;
-            }
-
-            return $link;
-        })->filter(fn ($link) => isBrokenLink($link))->toArray();
+        $content = collect($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
+                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;
+                }
+
+                return $link;
+            })
+            ->filter(fn ($link) => isBrokenLink($link))->toArray();
 
         $this->actualValue = $content;
 
diff --git a/src/Seo.php b/src/Seo.php
index 5e8c72f0..50009bd0 100755
--- a/src/Seo.php
+++ b/src/Seo.php
@@ -19,6 +19,8 @@ class Seo
      */
     public ProgressBar|null $progress;
 
+    public string $url;
+
     public function __construct(
         protected Http $http,
         protected Collection $successful,
@@ -29,6 +31,7 @@ public function __construct(
     public function check(string $url, ProgressBar|null $progress = null): SeoScore
     {
         $this->progress = $progress;
+        $this->url = $url;
 
         try {
             $response = $this->visitPage(url: $url);
@@ -66,6 +69,7 @@ private function runChecks(Response $response): void
                 'checks' => $checks,
                 'progress' => $this->progress,
                 'crawler' => $crawler,
+                'url' => $this->url,
             ])
             ->through($checks->keys()->toArray())
             ->then(function ($data) {
diff --git a/src/Traits/PerformCheck.php b/src/Traits/PerformCheck.php
index 46431e6b..b4200dc7 100644
--- a/src/Traits/PerformCheck.php
+++ b/src/Traits/PerformCheck.php
@@ -6,8 +6,12 @@
 
 trait PerformCheck
 {
+    public string|null $url = null;
+
     public function __invoke(array $data, Closure $next)
     {
+        $this->url = $data['url'] ?? null;
+
         if (! in_array('exit', $data)) {
             $result = $this->check($data['response'], $data['crawler']);
         }
diff --git a/src/helpers.php b/src/helpers.php
index 173fce9c..34826c26 100644
--- a/src/helpers.php
+++ b/src/helpers.php
@@ -135,3 +135,22 @@ function bytesToHumanReadable(int $bytes): string
         return round($bytes / (1000 ** $i), 2).' '.$units[$i];
     }
 }
+
+if (! function_exists('addBaseIfRelativeUrl')) {
+    function addBaseIfRelativeUrl(string $url, string|null $checkedUrl = null): string
+    {
+        if (! Str::startsWith($url, '/')) {
+            return $url;
+        }
+
+        if (config('app.url')) {
+            return config('app.url').$url;
+        }
+
+        if ($checkedUrl) {
+            return $checkedUrl.$url;
+        }
+
+        return $url;
+    }
+}
diff --git a/tests/Checks/Content/BrokenLinkCheckTest.php b/tests/Checks/Content/BrokenLinkCheckTest.php
index 5eb2cd71..2f4c3871 100644
--- a/tests/Checks/Content/BrokenLinkCheckTest.php
+++ b/tests/Checks/Content/BrokenLinkCheckTest.php
@@ -42,3 +42,16 @@
 
     $this->assertTrue($check->check(Http::get('vormkracht10.nl'), $crawler));
 });
+
+it('can run the broken link check on a relative url', function () {
+    $check = new BrokenLinkCheck();
+    $crawler = new Crawler();
+
+    Http::fake([
+        'vormkracht10.nl' => Http::response('<html><head></head><body><a href="/404">Vormkracht10</a></body></html>', 200),
+    ]);
+
+    $crawler->addHtmlContent(Http::get('vormkracht10.nl')->body());
+
+    $this->assertFalse($check->check(Http::get('vormkracht10.nl'), $crawler));
+});