[Feature] Exclude links from broken link check (#56)

* Exclude links from broken link check * Fix styling --------- Co-authored-by: Baspa <[email protected]>
vormkracht10 · Jan 26, 2024 · 2f98cbf · 2f98cbf
1 parent 4432444
commit 2f98cbf
Show file tree

Hide file tree

Showing 3 changed files with 56 additions and 9 deletions.
diff --git a/config/seo.php b/config/seo.php
@@ -188,5 +188,11 @@
  'broken_link_check' => [
  // Add status codes that should be considered as broken links. Empty array means all status codes starting with a 4, 5 or 0.
  'status_codes' => [],
+
+ // If you wish to skip running some checks on some paths, list the paths
+ // in the array below. You can use wildcards.
+ 'exclude_paths' => [
+ //
+ ],
  ],
 ];
diff --git a/src/Checks/Content/BrokenLinkCheck.php b/src/Checks/Content/BrokenLinkCheck.php
@@ -53,15 +53,7 @@ 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 or tel
- if (preg_match('/^mailto:/msi', $link) ||
- preg_match('/^tel:/msi', $link) ||
- filter_var($link, FILTER_VALIDATE_URL) === false
- ) {
- return false;
- }
-
- return $link;
+ return $this->isValidLink($link) && ! $this->isExcludedLink($link);
  })
  ->filter(function ($link) {
  return isBrokenLink($link) ? $link : false;
@@ -89,4 +81,38 @@ public function validateContent(Crawler $crawler): bool
 
  return true;
  }
+
+ private function isValidLink($link): bool
+ {
+ return ! preg_match('/^mailto:/msi', $link) &&
+ ! preg_match('/^tel:/msi', $link) &&
+ filter_var($link, FILTER_VALIDATE_URL) !== false;
+ }
+
+ private function isExcludedLink($link): bool
+ {
+ $excludedPaths = config('seo.broken_link_check.exclude_paths');
+ if (empty($excludedPaths)) {
+ return false;
+ }
+
+ foreach ($excludedPaths as $path) {
+ if ($this->linkMatchesPath($link, $path)) {
+ return true;
+ }
+ }
+
+ return false;
+ }
+
+ private function linkMatchesPath($link, $path): bool
+ {
+ if (str_contains($path, '*')) {
+ $path = str_replace('/*', '', $path);
+
+ return str_starts_with($link, $path);
+ }
+
+ return str_contains($link, $path);
+ }
 }
diff --git a/tests/Checks/Content/BrokenLinkCheckTest.php b/tests/Checks/Content/BrokenLinkCheckTest.php
@@ -106,3 +106,18 @@
 
  $this->assertTrue($check->check(Http::get('vormkracht10.nl/admin/dashboard'), $crawler));
 });
+
+it('can exclude certain paths from the broken link check', function () {
+ $check = new BrokenLinkCheck();
+ $crawler = new Crawler();
+
+ config(['seo.broken_link_check.exclude_paths' => ['https://vormkracht10.nl/excluded']]);
+
+ Http::fake([
+ 'vormkracht10.nl' => Http::response('<html><head></head><body><a href="https://vormkracht10.nl/excluded">Excluded Link</a></body></html>', 200),
+ ]);
+
+ $crawler->addHtmlContent(Http::get('vormkracht10.nl')->body());
+
+ $this->assertTrue($check->check(Http::get('vormkracht10.nl'), $crawler));
+});