fix: Get rid of non standar characters for a domain.

Author: adrorocker

src/Host/Host.php

@@ -47,7 +47,7 @@ public function tld(?string $tld = null): string|self
         $root = str_replace(strval($this->tld), '', strval($this->host));
         $root = explode('.', trim(strval($root), '.'));
 
-        $this->domain = end($root) . '.' . $this->tld;
+        $this->domain(end($root) . '.' . $this->tld);
 
         return $this;
     }
@@ -63,6 +63,11 @@ public function domain(?string $domain = null): string|self
         return $this;
     }
 
+    public function isValid(): bool
+    {
+        return !empty($this->tld());
+    }
+
     public function toString(): string
     {
         return strval($this->host);

src/Parse/PublicSuffixListParser.php

@@ -15,6 +15,23 @@ public static function parse(string $publicSuffixListPath): array
         unset($list[0]);
         unset($list[count($list)]);
 
-        return array_values($list);
+        $list = array_values($list);
+
+        $fn = function (string $tld) {
+            return boolval(
+                preg_match(
+                    "/^([a-z\d](-*[a-z\d])*)(\.([a-z\d](-*[a-z\d])*))*$/i",
+                    $tld
+                )
+            );
+        };
+
+        foreach ($list as $key => $tld) {
+            if ($fn($tld) === false) {
+                unset($list[$key]);
+            }
+        }
+
+        return $list;
     }
 }

src/Validator.php

@@ -25,8 +25,11 @@ public function validate(string $host): Host
     {
         $host = new Host($host);
 
-        $parts = $host->exploded();
-        $host->tld($this->getTld($parts));
+        $tld = $this->getTld($host->exploded());
+
+        if ($this->verifyTld($tld)) {
+            $host->tld($tld);
+        }
 
         return $host;
     }
@@ -54,11 +57,22 @@ protected function getTld(array $parts, ?string $tld = null): string
         unset($parts[count($parts) - 1]);
 
         foreach ($this->publicSuffixList as $key => $item) {
-            if (strpos($item, $current) !== false) {
+            if (preg_match('/\b' . preg_quote($current) . '\b/i', $item) === 1) {
                 return $this->getTld($parts, $current);
             }
         }
 
         return strval($tld);
     }
+
+    protected function verifyTld(string $tld): bool
+    {
+        foreach ($this->publicSuffixList as $key => $item) {
+            if (preg_match('/\b' . preg_quote($tld) . '\b/i', $item) === 1) {
+                return true;
+            }
+        }
+
+        return false;
+    }
 }

tests/Unit/FactoryTest.php

@@ -5,9 +5,7 @@
 
 
 test('factory make', function () {
-    $validator = Factory::make(
-        getPublicSuffixListPath()
-    );
+    $validator = getInstance();
 
     expect($validator)->toBeInstanceOf(Validator::class);
 });

tests/Unit/ValidatorTest.php

@@ -8,8 +8,17 @@
     $host = $validator->validate($url);
 
     it('is instance of host', fn () =>  expect($host)->toBeInstanceOf(Host::class));
+    it('is a valid domain', fn () =>  expect($host->isValid())->toBeTrue());
     it("is same as $url", fn () =>  expect($host->original())->toBe($url));
     it('com.mx is the TLD', fn () =>  expect($host->tld())->toBe('com.mx'));
     it('adro.com.mx is the domain', fn () =>  expect($host->domain())->toBe('adro.com.mx'));
     it('www.adro.com.mx is the host', fn () =>  expect($host->toString())->toBe('www.adro.com.mx'));
 });
+
+describe('Validator fail', function () {
+    $validator = getInstance();
+    $url = 'https://adro.is.a.rocker.and/he-rocks?or=not';
+    $host = $validator->validate($url);
+
+    it('is not a valid domain', fn () =>  expect($host->isValid())->toBeFalse());
+});