refactor(app): Standardize subdomain detection logic (#9751)

lonnieezell · datamweb · web-flow · commit 01a45beccff6 · 2025-10-10T02:36:50.000-05:00
* refactor(app): Standardize subdomain detection logic

* Update app/Config/Hostnames.php

Co-authored-by: Pooya Parsa &lt;pooya_parsa_dadashi@yahoo.com&gt;

* Update tests/system/Helpers/URLHelper/MiscUrlTest.php

Co-authored-by: Pooya Parsa &lt;pooya_parsa_dadashi@yahoo.com&gt;

* addressing review comments

* cs fix

* cs fix

* cs fix

* remove typo in docs ci-skip

---------

Co-authored-by: Pooya Parsa &lt;pooya_parsa_dadashi@yahoo.com&gt;
diff --git a/app/Config/Hostnames.php b/app/Config/Hostnames.php
@@ -0,0 +1,40 @@
+<?php
+
+namespace Config;
+
+class Hostnames
+{
+    // List of known two-part TLDs for subdomain extraction
+    public const TWO_PART_TLDS = [
+        'co.uk', 'org.uk', 'gov.uk', 'ac.uk', 'sch.uk', 'ltd.uk', 'plc.uk',
+        'com.au', 'net.au', 'org.au', 'edu.au', 'gov.au', 'asn.au', 'id.au',
+        'co.jp', 'ac.jp', 'go.jp', 'or.jp', 'ne.jp', 'gr.jp',
+        'co.nz', 'org.nz', 'govt.nz', 'ac.nz', 'net.nz', 'geek.nz', 'maori.nz', 'school.nz',
+        'co.in', 'net.in', 'org.in', 'ind.in', 'ac.in', 'gov.in', 'res.in',
+        'com.cn', 'net.cn', 'org.cn', 'gov.cn', 'edu.cn',
+        'com.sg', 'net.sg', 'org.sg', 'gov.sg', 'edu.sg', 'per.sg',
+        'co.za', 'org.za', 'gov.za', 'ac.za', 'net.za',
+        'co.kr', 'or.kr', 'go.kr', 'ac.kr', 'ne.kr', 'pe.kr',
+        'co.th', 'or.th', 'go.th', 'ac.th', 'net.th', 'in.th',
+        'com.my', 'net.my', 'org.my', 'edu.my', 'gov.my', 'mil.my', 'name.my',
+        'com.mx', 'org.mx', 'net.mx', 'edu.mx', 'gob.mx',
+        'com.br', 'net.br', 'org.br', 'gov.br', 'edu.br', 'art.br', 'eng.br',
+        'co.il', 'org.il', 'ac.il', 'gov.il', 'net.il', 'muni.il',
+        'co.id', 'or.id', 'ac.id', 'go.id', 'net.id', 'web.id', 'my.id',
+        'com.hk', 'edu.hk', 'gov.hk', 'idv.hk', 'net.hk', 'org.hk',
+        'com.tw', 'net.tw', 'org.tw', 'edu.tw', 'gov.tw', 'idv.tw',
+        'com.sa', 'net.sa', 'org.sa', 'gov.sa', 'edu.sa', 'sch.sa', 'med.sa',
+        'co.ae', 'net.ae', 'org.ae', 'gov.ae', 'ac.ae', 'sch.ae',
+        'com.tr', 'net.tr', 'org.tr', 'gov.tr', 'edu.tr', 'av.tr', 'gen.tr',
+        'co.ke', 'or.ke', 'go.ke', 'ac.ke', 'sc.ke', 'me.ke', 'mobi.ke', 'info.ke',
+        'com.ng', 'org.ng', 'gov.ng', 'edu.ng', 'net.ng', 'sch.ng', 'name.ng',
+        'com.pk', 'net.pk', 'org.pk', 'gov.pk', 'edu.pk', 'fam.pk',
+        'com.eg', 'edu.eg', 'gov.eg', 'org.eg', 'net.eg',
+        'com.cy', 'net.cy', 'org.cy', 'gov.cy', 'ac.cy',
+        'com.lk', 'org.lk', 'edu.lk', 'gov.lk', 'net.lk', 'int.lk',
+        'com.bd', 'net.bd', 'org.bd', 'ac.bd', 'gov.bd', 'mil.bd',
+        'com.ar', 'net.ar', 'org.ar', 'gov.ar', 'edu.ar', 'mil.ar',
+        'gob.cl', 'com.pl', 'net.pl', 'org.pl', 'gov.pl', 'edu.pl',
+        'co.ir', 'ac.ir', 'org.ir', 'id.ir', 'gov.ir', 'sch.ir', 'net.ir',
+    ];
+}
diff --git a/system/Helpers/url_helper.php b/system/Helpers/url_helper.php
@@ -17,6 +17,7 @@
 use CodeIgniter\HTTP\URI;
 use CodeIgniter\Router\Exceptions\RouterException;
 use Config\App;
+use Config\Hostnames;
 
 // CodeIgniter URL Helpers
 
@@ -534,3 +535,52 @@ function url_is(string $path): bool
         return (bool) preg_match("|^{$path}$|", $currentPath, $matches);
     }
 }
+
+if (! function_exists('parse_subdomain')) {
+    /**
+     * Parses the subdomain from the current host name.
+     *
+     * @param string|null $host The hostname to parse. If null, uses the current request's host.
+     *
+     * @return string The subdomain, or an empty string if none exists.
+     */
+    function parse_subdomain(?string $host = null): string
+    {
+        if ($host === null) {
+            $host = service('request')->getUri()->getHost();
+        }
+
+        // Handle localhost and IP addresses - they don't have subdomains
+        if ($host === 'localhost' || filter_var($host, FILTER_VALIDATE_IP)) {
+            return '';
+        }
+
+        $parts     = explode('.', $host);
+        $partCount = count($parts);
+
+        // Need at least 3 parts for a subdomain (subdomain.domain.tld)
+        // e.g., api.example.com
+        if ($partCount < 3) {
+            return '';
+        }
+
+        // Check if we have a two-part TLD (e.g., co.uk, com.au)
+        $lastTwoParts = $parts[$partCount - 2] . '.' . $parts[$partCount - 1];
+
+        if (in_array($lastTwoParts, Hostnames::TWO_PART_TLDS, true)) {
+            // For two-part TLD, need at least 4 parts for subdomain
+            // e.g., api.example.co.uk (4 parts)
+            if ($partCount < 4) {
+                return ''; // No subdomain, just domain.co.uk
+            }
+
+            // Remove the two-part TLD and domain name (last 3 parts)
+            // e.g., admin.api.example.co.uk -> admin.api
+            return implode('.', array_slice($parts, 0, $partCount - 3));
+        }
+
+        // Standard TLD: Remove TLD and domain (last 2 parts)
+        // e.g., admin.api.example.com -> admin.api
+        return implode('.', array_slice($parts, 0, $partCount - 2));
+    }
+}
diff --git a/system/Router/Attributes/Restrict.php b/system/Router/Attributes/Restrict.php
@@ -42,38 +42,6 @@
 #[Attribute(Attribute::TARGET_CLASS | Attribute::TARGET_METHOD | Attribute::IS_REPEATABLE)]
 class Restrict implements RouteAttributeInterface
 {
-    private const TWO_PART_TLDS = [
-        'co.uk', 'org.uk', 'gov.uk', 'ac.uk', 'sch.uk', 'ltd.uk', 'plc.uk',
-        'com.au', 'net.au', 'org.au', 'edu.au', 'gov.au', 'asn.au', 'id.au',
-        'co.jp', 'ac.jp', 'go.jp', 'or.jp', 'ne.jp', 'gr.jp',
-        'co.nz', 'org.nz', 'govt.nz', 'ac.nz', 'net.nz', 'geek.nz', 'maori.nz', 'school.nz',
-        'co.in', 'net.in', 'org.in', 'ind.in', 'ac.in', 'gov.in', 'res.in',
-        'com.cn', 'net.cn', 'org.cn', 'gov.cn', 'edu.cn',
-        'com.sg', 'net.sg', 'org.sg', 'gov.sg', 'edu.sg', 'per.sg',
-        'co.za', 'org.za', 'gov.za', 'ac.za', 'net.za',
-        'co.kr', 'or.kr', 'go.kr', 'ac.kr', 'ne.kr', 'pe.kr',
-        'co.th', 'or.th', 'go.th', 'ac.th', 'net.th', 'in.th',
-        'com.my', 'net.my', 'org.my', 'edu.my', 'gov.my', 'mil.my', 'name.my',
-        'com.mx', 'org.mx', 'net.mx', 'edu.mx', 'gob.mx',
-        'com.br', 'net.br', 'org.br', 'gov.br', 'edu.br', 'art.br', 'eng.br',
-        'co.il', 'org.il', 'ac.il', 'gov.il', 'net.il', 'muni.il',
-        'co.id', 'or.id', 'ac.id', 'go.id', 'net.id', 'web.id', 'my.id',
-        'com.hk', 'edu.hk', 'gov.hk', 'idv.hk', 'net.hk', 'org.hk',
-        'com.tw', 'net.tw', 'org.tw', 'edu.tw', 'gov.tw', 'idv.tw',
-        'com.sa', 'net.sa', 'org.sa', 'gov.sa', 'edu.sa', 'sch.sa', 'med.sa',
-        'co.ae', 'net.ae', 'org.ae', 'gov.ae', 'ac.ae', 'sch.ae',
-        'com.tr', 'net.tr', 'org.tr', 'gov.tr', 'edu.tr', 'av.tr', 'gen.tr',
-        'co.ke', 'or.ke', 'go.ke', 'ac.ke', 'sc.ke', 'me.ke', 'mobi.ke', 'info.ke',
-        'com.ng', 'org.ng', 'gov.ng', 'edu.ng', 'net.ng', 'sch.ng', 'name.ng',
-        'com.pk', 'net.pk', 'org.pk', 'gov.pk', 'edu.pk', 'fam.pk',
-        'com.eg', 'edu.eg', 'gov.eg', 'org.eg', 'net.eg',
-        'com.cy', 'net.cy', 'org.cy', 'gov.cy', 'ac.cy',
-        'com.lk', 'org.lk', 'edu.lk', 'gov.lk', 'net.lk', 'int.lk',
-        'com.bd', 'net.bd', 'org.bd', 'ac.bd', 'gov.bd', 'mil.bd',
-        'com.ar', 'net.ar', 'org.ar', 'gov.ar', 'edu.ar', 'mil.ar',
-        'gob.cl',
-    ];
-
     public function __construct(
         public array|string|null $environment = null,
         public array|string|null $hostname = null,
@@ -145,7 +113,7 @@ private function checkSubdomain(RequestInterface $request): void
             return;
         }
 
-        $currentSubdomain  = $this->getSubdomain($request);
+        $currentSubdomain  = parse_subdomain($request->getUri()->getHost());
         $allowedSubdomains = array_map('strtolower', (array) $this->subdomain);
 
         // If no subdomain exists but one is required
@@ -158,40 +126,4 @@ private function checkSubdomain(RequestInterface $request): void
             throw new PageNotFoundException('Access denied: subdomain is blocked.');
         }
     }
-
-    private function getSubdomain(RequestInterface $request): string
-    {
-        $host = strtolower($request->getUri()->getHost());
-
-        // Handle localhost and IP addresses - they don't have subdomains
-        if ($host === 'localhost' || filter_var($host, FILTER_VALIDATE_IP)) {
-            return '';
-        }
-
-        $parts     = explode('.', $host);
-        $partCount = count($parts);
-
-        // Need at least 3 parts for a subdomain (subdomain.domain.tld)
-        // e.g., api.example.com
-        if ($partCount < 3) {
-            return '';
-        }
-        // Check if we have a two-part TLD (e.g., co.uk, com.au)
-        $lastTwoParts = $parts[$partCount - 2] . '.' . $parts[$partCount - 1];
-        if (in_array($lastTwoParts, self::TWO_PART_TLDS, true)) {
-            // For two-part TLD, need at least 4 parts for subdomain
-            // e.g., api.example.co.uk (4 parts)
-            if ($partCount < 4) {
-                return ''; // No subdomain, just domain.co.uk
-            }
-
-            // Remove the two-part TLD and domain name (last 3 parts)
-            // e.g., admin.api.example.co.uk -> admin.api
-            return implode('.', array_slice($parts, 0, $partCount - 3));
-        }
-
-        // Standard TLD: Remove TLD and domain (last 2 parts)
-        // e.g., admin.api.example.com -> admin.api
-        return implode('.', array_slice($parts, 0, $partCount - 2));
-    }
 }
diff --git a/system/Router/RouteCollection.php b/system/Router/RouteCollection.php
@@ -1637,7 +1637,7 @@ private function checkSubdomains($subdomains): bool
         }
 
         if ($this->currentSubdomain === null) {
-            $this->currentSubdomain = $this->determineCurrentSubdomain();
+            $this->currentSubdomain = parse_subdomain($this->httpHost);
         }
 
         if (! is_array($subdomains)) {
@@ -1653,50 +1653,6 @@ private function checkSubdomains($subdomains): bool
         return in_array($this->currentSubdomain, $subdomains, true);
     }
 
-    /**
-     * Examines the HTTP_HOST to get the best match for the subdomain. It
-     * won't be perfect, but should work for our needs.
-     *
-     * It's especially not perfect since it's possible to register a domain
-     * with a period (.) as part of the domain name.
-     *
-     * @return false|string the subdomain
-     */
-    private function determineCurrentSubdomain()
-    {
-        // We have to ensure that a scheme exists
-        // on the URL else parse_url will mis-interpret
-        // 'host' as the 'path'.
-        $url = $this->httpHost;
-        if (! str_starts_with($url, 'http')) {
-            $url = 'http://' . $url;
-        }
-
-        $parsedUrl = parse_url($url);
-
-        $host = explode('.', $parsedUrl['host']);
-
-        if ($host[0] === 'www') {
-            unset($host[0]);
-        }
-
-        // Get rid of any domains, which will be the last
-        unset($host[count($host) - 1]);
-
-        // Account for .co.uk, .co.nz, etc. domains
-        if (end($host) === 'co') {
-            $host = array_slice($host, 0, -1);
-        }
-
-        // If we only have 1 part left, then we don't have a sub-domain.
-        if (count($host) === 1) {
-            // Set it to false so we don't make it back here again.
-            return false;
-        }
-
-        return array_shift($host);
-    }
-
     /**
      * Reset the routes, so that a test case can provide the
      * explicit ones needed for it.
diff --git a/tests/system/Helpers/URLHelper/MiscUrlTest.php b/tests/system/Helpers/URLHelper/MiscUrlTest.php
@@ -963,4 +963,40 @@ public function testUrlToMissingArgument(): void
 
         url_to('loginURL');
     }
+
+    #[DataProvider('provideParseSubdomain')]
+    public function testParseSubdomain(?string $host, string $expected, bool $useRequest = false): void
+    {
+        if ($useRequest) {
+            // create a request whose host will be used when passing null to parse_subdomain
+            $this->config->baseURL = 'http://sub.example.com/';
+            $this->createRequest('http://sub.example.com/');
+
+            $this->assertSame($expected, parse_subdomain(null));
+
+            return;
+        }
+
+        $this->assertSame($expected, parse_subdomain($host));
+    }
+
+    /**
+     * Provides test cases for parsing subdomains.
+     *
+     * @return array<string, array{0: string|null, 1: string, 2: bool}>
+     */
+    public static function provideParseSubdomain(): iterable
+    {
+        return [
+            'standard subdomain'          => ['api.example.com', 'api', false],
+            'multi-level subdomain'       => ['admin.api.example.com', 'admin.api', false],
+            'no subdomain (domain only)'  => ['example.com', '', false],
+            'localhost'                   => ['localhost', '', false],
+            'ipv4'                        => ['127.0.0.1', '', false],
+            'ipv6'                        => ['::1', '', false],
+            'two-part tld no subdomain'   => ['example.co.uk', '', false],
+            'two-part tld with subdomain' => ['api.example.co.uk', 'api', false],
+            'null uses request host'      => [null, 'sub', true],
+        ];
+    }
 }
diff --git a/user_guide_src/source/helpers/url_helper.rst b/user_guide_src/source/helpers/url_helper.rst
@@ -361,6 +361,21 @@ The following functions are available:
     This function works the same as :php:func:`url_title()` but it converts all
     accented characters automatically.
 
+.. php:function:: parse_subdomain($hostname)
+
+    :param  string|null  $hostname: The hostname to parse. If null, uses the current request's host.
+    :returns: The subdomain, or an empty string if none exists.
+    :rtype: string
+
+    Parses the subdomain from the given host name.
+
+    Here are some examples:
+
+    .. literalinclude:: url_helper/027.php
+
+    You can customize the list of known two-part TLDs by adding them to the
+    ``Config\Hostnames::TWO_PART_TLDS`` array.
+
 .. php:function:: prep_url([$str = ''[, $secure = false]])
 
     :param  string   $str: URL string
diff --git a/user_guide_src/source/helpers/url_helper/027.php b/user_guide_src/source/helpers/url_helper/027.php
@@ -0,0 +1,14 @@
+<?php
+
+// Outputs "blog"
+echo parse_subdomain('blog.example.com');
+
+// Outputs an empty string
+echo parse_subdomain('example.com');
+echo parse_subdomain('example.co.uk');
+
+// Outputs "shop" - correctly handles two-part TLDs
+echo parse_subdomain('shop.example.co.uk');
+
+// Outputs "shop.old"
+echo parse_subdomain('shop.old.example.co.uk');
