Move URL into separate file, handle source URL validation for sending webmentions

composer.lock

@@ -1024,16 +1024,16 @@
     "packages-dev": [
         {
             "name": "myclabs/deep-copy",
-            "version": "1.11.0",
+            "version": "1.11.1",
             "source": {
                 "type": "git",
                 "url": "https://github.com/myclabs/DeepCopy.git",
-                "reference": "14daed4296fae74d9e3201d2c4925d1acb7aa614"
+                "reference": "7284c22080590fb39f2ffa3e9057f10a4ddd0e0c"
             },
             "dist": {
                 "type": "zip",
-                "url": "https://api.github.com/repos/myclabs/DeepCopy/zipball/14daed4296fae74d9e3201d2c4925d1acb7aa614",
+                "url": "https://api.github.com/repos/myclabs/DeepCopy/zipball/7284c22080590fb39f2ffa3e9057f10a4ddd0e0c",
-                "reference": "14daed4296fae74d9e3201d2c4925d1acb7aa614",
+                "reference": "7284c22080590fb39f2ffa3e9057f10a4ddd0e0c",
                 "shasum": ""
             },
             "require": {
@@ -1071,7 +1071,7 @@
             ],
             "support": {
                 "issues": "https://github.com/myclabs/DeepCopy/issues",
-                "source": "https://github.com/myclabs/DeepCopy/tree/1.11.0"
+                "source": "https://github.com/myclabs/DeepCopy/tree/1.11.1"
             },
             "funding": [
                 {
@@ -1079,7 +1079,7 @@
                     "type": "tidelift"
                 }
             ],
-            "time": "2022-03-03T13:19:32+00:00"
+            "time": "2023-03-08T13:26:56+00:00"
         },
         {
             "name": "nikic/php-parser",
@@ -1568,16 +1568,16 @@
         },
         {
             "name": "phpunit/phpunit",
-            "version": "10.0.14",
+            "version": "10.0.16",
             "source": {
                 "type": "git",
                 "url": "https://github.com/sebastianbergmann/phpunit.git",
-                "reference": "7065dbebcb0f66cf16a45fc9cfc28c2351e06169"
+                "reference": "07d386a11ac7094032900f07cada1c8975d16607"
             },
             "dist": {
                 "type": "zip",
-                "url": "https://api.github.com/repos/sebastianbergmann/phpunit/zipball/7065dbebcb0f66cf16a45fc9cfc28c2351e06169",
+                "url": "https://api.github.com/repos/sebastianbergmann/phpunit/zipball/07d386a11ac7094032900f07cada1c8975d16607",
-                "reference": "7065dbebcb0f66cf16a45fc9cfc28c2351e06169",
+                "reference": "07d386a11ac7094032900f07cada1c8975d16607",
                 "shasum": ""
             },
             "require": {
@@ -1609,7 +1609,7 @@
                 "sebastian/version": "^4.0"
             },
             "suggest": {
-                "ext-soap": "*"
+                "ext-soap": "To be able to generate mocks based on WSDL files"
             },
             "bin": [
                 "phpunit"
@@ -1648,7 +1648,7 @@
             ],
             "support": {
                 "issues": "https://github.com/sebastianbergmann/phpunit/issues",
-                "source": "https://github.com/sebastianbergmann/phpunit/tree/10.0.14"
+                "source": "https://github.com/sebastianbergmann/phpunit/tree/10.0.16"
             },
             "funding": [
                 {
@@ -1664,7 +1664,7 @@
                     "type": "tidelift"
                 }
             ],
-            "time": "2023-03-01T05:37:49+00:00"
+            "time": "2023-03-13T09:02:40+00:00"
         },
         {
             "name": "sebastian/cli-parser",

index.php

@@ -3,8 +3,11 @@
 require_once __DIR__ . "/vendor/autoload.php";
 
 use Lewisdale\Webmentions\Endpoint;
+use Lewisdale\Webmentions\Exceptions\InvalidSourceException;
 use Lewisdale\Webmentions\Exceptions\InvalidTargetException;
 use Lewisdale\Webmentions\Exceptions\InvalidUrlException;
+use Lewisdale\Webmentions\Exceptions\SourceNotFoundException;
+use Lewisdale\Webmentions\Exceptions\TargetNotMentionedException;
 use Lewisdale\Webmentions\Gateways\SqliteGateway;
 use Lewisdale\Webmentions\Router\Request;
 use Lewisdale\Webmentions\Router\Response;
@@ -18,7 +21,13 @@ $router = new Router();
 $router->post("/send", function (Request $req, Response $response) {
     $mentioner = new Webmention();
     $source = $req->query["source"];
+    try {
         $mentioner->sendForPage($source);
+        return;
+    } catch (InvalidSourceException $e) {
+        $response->status_code = StatusCode::BadRequest;
+        return $e->getMessage();
+    }
 });
 
 $router->post("/endpoint", function (Request $req, Response $response) {
@@ -29,21 +38,22 @@ $router->post("/endpoint", function (Request $req, Response $response) {
 
     try {
         $endpoint->receiveWebmention($source, $target);
-    } catch (InvalidUrlException $e) {
+    } catch (InvalidUrlException) {
         $response->status_code = StatusCode::BadRequest;
         return "Source and target must be valid URLs";
-    } catch (InvalidTargetException $e) {
+    } catch (InvalidTargetException) {
         $response->status_code = StatusCode::BadRequest;
         return "Target must be on the domain lewisdale.dev";
-    } catch (\Lewisdale\Webmentions\Exceptions\SourceNotFoundException $e) {
+    } catch (SourceNotFoundException $e) {
         $response->status_code = StatusCode::BadRequest;
         return "Source URL was unreachable";
-    } catch (\Lewisdale\Webmentions\Exceptions\TargetNotMentionedException) {
+    } catch (TargetNotMentionedException) {
         $response->status_code = StatusCode::BadRequest;
         return "Source does not mention the target";
     }
 
     $response->status_code = StatusCode::Created;
+    return;
 });
 
 $router->get('/', fn($req, $res) => "<h1>Webmention server</h1>");

src/Endpoint.php

@@ -2,8 +2,6 @@
 
 namespace Lewisdale\Webmentions;
 
-use League\Uri\Exceptions\SyntaxError;
-use League\Uri\Uri;
 use Lewisdale\Webmentions\Exceptions\InvalidTargetException;
 use Lewisdale\Webmentions\Exceptions\InvalidUrlException;
 use Lewisdale\Webmentions\Exceptions\SourceNotFoundException;
@@ -25,29 +23,15 @@ class Endpoint
     {
     }
 
-    public function validateUrl(string $url): bool
-    {
-        try {
-            $uri = Uri::createFromString($url);
-            $scheme = $uri->getScheme();
-            $schemeValid = in_array($scheme, ["http", "https"]);
-
-            return $schemeValid && !!filter_var($url, FILTER_VALIDATE_URL);
-
-        } catch (SyntaxError $e) {
-            return false;
-        }
-    }
-
     public function receiveWebmention(string $source, string $target): void
     {
         // Validate that both source and target are actual domains
-        if (!$this->validateUrl($source) || !$this->validateUrl($target)) {
+        if (!Url::validateUrl($source) || !Url::validateUrl($target)) {
             throw new InvalidUrlException();
         }
 
         // Validate that the webmention target is a domain I care about
-        if (!str_contains($target, "https://lewisdale.dev")) {
+        if (!Url::matchesHost($target, "lewisdale.dev")) {
             throw new InvalidTargetException();
         }
 

src/Exceptions/InvalidSourceException.php

@@ -0,0 +1,11 @@
+<?php declare(strict_types=1);
+
+namespace Lewisdale\Webmentions\Exceptions;
+
+class InvalidSourceException extends \Exception
+{
+    function __construct()
+    {
+        parent::__construct("Source url is invalid");
+    }
+}

src/Url.php

@@ -0,0 +1,30 @@
+<?php declare(strict_types=1);
+
+namespace Lewisdale\Webmentions;
+
+use League\Uri\Exceptions\SyntaxError;
+use League\Uri\Uri;
+
+class Url
+{
+    public static function matchesHost(string $source, string $host): bool
+    {
+        $uri = Uri::createFromString($source);
+        return $uri->getHost() === $host;
+    }
+
+    public static function validateUrl(string $url): bool
+    {
+        try {
+            $uri = Uri::createFromString($url);
+            $scheme = $uri->getScheme();
+            $schemeValid = in_array($scheme, ["http", "https"]);
+            [, $tld] = explode(".", $uri->getHost() ?? "");
+
+            return !empty($tld) && $schemeValid && !!filter_var($url, FILTER_VALIDATE_URL);
+
+        } catch (SyntaxError) {
+            return false;
+        }
+    }
+}

src/Webmention.php

@@ -2,11 +2,13 @@
 
 namespace Lewisdale\Webmentions;
 
+use Lewisdale\Webmentions\Exceptions\InvalidSourceException;
 use Symfony\Component\DomCrawler\Crawler;
 use Symfony\Component\HttpClient\HttpClient;
 use Symfony\Contracts\HttpClient\HttpClientInterface;
 
-class Webmention {
+class Webmention
+{
     private readonly HttpClientInterface $client;
 
     function __construct()
@@ -14,7 +16,12 @@ class Webmention {
         $this->client = HttpClient::create();
     }
 
-    public function sendForPage(string $source) : void {
+    public function sendForPage(string $source): void
+    {
+        if (!Url::matchesHost($source, "lewisdale.dev")) {
+            throw new InvalidSourceException();
+        }
+        
         $urls = $this->getUrls($source);
 
         foreach ($urls as $target) {
@@ -22,7 +29,8 @@ class Webmention {
         }
     }
 
-    private function getUrls(string $url) : array {
+    private function getUrls(string $url): array
+    {
         $page = file_get_contents($url);
         $doc = new Crawler($page);
 
@@ -37,7 +45,8 @@ class Webmention {
         return $urls;
     }
 
-    private function sendWebmention(string $source, string $target) {
+    private function sendWebmention(string $source, string $target)
+    {
         $endpoint = $this->getWebmentionEndpoint($target);
 
         if ($endpoint) {
@@ -46,13 +55,14 @@ class Webmention {
                 [
                     "body" => [
                         "source" => $source,
-                    "target" => $target
+                        "target" => $target,
-                ]
+                    ],
                 ]);
         }
     }
 
-    private function getWebmentionEndpoint(string $url) : string | null {
+    private function getWebmentionEndpoint(string $url): string|null
+    {
         $response = $this->client->request('GET', $url);
         return EndpointParser::parse($response);
     }

tests/EndpointTest.php

@@ -49,14 +49,6 @@ class EndpointTest extends TestCase
         $endpoint->receiveWebmention($source, $target);
     }
     
-    #[TestWith(["https://my.url.com", true])]
-    #[TestWith(["my.url.com", false])]
-    public function testValidatesUrls(string $url, bool $expected)
-    {
-        $endpoint = new Endpoint($this->mockClient, $this->mockGateway);
-        $this->assertEquals($expected, $endpoint->validateUrl($url), "Expected $url");
-    }
-
     #[TestWith(["https://my-valid-source.url", "htt://my-invalid-target.com"])]
     #[TestWith(["my-invalid-source", "https://my-valid-target.com"])]
     #[TestWith(["http:///an-invalid-source", "an-invalid-target"])]

tests/UrlTest.php

@@ -0,0 +1,28 @@
+<?php
+
+
+use Lewisdale\Webmentions\Url;
+use PHPUnit\Framework\Attributes\TestWith;
+use PHPUnit\Framework\TestCase;
+
+class UrlTest extends TestCase
+{
+
+    #[TestWith(["https://lewisdale.dev/post/a-post", "lewisdale.dev", true])]
+    #[TestWith(["https://css-tricks.com/test/host", "lewisdale.dev", false])]
+    #[TestWith(["ht:/lewisdale.dev/not-a-valid-url", "lewisdale.dev", false])]
+    public function testItValidatesAHost(string $source, string $target, bool $expected)
+    {
+        $this->assertEquals($expected, Url::matchesHost($source, $target, $expected));
+    }
+
+    #[TestWith(["http://a-valid-http-url.com", true])]
+    #[TestWith(["ht://an-invalid-url.com", false])]
+    #[TestWith(["https://no-tld./", false])]
+    #[TestWith(["https://a-valid-https.com", true])]
+    #[TestWith(["ftp://wrong-protocol.com", false])]
+    public function testItValidatesAUrl(string $url, bool $expected)
+    {
+        $this->assertEquals($expected, Url::validateUrl($url));
+    }
+}