Merge pull request #749 from CamKem/fix/guzzle-issues

Fix: Parse URL's via Tokenization

app/Services/MetaData.php

@@ -5,8 +5,10 @@
 namespace App\Services;
 
 use DOMDocument;
-use GuzzleHttp\Exception\RequestException;
+use GuzzleHttp\Exception\TransferException;
+use GuzzleHttp\Psr7\Exception\MalformedUriException;
 use Illuminate\Http\Client\ConnectionException;
+use Illuminate\Http\Client\HttpClientException;
 use Illuminate\Support\Collection;
 use Illuminate\Support\Facades\Cache;
 use Illuminate\Support\Facades\Http;
@@ -97,12 +99,13 @@ private function getData(): Collection
             $response = Http::get($this->url);
 
             if ($response->ok()) {
-                $data = $this->parse($response->body());
+                $data = $this->parse(
+                    $response->body()
+                );
             }
-        } catch (ConnectionException) {
-            // Catch but not capture the exception
-        } catch (RequestException) {
-            // Catch but not capture the exception
+        } catch (HttpClientException|MalformedUriException|TransferException) {
+            // Catch but not capture all base exceptions for:
+            // Laravel Http Client, Guzzle, and PSR-7
         }
 
         return $data;

app/Services/ParsableContentProviders/LinkProviderParsable.php

@@ -15,50 +15,121 @@
      */
     public function parse(string $content): string
     {
-        return (string) preg_replace_callback(
-            '/(<(a|code|pre)\s+[^>]*>.*?<\/\2>)|(?<!src=")((https?:\/\/)?((localhost)|((?:\d{1,3}\.){3}\d{1,3})|[\w\-._@:%\+~#=]{1,256}(\.[a-zA-Z]{2,})+)(:\d+)?(\/[\w\-._@:%\+~#=\/]*)?(\?[\w\-._@:%\+~#=\/&]*)?)(?<!\.)((?![^<]*>|[^<>]*<\/))/is',
-            function (array $matches): string {
-                if ($matches[1] !== '') {
-                    return $matches[1];
-                }
-
-                $humanUrl = Str::of($matches[0])
-                    ->replaceMatches('/^https?:\/\//', '')
-                    ->rtrim('/')
-                    ->toString();
-
-                $isMail = (bool) preg_match('/^[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}$/', $humanUrl);
-                $isHttp = Str::startsWith($matches[0], ['http://', 'https://']);
-
-                if ((! $isMail) && (! $isHttp)) {
-                    return $matches[0];
-                }
-
-                $url = $isHttp ? $matches[0] : 'https://'.$matches[0];
-
-                $url = $isMail ? 'mailto:'.$humanUrl : $url;
-
-                $linkHtml = '<a data-navigate-ignore="true" class="text-blue-500 hover:underline hover:text-blue-700 cursor-pointer" target="_blank" href="'.$url.'">'.$humanUrl.'</a>';
-
-                if (! $isMail && $url) {
-                    $service = new MetaData($url);
-                    $metadata = $service->fetch();
-
-                    if ($metadata->isNotEmpty() && ($metadata->has('image') || $metadata->has('html'))) {
-                        $trimmed = trim(
-                            view('components.link-preview-card', [
-                                'data' => $metadata,
-                                'url' => $url,
-                            ])->render()
-                        );
-
-                        return $linkHtml.' '.preg_replace('/<!--(.|\s)*?-->/', '', $trimmed);
-                    }
-                }
-
-                return $linkHtml;
-            },
-            str_replace('&amp;', '&', $content)
+        $tokens = $this->tokenize($content);
+
+        if ($tokens === false) {
+            return $content;
+        }
+
+        $processedTokens = array_map(
+            fn (string $token): string => $this->processToken($token),
+            $tokens);
+
+        return implode('', $processedTokens);
+    }
+
+    /**
+     * Split the content into tokens based on spaces and newlines.
+     *
+     * @return list<string>|false
+     */
+    private function tokenize(string $content): array|false
+    {
+        return preg_split('/(\s|<br>)/', $content, -1, PREG_SPLIT_DELIM_CAPTURE);
+    }
+
+    /**
+     * Process a single token and convert valid URLs into HTML links.
+     */
+    private function processToken(string $token): string
+    {
+        $allowableAttachedCharacters = '{([<!,.?;:>)]}';
+
+        $trimmedToken = trim($token, $allowableAttachedCharacters);
+
+        if ($trimmedToken === '' || $trimmedToken === '0') {
+            return $token;
+        }
+
+        if (filter_var($trimmedToken, FILTER_VALIDATE_EMAIL)) {
+            $trimmedToken = "mailto:{$trimmedToken}";
+        } elseif (! $this->isValidUrl($trimmedToken)) {
+            return $token;
+        }
+
+        $humanUrl = Str::of($trimmedToken)
+            ->replaceMatches('/^(https?:\/\/|mailto:)/', '')
+            ->rtrim('/')
+            ->toString();
+
+        $linkHtml = "<a data-navigate-ignore=\"true\" class=\"text-blue-500 hover:underline hover:text-blue-700 cursor-pointer\" target=\"_blank\" href=\"{$trimmedToken}\">{$humanUrl}</a>";
+
+        $service = new MetaData($trimmedToken);
+        $metadata = $service->fetch();
+        if ($metadata->isNotEmpty() && ($metadata->has('image') || $metadata->has('html'))) {
+            $trimmedPreviewCard = trim(
+                view('components.link-preview-card', [
+                    'data' => $metadata,
+                    'url' => $trimmedToken,
+                ])->render()
+            );
+
+            $linkHtml .= $trimmedPreviewCard;
+        }
+
+        $leading = $this->getCharacters($token, $allowableAttachedCharacters, 'leading');
+        $trailing = $this->getCharacters($token, $allowableAttachedCharacters, 'trailing');
+
+        return $leading.$linkHtml.$trailing;
+    }
+
+    /**
+     * Extract leading or trailing punctuation/characters from a token.
+     */
+    private function getCharacters(string $token, string $allowableCharacters, string $direction): string
+    {
+        $pattern = match ($direction) {
+            'leading' => '/^(['.preg_quote($allowableCharacters, '/').']+)/',
+            'trailing' => '/(['.preg_quote($allowableCharacters, '/').']+)$/',
+            default => '',
+        };
+
+        if (preg_match($pattern, $token, $matches)) {
+            return $matches[1];
+        }
+
+        return '';
+    }
+
+    /**
+     * Validate if a token is a valid URL.
+     */
+    private function isValidUrl(string $token): bool
+    {
+        $urlComponents = parse_url($token);
+        if ($urlComponents === false || ! filter_var($token, FILTER_VALIDATE_URL)) {
+            return false;
+        }
+
+        $scheme = $urlComponents['scheme'] ?? null;
+        $host = $urlComponents['host'] ?? null;
+        if (! in_array($scheme, ['http', 'https'], true) || ! filter_var($host, FILTER_VALIDATE_DOMAIN, FILTER_FLAG_HOSTNAME)) {
+            return false;
+        }
+
+        foreach (['path', 'query', 'fragment'] as $part) {
+            if (isset($urlComponents[$part]) && preg_match('/[\s<>{}[\]]/', $urlComponents[$part])) {
+                return false;
+            }
+        }
+
+        if (isset($urlComponents['port']) && (preg_match('/^\d{1,5}$/', (string) $urlComponents['port']) === 0 || preg_match('/^\d{1,5}$/', (string) $urlComponents['port']) === false)) {
+            return false;
+        }
+
+        return (bool) preg_match(
+            '/((https?:\/\/)?((localhost)|((?:\d{1,3}\.){3}\d{1,3})|[\w\-._@:%+~#=]{1,256}(\.[a-zA-Z]{2,})+)(:\d+)?(\/[\w\-._@:%+~#=\/]*)?(\?[\w\-._@:%+~#=\/&]*)?)/i',
+            $token
         );
     }
 }

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____htt___example_com______htt___example_com__.snap

@@ -0,0 +1 @@
+htt://example.com

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http____2001_0db8_85a3_0000_0…_7334_______http____2001_0db8_85a3_0000_0____7334___.snap

@@ -0,0 +1 @@
+http://[2001:0db8:85a3:0000:0000:8a2e:0370:7334]

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http____2001_0db8_85a3_0000_0…__8080______http____2001_0db8_85a3_0000_0_____8080__.snap

@@ -0,0 +1 @@
+http://[2001:0db8:85a3:0000:0000:8a2e:0370:7334]:8080

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http____example_com______http____example_com__.snap

@@ -0,0 +1 @@
+http://.example.com

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http____example_com______http____example_com____2.snap

@@ -0,0 +1 @@
+http:///example.com

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http___exa__mple_com______http___exa__mple_com__.snap

@@ -0,0 +1 @@
+http://exa_mple.com

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http___example______http___example__.snap

@@ -0,0 +1 @@
+http://example

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http___example__com______http___example__com__.snap

@@ -0,0 +1 @@
+http://example..com

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http___example_com______http___example_com__.snap

@@ -0,0 +1 @@
+http://example=com

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http___example_com_abcd______http___example_com_abcd__.snap

@@ -0,0 +1 @@
+http://example.com:abcd

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http___example_com_this___that______http___example_com_this___that__.snap

@@ -0,0 +1 @@
+http://example.com?this<>=that

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http___example_com_this_that_…__that______http___example_com_this_that______that__.snap

@@ -0,0 +1 @@
+http://example.com?this=that#this<>=that

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http___example_com_👍______http___example_com_👍__.snap

@@ -0,0 +1 @@
+http://example.com/👍

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http__example_com______http__example_com__.snap

@@ -0,0 +1 @@
+http//example.com

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____http__example_com______http__example_com____2.snap

@@ -0,0 +1 @@
+http:/example.com

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____protocol___example_com______protocol___example_com__.snap

@@ -0,0 +1 @@
+protocol://example.com

tests/.pest/snapshots/Unit/Services/ContentProvidersTest/malformed_links_are_correctly_handled_by_content_parser_with_data_set____www_example_com______www_example_com__.snap

@@ -0,0 +1 @@
+www.example.com

tests/.pest/snapshots/Unit/Services/ContentTest/only_links_with_images_or_html_oEmbeds_are_parsed.snap

@@ -1,9 +1,9 @@
-Sure, here is the link: <a data-navigate-ignore="true" class="text-blue-500 hover:underline hover:text-blue-700 cursor-pointer" target="_blank" href="https://laravel.com">laravel.com</a> <div
+Sure, here is the link: <a data-navigate-ignore="true" class="text-blue-500 hover:underline hover:text-blue-700 cursor-pointer" target="_blank" href="https://laravel.com">laravel.com</a><div
     id="link-preview-card"
     data-url="https://laravel.com"
     class="mx-auto mt-2 min-w-full group/preview" data-navigate-ignore="true"
 >
-        <a href="https://laravel.com" target="_blank" rel="noopener noreferrer">
+<!--[if BLOCK]><![endif]-->        <a href="https://laravel.com" target="_blank" rel="noopener noreferrer">
         <div
             title="Click to visit: laravel.com"
             class="relative w-full bg-slate-100/90 border border-slate-300
@@ -25,4 +25,5 @@ Sure, here is the link: <a data-navigate-ignore="true" class="text-blue-500 hove
             <a href="https://laravel.com" target="_blank" rel="noopener noreferrer"
                class="text-xs text-slate-500 group-hover/preview:text-pink-600">From: laravel.com</a>
         </div>
+<!--[if ENDBLOCK]><![endif]-->
 </div>. Let me know if you have any questions.

tests/Unit/Services/ContentProvidersTest.php

@@ -200,6 +200,29 @@
     ],
 ]);
 
+test('malformed links are correctly handled by content parser', function (string $content) {
+    $provider = new App\Services\ParsableContentProviders\LinkProviderParsable();
+    expect($provider->parse($content))->toMatchSnapshot();
+})->with([
+    'http://example..com',
+    'htt://example.com',
+    'protocol://example.com',
+    'http//example.com',
+    'http://exa_mple.com',
+    'http://example',
+    'http://.example.com',
+    'http://example=com',
+    'www.example.com',
+    'http:/example.com',
+    'http:///example.com',
+    'http://example.com?this<>=that',
+    'http://example.com?this=that#this<>=that',
+    'http://[2001:0db8:85a3:0000:0000:8a2e:0370:7334]',
+    'http://[2001:0db8:85a3:0000:0000:8a2e:0370:7334]:8080',
+    'http://example.com:abcd',
+    'http://example.com/👍',
+]);
+
 test('only http or https urls are converted to links', function (string $content, string $parsed) {
     $provider = new App\Services\ParsableContentProviders\LinkProviderParsable();
 

tests/Unit/Services/MetaDataTest.php

@@ -3,8 +3,11 @@
 declare(strict_types=1);
 
 use App\Services\MetaData;
+use GuzzleHttp\Exception\TransferException;
 use GuzzleHttp\Promise\RejectedPromise;
+use GuzzleHttp\Psr7\Exception\MalformedUriException;
 use Illuminate\Http\Client\ConnectionException;
+use Illuminate\Http\Client\HttpClientException;
 use Illuminate\Http\UploadedFile;
 use Illuminate\Support\Facades\Cache;
 use Illuminate\Support\Facades\Http;
@@ -261,3 +264,21 @@
     expect(round(MetaData::CARD_WIDTH / MetaData::CARD_HEIGHT, 2))
         ->toBe(round(16 / 9, 2));
 });
+
+it('handles all exceptions', function (Exception $exception) {
+    $url = 'https://laravel.com';
+
+    Http::fake([
+        $url => fn ($request) => new RejectedPromise($exception),
+    ]);
+
+    $service = new MetaData($url);
+    $data = $service->fetch();
+
+    expect($data->isEmpty())->toBeTrue();
+})->with([
+    new ConnectionException('Connection error'),
+    new MalformedUriException('Malformed URI'),
+    new HttpClientException('Not Found'),
+    new TransferException('Transfer error'),
+]);