Update Laravel AI dependency and enhance URL detection PII redaction patterns

composer.json

@@ -29,7 +29,7 @@
     ],
     "require": {
         "php": "^8.4.0",
-        "laravel/ai": "^0.8.0"
+        "laravel/ai": "*"
     },
     "require-dev": {
         "laravel/pint": "^1.29.1",

src/InjectionGuard/composer.json

@@ -5,10 +5,18 @@
     "homepage": "https://intercept.promptphp.com",
     "type": "library",
     "license": "MIT",
+    "authors": [
+        {
+            "name": "Victor Ukam",
+            "email": "victorjohnukam@gmail.com",
+            "homepage": "https://github.com/veeqtoh",
+            "role": "Developer"
+        }
+    ],
     "require": {
         "php": "^8.4.0",
-        "laravel/ai": "^0.8.0",
-        "promptphp/intercept-support": "^0.1"
+        "laravel/ai": "*",
+        "promptphp/intercept-support": "*"
     },
     "autoload": {
         "psr-4": {

src/InjectionGuard/tests/Fixtures/PromptInjectionGuardTestProvider.php

@@ -4,47 +4,103 @@
 
 namespace PromptPHP\Intercept\InjectionGuard\Tests\Fixtures;
 
-use Laravel\Ai\Contracts\Gateway\TextGateway;
+use Laravel\Ai\Contracts\Gateway\StepTextGateway;
 use Laravel\Ai\Contracts\Providers\TextProvider;
+use Laravel\Ai\Gateway\TextGenerationLoop;
 use Laravel\Ai\Prompts\AgentPrompt;
 use Laravel\Ai\Responses\AgentResponse;
 use Laravel\Ai\Responses\StreamableAgentResponse;
 use RuntimeException;
 
 final class PromptInjectionGuardTestProvider implements TextProvider
 {
+    /**
+     * {@inheritDoc}
+     */
     public function prompt(AgentPrompt $prompt): AgentResponse
     {
         throw new RuntimeException('Not used in this test.');
     }
 
+    /**
+     * {@inheritDoc}
+     */
     public function stream(AgentPrompt $prompt): StreamableAgentResponse
     {
         throw new RuntimeException('Not used in this test.');
     }
 
-    public function textGateway(): TextGateway
+    /**
+     * {@inheritDoc}
+     */
+    public function useTextGateway(StepTextGateway $gateway): self
     {
-        throw new RuntimeException('Not used in this test.');
+        return $this;
     }
 
-    public function useTextGateway(TextGateway $gateway): self
+    /**
+     * {@inheritDoc}
+     */
+    public function textGenerationLoop(): TextGenerationLoop
     {
-        return $this;
+        throw new RuntimeException('Not used in this test.');
     }
 
+    /**
+     * {@inheritDoc}
+     */
     public function defaultTextModel(): string
     {
         return 'test-model';
     }
 
+    /**
+     * {@inheritDoc}
+     */
     public function cheapestTextModel(): string
     {
         return 'test-cheapest-model';
     }
 
+    /**
+     * {@inheritDoc}
+     */
     public function smartestTextModel(): string
     {
         return 'test-smartest-model';
     }
+
+    // ---- Provider interface ----
+
+    /**
+     * {@inheritDoc}
+     */
+    public function name(): string
+    {
+        return 'test-provider';
+    }
+
+    /**
+     * {@inheritDoc}
+     */
+    public function driver(): string
+    {
+        return 'test';
+    }
+
+    /**
+     * {@inheritDoc}
+     */
+    public function providerCredentials(): array
+    {
+        return [];
+    }
+
+    /**
+     * {@inheritDoc}
+     */
+    public function additionalConfiguration(): array
+    {
+        return [];
+    }
 }

src/PIIRedactor/composer.json

@@ -15,8 +15,8 @@
     ],
     "require": {
         "php": "^8.4",
-        "laravel/ai": "^0.8.0",
-        "promptphp/intercept-support": "^0.1"
+        "laravel/ai": "*",
+        "promptphp/intercept-support": "*"
     },
     "autoload": {
         "psr-4": {

src/PIIRedactor/src/Detectors/RegexDetector.php

@@ -4,6 +4,7 @@
 
 namespace PromptPHP\Intercept\PIIRedactor\Detectors;
 
+use Closure;
 use InvalidArgumentException;
 use PromptPHP\Intercept\PIIRedactor\Detectors\Contracts\Detector;
 use PromptPHP\Intercept\PIIRedactor\ValueObjects\Detection;
@@ -13,14 +14,16 @@
     /**
      * Create a new regex detector.
      *
-     * @param string $type       The detector entity type.
-     * @param string $pattern    The regex pattern to use for detection.
-     * @param float  $confidence The confidence level of the detection (default: 1.0).
+     * @param string   $type       The detector entity type.
+     * @param string   $pattern    The regex pattern to use for detection.
+     * @param float    $confidence The confidence level of the detection (default: 1.0).
+     * @param ?Closure $validator  A optional closure to validate detected values.
      */
     public function __construct(
         protected string $type,
         protected string $pattern,
         protected float $confidence = 1.0,
+        protected ?Closure $validator = null,
     ) {
         //
     }
@@ -40,7 +43,7 @@ public function type(): string
      *
      * @param string $text The text to analyze for sensitive values.
      *
-     * @return array<int, Detection>
+     * @return array<int, Detection> An array of Detection objects representing the detected sensitive values.
      */
     public function detect(string $text): array
     {
@@ -59,15 +62,24 @@ public function detect(string $text): array
             return [];
         }
 
-        return array_map(
-            fn (array $match): Detection => new Detection(
+        $detections = [];
+
+        foreach ($matches[0] as $match) {
+            $value = $match[0];
+
+            if ($this->validator !== null && ! ($this->validator)($value)) {
+                continue;
+            }
+
+            $detections[] = new Detection(
                 type: $this->type,
-                value: $match[0],
+                value: $value,
                 start: $match[1],
-                length: strlen($match[0]),
+                length: strlen($value),
                 confidence: $this->confidence,
-            ),
-            $matches[0],
-        );
+            );
+        }
+
+        return $detections;
     }
 }

src/PIIRedactor/src/PIIRedactor.php

@@ -373,19 +373,53 @@ protected function defaultDetectors(): array
                 EntityTypes::MAC_ADDRESS->value,
                 '/\b(?:[0-9A-Fa-f]{2}[:-]){5}[0-9A-Fa-f]{2}\b/'
             ),
+            // Pattern 1: URLs with scheme (http:// or https://).
             new RegexDetector(
                 EntityTypes::URL->value,
-                '/\b(?:https?:\/\/|www\.)[a-zA-Z0-9+&@#\/%?=~_|!:,.;]*[a-zA-Z0-9+&@#\/%=~_|]|\b(?<![a-zA-Z0-9@])(?:[a-zA-Z0-9-]+\.)+[a-zA-Z]{2,}(?:\/[a-zA-Z0-9+&@#\/%?=~_|!:,.;]*[a-zA-Z0-9+&@#\/%=~_|])?\b/'
+                '~\bhttps?://[^\s<>"{}|\\^`\[\]]+(?<![.,;:!])~i',
+                1.0,
+                function (string $url): bool {
+                    $url = rtrim($url, '.,;:!'); // trim trailing punctuation swallowed by regex.
+
+                    $parsed = parse_url($url);
+
+                    return $parsed !== false
+                        && isset($parsed['scheme'], $parsed['host'])
+                        && in_array($parsed['scheme'], ['http', 'https'], true);
+                }
             ),
+            // Pattern 2: URLs without scheme (www.example.com or example.com/path)
+            // More conservative to avoid false positives.
+            new RegexDetector(
+                EntityTypes::URL->value,
+                '~\b(?<![a-zA-Z0-9@])[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(?:\.[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)+(?:/[^\s<>"{}|\\^`\[\]]+)?(?<![.,;:!])~i',
+                0.9,
+                function (string $url): bool {
+                    // Bare domains in prose are not URLs.
+                    if (! str_contains($url, '/') && ! str_starts_with($url, 'www.')) {
+                        return false;
+                    }
+
+                    $url = rtrim($url, '.,;:!');
+
+                    // Prepend scheme so parse_url can validate the host.
+                    $parsed = parse_url('http://'.$url);
+
+                    return $parsed !== false
+                        && isset($parsed['host'])
+                        && str_contains($parsed['host'], '.');
+                }
+            ),
+
         ];
     }
 
     /**
      * Remove overlapping detections.
      *
-     * @param array<int, Detection> $detections Thre list of detections to process.
+     * @param array<int, Detection> $detections The list of detections to process.
      *
-     * @return array<int, Detection>
+     * @return array<int, Detection> The list of detections with overlaps removed.
      */
     protected function removeOverlaps(array $detections): array
     {
@@ -439,6 +473,7 @@ protected function priority(string $type): int
             EntityTypes::EMAIL->value        => 30,
             EntityTypes::IP_ADDRESS->value   => 20,
             EntityTypes::PHONE->value        => 10,
+            EntityTypes::URL->value          => 8,
             EntityTypes::MAC_ADDRESS->value  => 5,
             default                          => 0,
         };
@@ -570,7 +605,7 @@ protected function maskGeneric(string $value): string
      *
      * @param array<int, Detection> $detections The list of detections to summarise.
      *
-     * @return array<string, int>
+     * @return array<string, int> A summary of detected entities, with entity types as keys and counts as values.
      */
     protected function summariseEntities(array $detections): array
     {