Align entity types with model labels

Author: stevebauman

src/EntityFactory.php

@@ -41,14 +41,13 @@ public function fromDecodedOutput(array $entities, string $sourceText): array
     /**
      * Create entities from invalid JSON output prefixes.
      *
-     * The upstream CLI includes raw entity text in the JSON response without
-     * escaping it first. This fallback uses the byte offsets and ignores the
-     * echoed text field so quoted entity text can still be handled safely.
-     *
      * @return array<int, Entity>
      */
     public function fromOutputPrefixes(string $output, string $sourceText): array
     {
+        // The upstream CLI prints entity text directly into a JSON-like response
+        // without escaping it first. Read only the fields before the text value,
+        // then rehydrate entity text from the original byte offsets.
         preg_match_all(
             '/\{\s*"entity_group"\s*:\s*"(?P<type>[^"]+)"\s*,\s*"start"\s*:\s*(?P<start>-?\d+)\s*,\s*"end"\s*:\s*(?P<end>-?\d+)\s*,\s*"score"\s*:\s*(?P<score>-?(?:\d+\.?\d*|\.\d+)(?:[eE][+-]?\d+)?)\s*,\s*"text"\s*:\s*"/',
             $output,

src/EntityType.php

@@ -7,21 +7,12 @@
  */
 enum EntityType: string
 {
-    case Person = 'person';
-    case Email = 'email';
-    case Phone = 'phone';
-    case Address = 'address';
-    case CreditCard = 'credit_card';
-    case Date = 'date';
-    case IpAddress = 'ip_address';
-    case Url = 'url';
-
-    case PrivatePerson = 'private_person';
-    case PrivateEmail = 'private_email';
-    case PrivatePhone = 'private_phone';
+    case AccountNumber = 'account_number';
     case PrivateAddress = 'private_address';
-    case PrivateCreditCard = 'private_credit_card';
     case PrivateDate = 'private_date';
-    case PrivateIpAddress = 'private_ip_address';
+    case PrivateEmail = 'private_email';
+    case PrivatePerson = 'private_person';
+    case PrivatePhone = 'private_phone';
     case PrivateUrl = 'private_url';
+    case Secret = 'secret';
 }

tests/Feature/ClassifierTest.php

@@ -14,7 +14,7 @@
 
     expect($entities)->toHaveCount(1)
         ->and($entities[0])->toBeInstanceOf(Entity::class)
-        ->and($entities[0]->type)->toBe('email')
+        ->and($entities[0]->type)->toBe('private_email')
         ->and($entities[0]->text)->toBe('jdoe@example.com')
         ->and($entities[0]->start)->toBe(20)
         ->and($entities[0]->end)->toBe(36)
@@ -26,7 +26,7 @@
     $this->setFakePrivacyFilterEnvironment([
         'PRIVACY_FILTER_FAKE_MODE' => 'raw-text',
         'PRIVACY_FILTER_FAKE_NEEDLE' => 'John "JD" Doe',
-        'PRIVACY_FILTER_FAKE_TYPE' => 'person',
+        'PRIVACY_FILTER_FAKE_TYPE' => 'private_person',
     ]);
 
     $entities = (new Classifier(
@@ -36,7 +36,7 @@
     ))->entities('Contact John "JD" Doe today.');
 
     expect($entities)->toHaveCount(1)
-        ->and($entities[0]->type)->toBe('person')
+        ->and($entities[0]->type)->toBe('private_person')
         ->and($entities[0]->text)->toBe('John "JD" Doe');
 });
 

tests/Fixtures/privacy-filter

@@ -3,7 +3,7 @@
 
 $text = stream_get_contents(STDIN);
 $needle = getenv('PRIVACY_FILTER_FAKE_NEEDLE') ?: 'jdoe@example.com';
-$type = getenv('PRIVACY_FILTER_FAKE_TYPE') ?: 'email';
+$type = getenv('PRIVACY_FILTER_FAKE_TYPE') ?: 'private_email';
 $mode = getenv('PRIVACY_FILTER_FAKE_MODE') ?: 'json';
 $start = strpos($text, $needle);
 

tests/Unit/EntityFactoryTest.php

@@ -6,7 +6,7 @@
 it('creates entities from decoded output', function () {
     $entities = (new EntityFactory)->fromDecodedOutput([
         [
-            'entity_group' => 'email',
+            'entity_group' => 'private_email',
             'start' => 20,
             'end' => 36,
             'score' => 0.9876,
@@ -15,18 +15,18 @@
     ], 'Contact John Doe at jdoe@example.com.');
 
     expect($entities)->toHaveCount(1)
-        ->and($entities[0]->type)->toBe('email')
+        ->and($entities[0]->type)->toBe('private_email')
         ->and($entities[0]->text)->toBe('jdoe@example.com');
 });
 
 it('creates entities from output prefixes', function () {
     $entities = (new EntityFactory)->fromOutputPrefixes(
-        "[\n {\"entity_group\": \"person\", \"start\": 8, \"end\": 21, \"score\": 0.9876, \"text\": \"John \"JD\" Doe\"}\n]\n",
+        "[\n {\"entity_group\": \"private_person\", \"start\": 8, \"end\": 21, \"score\": 0.9876, \"text\": \"John \"JD\" Doe\"}\n]\n",
         'Contact John "JD" Doe today.',
     );
 
     expect($entities)->toHaveCount(1)
-        ->and($entities[0]->type)->toBe('person')
+        ->and($entities[0]->type)->toBe('private_person')
         ->and($entities[0]->text)->toBe('John "JD" Doe');
 });
 

tests/Unit/EntityTest.php

@@ -6,14 +6,14 @@
 
 it('creates an entity from a decoded payload', function () {
     $entity = Entity::from([
-        'entity_group' => 'email',
+        'entity_group' => 'private_email',
         'start' => 20,
         'end' => 36,
         'score' => 0.9876,
         'text' => 'ignored@example.com',
     ], 'Contact John Doe at jdoe@example.com.');
 
-    expect($entity->type)->toBe('email')
+    expect($entity->type)->toBe('private_email')
         ->and($entity->text)->toBe('jdoe@example.com')
         ->and($entity->start)->toBe(20)
         ->and($entity->end)->toBe(36)
@@ -22,19 +22,19 @@
 
 it('creates an entity from alternate type keys', function () {
     $entity = Entity::from([
-        'label' => 'person',
+        'label' => 'private_person',
         'start' => 8,
         'end' => 16,
         'score' => 0.9876,
     ], 'Contact John Doe today.');
 
-    expect($entity->type)->toBe('person')
+    expect($entity->type)->toBe('private_person')
         ->and($entity->text)->toBe('John Doe');
 });
 
 it('exposes entity details as arrays and json', function () {
     $entity = new Entity(
-        type: 'email',
+        type: 'private_email',
         start: 20,
         end: 36,
         score: 0.9876,
@@ -43,7 +43,7 @@
 
     expect($entity->length())->toBe(16)
         ->and($entity->toArray())->toBe([
-            'type' => 'email',
+            'type' => 'private_email',
             'start' => 20,
             'end' => 36,
             'score' => 0.9876,
@@ -78,7 +78,7 @@
 
 it('throws an exception when the decoded payload is invalid', function () {
     Entity::from([
-        'entity_group' => 'email',
+        'entity_group' => 'private_email',
         'start' => 20,
         'end' => 200,
         'score' => 0.9876,