feat: Improving on format keyword

Author: DannyvdSluijs

src/JsonSchema/ConstraintError.php

@@ -28,6 +28,7 @@ class ConstraintError extends Enum
     public const FORMAT_EMAIL = 'emailFormat';
     public const FORMAT_HOSTNAME = 'styleHostName';
     public const FORMAT_IP = 'ipFormat';
+    public const FORMAT_JSON_POINTER = 'jsonPointerFormat';
     public const FORMAT_PHONE = 'phoneFormat';
     public const FORMAT_REGEX= 'regexFormat';
     public const FORMAT_STYLE = 'styleFormat';
@@ -84,6 +85,7 @@ public function getMessage()
             self::FORMAT_EMAIL => 'Invalid email',
             self::FORMAT_HOSTNAME => 'Invalid hostname',
             self::FORMAT_IP => 'Invalid IP address',
+            self::FORMAT_JSON_POINTER => 'Invalid JSON pointer',
             self::FORMAT_PHONE => 'Invalid phone number',
             self::FORMAT_REGEX=> 'Invalid regex format %s',
             self::FORMAT_STYLE => 'Invalid style',

src/JsonSchema/Constraints/Drafts/Draft06/FormatConstraint.php

@@ -44,7 +44,7 @@ public function check(&$value, $schema = null, ?JsonPointer $path = null, $i = n
                 }
                 break;
             case 'date-time':
-                if (Rfc3339::createFromString($value) === null) {
+                if (!$this->validateRfc3339DateTime($value)) {
                     $this->addError(ConstraintError::FORMAT_DATE_TIME(), $path, ['dateTime' => $value, 'format' => $schema->format]);
                 }
                 break;
@@ -108,6 +108,12 @@ public function check(&$value, $schema = null, ?JsonPointer $path = null, $i = n
                     $this->addError(ConstraintError::FORMAT_HOSTNAME(), $path, ['format' => $schema->format]);
                 }
                 break;
+            case 'json-pointer':
+                if (!$this->validateJsonPointer($value)) {
+                    $this->addError(ConstraintError::FORMAT_JSON_POINTER(), $path, ['format' => $schema->format]);
+                }
+                break;
+                break;
             default:
                 break;
         }
@@ -167,4 +173,36 @@ private function validateHostname(string $host): bool
 
         return preg_match($hostnameRegex, $host) !== false;
     }
+
+    private function validateJsonPointer(string $value): bool
+    {
+        // Must be empty or start with a forward slash
+        if ($value !== '' && $value[0] !== '/') {
+            return false;
+        }
+
+        // Split into reference tokens and check for invalid escape sequences
+        $tokens = explode('/', $value);
+        array_shift($tokens); // remove leading empty part due to leading slash
+
+        foreach ($tokens as $token) {
+            // "~" must only be followed by "0" or "1"
+            if (preg_match('/~(?![01])/', $token)) {
+                return false;
+            }
+        }
+
+        return true;
+    }
+
+    private function validateRfc3339DateTime(string $value): bool
+    {
+        $dateTime = Rfc3339::createFromString($value);
+        if (is_null($dateTime)) {
+            return false;
+        }
+
+        // Compare value and date result to be equal
+        return true;
+    }
 }

src/JsonSchema/Rfc3339.php

@@ -6,27 +6,51 @@
 
 class Rfc3339
 {
-    private const REGEX = '/^(\d{4}-\d{2}-\d{2}[T ]{1}\d{2}:\d{2}:\d{2})(\.\d+)?(Z|([+-]\d{2}):?(\d{2}))$/';
+    private const REGEX = '/^(\d{4}-\d{2}-\d{2}[T ](0[0-9]|1[0-9]|2[0-3]):([0-5][0-9]):((?:[0-5][0-9]|60)))(\.\d+)?(Z|([+-](0[0-9]|1[0-9]|2[0-3])):([0-5][0-9]))$/';
 
     /**
      * Try creating a DateTime instance
      *
-     * @param string $string
+     * @param string $input
      *
      * @return \DateTime|null
      */
-    public static function createFromString($string)
+    public static function createFromString($input): ?\DateTime
     {
-        if (!preg_match(self::REGEX, strtoupper($string), $matches)) {
+        if (!preg_match(self::REGEX, strtoupper($input), $matches)) {
             return null;
         }
 
+        $input = strtoupper($input); // Cleanup for lowercase t and z
+        $inputHasTSeparator = strpos($input, 'T');
+
         $dateAndTime = $matches[1];
-        $microseconds = $matches[2] ?: '.000000';
-        $timeZone = 'Z' !== $matches[3] ? $matches[4] . ':' . $matches[5] : '+00:00';
-        $dateFormat = strpos($dateAndTime, 'T') === false ? 'Y-m-d H:i:s.uP' : 'Y-m-d\TH:i:s.uP';
-        $dateTime = \DateTime::createFromFormat($dateFormat, $dateAndTime . $microseconds . $timeZone, new \DateTimeZone('UTC'));
+        $microseconds = $matches[5] ?: '.000000';
+        $timeZone = 'Z' !== $matches[6] ? $matches[6] : '+00:00';
+        $dateFormat = $inputHasTSeparator === false ? 'Y-m-d H:i:s.uP' : 'Y-m-d\TH:i:s.uP';
+        $dateTime = \DateTimeImmutable::createFromFormat($dateFormat, $dateAndTime . $microseconds . $timeZone, new \DateTimeZone('UTC'));
+
+        if ($dateTime === false) {
+            return null;
+        }
+
+        $utcDateTime = $dateTime->setTimezone(new \DateTimeZone('+00:00'));
+        $oneSecond = new \DateInterval('PT1S');
+
+        // handle leap seconds
+        if ($matches[4] === '60' && $utcDateTime->sub($oneSecond)->format('H:i:s') === '23:59:59') {
+            $dateTime = $dateTime->sub($oneSecond);
+            $matches[1] = str_replace(':60', ':59', $matches[1]);
+        }
+
+        // Ensure we still have the same year, month, day, hour, minutes and seconds to ensure no rollover took place.
+        if ($dateTime->format($inputHasTSeparator ? 'Y-m-d\TH:i:s' : 'Y-m-d H:i:s') !== $matches[1]) {
+            return null;
+        }
+
+        $mutable = \DateTime::createFromFormat('U.u', $dateTime->format('U.u'));
+        $mutable->setTimezone($dateTime->getTimezone());
 
-        return $dateTime ?: null;
+        return $mutable;
     }
 }

src/JsonSchema/Tool/Validator/UriValidator.php

@@ -19,12 +19,16 @@ public static function isValid(string $uri): bool
             (\#(.*))?                                # Optional fragment
         $/ix';
 
-        // RFC 3986: Non-Hierarchical URIs (mailto, data, urn)
+        // RFC 3986: Non-Hierarchical URIs (mailto, data, urn, news)
         $nonHierarchicalPattern = '/^
-            (mailto|data|urn):                        # Only allow known non-hierarchical schemes
-            (.+)                                      # Must contain at least one character after scheme
+                (mailto|data|urn|news|tel):          # Only allow known non-hierarchical schemes
+                (.+)                                 # Must contain at least one character after scheme
         $/ix';
 
+        // Validation for newsgroup name (alphanumeric + dots, no empty segments)
+        $newsGroupPattern = '/^[a-z0-9]+(\.[a-z0-9]+)*$/i';
+        $telPattern = '/^\+?[0-9.\-() ]+$/'; // Allows +, digits, separators
+
         // RFC 5322-compliant email validation for `mailto:` URIs
         $emailPattern = '/^[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}$/';
 
@@ -40,7 +44,7 @@ public static function isValid(string $uri): bool
                 return false;
             }
 
-            // Validate path (reject illegal characters: < > { } | \ ^ `)
+            // Validate the path (reject illegal characters: < > { } | \ ^ `)
             if (!empty($matches[6]) && preg_match('/[<>{}|\\\^`]/', $matches[6])) {
                 return false;
             }
@@ -57,6 +61,14 @@ public static function isValid(string $uri): bool
                 return preg_match($emailPattern, $matches[2]) === 1;
             }
 
+            if ($scheme === 'news') {
+                return preg_match($newsGroupPattern, $matches[2]) === 1;
+            }
+
+            if ($scheme === 'tel') {
+                return preg_match($telPattern, $matches[2]) === 1;
+            }
+
             return true; // Valid non-hierarchical URI
         }